The Internet of Things

The Internet of Things

Foundation for Smart Cities, eHealth,
and Ubiquitous Computing
Edited by
Ricardo Armentano
Robin Singh Bhadoria
Parag Chatterjee
Ganesh Chandra Deka
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2018 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works
Printed on acid-free paper
International Standard Book Number-13: 978-1-4987-8902-8 (Hardback)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to
publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or
the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced
in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright
material has not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any
form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and
recording, or in any information storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://
www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923,
978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For
organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identifi-
cation and explanation without intent to infringe.
Library of Congress Cataloging‑in‑Publication Data
Names: Armentano, Ricardo, editor.

Title: The Internet of things : foundation for smart cities, eHealth and
ubiquitous computing / editors, Ricardo Armentano [and 3 others].
Other titles: Internet of things (Armentano)
Description: Boca Raton : CRC Press, 2018. | Includes bibliographical
references and index.
Identifiers: LCCN 2017016990| ISBN 9781498789028 (hardback : alk. paper) |
ISBN 9781315156026 (ebook)
Subjects: LCSH: Internet of things. | Ubiquitous computing. | Medical
informatics.
Classification: LCC TK5105.8857 I56 2018 | DDC 004.67/8--dc23
LC record available at https://lccn.loc.gov/2017016990
Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com
and the CRC Press Web site at

http://www.crcpress.com
Contents
Preface .............................................................................................................................................. ix
Editors .............................................................................................................................................. xi
Contributors ................................................................................................................................. xiii
Section I Introduction to the Internet of Things:

Definition and Basic Foundation
1. The IoT Vision from an Opportunistic Networking Perspective................................. 3
Zoran Bojkovic, Bojan Bakmaz, and Miodrag Bakmaz
2. DEFINE—An Architecture Framework for Designing IoT Systems ......................... 21

S.A. Srinivasa Moorthy
3. MAC and Network Layer Issues and Challenges for IoT ............................................ 49
Upena Dalal, Shweta Shah, and Jigisha Patel
4. Privacy and Security of Data in IoT: Design Principles and Techniques ................. 79
K.V. Arya and Vishnu Prabhakaran
5. Integration of Cloud Computing and IoT: Opportunities and Challenges .............. 99

Sara Ghanavati, Jemal Abawajy, and Davood Izadi
Section II Frameworks for the Internet of Things:

An Architectural Perspective
6. IoT Framework Based on WSN Infrastructure with Different Topological
Distributions ....................................................................................................................... 121
Marcel Stefan Wagner, Diogo Ferreira Lima Filho, Miguel Arjona Ramírez,
and Wagner Luiz Zucchi
7. Semantic Complex Service Composition within an IoT Ecosystem ........................ 151

Grigorios Tzortzis, Charilaos Akasiadis, and Evaggelos Spyrou
8. Enabling Smart Cities through IoT: The ALMANAC Way........................................ 173

Dario Bonino, Maria Teresa Delgado, Claudio Pastrone, Maurizio Spirito,
Alexandre Alapetite, Thomas Gilbert, Mathias Axling, Matts Ahlsen, Peter Rosengren,
Marco Jahn, Raphael Ahrens, Otilia Werner-Kytölä, and Jose Angel Carvajal Soto
v
vi Contents
9. NFC in IoT-Based Payment Architecture....................................................................... 203

Alak Majumder, Shirsha Ghosh, Joyeeta Goswami, and Bidyut K. Bhattacharyya
10. Trust Management in IoT ................................................................................................. 221

Avani Sharma, Pilli Emmanuel Shubhakar, and Arka Prokash Mazumdar
Section III Interdisciplinary Aspects of the Internet of Things
11. Highway Safety of Smart Cities ...................................................................................... 243

Sándor Szénási
12. Connected Bicycles: Potential Research Opportunities in Wireless

Sensor Network .................................................................................................................. 273
Sadik Kamel Gharghan, Rosdiadee Nordin, Nor Fadzilah Abdullah,
and Kelechi Anabi
13. Self-Adaptive Cyber-City System ................................................................................... 295

Iping Supriana, Kridanto Surendro, Aradea Dipaloka, and Edvin Ramadhan
14. The Role of Big Data, Cloud Computing, and Mobile Technologies

in the Development of IoT Ecosystems .......................................................................... 321
Janusz Wielki
15. Data Processing in IoT Using an Enterprise Service Bus ........................................... 337
Siddharth S. Prasad, Robin Singh Bhadoria, and Mohit Mittal
Section IV Future Research, Scope, and Case Studies of the

Internet of Things and Ubiquitous Computing
16. Infection Tracing in i-Hospital ........................................................................................ 349
Mimonah Al Qathrady, Ahmed Helmy, and Khalid Lmuzaini
17. Emergency Department, Sustainability, and eHealth: A Proposal to

Merge These Elements Improving the Sanitary System ............................................ 371
Andrea Malizia, Laura Morciano, Jacopo Maria Legramante, Pasqualino Gaudio,
Sandro Mancinelli, Francesco Gilardi, Carlo Bellecci, and Leonardo Palombi
18. The Contiki Operating System: A Tool for Design and Development of
IoT—Case Study Analysis ................................................................................................ 391
B. Venkatalakshmi, A. Pravin Renold, and S. Vijayakumar
19. Novel Trends and Advances in Health Care Using IoT: A Revolution....................405
Awanish Kumar and Archana Vimal
Contents vii
20. Mining Ubiquitous Data Streams for IoT ..................................................................... 421

Nitesh Funde, Meera Dhabu, Ayesha Khan, and Rahul Jichkar
21. IoT toward Efficient Analysis of Aging, Cardiometabolic, and

Neurodegenerative Diseases—An eHealth Perspective............................................. 441
Leandro Cymberknop, Parag Chatterjee, Diego Dujovne, Luis Romero,
and Ricardo Armentano
Index ............................................................................................................................................. 457

Preface
The Internet of Things (IoT) has emerged as a massive technology, which discovers new
fields that touch life especially in building protocols for smart cities and health care. It
poses a new trend of IoT, which thrives to proffer personalized services to the users by
developing a ubiquitous pool of services. Identifying these game-changing approaches
of IoT has led us to write this book, which would showcase a concrete state-of-the-art
research in these fields.
Research in the field of IoT cannot be confined to a specific area; it is enabled by the
handshaking of several domains of research such as sensors, networking, cloud comput-
ing, edge computing, big data, machine learning intelligence, security, and privacy. This
makes the domain of this book quite interdisciplinary where researchers from more than
15 countries in the world have shared their leading research in the field of IoT. All of the
21 chapters showcase the best research outcomes of the authors, having paramount signifi-
cance in the field of IoT.
Smart cities hold one of the key sectors of development led by IoT. Ranging from smart
transports to efficient management of citizens’ data and highway safety measures, IoT has
revolutionized every aspect for the development of a city. With a two-fold approach, this
book discusses the core concepts of IoT as well as different case studies related to applica-
tion for building smart cities with health care. We hope that this book would be a perfect
companion for all the researchers in IoT. The chapters on eHealth showcase interesting
research on developing comprehensive smart health care models.
With the advent of IoT, there has been a humongous increase in the amount of data gen-
erated by the devices. However, this gives more power to the systems, as the more these
data are analyzed, deeper insights into the users are obtained. Some chapters have also
discussed these aspects of data in IoT.
This book has the following four sections:
• Introduction to the Internet of Things: Definition and Basic Foundation

• Frameworks for the Internet of Things: An Architectural Perspective
• Interdisciplinary Aspects of the Internet of Things
• Future Research, Scope, and Case Studies of the Internet of Things and Ubiquitous
Computing
Ricardo Armentano
Robin Singh Bhadoria
Parag Chatterjee
Ganesh Chandra Deka
ix
Editors
Ricardo Armentano works as a distinguished professor at

the Favaloro University, Buenos Aires, Argentina. In addition,
he is the director of the PhD program (signal and image pro-
cessing) at the National Technological University, Buenos
Aires, Argentina and a Grade 5 Investigator of the URU/84/002
United Nations Program for the Development of Basic
Sciences. He received his engineering degree in 1984 and by
the end of 1994 he earned his first PhD in physiological
sciences from the University of Buenos Aires, Argentina. In
1999, he earned his second PhD degree from Université de
Paris VII Denis Diderot, Paris, France. He has acquired inter-
national recognition in the field of cardiovascular hemody-
namics and arterial hypertension and has profound research
and teaching experience in the fields of cardiovascular dynamics applied to an extensive
area of engineering in medicine and biology. He has more than 200 publications, including
books, book chapters, and peer-reviewed articles. Armentano is the vice-chair of the
Global Citizen Safety and Security Working Group of the International Federation of
Medical and Biological Engineering (IFMBE). He is a member of the EMBS IEEE since 1985
and chair of the Argentinean Chapter of the Engineering in Medicine and Biology Society
(EMBS), IEEE (2004). He was the chair of the 32nd International Conference of the EMBS
IEEE Buenos Aires 2010 and was the Latin-American IEEE EMB representative for 2015.
Robin Singh Bhadoria earned his PhD from Indian Institute

of Technology Indore, India. He has worked in different fields
such as data mining, frequent pattern mining, cloud comput-
ing, service-oriented architecture, and wireless sensor
networks. He has published more than 50 research articles in
several international and national conferences and journals
of high repute such as IEEE, Elsevier, and Springer that also
includes book chapters. Presently, he is serving as the associ-
ate editor of International Journal of Computing, Communications
and Networking (IJCCN) ISSN 2319-2720. In addition, he is
serving as an editorial board member for different journals of
the world. Presently, he is a professional member of different
professional research bodies such as IEEE (U.S.), International
Association of Engineers (IAENG) (Hong-Kong), Internet Society, Virginia (U.S.), and
International Association of Computer Science and Information Technology (IACSIT)
(Singapore).
xi
xii Editors
Parag Chatterjee works as a researcher at the UTN–GIBIO

(Research and Development Group in Bioengineering) and as
a professor at the postgraduate Department of Information
Systems Engineering in the National Technological University
(Universidad Tecnológica Nacional), Buenos Aires, Argentina. In
addition, he is a visiting scientist to the University of the
Republic (Universidad de la República), Montevideo, Uruguay.
In 2015, Chatterjee received his MSc degree with a first class
in computer science from the St. Xavier’s College, Kolkata
(University of Calcutta), India. His research has been
published in several international journals and was presented at IEEE International
Conferences and sessions of the Indian Science Congress. In addition, he delivered keynote
lectures and invited talks at highly reputed institutions such as the Indian Institute of
Science (IISc Bangalore, India), Jain University (India), Favaloro University (Argentina),
University of the Republic (Uruguay), and many international conferences. He has
cochaired sessions at international conferences such as SICC-2017 in Rome, and delivered
talks and moderated workshops during the IoT Week 2017, Geneva. Besides serving as an
editorial board member and reviewer of several international journals, Chatterjee is a
member of international scientific bodies such as Indian Science Congress Association,
Machine Intelligence Research Labs, and IEEE. His current research is focused on the
Internet of Things applied to health care.
Ganesh Chandra Deka works as the deputy director of train-

ing, under the Directorate General of Training and
Employment, Ministry of Skill Development and
Entrepreneurship, Government of India, New Delhi. His
research interests include big data, cloud computing, data
mining, NoSQL databases, and vocational education and
training. He has published more than 50 research papers in
various conferences, workshops, and international journals of
repute, including IEEE and Elsevier. He is editor-in-chief of
the International Journal of Computing, Communications, and
Networking (ISSN 2319-2720). So far, he has organized eight
IEEE International Conferences in India as the technical chair.
He is the coauthor for four textbooks on fundamentals of
computer science. He has edited five books (3 by IGI Global, U.S. and 2 by CRC Press/
Taylor & Francis Group) in the field of cloud computing, big data and NoSQL.
Contributors
Jemal Abawajy Kelechi Anabi

Parallel and Distributed Computing Faculty of Engineering and Built
Laboratory Environment
School of Information Technology Universiti Kebangsaan Malaysia
Deakin University Bangi, Malaysia
Burwood, Victoria, Australia
K.V. Arya
Nor Fadzilah Abdullah ABV-Indian Institute of Information
Faculty of Engineering and Built Technology and Management, Gwalior
Environment Gwalior, India
Universiti Kebangsaan Malaysia
Bangi, Malaysia Mathias Axling
CNet Svenska AB
Matts Ahlsen Stockholm, Sweden
CNet Svenska AB
Stockholm, Sweden Bojan Bakmaz
Faculty of Transport and Traffic Engineering
University of Belgrade
Raphael Ahrens
Belgrade, Serbia
User-Centered Ubiquitous Computing
Department
Fraunhofer FIT Miodrag Bakmaz
Sankt Augustin, Germany Faculty of Transport and Traffic
Engineering
University of Belgrade
Charilaos Akasiadis Belgrade, Serbia
Institute of Informatics and
Telecommunications
Carlo Bellecci
National Center for Scientific Research
Department of Industrial Engineering
“Demokritos”
University of Rome Tor Vergata
Athens, Greece
Rome, Italy
Mimonah Al Qathrady Bidyut K. Bhattacharyya

Computer Engineering Department of Electrical Engineering
University of Florida National Institute of Technology Agartala
Gainesville, Florida Agartala, India
Alexandre Alapetite Zoran Bojkovic

Data Science and Engineering Lab Department of Electrical Engineering
Alexandra Institute University of Belgrade
Copenhagen, Denmark Belgrade, Serbia
xiii
xiv Contributors
Dario Bonino Nitesh Funde

Pervasive Technologies Department of Computer Science and
Istituto Superiore Mario Boella (ISMB) Engineering
Turin, Italy Visvesvaraya National Institute of
Technology
Jose Angel Carvajal Soto Nagpur, India
Department Pasqualino Gaudio
Fraunhofer FIT Department of Industrial
Sankt Augustin, Germany Engineering
Leandro Cymberknop Rome, Italy
National University of Technology
Buenos Aires, Argentina
Sara Ghanavati
Upena Dalal Parallel and Distributed Computing
Department of Electronics and Laboratory
Communication Engineering School of Information Technology
Sardar Vallabhbhai National Institute Deakin University
of Technology, Surat Burwood, Victoria, Australia
Surat, India
Francesco Gilardi
Maria Teresa Delgado Department of Biomedicine and
Pervasive Technologies Prevention
Istituto Superiore Mario Boella (ISMB) University of Rome Tor Vergata
Turin, Italy Rome, Italy
Meera Dhabu Thomas Gilbert

Department of Computer Science and Data Science and Engineering Lab
Engineering Alexandra Institute
Visvesvaraya National Institute of Aarhus, Denmark
Technology
Nagpur, India
Sadik Kamel Gharghan
Aradea Dipaloka College of Electrical and Electronic
Department of Informatics Engineering Engineering Techniques
Faculty of Engineering Middle Technical University
Siliwangi University Baghdad, Iraq
Tasikmalaya, Indonesia
Shirsha Ghosh
Diego Dujovne Department of Computer Science
School of Informatics and and Engineering
Telecommunications National Institute of Technology
Diego Portales University Arunachal Pradesh
Santiago, Chile Yupia, India
Contributors xv
Joyeeta Goswami Diogo Ferreira Lima Filho

Department of Computer Science and Department of Electronic Systems
Engineering Engineering
National Institute of Technology University of São Paulo
Arunachal Pradesh São Paulo, Brazil
Yupia, India
Khalid Lmuzaini
Ahmed Helmy King Abdulaziz City for Science and
Computer and Information Science and Technology
Engineering (CISE) Department Riyadh, Saudi Arabia
University of Florida
Gainesville, Florida Alak Majumder
Department of Electronics and
Davood Izadi Communication Engineering
Parallel and Distributed Computing National Institute of Technology
Laboratory Arunachal Pradesh
School of Information Technology Yupia, India
Deakin University
Burwood, Victoria, Australia Andrea Malizia
Department of Industrial Engineering
Marco Jahn and
User-Centered Ubiquitous Computing Department of Biomedicine and
Department Prevention
Fraunhofer FIT University of Rome Tor Vergata
Sankt Augustin, Germany Rome, Italy
Rahul Jichkar Sandro Mancinelli

Persistent Systems Ltd. Department of Biomedicine and
Nagpur, India Prevention
Ayesha Khan Rome, Italy
Infocepts Technologies Pvt. Ltd.
Nagpur, India Arka Prokash Mazumdar
Department of Computer Science and
Awanish Kumar Engineering
Department of Biotechnology Malaviya National Institute of Technology,
National Institute of Technology Raipur Jaipur
Raipur, India Jaipur, India
Jacopo Maria Legramante Mohit Mittal

Department of Medicina dei Sistemi Department of Computer Science
University of Rome Tor Vergata Gurukul Kangri University
Rome, Italy Haridwar, India
xvi Contributors
Laura Morciano A. Pravin Renold

Department of Biomedicine and Department of Electrical and Electronics
Prevention Engineering
University of Rome Tor Vergata Velammal Engineering College
Rome, Italy Chennai, India
Rosdiadee Nordin Luis Romero

Faculty of Engineering and Built National University of San Juan
Environment San Juan, Argentina
Universiti Kebangsaan Malaysia
Bangi, Malaysia Peter Rosengren
CNet Svenska AB
Leonardo Palombi Stockholm, Sweden
Department of Biomedicine and Prevention
University of Rome Tor Vergata Shweta Shah
Rome, Italy Electronics and Communication
Engineering Department
Claudio Pastrone Sardar Vallabhbhai National Institute of
Pervasive Technologies Technology, Surat
Istituto Superiore Mario Boella (ISMB) Surat, India
Turin, Italy
Avani Sharma
Jigisha Patel Department of Computer Science and
Electronics and Communication Engineering
Engineering Department Malaviya National Institute of Technology,
Sardar Vallabhbhai National Institute of Jaipur
Technology, Surat Jaipur, India
Surat, India
Pilli Emmanuel Shubhakar
Department of Computer Science and
Vishnu Prabhakaran
Engineering
ABV-Indian Institute of Information
Malaviya National Institute of Technology,
Technology and Management, Gwalior
Jaipur
Gwalior, India
Jaipur, India
Siddharth S. Prasad
Maurizio Spirito
Discipline of Computer Science and
Emerging Trends and Opportunities
Engineering
Istituto Superiore Mario Boella (ISMB)
Indian Institute of Technology Indore
Turin, Italy
Indore, India
Evaggelos Spyrou
Edvin Ramadhan Institute of Informatics and
Department of Informatics Engineering Telecommunications
Faculty of Computer Science National Center for Scientific Research
Sriwijaya University “Demokritos”
Palembang, Indonesia Athens, Greece
Miguel Arjona Ramírez S.A. Srinivasa Moorthy

University of São Paulo Andhra Pradesh Electronics and IT Agency
São Paulo, Brazil Visakhapatnam, India
Contributors xvii
Iping Supriana S. Vijayakumar

School of Electrical Engineering and Department of Electrical and Electronics
Informatics Engineering
Bandung Institute of Technology Velammal Engineering College
Bandung, Indonesia Chennai, India
Kridanto Surendro Archana Vimal

School of Electrical Engineering and Department of Biotechnology
Informatics National Institute of Technology
Bandung Institute of Technology Raipur
Bandung, Indonesia Raipur, India
Sándor Szénási Marcel Stefan Wagner

Neumann János Faculty of Informatics University of São Paulo
Óbuda University São Paulo, Brazil
Budapest, Hungary
Otilia Werner-Kytölä
and
Faculty of Economics Department
Selye János University Fraunhofer FIT
Komárno, Slovakia Sankt Augustin, Germany
Grigorios Tzortzis Janusz Wielki

Institute of Informatics and Department of E-Business and Electronic
Telecommunications Economy
National Center for Scientific Research Opole University of Technology
“Demokritos” Opole, Poland
Athens, Greece
Wagner Luiz Zucchi
B. Venkatalakshmi University of São Paulo
Department of Electronics and São Paulo, Brazil
Communication Engineering
Velammal Engineering College
Chennai, India
Section I
Introduction to the
Internet of Things
Definition and Basic Foundation
1
The IoT Vision from an Opportunistic
Networking Perspective
Zoran Bojkovic, Bojan Bakmaz, and Miodrag Bakmaz
CONTENTS
1.1 Introduction ............................................................................................................................ 3
1.2 IoT-Enabling Communications Technologies ....................................................................4
1.2.1 Short-Range Communications Technologies .........................................................5
1.2.2 Cellular Technologies and IoT .................................................................................6
1.3 Software-Defined IoT ............................................................................................................6
1.3.1 Benefits of Software-Defined Networking ............................................................. 7
1.3.2 Software-Defined IoT Architecture and System Design ...................................... 8
1.3.3 Transition to Cloud Computing............................................................................... 9
1.4 Cloud Computing and IoT .................................................................................................... 9
1.4.1 Cloud of Things........................................................................................................ 10
1.4.2 Big Data Analysis ..................................................................................................... 11
1.5 Information-Centric Networking toward Efficient IoT Environment.......................... 12
1.5.1 Content-Based Naming ........................................................................................... 13
1.5.2 Information Distribution ........................................................................................ 15
1.5.3 In-Network Caching ................................................................................................ 16
1.6 Concluding Remarks ........................................................................................................... 18
References....................................................................................................................................... 18
1.1 Introduction
Currently, the Internet of Things (IoT) is still in the stages of development and deployment.
However, IoT will have an important impact on everyday life, just like Internet has today.
IoT allows us to interact and control systems as well as environments by creating digital
representation of the physical world. It is expected that IoT will have a huge influence to
the development not only on industry but our society as a whole. Differences in traffic
characteristics along with the energy constraints and specific features of the IoT commu-
nication environment are the motivation for research activities.
Inspired by the vision of Nikola Tesla “when wireless is perfectly applied the whole
earth will be converted into a huge brain, which in fact it is, all things being particles of
a real and rhythmic whole,” this chapter deals with IoT from the opportunistic network-
ing perspective. Starting from the summation of the technologies that are giving funda-
mentals for understanding IoT conception, this chapter continues with software-defined
IoT (SD-IoT), as a new solution. In software-defined networking, the control intelligence is
3
4 The Internet of Things
moved from data plane devices (switches, routers) and implemented in a logically central-
ized controller that interacts with data plane through standard interfaces. In accordance with
this, SD-IoT decouples the control logic from functions of the physical devices through a logi-
cally centralized controller that manages the devices via standard interface. Questions that
often arise are how to eliminate bottlenecks, with the idea to process the IoT data and on the
other hand, not to place a strain on the network. Fortunately, there exists a solution in Internet
traffic routing. Next, we deal with interaction of cloud computing and IoT, which can provide
intelligent participation and connection not only from rational, but also machine-type com-
munications. Special requirements go to interactions with sensors. In connection with this are
storage big data, huge computation power, and real-time audio and video streaming. Big data
analytics, as one of the trending research topics in science and technology communities when
dealing with IoT vision from opportunistic networking perspective, is presented. One of the
important challenges in IoT vision is how to move to real-time control for smart and connected
communities (SCC). Some opportunities of IoT in SCC are identified as mobile crowdsensing
and cyber-physical cloud computing. In the final part of this chapter, information-centric net-
working (ICN) will be invoked. Improved data dissemination efficiency and robustness in
communication scenarios indicate the high potential of ICN as a model that is different from
the traditional IP-centric one. This approach consists of the retrieval of content by names,
regardless of origin server location, application, and distribution channel, thus enabling
in-network caching/replication and content-based security.
1.2 IoT-Enabling Communications Technologies

The main task of the IoT is to enable ubiquitous objects to coordinate together. In this way, the
traditional objects become smart by exploiting pervasive computing, embedded devices, and
underlying communications technologies and protocols. Thus, world’s economy and quality
of everyday life will be significantly improved. On the other hand, emerging technologies and
different applications have to match the market demands as well as customers’ needs. Devices
are offered to fit the requirements of anywhere and anytime service availability. IoT-enabling
technologies and protocols contain the Internet and IP-based standards that will be funda-
mental in providing a common, well-paved road for the development and deployment of new
IoT applications (Al-Fuqaha et al., 2015).
IoT devices can be either stationary (e.g., home smart meters, thermostats, traffic signaliza-
tion, etc.) or mobile (e.g., fleet management devices, eHealth sensors, etc.). They can be con-
nected to the core Internet using either wired or wireless access links. Although the wired
solutions can provide higher data rates, reliability, security, and lower latency, they may not
be suitable for many IoT applications due to its cost, lack of scalability, and mobility support.
Wireless access networks can be either capillary/short range (e.g., ZigBee, Bluetooth, Wi-Fi, etc.)
or cellular (e.g., high speed packet access [HSPA], long term evolution [LTE], etc.). Capillary
solutions, mainly used for shared short-range links, are rather cheaper to roll out and gener-
ally scalable. However, small coverage, weak security, severe interference, and lack of univer-
sal infrastructure pose restriction on their applications. On the other hand, mobile systems
offer wide coverage, seamless connectivity, acceptable security, and ready-to-use infrastruc-
ture, making cellular networking a promising solution for IoT applications. After decades
of conceptual inception of the IoT, a variety of communication technologies has gradually
emerged, reflecting a large divrsity of application domains and communication requirements.
The IoT Vision from an Opportunistic Networking Perspective 5
1.2.1 Short-Range Communications Technologies

IEEE 802.15.4 is generally characterized by low data rate, high message throughput, low
power consumption, and low cost. Also, it provides a reliable and secure communica-
tion, but without quality of service (QoS) guarantees. This standard is the base for the
well-known ZigBee protocol, as they both focus on offering low data rate services over
power-constrained devices, and they build a complete network protocol stack for wire-
less sensor networks. The IEEE 802.15.4 Task Group has defined star and mesh network
topologies to suit different application requirements. These topologies consist of different
devices that can be classified as full function device (FFD) and reduced function device
(RFD). The FFD includes all features of the IEEE 802.15.4 standard and can serve as a
regular device or as a personal area network coordinator/controller. Also, the FFD can
serve as a relay in a mesh topology or as an end node with limited functions. On the other
hand, RFD implements the minimum required functionalities by the standard and can
only function as an end node such as home automation sensor.
In 2010, the Bluetooth Special Interest Group published their Core Specification 4.0, for
Bluetooth Low Energy (BLE) technology. BLE utilizes a short-range links with a minimal
amount of power (0.01 to 10 mW) and lower energy consumption compared with its previ-
ous versions. Almost all the latest versions of mobile platforms have been working to pro-
vide native support for BLE and are already everywhere in the market. Its coverage is 10
times that of the classic Bluetooth, whereas its latency is 15 times shorter. Compared with
ZigBee, BLE is more efficient in terms of energy efficiency, and with these characteristics
it is a good candidate for certain IoT applications (Chang, 2014). Implementation of BLE
nodes in the IoT environment requires invoking IPv6, which was not originally supported
by this standard. Usage of BLE for Internet applications poses challenges beyond IPv6
packet transport, including gateway operation, as well as application protocol efficiency
and security. A holistic solution for enabling and optimizing IPv6 over BLE is proposed
by Nieminen et al. (2014). Recently, the Bluetooth SIG announced release 5.0, which will
provide significantly increased speed, data-broadcasting capacity, and operation range.
Extending range will provide robust, reliable IoT connections that make full-home and
building and outdoor use cases a reality.
The IEEE 802.11ah Task Group was formed in 2010 to define an amendment to the 802.11
standard that operates in the sub-1-GHz license-exempt frequency spectrum to support
the following three use cases: sensors and meters, backhaul sensor and meter data, and
extended range Wi-Fi. The sensors and meters use case includes sub-use cases such as
smart grid, environmental monitoring, eHealth, home/building automation, and so on.
They all fit into IoT use cases that typically support a much larger number of devices per
access point for both indoor and outdoor environments with a longer transmission range
but at lower data rates than the use cases and requirements of conventional IEEE 802.11
(Park, 2015). IEEE 802.11ah (a.k.a. HaLow) is designed to support outdoor applications that
need a transmission range up to 1 km at 150 kb/s. For indoor applications, the long-range
capability can be used for energy-efficient sensor communications by using a very low
transmit power. IEEE 802.11ah provides energy-efficient medium access control protocols
and frame formats that are optimized for IoT applications, which need to support a large
number of nodes and infrequent small packet transmissions. The basic characteristics of
prominent short-range wireless technologies for IoT use cases are summarized in Table 1.1.
6LoWPAN (IPv6 over Low-Power Wireless Personal Area Networks) protocol was origi-
nally designed as an adaptation layer between IPv6 and 802.15.4 that provides a set of
mechanisms including fragmentation (RFC 4944), header compression (RFC 6282), and
TABLE 1.1
Characteristics of Short-Range Wireless Technologies for IoT
Standard IEEE 802.15.4 Bluetooth LE IEEE 802.11ah
Frequency band (MHz) 868/915/2,400 2,400 900

Data rates (Mb/s) 0.25 1 347
Coverage (m) 30 100 1,000
Number of nodes 65,000 6,000 6,000
optimized IPv6 neighbor discovery (RFC 6775). Recently, its application was expanded
over a variety of other radio networking technologies including BLE and IEEE 802.11ah.
6LoWPAN offers all the advantages from the IPv6 such as ubiquity, scalability, flexibility,
and end-to-end connectivity to the IoT environment.
1.2.2 Cellular Technologies and IoT

Although today all cellular technologies can support more or less some of the IoT applica-
tions; native support for massive communications is expected with development of future
generation of mobile systems, (i.e., fifth generation [5G]). Practically, in order to realize the
vision of IoT, emerging mobile networks need to provide connectivity to thousands of
devices, whereas, for example, standard LTE systems were originally designed to support
up to 600 connected users per cell. It is obvious that each connected thing will transmit
small data blocks sporadically. Current cellular architectures are not designed to simul-
taneously serve the aggregated traffic generated from a large number of devices. For
instance, current systems could easily serve five devices at 2 Mb/s each, but not 10,000
devices each requiring 1 kb/s (Boccardi et al., 2014).
Apart from higher capacity, low latency is another challenging issue for future cellular
systems. Generally, IoT comprises a vast amount of real-time applications with extremely
low-latency requirements. Motivated by the tactile sense of the humans, which can distin-
guish latencies on the order of 1 ms, 5G can be applied to managing and control scenarios,
implying a disruptive change from today’s content-driven communications. As stated by
Fettweis (2012), a 1-ms round-trip time for a typical tactile interaction requires a budget of
100 ms on the physical layer. This is far shorter than current cellular systems allow, miss-
ing the target by nearly two orders of magnitude.
1.3 Software-Defined IoT

It is well known that Internet has got great success, but it still shows some lacks. For
example, control is implemented by various routing and management protocols and is
embedded in every router/switch being hard to change. In that way, Internet infrastruc-
ture is developing slowly, whereas vendor-dependent interfaces make the networking
management very complex and unreliable. In order to support sustainable evolution,
efficient management, and heterogeneous application requirements, IoT architecture
should avoid these problems. It is clear that a change is needed in how networks are
designed and operated. Software-defined networks (SDNs) offer the ability to address
the abovementioned problems (Vaughan-Nichols, 2011). The main idea of SDNs inte-
gration in the IoT infrastructure is to extend software-defined approach from network
devices to sensor platforms and the cloud. This solution will provide well-defined service
application programming interfaces (APIs) in terms of data acquisition, transmission,
and processing.
1.3.1 Benefits of Software-Defined Networking

SDN approach supports the changing nature of future networks while taking advantage
of savings in equipment investments and operating costs. This will give operators the
agility to create or program highly flexible and dynamic networks capable of integrating
and monitoring terminals, intelligent machines, and smart devices while providing new
services.
In an attempt to overcome the inconvenience of vertical integration in today’s networks,
which means that the control and data planes are bundled together; SDN control logic is
separated from the underlying routers and switches (Zilberman et al., 2015). In this way,
the centralization of network control is promoted while increasing flexibility. SDN can
be considered to utilize a common and defined interface between a control plane and
date plane. The control plane is used for overall coordination, for example, routing and
error recovery, whereas the data plane usually manages packet-by-packet operations.
Comparison between functionality of the legacy router in classical network and the equiv-
alent SDN architecture is shown in Figure 1.1.
The separation of control plane from data plane would be realized through an API
between SDN switch agent and controller. Several processing functions will be carried
out such as buffering while headers are processed, header examination to identify switch-
ing action, and finally queuing packet for transmission. On the other hand, the control
plane manages the mechanisms by which the data-plane forwarding tables are created.
This includes running one or more routing protocols as well as exchanging local routing
information with other routers, together with deriving a local forwarding table. SDNs hold
great potential for application in optical networks (Channegowda et al., 2013), 5G mobile
networks (Cho et al., 2014), direct communications (Zhou et al., 2016), and in IoT environ-
ment (Liu et al., 2015).
Monitoring Host
Management Management plane
Routing protocols
Controller
Routing protocols Routing table
Control plane
Routing table
Forwarding table Switch
Forwarding table Data plane switching agent
switching
SDN enabled
Legacy router
router
FIGURE 1.1
Functionality of legacy router and equivalent SDN architecture.
1.3.2 Software-Defined IoT Architecture and System Design

SD-IoT architecture proposed by Liu et al. (2015) consists of physical infrastructure layer,
control layer, and application layer, as presented in Figure 1.2.
Physical infrastructure layer includes various physical devices, that is, sensors, actu-
ators, gateways, access points, routers, and servers. The physical devices transmit data
from one node to another and process them to extract required information. They interact
with control layer through standard interfaces, that is, a southbound interface. The control
layer acts between the infrastructure layer and application layer, managing the physical
devices with various functions through different southbound interfaces. Also, this layer
provides services to the application layer through APIs known as northbound interfaces.
The control layer provides data acquisition, transmission, and processing service. The
data acquisition service provides APIs for applications to specify their data requirements.
The controller obtains the required data from sensor platform. Data specification contains
general attributes: data type, targeted geographical areas, and time duration. In SD-IoT,
sensor platform is equipped with more than one sensor shared by many applications. The
sensor controller has a global overview of the underlying sensor platform. The network
is used to transmit data from the sensor platforms to servers in the cloud. To realize data
transmission, the network uses an SDN architecture. The controller steers packets to dif-
ferent destinations and schedules traffic to satisfy application requirements for network
quality and optimize the usage of network resource. Data-processing APIs allow appli-
cations to specify the required resources that include running submitted programs on
specific platform and deploying corresponding software entities. Finally, in application
Application layer
Smart city Smart transportation eHealth
Northbound interfaces
Data Data Data

Control layer
acquisition transmission processing
Sensor controller Network controller Cloud controller

Sensor Network Processing
Manage and Manage and Manage and
states states unit
optimization optimization optimization
monitoring monitoring monitoring
Southbound interfaces
Physical layer
Sensor platform Network Cloud platform
FIGURE 1.2
SD-IoT architecture.
layer, developers build sensing applications using provided APIs. They can customize
data acquisition, transmission, and processing. As the infrastructure is shared by multiple
applications, the overall costs are reduced.
In SD-IoT system design, some open problems related to the southbound interface, con-
trol layer, and so on often arise. By providing processing and transmission procedure in
the sensor platform, the sensor controller is decoupled from sensor platforms. To reduce
energy consumption at the sensor platform, middleware software can be implemented
in the controller. In order to decrease the frequency of interactions with sensor platforms
when they are not active, the controller is used. High scalability, performance, and robust-
ness can be achieved with a design of a logically centralized control layer. In order to
achieve these objectives, deploying multiple controllers is an obligation. The network con-
troller in SD-IoT provides end-to-end QoS-guaranteed data transmission. Anyhow, some
challenges need to be emphasized. Namely, the forwarding devices have a limited num-
ber of queues for QoS provisioning. Thus, it is difficult to support a great number of QoS
requirements. Next, it is a challenging issue to design efficient traffic-scheduling algo-
rithms to satisfy QoS requirements in a large-scale network environment. Ahmed and
Boutaba (2014) proposed a vertical approach to organize multiple controllers for wide-area
SDNs, which can also be extended to the control of sensor platforms and clouds.
1.3.3 Transition to Cloud Computing

It is well known that most objects in physical world are tiny not only in size but also in
capacity. In order to monitor these types of objects, a great number of sensors with the goal
to generate a huge amount of data to be stored and processed successfully are required. IoT
itself is not capable of storing, computing, or analyzing data collected from millions and
millions of sensors. Fortunately, this weakness is the potency of cloud computing in which
the storage and computation can be regarded as infinite. Thus, the IoT–cloud integration
is welcome. By providing infinite storage and computation, a shared pool of resources
becomes dynamically allocated and obtained when using any of IoT applications.
As an example, consider the importance of the cloud controller. In SD-IoT system, this
entity needs to decide how to map application service requests to the physical devices. The
constraints include server and storage capacity, as well as the type of software or platform.
If the cloud allows applications to store data and after that to rent virtual resources to pro-
cess them, the cloud controller needs to decide where to store the data and which servers
should be used to host virtual resources for post-data processing.
1.4 Cloud Computing and IoT

As a paradigm that is envisioned to involve all things in the world, IoT can never be
observed and treated as an isolated island. Its potential can never be fully explored before
the complete formation of cyber-physical space, where humans, computers, and smart
objects are pervasively interconnected. As a result of this process, the integration of IoT
with the existing communications systems and the corresponding enterprise networks and
cloud computing can be achieved. Social and mobile networks are included, too. Recently,
the integration of heterogeneous communications systems has become a main source
of networking innovation and has stimulated the proposition of novel interdisciplinary
concepts such as the Web of Things (Christophe et al., 2011), Social IoT (Atzori et al., 2011),
and Cloud of Things (CoT) (Distefano et al., 2012).
Starting from the fact that cloud computing can have virtually unlimited capabilities in
terms of storage and processing power, it represents a model for accessing a set of shared
and configurable resources that include networks, servers, storage facilities, and so on
(Mell and Grance, 2011). Therefore, IoT can be abstracted of its limitations regarding het-
erogeneity, connectivity, and security by cloud computing integration. Although concep-
tual analyses are abundant in the open literature, cloud computing and IoT integration
are still a challenging issue. Considering that objects are huge in number is a challenging
task because they are largely dynamic, their life spans vary, and the provided applications
require different QoS supports.
Taking into account relations between IoT and cloud computing, two different
approaches exist: bringing the cloud to the things and vice versa (Cavalcante et al., 2016).
The possibility of bringing the cloud to the things understands taking advantages offered
by cloud services. In that way, constraints of IoT such as storage, processing, energy con-
sumption are taken into account. This is of importance due to the energy limitation for the
IoT devices. To overcome this limitation, such devices could play the simple role of data
providers and send data to be processed and stored directly on the cloud. In the second
case, bringing the things to the cloud, it becomes a layer between the things and applica-
tions. In that way, CoT can be referred to everything as a service (XaaS), which represents
a flexible and scalable model, offering anything that could be consumed as cloud services.
An example of this approach is the concept of Sensing as a Service (SenaaS), which can
virtualize, share, and reuse sensing data to be ubiquitously consumed from the cloud.
1.4.1 Cloud of Things

The concept of IoT, with underplaying physical objects abstracted according to think-like
semantics, seems a valid starting point for orchestration of the various resources. In this
context, the cloud could play the central role to connect IoT with the Internet of People
through the Internet of Services. This new concept of pervasive and ubiquitous computing
can be addressed as the CoT, going beyond the interconnection and hyperlink of things
(Petrolo et al., 2015). It is based on the SenaaS model of IoT architecture, which involves
four conceptual layers (Perera et al., 2014): sensor and sensor owners’ layer, sensor publish-
ers, extended service providers, and sensor data consumers. First layer consists of sen-
sors, whereas the owners manage them and allow or not the publication in cloud. Sensor
publishers detect available sensors and communicate with the sensor owners. They also
get permission to publish sensors in cloud. In order to select required sensors, extended
service providers communicate with multiple sensor publishers. Finally, for consuming
sensed data, sensor data consumers need to be registered. The benefits of invoking this
model are sharing and reusing of sensed data, reduction of data acquisition cost, and col-
lect data previously unavailable.
However, there are some challenges that CoT needs to deal with such as heterogene-
ity in sensor types, protocols, and communication technologies, as well as interoperabil-
ity among different hardware and cloud solutions, and so on. In order to bridge the gap
between the disparate technologies, defining the abstraction layer is an obligation. The
technologies promote interoperability between IoT resources, information models, data
providers, and consumers, as well as simplify effective data access and integration.
In IoT, processing, storing, and representing big data are provided using the CoT
conception. Generally, some questions such as verifying, filtering, and analyzing data
still remain to be solved. One of the reasons is not only the lack of corresponding
standards, but also the heterogeneity of involved technologies. Second, the amount of
generated data requires the improvement as well as optimization of integration issues.
1.4.2 Big Data Analysis

Big data is the term that describes the large volume of data, both structured and unstruc-
tured, that inundates our life on a day-to-day basis. These data can be analyzed for insights
that lead to better decisions and strategic business moves. Big data and the trending
research in communications technology with great applications such as economy, climate,
health, social science, data mining, machine learning, and so on are continuing to grow
everyday using managing software tools. This will be possible owing to supporting the
networking and computers complexity when real-time services are required.
General architecture of big data analysis is shown in Figure 1.3. Volume, velocity, and
variety are common characteristics of big data (Lu et al., 2014). As it is not recommended
to move collected data for centralized storing, distributed big data is proposed, whereas
big data should be processed in parallel. In this way, new knowledge and innovation can
be mined in a reasonable amount of time. Big data processing is divided into two groups:
intraprocessing and interprocessing. In intra-big data processing, all data belong to the
same organization. On the other hand, if big data are part of different organizations, it will
be inter-big data processing. It should be emphasized that inter-big data processing will be
more challenging because big data sharing will first be executed before processing. Also,
many new security and privacy issues will arise during the big data sharing. The follow-
ing problems such as storage of high-volume data, process streaming, and analysis of real-
time data are identified.
Big data and IoT have been combined to work together. Namely, big data is one of the
promising solutions for IoT data management. Recently, a next-generation operational data
historian system is proposed by Huang et al. (2014). The main characteristic of that system is
being designed for long-time storage and analyze data, as well as process real-time queries.
Here, the system combines the advantages of a rational model and a time series data model
to support throughput and fast queries. First of all, it is of importance for sampled data to be
packaged forming regular time series, which are compressed. Also, irregular time series, as
well as hybrid grouping structures according to data characteristics, have to be taken into
Intra/inter big data processing
Distributed big data storing
Multi-source big data collecting

(wireless sensor networks, mobile crowdsensing, social networks)
FIGURE 1.3
General architecture of big data analysis.
consideration. In that way, compared with traditional databases, data-processing perfor-

mance in this system becomes improved by one to two orders of magnitude.
Two main challenges for big data analytics are data heterogeneity and decision-making.
Big data analytics enable to move from IoT to real-time control, important from many
SCC applications.
First of all, two applications of IoT in SCC are identified: mobile crowdsensing and
cyber-physical cloud computing. Mobile crowdsensing serves as a relay for data collec-
tion from large number of mobile sensing devices (Ganti et al., 2011). Based on the type
of shape being measured or mapped, mobile crowdsensing applications are categorized
into three groups: environmental, infrastructural, and social. Compared with the classi-
cal wireless sensor networks, mobile crowdsensing advantages include significantly more
storage, computation, and networking resources equipped with multimodality sensing
capabilities. Also, mobile devices avoid the cost and time of deploying large-scale sen-
sor networks. Next, IoT represents a networking infrastructure for cyber-physical sys-
tems (CPSs). These are smart networked systems with embedded sensors and actuators
designed to sense and interact with physical world including the human users. Moreover,
CPSs support real-time, guaranteed performance in safety-critical applications. A new
computing paradigm that can modify and rapidly build these systems is cloud computing.
Cyber-physical cloud computing is characterized by efficient use of resources, modular
composition, rapid development, and scalability. Cyber-physical cloud computing is of
great significance for numerous SCC applications such as smart grid, smart health care,
and smart disaster management.
The data relevant to SCC are heterogeneous by nature. Thus, it is a challenge how to
transform physical, biological, or social variables into an electrical signal. Taking into
account, massy data and multisource data streams are also needed. In order to success-
fully deal with data heterogeneity, the questions that often arise are related to quality and
accuracy improvement, unified data representation and processing, interpretation and
interoperability improvement, knowledge creation, and short-term and long-term storage
conduction (Sun et al., 2016).
As for decision-making, there are many sources of uncertainty that have to be consid-
ered in SCC (facing cities is incomplete, lack the data needed to specify the boundary
conditions with sufficient accuracy, etc.). This process can be improved under uncertainty
by understanding assessment, representation and propagation of uncertainty, developing
robust optimization methods, and designing optimal sequential approach.
1.5 Information-Centric Networking toward Efficient IoT Environment

The large-scale deployment of IP-based IoT solutions still provides challenging issues,
regardless of the great efforts and valuable achievements of standardization bodies, (e.g.,
Internet Engineering Task Force [IETF], Institute of Electrical and Electronics Engineers
[IEEE], etc.). The general constraints and challenges are related to the limited expressive-
ness of addressing, complex mobility and multicast support, and opportunistic massive
access under the strict reliability and security requirements.
Simultaneously with ongoing development of IoT environment, research community is
currently promoting a new communication paradigm that is fundamentally different from
traditional IP-centric networking. Having in mind the fact that the Internet is increasingly
used for information dissemination, rather than for peer-to-peer communication, this rev-
olutionary paradigm aims to reflect future needs better than the traditional architecture
(Ahlgren et al., 2012). Significant approach of several future Internet research activities is
known as ICN. In this concept, instead of accessing and manipulating information bypass-
ing of servers hosting them, appealing of the named information object in the center of
networking, from the viewpoint of information flow and storage, is applied.
Defining network service in terms of named information objects, independently of
where and how they are stored or distributed, caching information storage becomes the
fundamental part of network infrastructure. As a result, an efficient and application-
independent large-scale information distribution is enabled (Ahlgren et al., 2011). Here, a
user request can be sent directly to the network, independently of considering the original
content location.
During the last decade, several prospective ICN approaches have been proposed in the
open literature. Although characterized by different design issues, they are sharing com-
mon core principles, objectives, and architectural properties that can be summarized as
follows (Rao et al., 2014): content-based naming, name-based routing (NBR) and name res-
olution, and in-network caching.
Considering the main principles of ICN (i.e., unique location-independent naming,
content-based security, in-network cashing), this revolutionary paradigm can match a
wide set of IoT applications that are information-centric in nature (Amadeo et al., 2016). It
is obvious that ICN names can directly address heterogeneous IoT contents and services
independent from the location, especially in the presence of mobility. Moreover, by cach-
ing data closer to the user, ICN can significantly reduce traffic load, delay, and massive
access to resource-constrained devices. Consequently, scientific community is considering
IoT integration with ICN within the Internet Research Task Force (Pentikousis et al., 2015)
and through recently published research works (Baccelli et al., 2014; Amadeo et al., 2015).
However, ICN deployment in IoT environment is still in discussion phase.
Generally, ICN can be deployed as overlay over the existing infrastructure or as clean-
slate solution directly over physical layer instead of IP. Bearing in mind the fact, that most of
IoT devices are resource-constrained, overlay solution seems to be unsustainable due to its
complexity and overhead as a result of overlay management and encapsulation inside tra-
ditional protocols. On the other hand, a clean-slate solution can easily be deployed if there
is no need to maintain backward compatibility or to provide global connectivity. A most
likely short-term solution would allow coexistence with IP-based technologies. Similarities
between ICN hierarchical names and uniform resource identifiers of web resources could
facilitate such coexistence. The translation between them may be implemented easily in the
node (e.g., gateway) interconnecting ICN islands and the rest of the Internet. In what fol-
lows, some challenging issues regarding ICN and IoT integration will be analyzed.
1.5.1 Content-Based Naming

In ICN, instead of specifying a host pair for communication, a piece of information itself is
named, addressed, and matched independently of its location, being located anywhere in
the network. In that way, moving from the host-naming model to the information-naming
model contributes that the information retrieval becomes receiver driven as an interfer-
ing implication and benefit. In ICN, there is no possibility for data to be received unless
it is explicitly requested by the receiver. As for the network, it is responsible for locat-
ing the best source that will provide the desired information after sanded request. Both
hierarchical and flat names are possible, with the former appearing as uniform resource
identifier-like identifiers with variable lengths, whereas the latter comprises unique fixed-
length identifiers without semantic structure. As routing entries for contents might be
aggregated, hierarchical nature helps mitigate the routing scalability. On the other hand,
flat naming aggravates the routing scalability problem due to the absence of aggregation.
Moreover, security and privacy are closely related to the naming scheme. In the case of
hierarchical naming, security-related information is embedded into a specific field of the
content unit, thus requiring a public key infrastructure for integrity checks. Flat name
principle instead enables the self-certifying namespaces, allowing integrity checks with-
out the need for a public key infrastructure.
Naming IoT content enables information to be structured into scopes and allows users
to specifically request the desired content. This flexibility exploits the higher addressing
potential of ICN, allowing a name in the IoT context to identify not only content but also
a service or a device function. By offering name resolution at the network layer and for-
warding content by its name, ICN also has the potential to reduce the signaling overhead
in IoT environment. ICN nodes have the ability to identify requests for the same named
information, avoiding the need to forward them differently on the same path. In addition,
content becomes cached in traversing nodes, allowing requests to be satisfied by the first
available copy, preventing source overquerying, and supporting connectionless scenarios.
Moreover, in ICN, data can be transmitted to multiple consumers by supporting native
multisource dissemination.
An ICN-naming scheme for IoT should be highly expressive and scalable. As stated
before, flat names are typically obtained through hash algorithms applied to existing con-
tents and can hardly be assigned to dynamic and unpublished contents. On the other
hand, hierarchical names facilitate dynamic on demand contents (e.g., a parameter mea-
sured by a sensor), provided that naming conventions have been specified during the sys-
tem configuration. On account of that, hierarchical naming has been mainly considered
in the literature in order to support such properties (Baccelli et al., 2014; Shang et al., 2014).
The basic idea is to define a hierarchy of name segments that identifies the IoT application
(e.g., building automation systems, smart city) and the attributes that describe the related
contents/services (Figure 1.4).
However, hierarchical naming entails length constraints, for instance, to fit the maxi-
mum payload size of some protocols. In parallel, variable-length names make line-speed
name lookup extremely challenging. Especially under large-scale scenarios, naming schemes
should be designed together with processing techniques (e.g., name component encod-
ing) that accelerate name lookup (Zhang et al., 2015). This is of practical importance for
FIGURE 1.4
Examples of hierarchical naming for smart city application.
safety-critical applications (e.g., smart transport and eHealth) in which content access
latency reduction is crucial. By sharing a common name prefix for multiple contents/
services, hierarchical names scale better than flat names, as they facilitate the definition
of name aggregation rules, which is of importance for big data analytics. In IoT, the name
prefix can identify application types, physical locations, or other macro-categories that
broadly identify groups of data and services (Amadeo et al., 2016).
1.5.2 Information Distribution

Efficient information distribution should make use of any available data source to reduce
network load and latency and increase information availability. ICN architectures have to
solve the problem of how to retrieve data based on a location-independent identifier. There
are two major solutions to this problem: NBR and name resolution system (NRS). The
main difference between these two solutions is in the fact that the first one involves a path
from the provider to the requesting host, whereas the second is going to involve matching
an information name to a provider or any source that can supply that information. These
two functions can be integrated or independent. In the case of integrated functions, the
information request is routed to the provider, which subsequently sends the information
to the requesting host by utilizing the reverse path over which the request was forwarded.
The independent approach is characterized by the name resolution function, which does
not determine or restrict the path that is used on the provider–subscriber route. Whether
it is a NBR or a NRS approach, a list of common desirable properties are identified by Bari
et al. (2012) as follows:
• ICN routing mechanism should provide low-latency network level primitive oper-
ations for content (original, replica, or cached) registration, metadata update, and
deletion. It should be able to route a content request to the closest copy, based on
some network metric.
• Message propagation for name resolution and retrieval should not leave the net-
work domain that contains both the source and the content.
• The routing mechanism should provide guarantees on discovery of any existing
content, regardless of the content’s popularity and replication level.
• As the number of contents is in the order of trillions, any NBR/NRS scheme needs
to scale to at least this many contents and possibly beyond to accommodate future
growth. The trade-off between routing stretch (routing path length and minimum
length path ratio) and routing table size needs to be analyzed, while keeping in
mind the huge number of names and physical limitations imposed by memory
technologies.
• Ideally, the content retrieval process should be a one-step process, either by com-
bining name resolution and routing in a single step or by completely eliminating
the name resolution part.
Offered by ICN for content discovery, NBR and NRS schemes may suit specific IoT scenar-
ios, mainly depending on the content characteristics (e.g., popularity, dynamic generation)
and network features (e.g., infrastructureless vs. infrastructured). NBR, coupled with data
delivery performed by maintaining some soft-state at each Interest forwarder, is suitable to
access popular contents in infrastructured scenarios, whereas it is the only viable solution
in isolated networks with an alternately available infrastructure or ad-hoc environment.

Its inherent benefits are as follows:
• Robust and resilient retrieval through adaptive forwarding coupled with

in-network caching proposed by Amadeo et al. (2014).
• Easy resource discovery in infrastructureless networks through direct interdevice
communication by means of Interest packet broadcasting analyzed by Baccelli
et al. (2014).
The downside of deploying NBR is mainly related to the growth of the forwarding table
size and routing updates in the case of a huge number of names, and the overhead of
maintaining the soft state. However, name-prefix aggregation can successfully cope with
such challenges, together with adaptive forwarding. NRS is useful in infrastructured sce-
narios with unpopular and popular content. It can also be beneficial with off-path cach-
ing, that is, an alternately connected IoT source can push data in a predefined always-on
location (e.g., the cloud) accessible by consumers. In these cases, deploying a global name
resolution service based on a hierarchically organized distributed hash tables is desirable,
whereas additional scalability properties for name lookup can be obtained by using data
center capacities in the cloud, as suggested by Xylomenos et al. (2014). Moreover, NBR and
NRS solutions may complement each other, and together application of cloud computing,
multilevel distributed hash tables, name-prefix aggregation, and adaptive forwarding, an
effective discovery and delivery platform can be provided with the potential to scale even
for a huge number of IoT resources.
1.5.3 In-Network Caching

In order to improve the content distribution and network utilization, an important approach
of ICN is to provide ubiquitous and transparent in-network caching process (Zhang et al.,
2013). Although caching is not a wholly new technique, the lack of a unique identification
of identical objects makes it incompletely utilized in the current Internet architecture. As
mentioned earlier, ICN names content in a unified, consistent, and network-aware way.
This characteristic feature makes caching a general and transparent service, independent
of applications.
In the context of in-network caching, the crucial problems that arise are related to the
selection of the storage locations and of the specific content items to store in the different
caches. Generally speaking, there exist two opposite approaches that can be distinguished
for addressing such problems: coordinated and uncoordinated. The main property of
coordinated case is that routers exchange information to achieve a better estimation of the
contents popularity, avoiding at the same time the storage of too many content copies. In the
uncoordinated case, each cache operates autonomously using transparent route-caching
policy. Here, each router in the end-to-end path decides whether to cache transiting pieces
of content and which cache-replacing policy without interacting with other routers to per-
form. In the cases where the popularity of objects does not change frequently, the least
frequently used approach would achieve the highest performance, whereas the most pro-
posed solutions adopt the least recently used replacing policy (Kim and Yeom, 2013).
It is of interest to point out that the effectiveness of in-network caching is higher when
coordinated solution is applied. Also, it should be noted that the overhead required to
manage the coordination between caches may become extremely high. Consequently,
time uncoordinated solutions are now adopted more frequently. As a perspective solution,
C2 R5 S2
4: X
R6 R4 Content Y
3: Q(X)
3: Q(X)
4: X
2: X
2: X
2: X
2: X 1: Q
R1 ) R3
(X) 1: Q(X
1: Q(X) 1: Q(X)
C1 R2
S1 Content X
FIGURE 1.5
An example of in-network caching concept.
some trade-offs between the coordinated and uncoordinated approaches are proposed by
Sourlas et al. (2012) and Chai et al. (2013).
As an illustrative example of in-network caching concept, the network shown in
Figure 1.5 is considered. Suppose that consumer C1 sends an interest message requesting
a piece of content X published by server S1. Consequently, the request message Q(X) tra-
verses routers R1, R2, and R3. Server S1 receives the interest message issued by consumer C1
and forward the desired piece of content X. Content data will traverse nodes R3, R2, and R1
back, which will cache local copies. Now, assuming that consumer C2 generates an inter-
est message for the same piece of content, the corresponding request message Q(X) will
traverse routers R6 and R1. The router R1 can perceive that a copy of X existing in its cache
and there is no need to forward the interest message to the next hop toward S1. Instead R1
sends the content X to consumer C2 only via router R6.
In-network caching acquires special significance in IoT environment. Generally, caching
and related operations can be quite ineffective in terms of processing and energy con-
sumption. However, in-network caching is beneficial because it speeds up data retrieval
and increases its availability. It can reduce the number of lossy hops toward the source of
information by limiting the network load and the overall energy consumption. In addi-
tion, caching is viable as data generated by IoT devices typically have a small size and a
short lifetime.
Overall, IoT data can be cached in network routers and resource-constrained devices by
implementing caching decision and replacement policies that account for the peculiarities
of IoT traffic, for example, compatibly with freshness requirements and device capabilities.
Specifically, the indiscriminate cache everything everywhere approach is inefficient due to
the high level of content redundancy and poor utilization of the available cache resources.
Alternative caching decision policies (e.g., probabilistic) have been proposed by Vural et al.
(2014) for more efficient usage of the available caching resources, alleviating the load of
nodes and bandwidth consumption. Some of these policies address the space usage issue
by also considering the content popularity for topology-related centrality metrics. These
could also be feasible approaches for IoT, when the same content is requested by different
applications. However, they are suitable for static content, which is not entirely the case for
IoT, where content is usually transient.
1.6 Concluding Remarks

This chapter focuses on some most relevant networking solutions for IoT realization that
will play a vital role in the development of future communications systems. Reaching a
level of maturity that allows for practical realization of IoT, the enabling technologies will
be useful in many applications. This will help us to clear uncertainty that prevents a mas-
sive adoption. Application innovation will be accelerated when using software-defined
architecture. In that way, not only control and physical infrastructure management but
also cloud computing will be invoked as sustainable solution to overcome or alleviate
several challenges faced by the IoT paradigm. Moreover, information-centric approach is
advocated as inspiring networking solution for future Internet design, as well as IoT envi-
ronment as its indispensable part. Proposed opportunistic networking vision is a good
start, although many open challenges still remain. Also, designing better infrastructures
and services for the future smart communities remain open questions for the future work.
References
Ahlgren, B. et al., 2011. Content, connectivity, and cloud: Ingredients for the network of the future.
IEEE Communications Magazine 49 (7): 62–70.
Ahlgren, B. et al., 2012. A survey of information-centric networking. IEEE Communications Magazine
50 (7): 26–36.
Ahmed, R. and Boutaba, R., (2014). Design considerations for managing wide area software defined
networks. IEEE Communications Magazine 52 (7): 116–123.
Al-Fuqaha, A. et al., 2015. Internet of things: A survey on enabling technologies, protocols, and appli-
cations. IEEE Communications Surveys & Tutorials 17 (4): 2347–2376.
Amadeo, M. et al., 2014. Named data networking for IoT: An architectural perspective. Proceedings of
the EuCNC. Bologna, Italy, pp. 1–5.
Amadeo, M. et al., 2015. Information centric networking in IoT scenarios: The case of a smart home.
Proceedings of the IEEE International Conference on Communications. London, GB: 648–653.
Amadeo, M. et al., 2016. Information-centric networking for the internet of things: Challenges and
opportunities. IEEE Networks 30 (2): 92–100.
Atzori, L., Iera, A., and Morabito, G., 2011. SIoT: Giving a social structure to the internet of things.
IEEE Communications Letters 15 (11): 1193–1195.
Baccelli, E. et al., 2014. Information centric networking in the IoT: Experiments with NDN in the wild.
Proceedings of The ACM ICN'14. Paris, France, pp. 77–86.
Bari, M.F. et al., 2012. A survey of naming and routing in information-centric networks. IEEE
Communications Magazine 50 (12): 44–53.
Boccardi, F. et al., 2014. Five disruptive technology directions for 5G. IEEE Communications Magazine
52 (2): 74–80.
Cavalcante, E. et al., 2016. On the interplay of internet of things and cloud computing: A systematic
mapping study. Computer Communications 89–90: 17–33.
Chai, W.K. et al., 2013. Cache “Less for More” in Information-centric networks (extended version).
Computer Communications 36 (7): 758–770.
Chang, K.-H., 2014. Bluetooth: A viable solution for IoT? [Industry Perspectives]. IEEE Wireless
Communications 21 (6): 6–7.
Channegowda, M., Nejabati, R., and Simeonidou, D., 2013. Software-defined optical networks tech-
nology and infrastructure: Enabling software-defined optical network operations. IEEE/OSA
Journal of Optical Communications and Networking 5 (10): A274–A282.
Cho, H.H. et al., 2014. Integration of SDR and SDN for 5G. IEEE Access 2: 1196–1204.
Christophe, B. et al., 2011. The web of things vision: Things as a service and interaction patterns.
Bell Labs Technical Journal 16 (1): 55–61.
Distefano, S., Merlino, G., and Puliafito, A., 2012. Enabling the cloud of things. Proceedings of the
6th (IMIS). Palermo, Italy, pp. 858–863.
Fettweis, G., 2012. A 5G wireless communications vision. Microwave Journal 55 (12): 24–36.
Ganti, R.K., Ye, F., and Lei, H., 2011. Mobile crowdsensing: Current state and future challenges.
Huang, S. et al., 2014. The next generation operational data historian for IoT based on informix.
Proceedings of the ACM SIGMOD. Snowbird, UT: 169–176.
Kim, Y. and Yeom, I., 2013. Performance analysis of in-network caching for content-centric net-
working. Computer Networks 57 (13): 2465–2482.
Liu, J. et al., 2015. Software-defined internet of things for smart urban sensing. IEEE Communications
Magazine 53 (9): 55–63.
Lu, R. et al., 2014. Toward efficient and privacy-preserving computing in big data era. IEEE
Network 28 (4): 46–50.
Mell, P. and Grance, T., 2011. The NIST definition of cloud computing. Recommendations of the
National Institute of Standards and Technology. Gaithersburg, MD.
Nieminen, J. et al., 2014. Networking solutions for connecting bluetooth low energy enabled
machines to the internet of things. IEEE Network 28 (6): 83–90.
Park, M., 2015. IEEE 802.11ah: Sub-1-GHz license-exempt operation for the internet of things.
Pentikousis, K. et al., 2015. Information-centric networking: Baseline scenarios. IRTF Information-
Centric Networking Research Group (ICNRG) RFC 7476.
Perera, C. et al., 2014. Sensing as a service model for smart cities supported by internet of things.
Transactions on Emerging Telecommunications Technologies 25 (1): 81–93.
Petrolo, R., Loscri, V., and Mitton, N., 2015. Towards a smart city based on cloud of things, a
survey on the smart city vision and paradigms. Transactions on Emerging Telecommunications
Technologies (early view).
Rao, K.R., Bojkovic, Z.S., and Bakmaz, B.M., 2014. Wireless Multimedia Communication Systems:
Design, Analysis, and Implementation. Boca Raton, FL: CRC Press.
Shang, W. et al., 2014. Securing building management systems using named data networking.
IEEE Network 28 (3): 50–56.
Sourlas, V. et al., 2012. Autonomic cache management in information-centric networks. Proceedings
of the IEEE/IFIP NOMS 2012. Maui, HI, pp. 121–129.
Sun, Y. et al., 2016. Internet of things and big data analytics for smart and connected communi-
ties. IEEE Access 4: 766–773.
Vaughan-Nichols, S.J., 2011. OpenFlow: The next generation of the network? Computer 44 (8): 13–15.
Vural, P. et al., 2014. In-network caching of internet-of-things data. Proceedings of the IEEE ICC
2014. Sydney, NSW, pp. 3185–3190.
Xylomenos, G. et al., 2014. A survey of information-centric networking research. IEEE Communi-
cations Surveys & Tutorials 16 (2): 1024–1049.
Zhang, G., Li, Y., and Lin, T., 2013. Caching in information centric networking: A survey. Computer
Networks 57 (16): 3128–3141.
Zhang, Y. et al., 2015. ICN based architecture for IoT-requirements and challenges. IETF Internet
Draft.
Zhou, M. et al., 2016. Design and implementation of device-to-device software-defined networks.
Proceedings of the IEEE ICC. Kuala Lumpur, Malaysia, pp. 1–6.
Zilberman, N. et al., 2015. Reconfigurable network systems and software-defined networking.
Proceedings of the IEEE 103 (7): 1102–1124.
2
DEFINE—An Architecture Framework
for Designing IoT Systems
S.A. Srinivasa Moorthy
CONTENTS
2.1 Introduction ..........................................................................................................................22
2.2 DEFINE Framework—An Outline .................................................................................... 25
2.2.1 Domain ...................................................................................................................... 25
2.2.2 Characteristics of Domains .................................................................................... 26
2.2.2.1 Operating Environment ........................................................................... 26
2.2.2.2 Usage of the Product................................................................................. 27
2.2.2.3 Product Cost............................................................................................... 27
2.2.2.4 Product Development Process ................................................................ 28
2.2.2.5 Product Life................................................................................................ 28
2.2.2.6 Product Support Requirement ................................................................ 29
2.2.3 Environment ............................................................................................................. 29
2.2.3.1 Temperature ............................................................................................... 29
2.2.3.2 Humidity .................................................................................................... 30
2.2.3.3 Pressure ...................................................................................................... 30
2.2.3.4 Dust Ingress ...............................................................................................30
2.2.3.5 Fluid Ingress ..............................................................................................30
2.2.3.6 Electrical Power ......................................................................................... 31
2.2.4 Functionality ............................................................................................................ 31
2.2.4.1 Usability and Ergonomics........................................................................ 32
2.2.4.2 Manufacturability ..................................................................................... 32
2.2.4.3 Product Interconnect and Interface ........................................................ 32
2.2.4.4 Product Performance ................................................................................ 32
2.2.4.5 Product Safety ........................................................................................... 33
2.2.4.6 Product Sustenance................................................................................... 33
2.2.5 Implementation ........................................................................................................34
2.2.5.1 Development Process ...............................................................................34
2.2.5.2 Product Manufacturing Process .............................................................34
2.2.5.3 Product Verification and Validation Process......................................... 35
2.2.5.4 Product Support Process .......................................................................... 35
2.2.5.5 Product Documentation and Control..................................................... 36
21
2.2.6
Necessities ................................................................................................................. 36
2.2.6.1 Compliance to Global Standards ............................................................ 36
2.2.6.2 Product Reliability .................................................................................... 37
2.2.6.3 Product Accuracy ...................................................................................... 37
2.2.6.4 Design for Support .................................................................................... 37
2.2.6.5 Product Interoperability........................................................................... 38
2.2.6.6 Environmental Compliance..................................................................... 39
2.2.7 Enablers ..................................................................................................................... 40
2.2.7.1 Reference Hardware Platform................................................................. 40
2.2.7.2 Reference Software Platform ................................................................... 40
2.2.7.3 Operating System Used............................................................................ 41
2.2.7.4 OEM/ODM Components ......................................................................... 41
2.2.7.5 Intellectual Property and Patents ...........................................................42
2.2.7.6 Tools and Libraries....................................................................................42
2.3 Putting All Together ............................................................................................................42
2.4 Summary............................................................................................................................... 47
Bibliography .................................................................................................................................. 47
2.1 Introduction
The easiest way to describe an embedded system is to compare it with a computer.
A typical computer includes subsystems such as CPU, hard drives, display, and keyboard.
It can be graphically described as in Figure 2.1.
An embedded system is similar to a computer, but it is not so big. The CPU and all the
peripherals can be accommodated in a single printed circuit board as illustrated in Figure 2.2.
Embedded systems have started pervading in our daily life, and their dominance is
increasing with the passage of time. The embedded systems that are used by us in day-
to-day life are interconnected and can send data from the activities they monitor or can
Power supply
Hard
Monitor
drive
CPU
DVD Key-
drive USB and other board
interfaces
FIGURE 2.1
Block diagram of a typical computer.
DEFINE—An Architecture Framework for Designing IoT Systems 23
Power supply
Flash LC
display
CPU
RAM Keys
Digital and Analog
I/O
FIGURE 2.2
Block diagram of typical embedded system.
communicate to a central system that is called the Internet of Things (IoT). This central sys-
tem could be a server or a cloud platform.
One of the best examples of IoT in our daily life is the set-top box that we use to watch
cable TV. A set-top box starts sending information to the central sever about the channels
that we watch and the programs we frequently watch. This enables broadcasters to tailor
the programs for the individual users (an option that is available today only in IP TV!).
IoT is a collection of Things (which are basically embedded systems of various sizes and
capability) that interact with a real environment with an infrastructure mechanism to con-
nect them to the central repository through communication links, and a central storage
and process mechanism (Cloud with applications to process the data collected).
As the applications of IoT start expanding, one of the biggest challenges that the
designers face is the development of Things (embedded systems) to work in different
environments, to be used by different users, and to work with different communication
infrastructures. Designing systems with so many combinations is a bit difficult. Major
challenge faced by most designers is designing a Thing (a.k.a. embedded system) that
can work unattended 24 × 7 without any human intervention and also send data to
the central location. The ability to design a system like this calls for skills that spread
across every aspect of product design covering electrical, software, mechanical, manu-
facturing, and compliance. Although expertise for individual skills may be available for
implementation, their availability during the early stage of product design is always a
challenge for small companies and startups. Second, access to these skills at the early
stage is also an expensive proposition. So, a tool or a framework to help in the early stage
design will be really beneficial to the designers in the long course. This chapter describes
one such framework called DEFINE that enables the designers to develop the system
architecture in all aspects by clearly understanding the challenges involved in the Thing
(System) design.
Functionally, all the embedded systems are very similar in architecture with the main
CPU, associated peripherals, and the connected sensors. These are controlled by dedicated
software called Firmware running on them to do the necessary functions. However, for
an embedded system to work successfully and reliably in different applications scenario,
many factors need to be addressed.
Normally, new product ideas come from needs or problems and many a time, the user
having the need will not be a technical person. These needy users reach out to a designer
to solve their problem. So the solution to the problem will be quite abstract in the begin-
ning, and starting the design with such abstract inputs will definitely be very challenging.
In addition, most startups and small companies will not be having all the skills required
for a complete product design, especially skills such as enclosure design, manufacturing,
test engineering, and so on. This forces them to reach out to external experts for help at the
very beginning of the project. In such situations, the designers will be only validating their
concepts, and a firm decision to design a product would not have been taken. Involving
experts at this stage will be expensive. This reiterates the need for a tool or a framework
for the designers to do the initial design inexpensively.
This chapter will discuss a framework that will help the designers to develop the initial
design in a robust way. Before discussing the framework, it is important that designers
understand the concept of Product Lifecycle (PLC). A brief introduction of a typical PLC
is described to help the users in understanding the activities involved in each phase or
stage in the development of a product. A typical PLC starts with the Concept stage and
progresses toward the final Disposal stage. Figure 2.3 illustrated the typical life cycle of an
electronic product. The PLC has six distinct phases/stages starting from concept, design,
engineering, new product introduction, manufacture, and finally to the disposal stage.
Figure 2.3 clearly lists out the distinct activities in each stage. The key aspect of the PLC is
the interdependence of the different activities as well as concurrent (parallel) activities in
some stages.
With this understanding, let us see how the DEFINE framework will help the developers
in designing a robust and successful product.
Dictionary meaning of framework is the basic structure of something: a set of ideas or
facts that provide support for something. The DEFINE tool does exactly the same. It allows
the developer to design a product from abstract inputs so that it can meet all needs of
the user successfully. DEFINE framework essentially consists of elements and each
element has characteristics. Every characteristic addresses a need of the product.
Concept Design Engineering New product introduction Manufacture Disposal
Hardware System
design engineering
SCM and BoM DFx and test Pre

consolidation engineering compliance
Software Verification and

Concept Coding Manufacture Disposal
design validation
Design Tooling and Pilot

analysis vendor dev. manufacture
Mechanical Tooling and

design prototype
FIGURE 2.3
Modern product life cycle.
When all the elements and the characteristics are covered, the basic design require-
ment of a product can be easily derived.
As the basic requirements are captured completely through the DEFINE framework, the
designers are able to have a solid design for the product.
2.2 DEFINE Framework—An Outline

Based on the years of experience and study as a designer and a product manager, the DEFINE
framework was developed to help the designers to get their design (especially the architec-
ture) right when they design embedded products or IoT—Things. With explosive growth
of IoT, this framework will enable designers to quickly arrive at a cost-effective, reliable,
and robust design that is able to meet the Time to Market target easily.
Basic foundation of this DEFINE framework is the elements that are critical for the
working of Things. Things in a IoT network have to work in all kinds of environment and
communication systems, with any type of user and usage (sometimes misuse) unlike the
computers that typically work in a controlled and comfortable environment (most of the
time, computers work in the same space where humans work, which are comfortable).
This essentially forces the designers to develop a system that is robust, reliable, and eas-
ily usable. This is feasible only when the systems working from various aspects are well
understood by the designers.
• DEFINE is a framework that enables the designers to understand the various

aspects to develop a robust product in the easiest way.
• DEFINE is an intuitive framework and stands for six key elements that impact the
performance of the systems.
D Domain
E Environment
F Functionalities
I Implementation
N Necessities
E Enablers
These six key elements when understood properly will enable the designer to design a
robust, functional embedded system or a Thing.
Note: From now, Embedded System and Thing are interchangeably used for easy
understanding.
This chapter would cover the details pertaining to each element and how they contrib-
ute to the success of the product being designed.
2.2.1 Domain
The first element in the framework is Domain. Classical meaning of Domain is—A territory
over which a rule or control is exercised or A sphere of activity, concern or function. Domain defines
the working environment of a product or a Thing. Based on the type of the domain, the
product’s characteristics will change the way it needs to be designed and manufactured.
Most electronic products can be classified under the following domains:
• Avionics
• Automotive
• Consumer electronics
• Industrial systems
• Medical devices
• Military
• Networking and telecom
There are six common characteristics that differentiate these domains from each other.
Understanding and using these six characteristics will ensure that the product meets spe-
cific needs of any domain.
• First, product requirements will be captured in the right way by selecting the
domain.
• Second, these six characteristics will form the basis for rest of the other require-
ments of the product.
Later in this chapter, we will discuss the interrelationship between the requirements and
how these six characteristics impact them. Now, let us see the six characteristics of the
domains and how they can be used to develop the product requirements in the following
section.
2.2.2 Characteristics of Domains

The following six characteristics typically define the product usage, cost of the product,
working environment, manufacturing process, and so on.
Six characteristics of the domains are as follows:
1. Environment in which the product is expected to work—Operating environment.

2. Who uses the product and how it is used?—Usage of the product.
3. What is product sensitivity to cost?—Cost of the product.
4. Process used for developing the product?—Product development process.
5. How long the customer will use the product?—Product’s life.
6. What is the support needed for the product?—Product support needed during its
useful life.
2.2.2.1 Operating Environment

Operating environment describes the actual working environment of a product. It includes
parameters such as temperature, dust, humidity, and electrical power. However, these tend to
be specific to the domains in which the products are to be used. For example, in the automo-
tive domain, temperature of 120°C and power fluctuation from 2 to 16 V DC are very normal.
TABLE 2.1
Domain and Operating Temperature
Telecom and
Domain Avionics Automotive Consumer Industrial Medical Military Networking
Temperature Extremely Moderately Regular Moderately Regular Extremely Moderately

variation varying varying varying varying varying
Products need to be designed to withstand these types of variations. Defining the domain
allows the designer to understand the characteristics of the environment in which the product
has to work and factor them into the design.
Table 2.1 gives the domain and the temperature encountered.
2.2.2.2 Usage of the Product

Second characteristic is the usage of the product. This characteristic defines how the prod-
uct will be used and who will be using the product. The focus is more on the user and
how it has to be easy for a user to use the product. For example, let us consider the avion-
ics domain. Typical aircraft controls and displays are almost consistent across different
aircrafts. In addition, in most cases, other electronics subsystems also are common. This is
essentially done to minimize and standardize the training needs for operations (for both
pilots and maintenance engineers). Another example could be a basic health parameter
measuring devices (devices for measuring temperature, BP, sugar, etc.). Their user inter-
face has to be standard so that the users who have minimal exposure to different types of
equipment can be trained very quickly.
Table 2.2 gives the list of all the usage parameters for each domain.
2.2.2.3 Product Cost

Although cost of the product looks as an easy parameter, it needs to be analyzed with the
customer’s perspective. Customers view the product’s entire life and the money spent on
operation and maintenance. Let us take an example of an Inkjet Printer the cost of which
is generally based on the manufacturer’s strategy to reach the market (they sell the printer
cheap and sell the printer toner for a high cost). However, customers view these products
from the point of recurring costs such as the consumption of paper and the cost of toner
during the usage. Low initial cost and higher operating cost will result in product failure.
Most designers tend to take the short-term view of a product but not its entire life, which
results in a good product failing due to high operation and maintenance cost. At the same
time, you have products in the medical field in which some of them are disposable, and
the cost of disposable part will decide the success of the product. So, when the designer
TABLE 2.2
Domain and User Training Needs
Telecom and
Usage Training No training No training Partial Training Training Partial

needed training needed needed training
TABLE 2.3
Domain and the Cost of Products
Telecom and
Cost Very high Moderate Low Moderate High Very high Moderate
chooses the domain, this characteristic will allow them to understand the cost implication
on the product at a very early stage.
Table 2.3 gives the cost of different domain-based products for reference.
2.2.2.4 Product Development Process

For most of the readers, this will sound a little puzzling as to why development process
is being discussed in the domain characteristics. Products are typically developed on the
basis of the competence and the experience of the team. However, products that are used
in the Medical and Avionics domain need to mandatorily follow standard processes as
defined by the certifying agencies such as FAA (Federal Aviation Administration) and
FDA (Food and Drug Administration). These two agencies regulate the products that are
used in their respective markets. In addition, these products have to be approved by the
agencies before they are released to the users. This type of approval is required even in
industrial and automotive domains when safety of public life is involved. If the products
are developed for any of these domains, the designer needs to factor the compliance to the
process and the documentation needs that have to be submitted for the approval. The effort
to develop these types of products is large especially the documentation alone, which can
take up substantial time. Designers need to consider this when they develop the products.
2.2.2.5 Product Life

This is a parameter that defines the life of the product used by the customer. Once the
product’s life is over, customer disposes it and changes to a new one. Typically, life of a
product is adversely impacted by the obsolescence of the components used in the prod-
ucts. With the trend of shorter life of electronic components, product that has to be used
for a long period needs careful design. Today, the average life of a typical electronics com-
ponent is in the range of 7–8 years. So, when the designer is developing a product, one of
the key elements to be factored is the life expectancy of the overall product. Products used
in aviation, medical, automotive, and industrial domains have a product life at any point
between 12 and 18 years. Designers need to be careful in selecting the parts so that com-
ponent obsolescence does not impact the product life. Without a structured components
engineering process, selection of components can bring grief to the designers.
Table 2.4 gives the domain and the expected life with remarks.
TABLE 2.4
Domain and the Product’s Life
Telecom and
Life Very long Long Short Long Long Very long Long
Note: Very long—greater than 20 years, Long—between 10 and 15 years, Short—less than 5 years.
2.2.2.6 Product Support Requirement

Product support has direct correlation to the product life. A typical case is a product
the life of which is going to be 15 years, but the components used in the product may be
obsolete within 8–9 years of the first release. However, customers expect the manufac-
turer to support the product throughout its life. This essentially mandates the designers
to factor the support for the product to its full life, into its design. This implies that the
design has to address component obsolescence. This forces the designers to restrict the
use components of very few suppliers who have the policy of offering drop in compat-
ible upgrades for obsolete parts (Free scale was one vendor who practices this). Typically,
development teams do not have the expertise of component engineering, and this results
in the usage of wrong parts or shorter lifespan parts. One way to address this is to
identify the contract manufacturer (CM) in the initial stage of design as most CMs have
components engineering teams. Another approach is to replace the entire module that
has obsolete parts with new parts that have longer life spans. However, this approach
is expensive.
2.2.3 Environment
Second element in the DEFINE framework is the environment. This is the actual environ-
ment in which the product is actually used. In the domain section, we briefly discussed
about environment associated with domain, but that was restricted with respect to the
domain-specific parameters. Once we factor the domain, this section will help develop-
ers the ways to handle the impact of those parameters on the product in detail. There are
six important characteristics that will impact the performance of any product by way of
environmental issues. They are
1. Temperature
2. Humidity
3. Pressure
4. Dust
5. Fluid Ingress
6. Electrical (power, lightening, EMC/EMI, etc.)
2.2.3.1 Temperature
Temperature impacts any product in two ways. One is due to ambient (outside environ-
ment) temperature. The other is the temperature generated inside the product and its
impact on the working of the product. Once the designer chooses the domain, outer con-
tour of the temperature will be known and that will help the designer to understand
the impact of ambient temperature on the product. However, the temperature inside the
product which is due to the working of the components inside depends entirely on the
design. It also has a bearing on the product enclosure design by way of the material used
for packing of the product, efficiency of the circuit design, and thermal management.
For example, let us assume that we are designing an automotive product in which the
circuit will dissipate 30 W of power. Unless this heat is taken out of the system, product
will have issue by way of heating. For example, if the developer up-front chooses a plastic
enclosure, the decision can impact the performance of the product due to heating. So it is
important that the designers understand the real working environment with respect to
temperature and how it can be handled.
Table 2.1 helps in understanding the effect of temperature on products.
2.2.3.2 Humidity
Similar to temperature, humidity also plays a vital role in the design process. Humidity
becomes critical if the Thing or system uses mechatronics parts or high voltage circuitry, in
which presence of humidity may lead to arcing that may result in fire. Solution to humid-
ity is a bit complicated, and it starts from the enclosure design, extending up to product
manufacturing process. Humidity also mandates strict adherence of design rules (espe-
cially in the case of electrical interconnections) to ensure that there is no arcing due to
presence of humidity. Sometimes, special processes such as Conformal Coating have to be
used. Humidity is an unavoidable problem, and the solution is a bit complex and needs to
be understood upfront in the design solution to be implemented.
2.2.3.3 Pressure
Third characteristic in the environment is the pressure. This is very critical in the case of
IoT things as most of these will have sensors that depend on pressure for working (water,
air, and other flow measurement). Atmospheric pressure also decides the heat dissipation
from a product. As the altitude increases, atmospheric pressure decreases, which leads
to less and inefficient heat transfer from the product. This gets aggravated further if the
system uses fans for cooling, as performance of the fan deteriorates with the increase in
altitude. So, it becomes critical to know the pressure of the locations in which the product
has to work for the successful working of the product. IoT systems when deployed over
a wide area have to work in different altitudes and locations, which make the pressure a
critical parameter in development. It becomes even more critical if the sensors used are
also dependent on pressure.
2.2.3.4 Dust Ingress

Dust is another factor that impacts all electronics systems. Dust impacts systems more
with moving parts and the mechanical movement. If the Thing that is being designed has
motors and other moving devices, then dust becomes the biggest threat to the product.
Designing a dust proof product is highly challenging. It becomes critical when the system
generates lots of heat. A system-generating heat needs ventilation by way of holes in the
enclosure but detrimental to dust prevention. A good design should satisfy both the dust
prevention and heat ventilation requirements, and the need has to be identified right in the
beginning stage of the design process.
2.2.3.5 Fluid Ingress

Fourth characteristic in the environment is the fluid ingress. Fluid means all kinds of flu-
ids from water to oil. If the products have components that are mechanical and optical
(like cameras and lenses), fluid ingress becomes a critical issue. Printed circuit materials
are sensitive fluids and can be damaged by fluids. For example, if designer is developing a
WaterFlow Meter to measure the flow of water in a reservoir, the Thing/System will need to
work in an environment that has lots of water. In such environment, the water vapor will
impact the product adversely. Another situation could be when the system is installed in
the open air and exposed to rain. So, the designers have to factor the fluid ingress in the
system as a parameter, which will impact the design of the system enclosure.
2.2.3.6 Electrical Power

The final environment characteristic to be considered is the electrical power that powers
the system (Thing). Conventional systems operate either on AC mains electrical power
or DC power that is derived from AC mains and batteries (charged through mains and
provides power to the system when there is no mains power). However, in the case of IoT
Things, they may be installed in an area where there is no power, and they may need to
work on solar or other nonconventional power supply. This criterion has to be identified
upfront while designing the system, as finding a power solution later to the designed sys-
tem is not an easy option. One of the key challenges in this scenario is considering the use
of nonconventional energy. The system software has to play a vital role in managing the
power, and adding power is never an easy job once the design is frozen.
Table 2.5 gives an interdependence matrix for the six parameters against the domains.
2.2.4 Functionality
Third key element in the DEFINE framework is the functionality. This element defines the
core functions to be performed by the system (Thing). Six key characteristics that are to
be captured or identified at the beginning of a successful product design are listed below:
1. Usability and ergonomics

2. Manufacturability
3. Product interconnect and interfaces
4. Product performance
5. Product safety
6. Product sustenance
TABLE 2.5
Domain and the Six Key Parameters
Telecom and
Temperature High High Low High Low High High

Humidity High Moderate Low High High High High
Pressure High Low Low High Moderate High Moderate
Dust ingress High High Low High High High Moderate
Fluid ingress High High Low Moderate Moderate High High
Electric High High Low Low Low High Moderate
power
2.2.4.1 Usability and Ergonomics

This characteristic essentially covers two key interrelated elements of the product: the user
and the usage. Usability defines how the user can learn and use the system. Ergonomics
ensures that the product is convenient and safe to use. Usability describes the flow of
actions while using the device and how the system interacts with the user. Ergonomics
details the shape of the product, colors used, audio and display used by the system, and
location of the system in the operating environment. Both these are key elements in ensur-
ing that user gets the best comfortable experience while using the product. One of the best
examples is the Point of Sale Terminal (which is a Thing) in which the buttons are used
and the display plays a very crucial role when user interacts with the system.
2.2.4.2 Manufacturability
Biggest success factor for a product is the volume sale of the product. This depends mainly
on how easily and cost effectively the product can be produced. Most designers think about
manufacturability in the end of the design cycle. This invariably leads to a heavy price for
the product’s manufacturing, resulting in an increased product cost. Manufacturability
becomes critical with increased use of CM. Designers have to plan for the product’s manu-
facturing early on in the development. This is because most CMs have design rules for
optimal cost and maximum yield in manufacturing. These have to be implemented in the
design right from the beginning to meet the design objectives. Delaying the identifica-
tion of manufacturing partner till the end of design will lead to reduced yield and esca-
lated cost. Identification of the CM and using their design for x process (DFx—x stands for
assembly, manufacture, test, cost, support, etc.) is critical for the successful manufacturing
of the product. As very few established CMs tend to have a well-defined DFx process,
evaluation of CM becomes a key aspect to be addressed as part of manufacturability.
2.2.4.3 Product Interconnect and Interface

Interconnection to the outside world is the key to success of any IoT product. The intercon-
nect mechanism may vary from pure software protocols to power and data communication
to IoT Thing. When a product uses subsystems such as power supply, communication mod-
ules, and so on, their interface is critical for the product to succeed. Second, if the product
contains mechatronics items such as Solenoids, Print Heads, Motors, and so on, the power
needs are different, and selection of right power supply (which is again a bought out sub-
system) becomes critical. Most designers fail to address these types of issues and face prob-
lems later in the field when they install the products. Sometimes, the subsystems may be
fine, but the interconnect assemblies consisting of the cable harness and connectors may
be inadequate for the need and they have to be designed properly. In the wireless systems,
software driving the communication port may have interconnect and interoperability
issues at protocol level and need to be addressed in the design. Hence, it is important that
the product interconnect issues are addressed properly at the start of the design.
2.2.4.4 Product Performance

This characteristic may initially seem to be bit abstract and generic. However, most elec-
tronic products tend to slow down in performance as their usage starts increasing. Most
designers tend to look at the current needs of a product as the requirement instead of the
longer life span of the product’s whole life, while designing and developing the system.
They later realize that the performance of the system starts failing as the usage of the
system increases. Poor product performance may be because of low CPU speed, insuf-
ficient memory (both FLASH and RAM), bad power management, poorly written, and
tested software. Even when all these aspects are well designed, poor installation can ruin
the system’s performance. In extreme cases, poor communication infrastructure could
be the root cause for bad performance. This essentially means that the designers need to
have a holistic view of the solution and ensure that performance parameters are defined
and captured properly. For example, best performance measured for a small home router
is the number of packets it can process per second. In a typical router design, the perfor-
mance depends on the CPU memory, CPU speed, and the CPU and I/O Bus width (32 or
64 bits). By defining the performance figure, rest of the components can be derived.
2.2.4.5 Product Safety

Another key element is the safety of the product. When the system is installed and work-
ing, safety of the user and the environment is critical. Product safety could be right from
safety of the user in terms of shape and size to electrical safety (by way of no electrical
shock to the users). When the system uses chemicals for its functionality (which is com-
mon in medical and analytical instruments), it is important that the chemicals are safe
to use, and any accidental emission or spillage will not cause any problems to the user
or environment. Systems that use LASER and X-RAY need to be carefully designed so
that they are safe to the user and the environment. When a system (Thing) is developed,
designer needs to understand the safety aspects and also the type of users and design the
system with adequate safety measures. System producing low-frequency sound is also a
health hazard if exposed for long duration. Designers need to understand issues similar
to these when designing the system. Most electronic packaging design rules cover safety
aspect in great detail.
2.2.4.6 Product Sustenance

Most designers tend to think that once the product goes to the manufacturing line, their
responsibility is over and completely overlooks the lifespan of the product and the user
support. To accommodate product sustenance, designers have to factor the product’s life
and how it can be sustained even in the case of nonavailability of the original design team,
components used as also the technology obsolescence. This is very critical in Avionics and
Military equipment as their life extends up to 25 years and more. During the procurement,
one of the critical conditions they need to evaluate is how the product will be supported
by the manufacturer during the life of the product. This effectively depends on six factors
that will ensure the product’s long life without any support issues.
1. Component engineering—Ensuring parts used in the product have support for

availability of upgrade.
2. Stable manufacturing technology—Product uses stable manufacturing tech-
nology that will stay and support the product manufacturing life.
3. Well-established Test and Calibration process that is well documented so that
even if the original design team is no longer available, it can still be used for the
support of product.
4. Systematic supply chain process that ensures all components obsolescence is well
tracked. Availability of alternate strategy for component procurement that may include
strategic stocking of parts in the case of obsolete parts through Time Buy mode.
5. Proper version controlled product documentation from the original design team
and maintained using Product Life Cycle Management tools.
6. Established Vendor Development program for alternate vendors and distributed
supplier base for continuous availability of parts.
2.2.5 Implementation
Fourth element in the DEFINE Framework is the Implementation. It covers the complete
process sequence from the Design Phase to Support Phase of a product. As this element
is the most critical element that decides the success or failure of the product, it should be
given highest priority by the designers for a successful product.
There are six processes that describe Implementation and they are as follows:
1. Development process
2. Manufacturing process
3. Verification and validation process
4. Calibration process
5. Product support process
6. Documentation and control process
2.2.5.1 Development Process

Most designers tend to treat this step as a burden and ignore it in the product development
effort. However, as the development matures and product goes into testing, they realize
that not following Product Development Process leads to extra costs and also the failure of
the product in many cases. Most developers get confused between Product Development
Process with Certification Processes such as ISO9000 that demand a large amount of docu-
mentation. In reality, basic development process has a simple method for controlling the
different versions of artifacts developed and how they are released to the field and manu-
facturing. If the product developed belongs to either medical or avionics (what is known
as Regulated Industry product), it is mandatory that the development processes have to
comply to FDA and FAA guide lines and certifications such as AS9100C and ISO135485
are mandatory. Product Development Process needs some amount of effort, and designers
need to factor them in the development cycle to avoid future problems. They also need to
stick to a simple development process for product success.
2.2.5.2 Product Manufacturing Process

With the increased trend toward CM also known as Electronic Manufacturing Services,
Product Manufacturing Process has transformed and is different from conventional in-house
manufacturing. When the products belong to either Avionics or Medical domain, rigor of the
FDA/FAA mandated process becomes a necessity even in manufacturing. Most CMs follow
a very strict process in their manufacturing line. Understanding them is important for the
designers to deliver an easy-to-manufacture product. As most CMs operate on very thin
margins, any additional work that they need to do include to make the product manufacture
worthy will cost a lot. Essentially it means if the same job is done as part of the design will
cost a lot less. In addition, regulated industry product manufacturing also mandates cor-
responding certifications for manufacturing, thereby increasing the cost. In the recent past,
most automotive vendors do mandate their CM to follow ISO16949 certification for manu-
facturing, to ensure the quality of the parts produced. All these compliances add additional
cost, and designers need to understand their implication well during the design cycle to
ensure that the product is within the cost and performance norm set. In the manufacturing,
the focus is based on tracking the components used, and tracking of the products produced
so that their performance in the field can be monitored and any fault can be tracked back
to the root cause of failure up to components level. Use of CM thus becomes rigid, which is
different from the in-house manufacturing. Strict adherence to the process laid out by the
CM/Electronic Manufacturing Services vendors is a must for the success of the product.
This essentially forces the product designers to choose the CM early in the design cycle.
2.2.5.3 Product Verification and Validation Process

This is the most crucial part of product development step, and failure to do a proper verifica-
tion and validation (V&V) leads to failure of the product in the field. Most designers think
this process is the job of the manufacturing team by confusing it with testing. However, V&V
is a process integrated into the product development. Regulated industry products need
to have documented V&V process. The V&V process should include the test cases against
every product requirement with both success and failures recoded for submission as formal
document to the regulatory agencies. If this practice is followed, it will help the designers
to develop a robust product, even if it is not mandatory. This will also result in having a
product manufactured with very few problems in the field, leading to customer satisfaction
and product success. Although the process sounds complex, in practice it is a very simple
to-do-list that needs to be religiously followed, and all the test cases have to be executed with
success in all the cases. Simplest technique for a comprehensive V&V is to create a test case
for every product requirement. This step ensures that the product fairs well with very few
problems when installed in the field and thereby reducing field repair expenses.
2.2.5.4 Product Support Process

Key to success of any product is the support that product gets after it is deployed in the
field (bought by customers). Product Support process is not just the logistics associated
with the product support, but it basically depends on how the product is designed to sup-
port quick and timely correction of the problem.
Some of the aspects that help in product support are as follows:
• Ability to upgrade the product software easily without a service engineer attend-
ing to it.
• If the product is connected through wireless network, how the Over The Air (OTA)
update can be implemented.
• Resources, such as FLASH memory size and RAM, that are additionally needed
for the remote product update.
• Ability to diagnose the product remotely and isolate the problem so that the sup-
port engineers can carry the right spares avoiding delay in solving the problem.
• When the product uses consumables or needs periodic replacements of sensors

and other devices, how the support will be implemented in the system to forewarn
the user as well as manufacturer about the impending replacement?
These issues have to be addressed right in the beginning of the design, so that product is
supportable, and the resources needed for this are accounted in the product costing.
2.2.5.5 Product Documentation and Control

Although this may be covered under the product support, primarily it is kept as a separate
element due to the following reasons:
• Regulated industry products and products with long life mandate detailed
documentation.
• When the product is manufactured through CMs, documentation is the only key
for success as the product may have multiple versions in the field, and they have
to be maintained.
These are seen as luxuries by the designers, and they tend to avoid them but later when the
product hits the market they struggle to meet the field needs and find it extremely difficult
to correct the problems. It is important that minimal set of documents that need to be created
has to be planned during the design phase and the time for doing that has to be factored in.
2.2.6 Necessities
Fifth element in the DEFINE framework is the Necessities. The components that fall under
Necessities are the must be there elements in the product and are not negotiable. Every prod-
uct should have Necessities which consists of the following six elements:
1. Compliance to global standards

2. Product reliability
3. Accuracy (meeting the specification throughout the life)
4. Design for support (entire life cycle of the product)
5. Interoperability
6. Compliance to environment standards
2.2.6.1 Compliance to Global Standards

With the increased globalization, concept of designing a product for specific locations
has become obsolete. With the market being wide, designing a product meeting a cus-
tom standard makes them uncompetitive. Second, local products can easily lose out to
products complying with global standards. Unless the compliance to global standard is
designed into a product early in the design cycle, implementing them at a later period will
never be economical and may not be feasible in some cases. This compliance requirement
also calls for the designers to understand the global standards so that the product can be
designed to meet the standards. This typically needs in-depth knowledge of standards
and domain that will come only with experience and does cost a lot of money to get. This
activity has to be factored in to the design cost up front.
2.2.6.2 Product Reliability

One of the least understood aspects of an electronic product is the reliability of the
product. Reliability is not something that happens by itself. It has to be designed into
the reliability figure, and one of the main design goals should be the reliability figure. For
most systems, reliability is typically described by a figure called Mean Time Between
Failure. In simple terms, it is estimated as the duration for which the product will run
continuously without any breakdown (Please note that Software also can contribute to
failure!). Basically reliability is a predicted value that is derived by using the reliability
figures collected over a period of time from the components used in different systems.
However, designer will get pointers about the reliability of the parts by the way they are
constructed. Parts with mechanical movements will have lower MTBF value (Relays,
Solenoid, motors, etc.). In addition, working environment also contributes to the reli-
ability. A dusty environment with parts that have moving elements is a good candidate
for lower reliability. With the IoT, reliability becomes a critical element as entire system
performance is dependent on the Things—one good example in which Reliability is
the key in the case of wearable diagnostic devices. These devices actually monitor the
health of the patients and have to be extremely robust and reliable. Unless a proper
reliability calculation is done and right parts are chosen, the product may not meet its
ultimate goal.
2.2.6.3 Product Accuracy

Accuracy of a product is defined as its ability to maintain the precision and tolerance for
which it was originally designed throughout its life. Designers should be aware of the fact
that the electronics components especially passive components (resisters, capacitors, etc.)
tend to vary from their original value, as the product ages with time. This results in reduced
system performance and errors creeping into the measurements. With the increased use of
semiconductors, the accuracy can be maintained to a fair degree. However, analog circuits
invariably use resisters and capacitors and end up impacting the accuracy. One way of
mitigating is to periodically calibrate the system against a defined standard and ensure
that the system’s precision is maintained to the designed value. However, if calibration has
to be facilitated it needs to be incorporated in the design up front. Although options for
automatic calibration in systems are feasible, it will not be possible to implement them in
all the products. Another challenge is that calibration is a special process, and knowledge
of the system is needed. This mandates that there should be a separate process for calibra-
tion, and system has to be calibrated with special equipment. This also means that the
manufacturer has to plan for infrastructure to meet the need for the calibration of equip-
ment. Most designers miss this point and end up with a design that has to be discarded
after the system goes out of tolerance measurement. One another key aspect is the need for
a centralized system that keeps track of the system’s age and ensures that every system is
calibrated, a task most designers and companies tend to miss. So it is essential that a plan
for the system calibration to maintain its accuracy has to be made while the product is
designed.
2.2.6.4 Design for Support

Fourth element in the Necessities is the plan for designing the support for the product
throughout its life. As we have seen, identifying the domain does give a fair idea of what
the product life is and how the product will be used. With globalization, usage patterns
change with countries and culture. Best example is the mobile phones. In North America,
faulty mobile phone (especially low- and middle-priced ones) gets discarded, and user
goes for a new phone. While in India, the users go in for fixing the problem. This essen-
tially forces the phone manufacturer to have different strategy for supporting the prod-
uct. One of the essential elements in the product design is to have a easy disassemble
and assemble plan for a product for repair and rectification. This becomes very important
when the product is water resistant. These water-resistant products need specialized tools
as well as additional components to be repaired to function as robust as the original. This
calls for designers to understand the support needs and implement them in the design so
that it is easy for the product support team to open, fix, and close the products that need
to be repaired. With the increased use of Personal Electronics as part of the IoT, Design for
Support becomes very crucial.
2.2.6.5 Product Interoperability

This is another key aspect of any product especially with reference to IoT. Although regu-
lar standalone products may not have this issue of interoperability, all IoT products will
have to support interoperability as their functionality depends on networking with other
devices and servers. With IoT, interworking becomes very critical using the publically
available communication and networking infrastructure. Interoperability has three critical
elements to address. They are as follows:
1. Networking interface (wireline, wireless, etc.)

2. Networking protocols (TCP/IP, GPRS, LTE, and new [like Constrained Application
Protocol]) or proprietary protocol defined by the manufacturers
3. System’s operational power and associated aspects like Interference/Emission,
and so on
These aspects vary from market to market, and designers have to be aware of the place
where the product will be sold and supported. Despite higher level protocols being
the same, minor location-specific variations can still stall the work. Designers have to ensure
that these aspects are fully covered in their design so that later discovery does not create
any problems. Typical problems that were seen in these types of situations are the lack of
provision of additional resources (memories) in the system that prevents the designers to
implement the required changes in the software. So, they need to plan for the product
interoperability and also plan for additional resources that may be needed when they
have to implement the new requirements. This essentially means that they have a full-
scale structure of the product and its market at the initial stage and plan accordingly.
The other hard aspect of interoperability is the product operating voltage and how it is
derived. Main challenge is when the Thing has to interact with the mains frequency that
varies from country to country. Also, some applications such as Avionics use 400 Hz
that is much higher than the conventional mains frequency. This calls for a completely
different power supply design for the product. Similarly, unattended installations need
power derived from different sources such as solar, wind, and so on and the system need
to be designed for these solutions too. Similarly, when the product is a medical device,
the requirements are different and the focus is on safety. So designers have to plan for the
interoperability requirement up front during the design phase so that changes in the later
stage are not needed.
2.2.6.6 Environmental Compliance

This environment is different from the products working environment. Here, we are
discussing the impact of the product on the global environment or compliance to green
initiatives. With the increased awareness about the impact of electronic waste on our envi-
ronment due to use of lead or associated materials in the manufacturing processes, most
countries have now drafted compliance laws covering the following three aspects:
1. Reduction of e-waste by encouraging recycling as well as use of recyclable materials.

2. Elimination of lead and similar toxic material that are used in the manufacturing
process of electronic products.
3. Compliance to using only approved chemicals and raw materials.
The three universal standards with respect to Environmental Compliance are as follows:
1. RoHS—Reduction of Hazardous Substance (this typically applies to components

which are treated with alternate material reducing the usage of lead to bare
minimum).
2. WEEE Directive—Waste Electronics and Electrical Equipment which that man-
dates that all the product design should be recyclable, and only a miniscule per-
centage of the total material used in a product can be nonrecyclable.
3. REACH—Registration, Evaluation, Authorization, and Restriction of Chemicals is
a directive that very clearly specifies the process of what materials have to be used
and how they should be recorded and submitted along with the product, as doc-
umentation in compliance with the REACH directives. Sole purpose of REACH
is to address the production and use of chemical substances, and their potential
impacts on both human health and the environment.
Environmental compliance is an activity that needs to done in the design phase. Although
RoHS compliance can be met by selecting the process and the components to meet the
standards, WEEE and REACH need proper selection and design of the system to meet
these needs. This calls for specialized skills as well as knowledge of the compliance rules
and how to design products to comply with the rules. Primary reason for the compli-
ance is that any product that does not meet these directives will not be allowed for use.
Figure 2.4 shows how the e-waste standards stack up against the compliance processes.
Remove
RoHS
Reduce
Reuse
WEEE
Recycle
Dispose
FIGURE 2.4
Green laws and the level of recyclability.
2.2.7 Enablers
Last element of the DEFINE framework are the Enablers. These are items that enable the
designers to execute the design faster, better, and reliable. In addition, enablers also allow
the designers to evaluate up front the options available, take decisions that ensure that the
product meets the design objectives and makes it successful. Enablers are the external ele-
ments that could include Reference Design, Reference Software and Libraries, provided by
the device vendors, tool vendors, open-source platforms, and so on.
There are six aspects which constitute the enablers and they are as follows:
1. Reference hardware platforms

2. Reference software platforms
3. Operating systems
4. OEM/ODM components
5. Intellectual properties and patents
6. Tools and libraries
2.2.7.1 Reference Hardware Platform

With the increased pressure on Time to Market, component vendors are finding multiple
ways to help the designers. One of them is the Reference Hardware Platforms. The device
vendor supplies the complete working circuit design with complete design files. This helps
the designers to evaluate the capabilities of the device before committing to design. It has
to be kept in mind that reference designs are not final product designs, but they form a
foundation to the product design. However, most reference design does not undergo the
rigorous design and engineering analysis that product design demand and caution have to
be exercised. One good example will be the timing tolerance design that a product design
need to undergo so that the circuit works properly even when the external parameters like
Temperature, Humidity, and Power Supply change. Most reference designs are mainly
catering to functional demonstration of the devices only. Although the availability of refer-
ence platform helps in the design process, using the reference design without any design
analysis will bring grief to the designers. When an embedded architecture is selected,
reference platform does help but designers have to clearly evaluate their needs against the
adequacy of the reference platform before selecting the final design.
2.2.7.2 Reference Software Platform

Similar to reference hardware platform, vendors also provide Reference software. Most of
the reference solutions will be based on the open-source software (Linux) and drivers for
the peripherals that are in the processor. Unlike the hardware platform, Software vendor
supplies only basic things. However, use of Reference Software will accelerate the devel-
opment time. But the actual system software has to be developed only by the designers.
Biggest caveat is that most reference software are not fully debugged as their purpose is
only to demonstrate the functioning of the devices and not the actual software itself. Blindly
using the reference software platform will lead to a system with bugs and will lead to a
suboptimal system with low performance. Most young designers do not understand this
aspect as they believe the vendor supplied software to be robust which may not be actually
true. In the context of IoT, real-time performance is very critical and using the reference
does not guarantee that. Designers can use the reference platforms for comparison, but they
should estimate the effort for the main software development. This aspect gets complicated
when the product is a Medical or Avionics or Safety critical industrial product in which
the software need to be completely validated independently with the target hardware (the
actual product hardware). With IoT taking a big lead in the industrial automation, software
platform selection is very important and need to be carefully selected.
2.2.7.3 Operating System Used

Third important element in the Enablers is the Operating system used. Although more than
50% do not use any commercial operating systems but use home grown schedulers or mono-
lithic software. However, with the designers having the tendency to use the open-source
hardware platforms like Raspberry Pi or Arduino, they use the OS that comes with them.
Primary reason for this is the ready availability of HW and SW platforms for development.
However, a cost-effective product in large volume cannot be realized out of the open-source
hardware platforms. Designers need to keep in mind that when a product has to be made
in volumes, the best and economical way is to have dedicated hardware and the software to
go with it. Second important issue is that the designers tend to ignore implications of use of
open-source software. When an open-source license is used, it is important to understand
what things are already covered and what needs to be covered, to avoid future litigations in
terms of infringements. Most designers tend to miss this point and run into legal issues later.
The issue gets complicated when the product belongs to regulated industries. In these cases,
the OS has to be validated on the target hardware separately and submitted for compliance
to agency (FDA/FAA) regulations. If the designers choose off the shelf operating system,
then they have to ensure that the OS is available for their chosen processor. In effect, ele-
ments in the Enabler segment are interrelated and have impact on each other. In addition, the
designers need to look for five key elements in an operating system. They are:
1. Performance parameter of the OS

2. Resources needed by the OS
3. Availability of Drivers and networking protocol stack support for the OS
4. Security of the OS from hacking
5. Licensing cost of the OS and the cost of maintenance for the entire life of the product
It can be seen that OS plays a big part in the software development stage of the system and
also on the performance of the system. Due care has to be taken to select the right OS and
its support elements for the success of the product.
2.2.7.4 OEM/ODM Components

OEM/ODM (Original Equipment Manufacturer/Original Device Manufacturer) parts are
basically subsystems (products) which are bought as a subsystem and integrated into the
products. Best examples in the case of IoT products are Power Supplies and Wireless mod-
ules. These are basically finished products manufactured by OEM/ODM vendors that can
be used in designs. Although use of OEM/ODM parts reduces the design cycle, we need
to remember that cost of OEM part will be more than the same designed part of the circuit.
Other than parts like Power Supplies that need specialized skills, OEM/ODM parts make
economic sense when the volumes are less, but expensive when the volumes are high.
Another challenge that designers normally miss is the need for understanding the OEM/
ODM part completely so that it can be successfully used the products. Most of the times
designers miss the critical elements of the EOM/ODM part and product runs into trouble.
Selection of OEM/ODM has to be based the following parameters;
1. Cost
2. Performance specification
3. Flexibility
4. Compliance to standards
5. Lead time availability
2.2.7.5 Intellectual Property and Patents

With the liberal patenting regime, most companies and individuals patent their ideas as
soon as they get them. Although majority of the patents will not be useful to create a product
but they ensure others from using the idea. These patents are called Defensive Patents. Second
technique is to start the litigation against a company that uses one of their ideas only after the
company brings the product to the market. Basic idea is to cause huge economic loss forcing
the company to discontinue or pay a hefty royalty and in both the cases company stands to
incur losses in the business. This essentially means designers have to validate their ideas
against the existing patent to ensure that they do not fall prey to the IP-related issues. It is
also imperative that if the idea being used by the company is unique it has to be patented so
that the product is protected from copycats. Many designers are not aware of this and end up
losing their idea to copies and incur losses. Challenge is that the IP and Patent search and
filing is expensive but this step is essential for success and survival of the product.
2.2.7.6 Tools and Libraries

Last element in the DEFINE framework is the Tool and SW libraries that are used in the
product development. Tools enable the designer to develop and debug the product or design
quickly in short time and bring out the product to the market. Tools are either supplied
by the device and OEM vendors or by third party suppliers. Best example for third-party
suppliers of tools are Compiler Vendors who supply for different processors that are far
more compact and optimized compared with the compilers supplied by the device vendors.
This is just one example. However, what is important is the selection of the right tool. Also
with the increased use of user interfaces, creating video and graphics-related utilities have
became expensive. Most designers prefer to buy the libraries that are proven in performance
and use them in the design. Tools also decide the programming languages that are used.
Normally IoT and embedded products use C language for higher performance and reliabil-
ity. In essence, choice of the tools makes the product development process simple and easy.
2.3 Putting All Together

How this framework can be used by the designers is the next step. Although actual frame-
work has lot of good guideline and check list, templates explaining them need spate book!
However, to put it in a nut shell, Tables 2.6 through 2.8 are given. The table lists the product
development life cycle and the six elements of DEFINE framework. Easiest way to use the
table is to look at each element and by looking at the density of the starts in the activities,
development effort can be estimated.
TABLE 2.6
DEFINE Parameters and Life Cycle Activities to Power Supply and Printed Circuit Boards
Product Life Cycle Activities
Product Power Printed
Elements Material Packaging Supply Components Circuit Board
Domain Operating environment ∗ ∗

Usage ∗
Cost ∗ ∗ ∗
Process ∗ ∗
Product life ∗ ∗ ∗
Product support ∗ ∗ ∗
Environment Temperature ∗ ∗ ∗ ∗
Humidity ∗ ∗ ∗ ∗
Pressure ∗
Dust ∗ ∗ ∗
Fluid ingress ∗ ∗ ∗
Electrical power ∗ ∗ ∗ ∗
Functionality Usability and ergonomics ∗ ∗ ∗
Manufacturability ∗ ∗ ∗ ∗
Interconnect and interface ∗ ∗
Performance ∗ ∗ ∗ ∗
Safety ∗ ∗ ∗ ∗
Product sustenance ∗ ∗
Implementation Development process ∗ ∗
Manufacturing process ∗ ∗ ∗ ∗
Verification and
validation
Calibration ∗
Support ∗ ∗ ∗
Document and control ∗
Necessities Compliance ∗ ∗ ∗ ∗ ∗
Reliability ∗ ∗ ∗ ∗ ∗
Accuracy ∗ ∗ ∗
Support ∗ ∗ ∗
Interoperability ∗ ∗
Environmental ∗ ∗ ∗ ∗ ∗
compliance
Enablers Hardware platform ∗ ∗
Software platform
Operating systems
OEM/ODM parts ∗ ∗ ∗ ∗
Intellectual properties ∗
Tools ∗ ∗ ∗
TABLE 2.7
DEFINE Parameters and Life Cycle Activities to System Software and Manufacturing Process
Wires and Thermal User System Manufacturing
Elements Cables Design Interface Software Process

Usage ∗ ∗ ∗
Cost ∗
Process ∗ ∗ ∗ ∗
Product life ∗ ∗
Product support ∗ ∗ ∗
Environment Temperature ∗ ∗
Humidity ∗ ∗
Pressure ∗
Dust ∗ ∗ ∗
Fluid ingress ∗ ∗ ∗ ∗
Electrical power ∗
Manufacturability ∗ ∗ ∗
Interconnect and interface ∗ ∗
Performance ∗ ∗ ∗ ∗ ∗
Safety ∗ ∗
Product sustenance ∗
Implementation Development process ∗ ∗ ∗ ∗
Manufacturing process ∗ ∗
Verification and ∗ ∗ ∗
validation
Calibration ∗ ∗ ∗ ∗
Support ∗ ∗
Document and control ∗ ∗ ∗ ∗ ∗
Necessities Compliance ∗ ∗ ∗ ∗
Reliability ∗ ∗ ∗
Support ∗ ∗ ∗ ∗
Interoperability ∗ ∗ ∗
Environmental ∗ ∗ ∗
compliance
Enablers Hardware platform ∗ ∗
Software platform ∗ ∗
Operating systems ∗ ∗
OEM/ODM parts ∗ ∗ ∗ ∗
Intellectual properties ∗ ∗ ∗
Tools ∗ ∗ ∗ ∗ ∗
TABLE 2.8
DEFINE Parameters and Life Cycle Activities to Product Repair, Support, and Disposal
Testing Product Product Product Product
Elements Process Compliance Repair Support Disposal

Usage ∗ ∗ ∗
Cost ∗ ∗ ∗ ∗
Process ∗ ∗ ∗
Product life ∗ ∗ ∗
Product support ∗ ∗ ∗ ∗
Environment Temperature
Humidity ∗ ∗
Pressure
Dust ∗ ∗ ∗
Fluid ingress ∗ ∗
Electrical power ∗ ∗ ∗
Manufacturability ∗ ∗ ∗ ∗ ∗
Interconnect and interface ∗ ∗ ∗ ∗ ∗
Performance ∗ ∗
Safety ∗ ∗
Product sustenance ∗ ∗
Implementation Development process ∗ ∗ ∗
Verification and ∗
validation
Calibration ∗
Support ∗ ∗ ∗
Document and control ∗ ∗ ∗ ∗ ∗
Reliability ∗ ∗ ∗ ∗
Support ∗ ∗ ∗ ∗
Environmental ∗ ∗
compliance
Enablers Hardware platform ∗ ∗ ∗ ∗ ∗
Software platform ∗ ∗ ∗
Operating systems ∗
OEM/ODM parts ∗ ∗ ∗ ∗ ∗
Intellectual properties
Tools ∗ ∗ ∗ ∗ ∗
TABLE 2.9
DEFINE Parameters and Life Cycle Activities to Consumer, Industry, and Medicine
Elements Avionics Auto Consumer Industrial Medical

Usage ∗ ∗
Cost ∗ ∗ ∗ ∗
Process ∗ ∗
Product life ∗ ∗
Product support ∗ ∗ ∗ ∗
Environment Temperature
Humidity ∗ ∗
Pressure
Dust ∗ ∗ ∗
Fluid ingress ∗ ∗
Electrical power ∗ ∗ ∗
Manufacturability ∗ ∗ ∗ ∗ ∗
Interconnect and interface ∗ ∗ ∗ ∗ ∗
Performance ∗ ∗ ∗
Safety ∗ ∗ ∗
Product sustenance ∗
Implementation Development process ∗ ∗ ∗
Verification and ∗
validation
Calibration ∗
Product support process ∗ ∗ ∗ ∗ ∗
Document and control ∗ ∗ ∗ ∗
Reliability ∗ ∗ ∗ ∗
Accuracy ∗ ∗
Support ∗ ∗ ∗
Environmental ∗ ∗
compliance
Enablers Hardware platform ∗ ∗ ∗ ∗ ∗
Software platform ∗ ∗ ∗
Operating systems ∗
OEM/ODM parts ∗ ∗ ∗ ∗ ∗
Intellectual properties
Tools ∗ ∗ ∗ ∗ ∗
2.4 Summary
In this chapter, we have seen how the DEFINE framework helps in designing an electronic
product including IoT. Although this approach can also be used for generic electronic prod-
ucts, this is very useful for IoT as the Things are expected to work in any kind of environment
that needs careful design right from the beginning. To help the designers better, the author
has provided a series of four tables that help the developers to estimate the effort and cost
needed for the product development. Tables 2.6 through 2.8 indicate the impact of DEFINE
elements on the product development. They are very important and allow the user to know
the dependency. Table 2.9 helps one to understand the dependence of DEFINE elements on
the domains that discussed. These two table sets will help the developers to quickly under-
stand the complexity of the effort involved and also to ensure that the product meets the
expectation of the end users.
Bibliography
Boothroyd, G. (2005) Assembly Automation & Product Design. Boca Raton, FL: CRC Press/Taylor &
Francis Group.
Cundy, D. R. and R. S. Brown. (1997) Introduction to Avionics. London, UK: Pearson Education as
Prentice Hall.
Fries, R. C. (2001) Handbook of Medical Device Design. New York: Marcel Dekker.
Haik, Y. (2005) Engineering Design Process. Boston, MA: Brooks/Cole, a Part of Cengage Learning.
Harper, C. A. (1997) Electronic Packaging & Interconnection Handbook. New York: McGraw-Hill.
Kalpakjian, S. and S. R. Schmid. (2001) Manufacturing Engineering and Technology. London, UK:
Pearson Education.
Noergaard, T. (2005) Embedded System Architecture: A Comprehensive Guide to Engineers & Programmers.
Burlington, MA: Newnes: An Imprint of Elsevier.
O’Connor, P. D. T. with D. Newton and R. Bromley. (2006) Practical Reliability Engineering. Hoboken,
NJ: John Wiley & Sons.
Pecht, M. G. (Ed.). (2004) Parts Selection and Management. Hoboken, NJ: John Wiley & Sons.
Ulrich, K. T. and S. D. Eppinger. (2005) Product Design and Development. New York: McGraw-Hill.
3
MAC and Network Layer Issues
and Challenges for IoT
Upena Dalal, Shweta Shah, and Jigisha Patel
CONTENTS
3.1 Introduction .......................................................................................................................... 50
3.2 IoT and TCP/IP Stack .......................................................................................................... 50
3.2.1 Importance of Adaptation Layer below IPv6....................................................... 51
3.2.2 IP and Non-IP IoT Devices ..................................................................................... 52
3.3 Challenges at MAC and Network Layers Due to Physical Layer Aspects .................. 52
3.4 MAC Layer Protocols with Key Features and Elimination of Challenges .................. 53
3.4.1 IEEE 802.11ah—Sub-GHz Wi-Fi.............................................................................54
3.4.2 Bluetooth Low Energy ............................................................................................ 56
3.4.3 IEEE 802.15.4e ........................................................................................................... 58
3.4.4 ZigBee and ZigBee Low-Energy Smart above IEEE 802.15.4 ............................. 60
3.4.4.1 ZigBee Network and Management ........................................................ 60
3.4.4.2 ZigBee Network Formation Procedure .................................................. 61
3.4.5 WirelessHART .......................................................................................................... 62
3.4.6 LoRaWAN .................................................................................................................64
3.4.6.1 LoRaWAN Architecture and Operation ................................................65
3.5 Other MAC Layer Protocols ............................................................................................... 67
3.5.1 Z-Wave ....................................................................................................................... 67
3.5.2 G.9959......................................................................................................................... 68
3.5.3 Weightless ................................................................................................................. 69
3.6 Network Layer Protocols with Key Features and Elimination of Challenges ............ 70
3.6.1 Challenges in IoT Networking ............................................................................... 71
3.6.2 Network Layer Routing Protocols ......................................................................... 71
3.6.2.1 RPL .............................................................................................................. 71
3.6.2.2 CORPL ........................................................................................................ 73
3.6.3 Network Layer Encapsulation Protocols .............................................................. 73
3.6.3.1 6LoWPAN ................................................................................................... 73
3.6.3.2 6LoWPAN Use Case.................................................................................. 74
3.7 Conclusion ............................................................................................................................ 75
References....................................................................................................................................... 76
49
3.1 Introduction
Internet of Things (IoT) and its protocols are the emerging areas of research. The fast
growth of the mobile Internet, micro-gazettes, ultra-high-speed computing, and machine-
to-machine (M2M) communication systems has enabled the IoT technologies. Any com-
munication system utilizes a set of rules and standards for networking to control the data
exchange with reliable reception. Till now, the networking was among the traditional com-
puters or smart mobiles, whereas IoT technologies allow Internet protocol (IP)-based con-
nectivity of things. Things are the devices or machines and not strictly computers. Those
act smartly and make collaborative decisions that are beneficial to specific applications. IoT
allows things to sense, process, or act by allowing them to communicate in coordination
with others in order to make decisions.
IoT is nothing but convergence among a number of disciplines and forms the heteroge-
neous system. The underlying technologies, such as wireless technologies for small and
large networks, ubiquitous computing, embedded sensors, light communication, cellu-
lar technologies, and IPs, allow IoT to provide its significance; however, they impose lots
of challenges and introduce the need for either modifications in the existing standards/
specifications or development of specialized standards and communication protocols.
Especially, the critical design of protocols is required at the IP level because the medium
access control (MAC) is normally the underlying technology specific, that is, Wi-Fi has its
own specific MAC frame format and handling mechanisms at data link layer. However, few
new MAC protocols are introduced, or the olders are optimized as per IoT challenges and
requirements. Up to a certain extent, the major challenges of IoT are handled by the lower
layers—especially MAC and Network implemented over specific wireless technology. The
chapter deals with challenges as well as IoT protocols that are suitable at the data link layer
and the network layer (IEEE 802.11ah—sub-GHz Wi-Fi) (Home of RF and Wireless Vendors
and Resources).
3.2 IoT and TCP/IP Stack

IoT follows TCP/IP protocol architecture that becomes customized as per the application
requirements. The challenges of IoT in general are mobility, reliability, scalability, manage-
ment, availability, interoperability, security, and privacy. These challenges are because of a
large number of different platforms/technologies used in IoT systems.
One of the disadvantages of the TCP/IP stack is that it is fairly complex. Conventional
IP protocol as per the standard results in larger size data packets and therefore requires
a significant amount of processing power, sufficient memory, more processing time,
and expensive devices. What we require from IoT is sometimes fast and near real-time
responses. For these reasons, for many well-established technology-based networks,
such as the ZigBee/Bluetooth, the protocols must be simple, fast, and energy efficient;
hence, new protocols for adaptation of IP are the requirement of time (IEEE 802.11ah—
sub-GHz Wi-Fi) (Home of RF and Wireless Vendors and Resources). Protocol stack
for web-based application and typical equivalent protocol stack for IoT is given in
Table 3.1.
MAC and Network Layer Issues and Challenges for IoT 51
TABLE 3.1
Comparison of Web-Based and Typical IoT-Based Protocol Requirements and Solutions
For Web Application For IoT Application Hybrid Scenario or

(Hundreds/thousands (Tens of bytes to Wireless Embedded
of bytes to handle) handle) Internet
• Inefficient content • Efficient objects • Both the Web and IoT
encoding • Efficient Web features are combined
• Huge overhead and • Optimized IP
difficult parsing access
• Requires full
Internet devices
Application XML Web Objects Web Objects
HTTP CoAP CoAP
TLS DTLS DTLS
Transport TCP (and/or UDP) UDP UDP
Network IPv6 6LoWPAN IPv6 [RPL]
• Routing activity over
• Encapsulation activity 6LoWPAN
(Acting as adaptation layer)a
Host to Network Interface Existing Wireless technologies like IEEE 802.15.4, IEEE 802.11ah, Bluetooth low
including Data link and power or other MAC layer Technologies with physical layer specifications
Physical [Few wireless technologies use dedicated licensed spectrum while few use
shared/unlicensed spectrum]
a IPv6 is an Internet Layer protocol for packet-switched internetworking and provides end-to-end datagram
transmission across multiple IP networks. 6LoWPAN is an acronym of IPv6 over Low-Power Wireless Personal Area
Networks. It gains popularity day-by-day and hence selected here to give an example. It is an adaptation layer for
things to be communicated via IPv6. IPv6 must be supported with specific routing protocols like RPL (described
later in this chapter). All these protocols working together is a good example of Embedded Wireless Internet.
3.2.1 Importance of Adaptation Layer below IPv6

The adaptation to IP by wireless-embedded devices is challenging due to several reasons:
1. Battery-powered wireless devices/things make and break the connections fre-

quently, whereas IP is based on always connected devices.
2. Sometimes, it is difficult to route the traffic of things in a large coverage area via
multihop wireless mesh networks with cost efficiency.
3. Low-power wireless networks have low bandwidth (few kbps) and frame size.
IPv6 requires that every link must have a maximum transmission unit of 1280 bytes
or greater; otherwise, link-specific fragmentation and reassembly must be pro-
vided below IPv6.
4. Multicasting is not supported basically in IEEE 802.15.4 or such technologies, but
it is essential in many IPv6 operations.
5. Standard protocols may not perform well with TCP/IP directly. For example, TCP
is not able to distinguish between packet losses due to congestion and those due
to channel error and, hence, required to define how to carry IP datagrams over
existing links and also perform necessary configuration functions to form and
maintain an IPv6 subnet.
Three working groups formed by The Internet Engineering Task Force (IETF) defined an
adaptation layer for IoT. Widely considered protocols, IEEE 802.15.4 and 6LowPAN, are
Routing
TCP/IP
ZigBee
t
Protocol h Network layer
ZigBee
Cellular
Standard
BT SW
conversion
6LoW
e RFID/
IP
PAN
protocols r NFC
T
e Encapsulation
TNS and others c or adaptation
h
LTE 2G/3G Ethernet Wi-Fi WiMax Bluetooth IEEE 802.15.4 MAC layer
From simple to complex system Physical layer
FIGURE 3.1
Representing major MAC and network layer protocols working in coordination along with a level of complexity.
explained in detail in Sections 3.4.3 and 3.4.6, respectively, to understand the requirement
of adaptation layer. Figure 3.1 shows the various protocols suitable at the physical layer
(PHY)/MAC and also the suitable protocols above specific MAC. 6LowPAN has differ-
ent roles. It is a thin layer above IEEE 802.15.4 acting as an adaptation layer and creating
embedded Internet, whereas it is whole and sole above Bluetooth.
In addition to the comments for MAC and network layer, a few comments are there for
the transport layer. As TCP/IP laid the foundation for the Internet, IoT communication
network mainly employs TCP and UDP protocols. Compared with UDP protocol, the
TCP protocol is more complex, which makes it difficult to employ over the devices with
constrained resources. Most of IoT uses UDP protocol, but as UDP is not stable, special
techniques are used to incorporate it with application layer (Patil and Lahudkar, 2016).
3.2.2 IP and Non-IP IoT Devices

Heterogeneity of devices and applications due to different vendors can be adopted in
modern IoT system due to open standards of TCP/IP protocol suite. The IoT is nothing but
connecting things to the Internet and hence the things/devices need a logical connection,
that is, an IP address and must use the IP suite to be able to exchange data with devices
and servers over the Internet. Devices in a local network can use non-IP protocols to com-
municate within the local network. These non-IP devices can be communicated to Internet
via an Internet gateway. The gateway communicates with both local devices using a non-
IP protocol and other devices on the Internet using IP. The gateway here is an application
layer gateway because it needs to restructure the data coming in from the local network
with a TCP/IP stack to enable communication with an Internet service.
3.3 Challenges at MAC and Network Layers Due to Physical Layer Aspects
There are hundreds of protocols supported by IoT. Of these, wireless protocols play an
important role in IoT development. Complexity of MAC and Network layers are depen-
dent upon wireless technology adopted at the physical layer; thus, there are few challenges
at the physical layer.
• Adopting Present Techniques of Wireless Communication—M2M devices are normally

connected in a group and communicate on the basis of various wireless protocols
such as ZigBee, Wi-Fi, and Bluetooth. For Internet access, they can have direct
access as per facility or they can use Internet gateways or GPRS or EDGE as per
their availability. That is why existing wireless protocols are adopted in IoT with
required minor modifications.
• Limitation in Current Technique—Device makers are proving to get to do with
existing communication protocols that do not satisfy the operational require-
ments of M2M. Having multiple set of rules also obstructs standardiza-
tion, bandwidth requirements, power requirements, and the number of node
limitations.
• Mobility and Handovers to Manage—It is a big challenge as both these tasks are
related to location identification or tracking tasks and for that the efficient com-
puting is also required. Near real-time systems are required in order to take quick
decisions. To maintain seamless connectivity over large coverage areas, hard and
soft handovers are to be managed by central control units. This is required for
continuous communication as well as IP management also.
• Spectrum/Resource Management—Applications have to be operative within the lim-
ited bandwidth satisfying as many numbers of users or nodes intended to add.
Also, channel assignment in near real time also requires fast computing.
Various organizations started to find out the compatibility of the various wireless technol-
ogies for IoT or M2M applications with the goal of embedding M2M communications in
the upcoming 5G systems. Similarly, they started enhancing the MAC and network layer
requirements. Few widely considered protocols of our focal domains are described in the
consecutive sections (Reiter, 2012).
3.4 MAC Layer Protocols with Key Features and Elimination of Challenges
MAC layer has the specific challenges to deal with are as follows:
• Data throughput: Can be optimized by managing collisions in contention-based

protocols, by managing the access of users to the available channel, by plan-based
frameworks, the control overhead, and so on.
• Scalability: Node density is expected to increase or decrease, and it should be
handled without disturbing the operation of the network and with an energy-
efficient solution.
• Latency: Near real-time solutions are expected that can be achieved by reducing
the channel-access latency or other fast decision-making strategies.
• Coexistence: Coexistence of licensed and unlicensed systems, interworking aspects,
adaptation, and so on are to be managed.
• Cost-effective solution: Cost is always a trade-off with bandwidth, quality of service
(QoS), power consumption, and speedy response of the system. Optimized solution
is required finally.
Various wireless technologies with their key features are highlighted in this section.
Sometimes, it is difficult to write only MAC layer issues without mentioning the key physi-
cal layer aspects, and hence as per requirement, the PHY is also described in the section
(Rajandekar and Sikdar, 2015).
3.4.1 IEEE 802.11ah—Sub-GHz Wi-Fi

The two primary limitations of IEEE 802.11 technologies have opened the door for 802.11ah
technology to be applied for M2M communication:
• The absence of power saving mechanisms: The energy constraints of wireless sensor
networks are not considered in the IEEE 802.11 standard.
• Unsuitable bands: Existing Wi-Fi bands require the use of intermediate nodes due to
their short wireless range and high obstruction losses, and thus the complex scenario.
The IEEE 802.11ah standardization task group has developed a global Wireless LAN stan-
dard that allows wireless access using carrier frequencies below 1 GHz in the industrial,
scientific, and medical (ISM) band. It is an unlicensed, shared spectrum range that will
help Wi-Fi-enabled devices to get guaranteed access for short-burst data transmissions.
The channels of IEEE 802.11ah are defined on the basis of spectrum available in a given
country. The basic channel width is 1 MHz, but it is possible to bond two adjacent channels
together to form a 2 MHz channel for higher data throughput. IEEE 802.11ah is interoper-
able with 1 MHz and 2 MHz bandwidth modes globally, which results in new application
and also compatible with 4, 8, and 16 MHz bandwidths for advanced data rate applications.
Key features of this standard are as follows:
• Wide coverage, that is, extended range of its access point (AP) device (up to 1 km)
because of the enhanced propagation and diffusion of 900 MHz radio waves
through strong obstacles such as walls and ceiling. Please refer to Figure 3.2.
• Less power consumption due to below 1 GHz, that is, power efficient.
• Unlicensed band, that is, no regulatory issues.
• Almost clear coexistence issues.
• Compatible with existing wireless standard versions, for example, IEEE 802.11a/
b/g/n.
900 MHz
2.4 GHz
5 GHz
FIGURE 3.2
Extended range compared with older versions of Wi-Fi.
TABLE 3.2
IEEE 802.11ah Modulation and Coding Schemes for 2-MHz Bandwidth Channels for Different
Guard Intervals Used with OFDM-Based Physical Layer and Spatial Stream
Data Rate (Mbps)

Index Modulation Technique Code Rate Moderate GI Short GI
0 BPSK 1/2 0.65 0.72

1 QPSK 1/2 1.3 1.44
2 QPSK 3/4 1.95 2.17
3 16-QAM 1/2 2.6 2.89
4 16-QAM 3/4 3.9 4.33
5 64-QAM 2/3 5.2 5.78
6 64-QAM 3/4 5.85 6.5
7 64-QAM 5/6 6.5 7.22
8 256-QAM 3/4 7.8 8.67
9 256-QAM 5/6 – –
Note: Moderate GI = 8 µs, Short GI = 4 µs (Similar table are there for 4, 8, and 16 MHz channels).
For, IEEE 802.11ah, a new PHY and MAC design is suggested to maintain extended range
of Wi-Fi and the IoT. PHY design improves link budget compared with 2.4-GHz technol-
ogy. In order to handle the trade-off between range, throughput, and energy efficiency,
IEEE 802.11ah found the solutions as follows:
• Different sets of modulation and coding schemes with mandatory coding schemes
include binary conventional coding and optional coding scheme low-density par-
ity check.
• Number of spatial streams.
• Different duration of the Guard Interval.
Table 3.2 shows the 2 MHz mode with PHY requirements.

To support M2M communication apart from the abovementioned physical layer aspects,
the following requirements are specified by IEEE 802.11ah:
• 8191 devices at most can be associated with an AP in a tree structure (hierarchy).

• At least 100 kbps data rate.
• One-hop network topologies.
• Short and sporadic data transmissions
• Packet data size ~100 bytes
• Packet interarrival time >30 s
• Lowered energy consumption by adopting power-saving strategies. These
may include the adoption of modified IEEE 802.11v power-saving features as
well as enhanced ultra-low-power consumption strategies, such as Radio On
Demand (ROD).
MAC layer management is done by AP by sending power-save poll frames and enhanced
distributed channel access (EDCA) mechanisms (Aust et al., 2012; Tian et al., 2016) (IoT pro-
tocols you need to know about) (IEEE 802.11ah-sub GHz WiFi).
3.4.2 Bluetooth Low Energy

The Bluetooth protocol is enhanced with low power as the primary design consideration.
Its major role is in wearable IoT designs, which are system on chip (SOC) type. To address
the limitations of earlier versions and for interoperability, the Bluetooth Special Interest
Group (SIG) introduced Bluetooth Low Energy (BLE) for short-range communication, also
known as Bluetooth Smart. BLE continues to operate in 2.4-GHz ISM band with a band-
width of 1 Mbps. Few main features of BLE are as follows:
• Robust architecture supporting adaptive frequency hopping with a 32 bit CRC.

• A low data rate makes it suitable for applications in which only state information
has to be exchanged.
• Due to burst transmission of small blocks of data, ignoring duplicate packets, due
to client/server mode along with master/slave, and so on, the host processor can
operate in a low power mode frequently.
• Reduced time for connection setup to data exchange to within a few ms.
• It supports broadcaster mode—just transmits the advertising packets—without
undergoing a connection procedure.
• The link layer is optimized for quick reconnections, thereby reducing power.
A BLE device is not compatible with conventional Bluetooth device as it is a different

technology. However, Bluetooth dual-mode devices support both BLE and classic/
conventional Bluetooth. Figure 3.3 represents the major differences in the Bluetooth pro-
tocol layered architecture for conventional, dual-mode and low-energy Bluetooth devices.
Here, the major change in MAC layer is observed along with changes in PHY.
Classic or BR/EDR Dual mode or BR/EDR/LE Single mode or LE
SPP SPP GAP GATT GAP GATT Profiles
RFCOMM RFCOMM SMP ATT SMP ATT
Protocols
L2CAP L2CAP L2CAP
Link
Link manager Link layer Link layer
manager
Controller
interface
BR/EDR PHY BR/EDR + LE PHY LE PHY
FIGURE 3.3
Protocol architecture difference in various versions of Bluetooth technology. GAP—Generic Access Profile,
SPP—Serial Port Profile, GATT—Generic Attribute Profile, RFCOMM—Serial cable emulation protocol based
on ETSI TS 07.10, SMP—Security Manager Protocol, ATT—Attribute Protocol, BR/EDR—Basic Rate/Enhanced
Rate, LE—Low Energy.
Standard Bluetooth 4.2 core specifications are as follows:
• Range: 50–150 m (Smart/BLE)

• Data Rates: 1 Mbps (Smart/BLE)
• Low latency connection (3 ms)
• Low power (15 mA peak transmit, 1-µA sleep)
• Designed to send small packets of data (opposed to streaming) via Connect→
transmit→disconnect→sleep states
• Security (128 bit AES CCM)
• Modulation (GFSK @ 2.4 GHz)
• Adaptive frequency hopping
• 24 bit CRC
• Output power: ~10 mW (10 dBm)
• 40 Channels—2 MHz spacing
• Frequency hopping in connections: pseudo-random, set in the connection request
• Transmit power (−20 to +10 dBm)
• Receive sensitivity (−70 dBm)
The various states and hence the processes of link layer are shown in Figure 3.4 in form of
link layer state machine for BLE (Gomez et al., 2012; Decuir, 2010).
Bluetooth low-energy beacons can be studied separately.
Scanning
Advertising Standby Initiating
Connection received
as role of slave Connection successful
as role of master
Connection
Bidirectional and unidirectional

activities can be identified from the
state machine
FIGURE 3.4
State diagram of link layer state machine.
3.4.3 IEEE 802.15.4e

The technologies that allow the location and identification of physical objects will be
important in the context of IoT. IEEE 802.15.4 (2003) is a standard specifying physical layer
and media access control for Low-Rate Wireless Personal Area Networks (LR-WPANs).
It forms the basis for the ZigBee, WirelessHART, and many other specifications. It can also
be used with 6LoWPAN and standard IP protocols to build a wireless-embedded Internet,
for example, wireless sensor network (WSN) providing an autonomous and intelligent
connection between the physical and virtual worlds.
IEEE 802.15.4 supported two physical layers originally, one with 868 and 915 MHz fre-
quency bands and the other with 2.4 GHz band. IEEE 802.15.4 standard has well-defined
frame format, headers and deals with the techniques how nodes can communicate with
each other. The physical layer is based on Direct Sequence Spread Spectrum (DSSS)
techniques. Twenty-seven channels were supported across three bands. One channel in
868 MHz band 10 channels in 915 MHz band and 16 channels in 2.4 GHz band are sup-
ported. The three bands provide a transmission rate of 20, 40, and 250 kb/s respectively.
In the later versions, many additional bands were added.
The IEEE 802.15.4 supports two classes of devices for the MAC layer: full function
devices (FFDs) and the reduced function devices (RFDs). All Personal Area Networks
(PAN) consist of at least one FFD that acts as the PAN coordinator to maintain the
PAN. RFDs are responsible for obtaining data directly from the environment and send-
ing them to a PAN coordinator. As the traditional frame formats are not suitable for
low-power multihop networking in IoT due to their overhead. In 2008, IEEE802.15.4e
was created to extend IEEE802.15.4 and support low-power communication to meet
IoT requirements. This is especially the MAC amendment for industrial application. By
time synchronization and channel hopping, it enables high reliability and low cost. In
contrast, IEEE 802.15.4g Smart Utility Networks (SUN) Task Group is given the task of
PHY amendment in 802.15.4 to provide a standard that facilitates very large-scale process
control applications.
Specific MAC features of IEEE 802.15.4e can be summarized as follows:
• Frame structure and power saving: IEEE 802.15.4e frame structure is designed for
scheduling and telling each node what to do. A node can sleep, transmit, or receive
information.
• Sleep mode: The node turns off its radio to save power and to store all messages
that it needs to send at the next transmission opportunity.
• Transmit mode, it sends its data and waits for an acknowledgment.
• Receive mode: The node turns on its radio before the scheduled receiving time,
receives the data, sends an acknowledgment, turn off its radio, delivers the
data to the upper layers, and goes back to sleep.
• Synchronization: Necessary to maintain nodes connectivity to their neighbors and
the gateways. Two approaches can be used for synchronization: acknowledgment
based or frame based.
• Acknowledgment-based mode can be used for maintaining the connectivity/
communication as well as guaranteed reliability simultaneously.
• Frame-based mode: The nodes are not communicating, and hence, they send an
empty frame at prespecified intervals of about 30 s typically.
• Scheduling: There is no standardization for how the scheduling is done. It can

be centralized by a manager node that is responsible for scheduling, informing
others about the schedule. Other nodes have to just follow the schedule.
• Channel hopping: IEEE802.15.4e introduces channel hopping for time slotted access
to the wireless medium and hence frequency diversity. It requires changing over
the frequency channel using a predetermined random sequence. This reduces the
effect of interference and multipath fading.
• Network formation: Network formation phase includes advertisements and joining
components. A new device will listen to advertisement commands and upon receiv-
ing at least one such command, it will send a join request to the advertising device.
• In a centralized system, this join request is routed to the manager node and
processed.
• In distributed systems, join requests are processed locally.
Once a device joins the network, and it is fully functional, the formation phase is disabled
and will be activated again if it receives another join request.
The MAC layer of IEEE 802.15.4 has the following features: association and disassocia-
tion, channel access mechanism, acknowledged frame delivery, frame validation, guar-
anteed time slot management, and beacon management. The MAC frame format of IEEE
802.15.4e is shown in Figure 3.5 that is almost the same as the general MAC of IEEE 802.15.4.
There are four types of MAC frames: data frames, MAC command frames, acknowledg-
ment frames, and beacon frames. The size of frame has to be less than 127 bytes consider-
ing the size constraint of the physical layer payload. The sequence number field is used to
match the received acknowledgment frames. The frame check sequence is a 16 bit Cyclic
Redundancy Checks (CRC). In the multipurpose format size of source, PAN identifier is
varied. The multipurpose frame structure provides flexibility for the various purposes.
Format supports short and long form of frame control field and allows for all MHR fields to
be present or elided as specified by the generating service (IEEE 802.15.4, 2011; Montenegro
et al., 2007; Patil and Lahudkar, 2016).
Bytes 0/1/5/6/1/0/
0/1 0/2 0/1/2/8 0/2 0/1/2/8 Variable Variable 2
1/2 14
Destination Source
Destination Source Information
PAN PAN
identifier
address address elements
identifier Auxiliary
Frame Sequence Frame
security FCS
control number payload
header
Addressing fields Header Payload
IEs IEs
MAC
MHR MFR
payload
FIGURE 3.5
IEEE 802.15.4e MAC layer frame—general and multipurpose formats.
The IEEE standard supports two types of channel access mechanisms: (1) in the case
of non-beacon networks, the devices use unslotted Carrier Sensing Multiple Access with
Collision Avoidance (CSMA-CA); (2) in the case of beacon-enabled networks, a slotted ver-
sion of CSMA-CA is used.
3.4.4 ZigBee and ZigBee Low-Energy Smart above IEEE 802.15.4

ZigBee, like Bluetooth, is a well-established standard protocol. ZigBee PRO and ZigBee
Remote Control are among other available ZigBee profiles, are based on the IEEE 802.15.4
protocol operating in the unlicensed frequency range.
ZigBee has some significant advantages such as low-power operation, high security,
robustness, and high scalability with more number of nodes. It can take advantage of both
wireless control and sensor networks in M2M and IoT applications.
The latest version of ZigBee is version 3.0, which is essentially the unification of the vari-
ous ZigBee wireless standards into one forming the foundation for IoT. This standard pro-
vides seamless interoperability. ZigBee supports mesh routing based on AODV protocol
(Patil and Lahudkar, 2016).
ZigBee 3.0 certification is now available. Standard: ZigBee 3.0 based on IEEE802.15.4
• Frequency: 2.4 GHz

• Range: 10–100 m
• Data Rates: 250 kbps
• Maximum number of nodes in the network: 1024
• Encryption: 128 bit AES
Few plus points for ZigBee are reliability and robustness, low power, scalability, and
security.
3.4.4.1 ZigBee Network and Management

As shown in Figure 3.6, ZigBee network comprises coordinator, router, and end devices.
3.4.4.1.1 Coordinator
In the beginning, the coordinator needs to be installed for establishing ZigBee network
service. It starts a new PAN; once started, other ZigBee components, router, and end
devices can join PAN. The coordinator is responsible for selecting the channel and PAN
ID. It can help in routing the data through the mesh network and allows join request from
router and end device. It is mains powered (AC) and support child devices. It will not go
to sleep mode.
3.4.4.1.2 Router
A first router needs to join the network; then it can allow other routers and end devices to
join the PAN. It is mains powered and support child devices. It will not go to sleep mode.
3.4.4.1.3 End-Devices
They neither allow other devices to join the PAN nor can assist in routing the data through
the network. They are battery powered and do not support child devices. These may sleep;
ZigBee router
ZigBee coordinator
ZigBee end devices

Mesh networking of
IEEE 802.15.4e
FIGURE 3.6
ZigBee network components.
hence, battery consumption can be reduced to a great extent. Star and mesh topologies
are supported in ZigBee mesh routing. PAN ID (a 16 bit number) is used to communicate
between ZigBee devices. The Coordinator will have the PAN ID set to zero always, and all
other devices will receive a 16 bit ID address. Thus, there are two main steps in completing
ZigBee Network Installation—forming the network by the Coordinator and joining the
network by Routers and End devices.
3.4.4.2 ZigBee Network Formation Procedure

Formation procedure indirectly exhibits the capability of the MAC. Also, indirectly, it
exhibits scalability features as well. Coordinator searches for suitable free RF channel,
not interfering with Wireless LAN frequencies in use. This is done on all the 16 channels.
It is also referred to as energy scan. Coordinator then starts the network formation by
assigning a PAN ID to the network. Assignment is done in two ways—manual (pre-
configured) and dynamic (obtained by checking other PAN IDs of networks already
in the operation nearby to avoid PAN ID conflict). Here Coordinator also assigns a net-
work address to itself, that is, 0×0000. Now coordinator completes self-configuration and
is ready to accept request queries from routers and end devices who wish to join the
PAN. In addition to the above, the coordinator sends broadcast beacon request frame on
remaining quiet channels. This is beacon scan or PAN scan. By this, coordinator receives
PAN ID of routers and end devices present nearby. It also comes to know whether routers
or end devices allow join or not. Now routers or end devices can join by sending associa-
tion request to the coordinator. The coordinator will respond to association response.
There are two ways of ZigBee network joining
1. MAC association—implemented by the device underlying MAC layer

2. Network rejoin—implemented by the network layer
Few more protocols related to data link layer are mentioned in brief.
3.4.5 WirelessHART
HART is an acronym for highway addressable remote transducer. The HART Protocol
makes use of the Bell 202 Frequency Shift Keying (FSK) standard to superimpose digital
communication signals at a low level on top of the 4–20 mA. One can say that it is digi-
tal over analog. Due to phase continuous digital FSK, no interference with the 4–20 mA
signal. HART technology is a master/slave protocol. The HART Protocol communi-
cates at 1200 bps without interrupting the 4–20 mA signal and allows a host application
(master) to get two or more digital updates per second from a smart field slave device.
Approximately 30 million HART devices are in use worldwide, and HART technology
is the most widely used field communication protocol for intelligent process instru-
mentation. Due to capability of wireless communication, it becomes competent for IoT
application.
WirelessHART is a data link protocol that operates above IEEE 802.15.4 PHY and adopts
time-division multiple access (TDMA) in its MAC. It is secure and reliable and uses
advanced encryption techniques for messages calculating the integrity to offer reliability.
The architecture, as shown in Figure 3.7, consists of a network manager, a security man-
ager, a gateway to connect the wireless network to the wired networks, wireless devices as
field devices, APs, routers and adapters. The standard offers end-to-end, per-hop, or peer-
to-peer security mechanisms.
The specifications and characteristics of WirelessHART are as follows:
• Radios comply with IEEE 802.15.4-2006 and use 2.4 GHz license free frequency
band. They employ DSSS technology and channel hopping for security and reli-
ability, as well as TDMA-synchronized, latency-controlled communications between
devices on the network.
• It can coexist with other wireless networks. Provision of blacklisting avoids fre-
quently used channels; thus, network optimizes bandwidth and radio time. Time
synchronization is incorporated for on-time messaging.
• Self-healing type of property of these networks can be exhibited. This is because
the network
• Adjusts communication paths for optimal performance.
• Monitors paths for degradation and repairs itself.
• Finds alternate paths around obstructions.
Core network
(wired)
Network
manager
WirelessHART
gateway
Access point
FD
WirelessHART
adaptor
WirelessHART
device
FIGURE 3.7
WirelessHART architecture. FD—Field Device, AP—Access point.
Mesh network topology and multiple APs are supported.

Each WirelessHART network includes three main elements as shown in Figure 3.7:
• HART-enabled wireless field devices (with a WirelessHART adapter attached to it) are
connected to process or plant equipment.
• Gateways enable communication between wireless field devices (through AP) and
host applications connected to a core network or other plant communications
network.
• Network manager is responsible for network configuration, scheduling communica-
tions between devices, managing routes, and monitoring network functionality.
The network manager can be integrated into the gateway, host application, or pro-
cess automation controller.
TCP/IP WirelessHART
Application Command oriented. Predefined data types and application procedures
TCP Auto segmented transfer of large data sets. Reliable stream transport.
Redundant paths
IP
Mesh network
Byte oriented, Token, TDMA, Channel hopping
Master/slave protocol
Network access IEEE 802.15.4
Analog and Digital signaling
4–20 mA IEEE 802.15.4 (2, 4 GHz)
Wired (FSK/PSK/RS485) Wireless 2.4 GHz
FIGURE 3.8
WirelessHART-layered architecture in comparison with TCP/IP stack.
This technology has been proven in field trials and real plant installations across various
process control industries.
• Each device in the mesh network can serve as a router for messages from other
devices. This extends the range of the network and provides redundant commu-
nication routes.
• The network manager determines the redundant routes based on latency, effi-
ciency, and reliability. Consequently, data are automatically rerouted to follow a
known good, redundant path with no loss of data in the case of nonsuitable path.
• The mesh design also makes adding or removing devices easy.
• For flexibility to meet different application requirements, the WirelessHART
standard supports multiple messaging modes: (1) one-way publishing of pro-
cess and control values, (2) spontaneous notification by exception, (3) ad-hoc
request/response, and (4) autosegmented block transfers of large data. Figure 3.8
represents the protocol stack (Kim et al., 2008).
3.4.6 LoRaWAN
Long-range wide area network (LoRaWAN) is a new wireless technology designed for long
range, low power wide area networks (LPWANs) with low cost, mobility, security, and
bidirectional communication features for future needs of IoT applications. It is based on
Semtech’s long range (LORA) radio modulation technique and a specific MAC protocol. It is
optimized for low power consumption and designed for scalability feature with millions of
devices. It is location-free and energy-harvesting technology and supports redundant opera-
tion. In the OSI Reference Model, LoRa represents PHY, and LoRaWAN roughly maps pri-
marily a data link (specifically MAC) layer, with only some components of a network layer.
LoRa transmissions work by chirping, breaking the chips in different places in terms
of time and frequency to encode a symbol. In LoRaWAN, Spreading Factor (SF) refers
to chirp rate. Different SFs can be decoded in the same frequency channel at the same
time. The fact that LoRa transmissions jump from one place to another at a particular time
means bit strings are changing.
3.4.6.1 LoRaWAN Architecture and Operation

LoRaWAN networks use a star of stars topology (as shown in Figure 3.9) with a hierarchy,
in which gateways are a transparent bridge to relay messages between end-devices and a
central network server in the backend. Gateways are connected to the network server via
standard IP connections, whereas end devices use single-hop wireless communication to
one or many gateways. Gateways forward messages between nodes and applications in
the backend. Nodes use a single hop wireless connection to one or more gateways. Many
end devices in the network could be trying to communicate with the gateway at the same
time, but the gateway would not be able to hear or understand them all at a time. To sup-
port a wide range of applications, LoRaWAN has several different classes of end point
devices, the capabilities of each are shown in Figure 3.10.
1. Bidirectional end devices (Class A): They allow for bidirectional communications
whereby the uplink transmission of each end device is accompanied by two short
downlink receive windows. The end device can transmit on the basis of its own
needs with a small variation based on a ALOHA. Downlink communications
from the server at any other time will have to wait until the next scheduled uplink.
2. Bidirectional end devices with scheduled receive slots (Class B): In addition to the Class
A random receive windows, Class B devices open extra receive windows at sched-
uled times. Network may send downlink packet to nodes at any receive slot. To
open its receive window at the scheduled time end device receives a time syn-
chronized Beacon from the gateway. This allows the server to know when the
end-device is listening.
3. Bidirectional end devices with maximal receive slots (Class C): End devices of Class C
have nearly continuously open received windows, only closed when transmitting.
The IOT
Network server
G
Your
application
G
Gateways
End device
G
FIGURE 3.9
LoRaWAN topology.
Class A: Receiver Initiated Transmission (RIT) strategy
Packet End devices sleeps Rx slot 1 Rx slot 2
End device Up Down Down
Base station
Received by all
Class B: Coordinated Sampled Listening (CSL)
Rx Rx Rx Rx Rx Rx Rx Rx
End device
Beacon
Beacon
128 seconds
Base station
Class C: Continuous Listening (CL)
Transmit RX1
RX2 RX2
Transmit on air RECEIVE_DELAY1 Extends to next uplink

time
RECEIVE_DELAY2
FIGURE 3.10
LoRaWAN devices with operative capabilities.
Similar to all wireless systems, the gateway cannot find out anything when it is carry-
ing. So, when the gateway sends a message, it disconnects the receivers and is momen-
tarily off the air. For this reason, transmissions and downlinks do not happen often in
LoRaWAN. On account of this, LoRaWAN is best suited for uplink focused networks.
Nearly all LPWAN systems, including LoRaWAN, have multiple receive channels. Most of
the LoRaWAN systems can receive eight messages simultaneously, across any number of
frequency channels.
The selection of the data rate is a trade-off between communication range and message
duration. Due to the spread spectrum technology, communications at different data rates
create a set of virtual channels (without interference between channels) increasing the
capacity of the gateway. LoRaWAN data rates range from 0.3 to 50 kbps with an adaptive
data rate (ADR) technique. It supports a coverage range of 2–5 km (urban environment),
15 km (suburban environment). Devices and applications have a 64 bit unique identi-
fiers, DevEUI and AppEUI, respectively. When a device joins the network, it receives
a dynamic (nonunique) 32 bit device address DevADDR (Tian et al., 2016) (LoRaWAN)
(What is LoRaWAN?).
3.5 Other MAC Layer Protocols

3.5.1 Z-Wave
Z-wave is intended to provide a simple and reliable method to wirelessly control lighting,
security systems, home cinema, automated window treatments, swimming pool, garage,
home access controls, and suitable for IoT applications using a Z-Wave gateway or central
control device. Z-Wave is a low power MAC protocol. It covers about 30–100 m range. The
protocol-layered architecture is shown in Figure 3.11.
Z-Wave is designed to provide reliable, low latency transmission of small data packets
at data rates up to 100 kbps. Message ability to hop is up to four times between nodes. It
uses 15 unlicensed ISM band. It operates at 868.42 in Europe, at 908.42 MHz in the United
States and Canada and uses other frequencies in other countries depending on their regu-
lations. Modulation is by Manchester channel encoding. The MAC and PHY are described
by ITU-T G.9959 and fully backward compatible. It uses CSMA/CA for multiple access
and ACK messages for reliable transmission. It follows a master/slave architecture in
which the master sends commands to slave and handles scheduling of the whole network.
Z-wave uses source-routed mesh network architecture. The devices can communicate
by using intermediate nodes to route around overcoming various obstacles or radio dead
spots that might occur in the multipath environment. A message from one to another node
can be delivered successfully even if the two nodes are not within range, providing that a
third node can communicate with both these nodes.
The simplest Z-wave network is a primary controller and one controllable device.
Additional devices can be added at any time, may be as secondary controllers such as tra-
ditional hand-held controllers, key-fob controllers, wall-switch controllers, and PC applica-
tions specifically designed for management and control of a Z-wave network. A Z-wave
network can include up to 232 devices with the option of bridging networks if more devices
are required.
Each Z-wave network is identified by a NetID (also called HomeID), and each device is
identified by a NodeID/an address. The NetID is the common identification of all nodes in
OSI TCP/IP protocol stack Z-wave
Application HTTP, RTP, FTP, etc. Device and command classes
Transport TCP UDP ICMP Routing layer
Network IP Transfer layer
Data link Ethernet MAC Proprietary MAC
Physical Ethernet PHY Proprietary PHY
FIGURE 3.11
Z-wave protocol stack.
a logical Z-wave network. The NetID is 4 bytes (32 bits) long and is assigned to each device
by the primary controller, when the device is included into the network. Nodes with dif-
ferent NetIDs cannot communicate with each other. The NodeID is 1 byte (8 bits) long and
unique in its network.
Z-wave units can operate in power save mode reducing power consumption. Z-wave
units to be able to route unsolicited messages cannot be in sleep mode. So, battery-operated
devices are not designed as repeater units. As Z-wave assumes that all devices in the
network remain in their original detected position, mobile devices, such as remote
controls, are excluded (Bojkovic et al., 2016; The IoT powered by Z-wave).
3.5.2 G.9959
G.9959 is a short-range narrowband communication technology with PHY and MAC layer
specifications and protocol from ITU, designed for low bandwidth and low cost, half-duplex
reliable wireless communication. The reference model is shown in Figure 3.12. It is designed
for critical real-time applications to deal with reliability and low power consumption.
The PHY and MAC layers of G.9959 are as follows, and also MAC indicates how to use
these for IPv6 transport. G.9959 PHY recommendations specify sub-1-GHz RF transceiv-
ers. The PHY layer provides bit rate adaptation, that is, data flow control between the MAC
and PHY. It encapsulates transmit MAC PDUs into the PHY frame and adds PHY-related
control and management overhead. This is all via PMI (Figure 3.12). The PHY layer pro-
vides encoding of the PHY frame content including header and payload and modulates
these encoded frames for transmission over the channel.
PHY is responsible for (1) assignment of an RF profile to a physical channel, (2) activation
and deactivation of the radio transceiver, (3) data frames transmission and reception, (4) clear
channel assessment, (5) frequency selection, and (6) link quality for received frames.
The MAC layer of G.9959 includes (1) unique network identifiers that allow 232 nodes
to join one network, (2) unique channels access mechanism, (3) collision avoidance mech-
anism and backoff time in the case of collision, (4) acknowledgements and automatic
OSI reference model
IPv6
Network layer
A-interface
Data link layer MAC
PMI G.9959
Physical layer PHY
MDI
Transmission medium
FIGURE 3.12
Protocol reference model of an ITU-T G.9959 transceiver. PMI—Physical medium-independent interface,
MDI—Medium-dependent interface.
TABLE 3.3
M Flag Support and Interpretation for RA
M Flag Support M Flag Value Required Node Behavior

No ignore Node must use linklayer-derived addressing
Yes 0 Node must use linklayer-derived addressing
1 Node must use DHCPv6 based addressing
Node must comply fully with RFC6775
retransmissions of MAC PDUs to guarantee reliability, (5) dedicated wakeup pattern that
allows nodes to sleep when they are out of communication and hence saves their power,
and (6) frame validation. G.9959 defines how a unique 32 bit HomeID network identifier is
assigned by a network controller and how an 8 bit NodeID host device identifier is allocated
to each node. This forms the basis for IPv6 support. Specific M-Flag support may or may
not be there for node addressing. By skipping support for the M flag, a cost-optimized node
implementation may save memory. The M flag must be interpreted as defined in Table 3.3.
The G.9959 Segmentation and Reassembly (SAR) layer does the same for larger data-
grams as usual and routing layer also provides a similar feature for routed communi-
cation. An IPv6 routing stack communicating over G.9959 may utilize link-layer status
indications such as data delivery confirmation and ACK timeout from the MAC layer
(Brandt and Buron, 2015).
3.5.3 Weightless
Weightless is another wireless low-power WAN (LPWAN) technology for IoT applica-
tions designed by a nonprofit global organization Weightless Special Interest Group
(SIG). Weightless is both the name of a group and the technology. It is to note that the
technology is an open global standard rather than a proprietary technology in order to
guarantee low cost and low risk, and to maximize user choice and ongoing innovation.
It is capable of long range, long battery life, QoS, ultra-low cost, and high capacity IoT
products to be deployed rapidly and economically.
Three different weightless open standards are available: Weightless-N, Weightless-P,
and Weightless-W. Weightless-N was first developed to support low cost and low-power
M2M communication. P and N versions use ultra-narrow bands in the sub-1-GHz ISM fre-
quency band, whereas Weightless-W provides the same features but uses white spaces of
TV frequencies. The modulation scheme is differential binary phase shift keying (DBPSK)
to transmit within narrow frequency bands using a frequency hopping algorithm for
interference mitigation. It provides encryption and implicit authentication using a shared
secret key to encode transmitted information via a 128-bit AES algorithm. Each Standard
is designed to be deployed in different use cases depending upon a number of main priori-
ties and summarized in Table 3.4.
Similar to other wireless systems, weightless system consists of two main parts, namely
network subsystems and user terminals. It consists of base station, Internet backbone
systems on the network side and terminal having weightless radio module along with
a protocol stack. Weightless system frame structure is similar to WiMAX system frame
structure having TDD topology. The weightless system operates with single carrier modu-
lation, broadband downlink, narrowband uplink, direct sequence spreading, long frame
size, and frequency hopping at the frame rate.
TABLE 3.4
Comparison of Weightless Protocol Versions
Weightless-N Weightless-P Weightless-W (Where

(for Cost Effective (for High Performance TV White Space is
Solution and One-Way and Two-Way Available, Two-Way
Communication) Communication) Communication)
Feature set Simple Full Extensive
Range >5 km >2 km >5 km
Battery life 9–10 years 3–8 years 3–5 years
Terminal cost Very low Low Moderate
Network cost Very low Medium Medium
Weightless frame consists of a preamble followed by downlink part and uplink part.
Downlink refers to transmissions from base station to terminals, whereas uplink refers to
transmissions from terminals to the base station. Usually weightless frame is of duration
of about 2 seconds. The frame is divided into slots that are the smallest possible length of
resource allocation to the terminals. Mobility is supported by the network by automatic
routing terminal messages to the correct the destination.
Other PHY/MAC technologies suitable for IoT applications are DECT/ULE, HomePLUG,
IEEE 802.3 Ethernet, IEEE 802.16 WiMAX, ultra-wide band (UWB), infrared data associa-
tion (IrDA), near-field communication (NFC), and so on (What is weightless?).
3.6 Network Layer Protocols with Key Features

and Elimination of Challenges
The main functions of the network layer are to enable the correct use of the MAC sub-
layer and provide a suitable interface for the upper layer, mainly transport and applica-
tion layer. Routing and encapsulation of IP are the two basic aspects of the network layer.
It also needs to control fragmentation as per the capabilities of the network and exhibits
Internetworking.
On the one hand, it is necessary to manage network layer data units coming from the
payload of the application layer and their routing according to the existing topology. On
the other hand, layer control is required to handle configuration of new devices and estab-
lish new networks. It is necessary to determine a neighboring device belongs to the par-
ticular network and also discovering new neighbors and routers. The control also detects
the presence of a receiver to allow direct communication and MAC synchronization.
Generally, the routing protocol used by the network layer of IoT is based on Ad Hoc
On-Demand Distance Vector (AODV) protocol, and algorithm indirectly highlights
functions associated at network layer. To find the path to destination, source device
broadcasts a route-request to all of its neighbors in the dynamic topology. The neighbors
in turn broadcast the route-request to their neighbors until the destination is reached.
Once the destination is reached, destination device sends route-reply via unicast trans-
mission following the lowest cost path back to the source. Once the source receives the
route-reply, it will update its routing table for the destination address of the next hop
in the path and total path cost. There are few routing protocols described soon (Patil
and Lahudkar, 2016; Salman, 2008).
3.6.1 Challenges in IoT Networking

IoT system may be deployed in harsh environments such as forests, underwater, battle
fields, and consequently has to rely on wireless mesh technologies for communication,
hence the following challenges:
• IoT nodes must not be always ON to conserve the battery power. This brings more
challenges to the TCP/IP protocol architecture. Mesh networks typically adopt
the multilink subnet model that is not supported by the original IP addressing
scheme.
• Broadcast and multicast are expensive on a battery-powered network because
single multicast will involve a series of multihop forwarding, resulting in waking
up of many nodes.
• A scalable routing mechanism is necessary for IP communications to happen
over the mesh networks. Also, communication between non-IP and IP networks
is required.
• The reliable and in-order byte stream delivery as in TCP is often not suited for
applications that require customized control and prioritization of their data
(Salman, 2008).
3.6.2 Network Layer Routing Protocols

3.6.2.1 RPL
Routing Protocol for Low-Power and Lossy Networks (RPL) provides a mechanism to
control different traffic inside the Low-Power and Lossy Networks (LLNs) supporting
specifications on routing metrics, objective functions, and security. In LLN, both the
routers and their interconnections are constrained. LLN routers typically operate with
constraints on processing power, memory, and battery power, whereas interconnects are
constrained by high loss rates, low data rates, and instability. LLNs comprise tens to thou-
sands of routers.
Supported communications are as follows:
• Point-to-point (between devices inside the LLN)

• Point-to-multipoint (from a central control point to a subset of devices inside the
LLN)
• Multipoint-to-point (from devices inside the LLN toward a central control point)
RPL is developed by routing over low power and lossy networks (ROLL) working group
of IETF, who did a detailed analysis of the routing requirements focusing on varieties of
IoT applications. The objective of the group was to design a routing protocol for LLNs,
supporting a variety of link layers without specific assessment on the link layer, sharing
the common link characteristics of being low bandwidth, lossy and low power. Link layer
could either be wireless such as IEEE 802.15.4, IEEE 802.15.4g, IEEE 802.11ah, or Power Line
Communication (PLC) using IEEE 802.15.4 such as IEEE P1901.2.
Routing issues are very challenging especially when the low-power and lossy radiolinks,
the battery-powered nodes, the multihop mesh, and the dynamic topologies due to mobility.
The solutions should take into account the specific application requirements, along with
IPv6 behavior and its adaption mechanisms. LLNs must be able to build up network routes
quickly, to distribute routing knowledge among nodes, and to adapt the topology efficiently.
In RPL, typically, the nodes of the network are connected through multihop paths to a
small set of root devices, which are usually responsible for data collection and coordina-
tion duties. RPL is a distance vector routing protocol for LLNs that makes use of IPv6, and
network devices running the protocol are connected like a spanning tree. For devices, an
optimized Destination Oriented Directed Acyclic Graph (DODAG) is created taking into
account link costs, node attributes/status information, and an Objective Function (OF) that
defines the way of routing metric computation. The graph is routed at a single destination.
OF specifies how routing constraints and other functions are taken into account during
topology construction (Al-Fuqaha et al., 2015).
In some cases, a network has to be optimized for different application scenarios, situations,
and deployments, and in order a DODAG may be constructed as per the expected number of
transmissions or the current status of battery power of a node. RPL allows building a logical
routing topology over an existing physical infrastructure. RPL Instance defines an OF for a
set of one or more DODAGs. The protocol tries to avoid routing loops and compute a node’s
position relative to other nodes with respect to the DODAG root. This position is called a
Rank that increases if nodes move away from the root and vice versa. The Rank may be in
terms of a simple hop-count, or it may be calculated with respect to other constraints.
The RPL protocol defines four types of control messages with respect to DODAG for
topology maintenance and data exchange.
1. DODAG information object (DIO): It is the main source of routing control informa-
tion. It may store information such as the current Rank of a node, the current RPL
Instance, the IPv6 address of the root, and so on.
2. Destination advertisement object (DAO): It enables the support of down traffic and is
used to propagate destination information upward along the DODAG.
3. DODAG information solicitation (DIS): It makes possible for a node toward requir-
ing DIO messages from a reachable neighbor.
4. DAO-ACK: It is sent by a DAO recipient in response to a DAO message.
The RPL specification defines all four types of control messages as ICMPv6 information
messages with a requested type of 155.
Another important fact about the protocol’s design is the maintenance of the topology.
As most of the devices in LLNs are typically battery powered, it is crucial to restrict the
number of sent control messages over the network. Many routing protocol broadcast con-
trol packets at a fixed time interval that causes waste of energy when the network is stable.
RPL adapts the sending rate of DIO messages by extending the Trickle algorithm. Number
of control messages will be less over stable links whereas RPL is made to send control
information more often in dynamic topologies.
3.6.2.1.1 RPL—Mode of Operation

RPL specifies how to build a DODAG sometimes referred to as a graph in the rest of this
section using an OF. The objective function operates on a combination of metrics and
constraints to compute the best path. The objective function does not necessarily specify
the metric/constraints but dictates some rules to form the DODAG (e.g., the number of
parents, backup parents, use of load-balancing). There could be several OFs in operation
on the same node in the mesh network because deployments vary greatly with different
objectives and a single mesh network may need to carry traffic over different path quality.
For example, several DODAGs may be used with the objectives
1. Find paths with best Expected Transmissions (ETX) values (metric) and avoid
non-encrypted links
2. Find the best path in terms of latency (metric) while avoiding battery-operated
nodes
The graph built by RPL is a logical routing topology. The network administrator may
decide to have multiple routing topologies (and hence graphs) active at the same time to
carry different traffic with different requirements. A node in the network can join one or
more graphs (in this case we call them RPL Instances) and mark the traffic according to the
graph characteristic to support QoS aware and constraint-based routing. The marked traf-
fic flows up and down along the edges of the specific graph.
Each RPL message has a level of security (32 bit and 64 bit MAC and ENC-MAC modes
are supported). The algorithms (CCM and AES-128) in use are indicated in the protocol
messages (Patil and Lahudkar, 2016).
3.6.2.2 CORPL
An extension of RPL is cognitive and opportunistic RPL (CORPL) that is designed for
cognitive networks and uses DODAG topology generation technique but with two
modifications to RPL. In cognitive networks the primary and secondary users are sup-
ported. Secondary users are called opportunistic users. An opportunistic forwarding
approach is used in CORPL to meet the utility requirements of secondary network to
forward the packet by choosing set of multiple forwarders and choosing best next hop
out of them in coordination among the nodes. DODAG is built in the same way as RPL.
Each node maintains a forwarding set instead of its parent only and updates its neighbor
with its changes using DIO messages. Based on latest information, each node dynami-
cally updates its neighbor priorities in order to construct the forwarder set (Patil and
Lahudkar, 2016).
3.6.3 Network Layer Encapsulation Protocols

3.6.3.1 6LoWPAN
6LoWPAN is the first and most commonly used encapsulation protocol. 6LoWPAN is an
open standard defined in RFC 6282 by IETF and represents an IPv6 adaptation layer
allowing short packet networks to exchange IP packets. Core protocols for 6LoWPAN
architecture have already been specified. It efficiently encapsulates IPv6 long headers in
small packets. The specification/protocol supports different address lengths, low band-
width, different topologies, low cost, scalable networks, mobility, unreliability, and long
sleep time. The standard provides header compression to reduce transmission overhead,
fragmentation to meet the required frame length (Patil and Lahudkar, 2016). 6LoWPAN
uses four types of headers in various frames:
• No 6LoWPAN header (00): Any frame that does not follow 6LoWPAN specifica-
tions is discarded.
• Dispatch header (01): It is used for multicasting and IPv6 header compressions.
• Mesh header (10): These headers are used for broadcasting.
• Fragmentation header (11): These headers are used to break long IPv6 header to fit
into fragments of maximum say 128 byte length.
3.6.3.2 6LoWPAN Use Case

To send IPv6 packets over, for example, 802.15.4 efficiently, one needs to concentrate on
the issues arising from the underlying MAC and PHY protocols as discussed earlier. In
the case of Ethernet links, a packet with the size of the IPv6 maximum transmission unit
(1280 bytes) can be sent easily as single frame over the link. In the case of 802.15.4, because
it supports a maximum packet size of 127 bytes, 6LoWPAN acts as an adaptation layer
between the IPv6 networking layer and the 802.15.4 link layer. It solves the issue by frag-
menting the IPv6 packet at the sender and reassembling it at the receiver. 6LoWPAN also
provides a compression mechanism to reduce the IPv6 headers sizes to reduce transmis-
sion overhead, which, in turn, reduces the energy consumption as the fewer bits are sent
over the wireless channel. It provides a very efficient low overhead mechanism for for-
warding multihop packets in a mesh network.
A typical example of IoT network based on 6LoWPAN is shown in Figure 3.13. The
6LoWPAN network is connected to the IPv6 network using an edge router. The edge router
handles three actions:
1. The data exchange between 6LoWPAN devices and the Internet/IPv6

2. Local data exchange between devices within the 6LoWPAN
3. The generation and maintenance of the 6LoWPAN network
One 6LoWPAN network may be connected to other IP networks through one or more edge
routers that forward IP datagrams between different media (Al-Fuqaha et al., 2015).
Two other device types are included inside the typical 6LoWPAN network shown in
Figure 3.13: routers and hosts. Routers route the data destined to another node in the
6LoWPAN network. Routers form the distributed network scenario. Hosts are the end
devices (or nodes in case of sensor networks) and are not able to route data to other devices
in the network. Host can also be a sleepy device, waking up periodically to check its parent
(a router) for data. This enables very low power consumption. 6LoWPAN network proto-
col architecture is represented in Figure 3.14 in comparison with TCP/IP protocol stack.
6LoWPAN connects more things to the cloud. A powerful feature of 6LoWPAN is that
although originally designed to support IEEE 802.15.4 in the 2.4 GHz band, it is now being
adopted over a variety of other networking media including sub-1 GHz low-power RF,
Bluetooth Smart, Power Line Control (PLC), and low power Wi-Fi (Patil and Lahudkar,
2016; Lora Alliance: Wide Area Networks for IoT) (Thread usage of 6LoWPAN).
Internet
Server Server
6LoWPAN
edge router
R R Router
WSN based
on 802.15.4
R R
H R
H R H R Router
PC
H
H Host
6LoWPAN network
FIGURE 3.13
An example of an IPv6 network with a 6LoWPAN mesh network.
TCP/IP Protocol stack 6LoWPAN protocol stack
HTTP RTP Application Application

TCP UDP ICMP Transport UDP ICMP
IP Network IPv6 with LoWPAN
Ethernet MAC Data link IEEE 802.15.4 MAC
Ethernet PHY Physical IEEE 802.15.4 PHY
FIGURE 3.14
Protocol stack comparison of TCP/IP and 6LowPAN.
3.7 Conclusion
After studying the various challenges of MAC and Network layer, one has the scope
to optimize or enhance the various protocols described in this chapter for IoT applica-
tions. The new protocols for IoT are developed on the basis of the situation and require-
ments. Major developments required are up to IP level; thereafter, everything is set in the
Internet. This shows the importance of developing PHY-MAC and relevant network layer
along with adaptation layer. These protocols may be non-IP. Internet of local low power
networks is a future of this whole world that will bring the revolution in human lives. Of
course upper layers are equally important along with the security aspects over these basic
aspects. They add quality of services in the application.
References
Al-Fuqaha, A., Guizani, M., Mohammadi, M., Aledhari, M., and Ayyash, M. (2015), Internet of things:
A survey on enabling technologies, protocols and applications. IEEE Communications Surveys
Tutorials, 17: 2347–2376. doi:10.1109/comst.20152444095
Aust, S., Prasad, R. V., and Niemegeers, I. G. (2012), IEEE 802.11 ah: Advantages in standards and fur-
ther challenges for sub 1 GHz Wi-Fi. In 2012 IEEE International Conference on Communications
(ICC), pp. 6885–6889.
Bojkovic, Z., Bakmaz, B., and Miodrag, B. (2016), Security Aspects in Emerging Wireless Networks, 1: 158–165.
Brandt, A. and Buron, J. (2015), Transmission of IPv6 Packets over ITU-T G.9959 Networks. IETF RFC
7428, Retrieved from http://www.ietf.org/rfc/rfc7428.txt
Decuir, J. (2010), Bluetooth 4.0: Low Energy, Presentation slides, Retrieved from: http://chapters.
comsoc.org/vancouver/BTLER3.pdf
Gomez, C., Oller, J., and Paradells, J. (2012), Overview and evaluation of Bluetooth low energy: An
emerging low-power wireless technology. Sensors, 12(9), 11734–11753.
Home of RF and Wireless Vendors and Resources, Retrieved from http://www.rfwireless-world.com
IEEE 802.11ah-subGHz Wi-Fi, Retrieved from http://www.radio-electronics.com/info/wireless/
wi-fi/ieee-802-11ah-sub-ghz-wifi.php
IEEE 802.15.4, (September 5, 2011), IEEE Standard for Local and metropolitan area networks-Part
15.4: Low-Rate Wireless Personal Area Networks (LR-WPANs), http://standards.ieee.org/
getieee802/download/802.15.4–2011.pdf
IOT protocols you need-to know about Retrieved from http://www.rs-online.com/designspark/
electronics/knowledge-item/eleven-internet-of-things
Kim, A., Hekland, F., Doyle, P., and Petersen, S. (2008), When HART goes wireless: Understanding
and implementing the Wireless HART standard in IEEE International Conference on Emerging
Technologies and Factory Automation (ETFA 2008), pp. 899–907.
Lora Alliance: Wide Area Networks for IOT, Retrieved from: https://www.lora-alliance.org/
What-Is-LoRa/Technology
LoRaWAN, Retrieved from: https://www.thethingsnetwork.org/wiki/LoRaWAN/Overview
Montenegro, G., Hui, J., Kushalnagar, N., and Culler, D. (2007), Transmission of IPv6 Packets over
IEEE 802.15.4 Networks. IETF RFC 4944, 1–30, https://tools.ietf.org/html/rfc4944
Patil, B., Nieminen, J., Isomaki, M., Savolainen, T., Shelby, Z., and Gomez, C. (2015), IPv6 over
Bluetooth low energy. IETF RFC 7668, http://www.ietf.org/rfc/rfc7668.txt
Patil, K. and Lahudkar, S. L. (2016), A survey of MAC layer issues and application layer protocols for
machine to machine communications, 3(2), 742–747
Rajandekar, A. and Sikdar, B. (2015), A survey of MAC layer issues and protocols for machine-to-
machine communications. IEEE Internet Things J., 99: 175–186. http://dx.doi.org/10.110g/
JIOT.2394438
Reiter, G. (2012), White paper “Wireless connectivity for IoT”, Texas Instruments. http://www.
ti.com/lit/wp/swry010/swry010.pdf
Salman, T. (2008), Networking protocols and standards for internet of things. http://www.cse.wustl.
edu/~jain/cse570–15/ftp/iot_prot
The Internet of Things powered by Z-wave, Retrieved from http://z-wavealliance.org/
Thread Usage of 6LoWPAN, Retrieved from https://www.silabs.com/SiteDocs/white-papers/
Thread-Usage-of-6LoWPAN.pdf
Tian, L., Famaey, J., and Latre, S. (2016), Evaluation of the IEEE 802.11ah restricted access window
mechanism for dense IoT networks. IEEE Conference world of wireless mobile and multimedia
networks, Coimbra, Portugal, pp. 21–24. doi:10.1109/WoWMoM.2016.7523502
Transmission of IPV6 Packets over ITU-T g.9959 Networks, Retrieved from http://www.ietf.org/rfc/
rfc7428.txt
What is LoRaWAN? Retrieved from http://www.link-labs.com/what-is-lorawan/
What is Weightless? Retrieved from http://www.weightless.org/about/what-is-weightless
4
Privacy and Security of Data in IoT:
Design Principles and Techniques
K.V. Arya and Vishnu Prabhakaran
CONTENTS
4.1 Introduction ..........................................................................................................................80
4.2 Issues in IoT ..........................................................................................................................80
4.2.1 Security Issues .......................................................................................................... 82
4.2.2 Communication Issues ............................................................................................ 82
4.2.3 Security Implementation in IoT .............................................................................83
4.3 Internet of Insecure Things ................................................................................................83
4.3.1 New IoT Threats and Attacks.................................................................................83
4.3.2 IoT Security Countermeasures............................................................................... 85
4.4 Challenges in Securing IoT Technologies ........................................................................85
4.4.1 Types of Threats in IoT ............................................................................................ 86
4.4.1.1 Spoofing Attacks .....................................................................................86
4.4.1.2 Non-Blind Spoofing................................................................................87
4.4.1.3 Blind Spoofing .........................................................................................87
4.4.1.4 Man in the Middle Attacks .................................................................... 87
4.4.1.5 Denial of Service Attacks ....................................................................... 87
4.4.1.6 Data Tampering.......................................................................................87
4.4.1.7 Repudiablility ..........................................................................................88
4.4.1.8 Information Disclosure .......................................................................... 88
4.4.1.9 Network Intrusions ................................................................................ 88
4.4.1.10 Social Engineering Attacks .................................................................... 88
4.4.2 Security Requirements ............................................................................................ 89
4.4.2.1 Data Security Requirements.................................................................. 89
4.4.2.2 System Security Requirements ............................................................. 89
4.4.2.3 Network Security Requirements .......................................................... 89
4.4.3 Privacy Principles in IoT Technologies ................................................................. 89
4.4.3.1 Privacy from the IoT Organization’s Perspective .............................. 90
4.4.4 Privacy at the Granular Level ................................................................................ 91
4.5 Methods of Providing Security.......................................................................................... 93
4.5.1 Authenticity .............................................................................................................. 93
4.5.2 Redesigns and Fixes ................................................................................................ 93
4.5.3 Firewalling and IPS ................................................................................................. 94
4.5.4 Secure Booting ......................................................................................................... 94
4.5.5 Access Control .......................................................................................................... 94
79
4.6 Building Safer IoT ................................................................................................................ 94

4.6.1 Cloud Security and the IoT ..................................................................................... 95
4.6.2 Security versus Efficiency....................................................................................... 95
4.6.3 Defend against IP Spoofing .................................................................................... 95
4.7 Security Issues in IoT Products .......................................................................................... 95
4.7.1 Unauthorized Access............................................................................................... 96
4.7.2 Encryption ................................................................................................................ 96
4.7.3 Updates and Patches................................................................................................ 96
4.7.4 Lack of Experience ................................................................................................... 97
4.8 Conclusion ............................................................................................................................ 97
Bibliography .................................................................................................................................. 98
4.1 Introduction
The Internet of Things (IoT) is moving from a unified structure to a perplexing system
of decentralized shrewd gadgets. This movement guarantees altogether new administra-
tions and business opportunities. An inexorably associated world will see the developing
systems administration and cloud enablement of a wide range of physical gadgets from
machines through cars to home apparatuses.
The IoT includes the expanding predominance of articles and elements—referred to, in
this setting as things—furnished with one-of-a-kind identifiers and the capacity to con-
sequently exchange information over a system. A significant part of the expansion in IoT
correspondence originates from figuring gadgets and inserted sensor frameworks utilized
as a part of mechanical machine-to-machine (M2M) correspondence, vehicle-to-vehicle
communication and wearable processing gadgets, and so on.
The principle issue is that on the grounds that systems administration machines and
different articles are moderately new, security has not generally been considered in item
plan. IoT items are frequently sold with old and non-patched implanted working frame-
works and programming. Besides, buyers frequently neglect to change the default pass-
words on shrewd gadgets—or in the event that they do transform them, neglect to choose
adequately solid passwords. To enhance security, an IoT gadget that should be specifically
open over the Internet ought to be fragmented into its own particular system and has sys-
tem access limited. The system section ought to then be checked to distinguish potential
bizarre activity, and move ought to be made if there is an issue.
4.2 Issues in IoT

A quarter century, on the off chance that you let me know my telephone could be utilized
to take the secret word to my e-mail account or to take a duplicate of my unique mark
information, I would have snickered at you and said you observe a lot of James Bond. Be
that as it may, today, in the event that you let me know that programmers with malevolent
expectations can utilize my toaster to break into my Facebook account, I will freeze and
rapidly pull the fitting from the detestable machine.
Privacy and Security of Data in IoT: Design Principles and Techniques 81
Devices (billions) IP V4 consumer (billions)

60 8000
7000
50
6000
40 5000
30 4000
3000
20
2000
10 1000
0 0
2003 2008 2010 2015 2020 2003 2008 2010 2015 2020
FIGURE 4.1
Evolution of IoT (in terms of number of devices and number of consumers) over the years.
Welcome to the period of the IoT, in which digitally associated gadgets are infringing on
each part of our lives, including our homes, workplaces, cars, and even our bodies. With
the appearance of IPv6 and the wide arrangement of Wi-Fi systems, IoT is developing at a
perilously quick pace, and scientists assess that by 2020, the quantity of dynamic remote-
associated gadgets will surpass 40 billion as observed in Figure 4.1.
In any case, IoT brings up essential issues and presents new difficulties for the security
of frameworks and forms and the protection of people. Some IoT applications are firmly
connected to delicate foundations and vital administrations, for example, the dispersion
of water and power and the reconnaissance of advantages. Different applications handle
delicate data about individuals, for example, their area and developments, or their well-
being and buying inclinations. Trust in and acknowledgment of IoT will rely on upon the
insurance it gives to individuals’ protection and the levels of security it sureties to frame-
works and procedures.
The upside is that we can do things we at no other time envisioned. In any case, as with
each good thing, there is a drawback to IoT: It is turning into an undeniably alluring focus
for cybercriminals as indicated in Figure 4.2. More associated gadgets mean more assault
vectors and more potential outcomes for programmers to target us; unless we move quickly
to address this rising security concern, we will soon be confronting an inescapable fiasco.
The IoT is based on a wide range of semiconductor advances, including power adminis-
tration gadgets, sensors, and chip. Execution and security necessities change significantly
Server Fake
server
An auditor is A fake server is used
listening. to initiate unplanned
Command can Network activities by sending
leak the commands to random
information Auditor destinations which
about the can be used to
infrastructure Device Fake initiate physical
or operations device attack
FIGURE 4.2
A typical information leakage scenario.
starting with one application then onto the next. One thing is consistent, nonetheless. The
more prominent the volume of delicate information we exchange over the IoT, the more
prominent the danger of information and fraud, gadget control, information adulteration,
IP robbery, and considerably server/system control. The issues in IoT can be classified in
different ways in the following sections.
4.2.1 Security Issues

In all the business processes, the required degree of reliability is very high. The following
security and privacy requirements are supposed to be the part of secure system (Fabian
and Gunther, 2007):
Power to handle attacks: The system is designed in such a way that it is capable of
avoiding any point of failure and should reconfigure itself to avoid node failures.
Data authentication: In principle, the system has the mechanism to authenticate the
retrieval of address and information.
Access control: System must be capable of implementing access control on the given
data.
Privacy of Client: By very design, the measures to be taken so that the information
provider can only interpret the information based on the need of particular
customer.
Security issues in IoT are classified depending on the different layers of IoT architecture.
In network layer, a number of security issues can arise. For example, Denial of Service,
man-in-the-middle attacks, and so on. Due to an enormous heterogeneous volume of
data exchange, a lot of network security issues come into picture. This apparently leads to
network congestion or sometimes even data eavesdropping too. In the application layer,
because of the integration between different technologies, the application may face extra
issues like replay attacks or data tampering. The key exchange mechanism should be secure
enough in order to prevent it from any intruder. Apart from these, the attacker might gain
the encryption key based on the time taken to encrypt the data, which is known as tim-
ing attack, must be addressed. The attacker can take control of all the nodes in the system
which is called Node Capture attack.
4.2.2 Communication Issues

IoT communication can be stated in two different ways, namely, inter- and intra-domain.
In both the cases, a lot of issues can be identified. To be precise, there is no standard
protocol or framework for IoT communication. This essentially means incapability to
merge different IoT technologies together. The lack of standard universal agreement
makes the system difficult to realize in all practical aspects. The existing protocols do
not provide the actual meaning to the IoT, as the IoT is much different, in which different
technologies of different functionalities are combined, which makes the system hetero-
geneous. As IoT systems experience heterogeneity, intracommunication process becomes
an unrealizable task. Moreover, effective mechanisms should be handled to take care of
the traffic explosion caused by exchange of different IoT devices. There is a possibility to
maintain interdomain communication for a long period, thus energy efficient protocols
have to be considered.
4.2.3 Security Implementation in IoT

A light-weight protocol, Message Queue Telemetry Transport, which is especially designed
for IoT can be implemented. The key feature here is to implement a broker. There are different
sorts of brokers available like HiveMQ, IBM Message Sight, and so on. The purpose of these
brokers is to help in obtaining the information from the things. All these brokers vary in their
features and some of them implement additional features on top of standard Message Queue
Telemetry Transport functionality. A notable implementation has been provided by using
Wired Drone Docking system that can perform an assortment of capacities careful coopera-
tion with gadgets in existing IoT environment, and it can be connected to industrial security.
4.3 Internet of Insecure Things

The Rapid7 exploration is the recent one to find the alarming security issues. This is con-
nected to IoT similar to regular gadgets—including clothes washers, indoor regulators,
and cars—that have processing and system capacities implanted into them. The Rapid7
scientists concentrated on electronic devices as they are generally utilized and given the
seriously individual uses IoT gadgets could serve. The scientists went ahead to caution that
the bugs they found could do substantially more than permit eavesdroppers to attack the
proprietors’ close to home protection. A portion of the additionally terrifying vulnerabili-
ties found on IoT gadgets has brought IoT security further up the pile of issues that should
be tended to rapidly. The Internet of Things is a term not without buildup and metaphor,
but rather at its heart is an idea that is as of now picking up energy. Innovation and infor-
mation transfers firms are connecting things as various as cell phones, cars, mechanical
sensors, and family apparatuses to the web, empowering intercommunication and inde-
pendent M2M information exchange. The quantity of associated buyer items as of now
counts in the billions, and investigators gauge it will hit 6.4 billion overall this year. There
is a lot of business potential in these extending associations, yet there is a major drawback:
the quantity of conceivable IoT security vulnerabilities is becoming exponentially increas-
ing too. The IoT makes basically every part of data administration more confounded: there
are more gadgets, more vulnerability, and more data that is streaming. That makes IoT
information security more muddled, as well.
It just opens up new pathways
Experts say that there are IoT security vulnerabilities along the entire chain: Someone
could spoof a sensor connecting to transmit data, an organization with poor authentication
could inadvertently allow a hacker onto its network, and an insecure connection could
allow a cybercriminal in.
4.3.1 New IoT Threats and Attacks

As the technology develops, there are much more threats that come into limelight. Thanks
to numerous researches that have provided a way to distinguish the new IoT threats. The
following gives an insight of how IoT threats can be classified based on different factors.
• Vehicles or human body can be manipulated

• Possibility of wrongly diagnosing and treating patients
• Possibility to gain physical access to any building

• Loss of vehicle control
• Critical information such as warning of floods can be blocked
• Critical infrastructure damage can occur
• Sensitive information such as credit card can be stolen
• Unauthorized tracking of people’s locations
• Manipulation of financial transactions
Let us see how a car could be controlled by a hacker. We assume the car has the ability to
make use of GPS and judge the route based on the traffic information. This can also mean
that hackers can gain control over your car. To be precise, a hacker can control the door
opening, flashing headlights, and so on. The scenario has been explained in Figure 4.3.
Apart from all these threats, there are many threats that are being discovered every day
or minute rather.
Attacks
Take control Steal information Disrupt service
Control for smart door Pacemakers can be

locks and lighting system attacked remotely.
can be vulnerable High capacity
Personal fitness insulin pumps can
device can tell a be vulnerable
hacker about your
Vehicles location
Door locks have been unlocked
Home remotely. Infotainment systems
offer multiple ways into a car’s
Malware infected electronics. Hacked vehicle
refrigerator control system can allow
has sent spam remote control of brakes
FIGURE 4.3
IoT dangers—car attacks.
Application
Network
Access point
Perception
Sensors
FIGURE 4.4
Ensuring security in trilayer of IoT architecture.
4.3.2 IoT Security Countermeasures

Security measures have to be implemented in all three layers starting from physical,
network, to application layers as shown in Figure 4.4. Proper authentication measures
have to be provided such as the one discussed in Zhao et al. (2011). A feature extraction
method has to be combined with hash function to avoid collisions between IoT terminal
nodes and platforms. This method emphasizes the authentication process when the data
are exchanged from IoT platform to terminal nodes.
To provide an accurate access control is as vital as authenticating for security. This is
given in the form of Identity Authentication and Capability based Access Control (Mahalle
et al., 2013). This method uses timestamp in the authentication messages between various
gadgets. Thus, it prevents one of the most popular attacks man in middle.
4.4 Challenges in Securing IoT Technologies

As indicated by an extraordinary number of researchers, gadgets in the IoT world get an
assorted cluster of potential security dangers. Some of these dangers are customary dan-
gers for standard PCs. For example, new smart TVs that empower clients to surf the Internet
buy other things and offer photographs by means of social media could put information
away. More sensitive data such as card data, passwords, and individual information can
encourage data tampering. The idea of Multiple Points of Vulnerability in the security of IoT
framework, actually, incorporates the IoT item, installed programming, information inside
the gadget, correspondence channels, information collection stage, to examine the sensor
Diverse threats
Multiple
Internet of
points of Resources
Things
vulnerability
Discrepancy in
IoT quality
FIGURE 4.5
Common security characteristics of IoT products.
information. Giving security to these surfaces is a genuine test for organizations here.
The common characteristics of IoT security products are given in Figure 4.5.
In all actuality, there is an incredible inconsistency in the nature of IoT gadgets overall,
as some of them are modern and costly and others are shabby and expendable. Security
vulnerabilities in one gadget might be utilized to encourage assaults on different gadgets
sometime later in a given customer’s system. There can be many challenges that arise in
deploying security in IoT architecture. For example, some of the IoT systems are badly
designed. These are done with the help of protocols that give rise to complex configu-
rations. There is a lack of mature IoT technologies and business processes. There is no
proper guidance for lifecycle maintenance and management of IoT devices. Furthermore,
the IoT presents one-of-a-kind physical security concerns. IoT security concerns are mind
boggling and not generally promptly apparent. Some security concerns are not promptly
identifiable and some worries are not feasible by essentially implementing secrecy assur-
ances, character, or area to exchanges. Numerous IoT designers are not yet acquainted
with secure improvement best practices. The race to make new IoT-based capacities will
probably bring about restricted spotlight on the security of the new usefulness being
made. There is an absence of models for confirmation and approval of IoT edge gadgets.
Review and logging principles are not characterized for IoT parts. Confined interfaces are
accessible to collaborate with IoT gadgets onto security gadgets and applications. There is
no emphasis yet on recognizing techniques for accomplishing situational consciousness of
the security stance of an association’s IoT resources. A security standard for stage setups
including virtualized IoT stages supporting multi-tenancy is juvenile.
4.4.1 Types of Threats in IoT

The threats in IoT can be classified into many types.
4.4.1.1 Spoofing Attacks

When a malicious user impersonates the authorized user, he falsifies the data. Many of the
existing IoT protocols do not authenticate the messages from a source or destination, results
of which make the system more vulnerable to spoofing attacks. The spoofing attacks can
be of different types like IP spoofing, e-mail spoofing, or network spoofing. To understand

the concept better, let us go to reality where we get some of the phishing mails asking to
update the profiles like Paypal. Many of us are aware of the fact that it is from the spoofed
site. But some of us may fall into the trap of these websites and click the link to navigate
to the spoofing site. This spoofing site is well designed such that it looks exactly similar to
the original site. Thus, the attackers have the higher probability of stealing the sensitive
information like credit or debit card details. There is a very popular spoofing occurs in the
name of IP spoofing. This necessarily implies that the attacker tries to steal the information
from a trusted computer network. IP spoofing is always carried out by the hacker for mali-
cious actions. The attacker first modifies the source address field in the transmitted packet.
When the destination opens up the connection for the forged source address, it can lead to
a large number of attacks. This IP spoofing can be classified as many.
4.4.1.2 Non-Blind Spoofing

This is a kind of attack that takes place when the hacker and the victim share the same
subnet. This hacker can bypass any authentication measures with the help of a mechanism
called session hijacking. Further, the stream of data can be corrupted. With the help of any
advanced tool, the hacker can reestablish the correct stream of data.
4.4.1.3 Blind Spoofing

The attackers can target the host with a bunch of data packets to obtain the correct sequence
numbers. These sequence numbers are really useful to assemble the data packets for the
attacker. However, modern systems use random numbering for sequences hence it is a bit
difficult for the attacker to guess the correct one. With the help of obtaining the sequence
numbers, the attacker can inject the data into the packets as it requires nonauthentication
connection establishment.
4.4.1.4 Man in the Middle Attacks

The purpose of this attack is to interrupt the legal communication between two legal par-
ties. Without having the exact knowledge about the machine, the hacker can send spoofed
packets into the stream of data. The attacker can fool the host and can modify or delete the
packets by forging the original identity.
4.4.1.5 Denial of Service Attacks

Within zero time, the attacker can flood the host machine with a large number of requests.
The attacker can spoof the IP address of the victim which in turn makes it difficult to be
traced. First, the attackers can create a list of more vulnerable hosts which is obtained with
the help of scanning the Internet and convince them to install any type of attack programs
which can then be exploited to obtain the full access.
4.4.1.6 Data Tampering

This is the deliberate modification of data. The interesting fact is that the tampered data
will always not be detected in the first place. These data can be tampered in any ways
like it can be tampered when it is in the channel or while it is being stored in the system.
The main motive of the attackers is to target the data packets that are not encrypted.
A good measure is to encrypt the data whenever and wherever possible.
4.4.1.7 Repudiablility
It is the action of denying something that has already occurred. It is like denying the recep-
tion of some receipt or service and asking for another one. This can be prevented with the
help of digital signatures and hashes.
4.4.1.8 Information Disclosure

The severity of this depends on the type of sensitive data are being disclosed. For example,
medical or credit card data will attract high severity. This could be a serious threat in case
of IoT applications. This is somewhere related to data tampering as disclosure is the very
first step in data tampering. So keeping the data tempering out of the context will help us
to get rid of information disclosure attacks.
4.4.1.9 Network Intrusions

IoT gadgets bring the guarantee of business streamlining, remote patient checking, help
with discovering parking spots, and a large group of different advantages, some not yet
even imagined. Most of the devices are built using low-cost hardware which makes them
vulnerable to security risks. One of such risks is network intrusions. Due to the inexpensive
nature of the IoT devices, it becomes really difficult for them to report or log any informa-
tion about whether the hacker has penetrated into the network or when the network secu-
rity has been compromised. The first phase to overcome this problem is to use detection
software for the network intrusion. For instance, let us consider a hacker tries to gain access
control of any IoT device. After discovering the IoT device on the network, this hacker can
use brute force attack to find out the correct log in information. When the hacker floods
with multiple login attempts, most embedded systems should just process and drop the
request each time when the login information is matched with the available data. These
embedded systems just consider the next request as a normal request. This essentially
means that the embedded system is not aware of the fact that it is under network intrusion
attack. Hence, it cannot report to the security management system. When the users have a
strong password which cannot be easily guessed by attacker, he may not succeed in any of
his attempts. Unfortunately, users have the habit of setting the easily guessed passwords.
For example, many of the passwords set by the users have user or admin in it. This opens up
many security risks. Intrusion-detection software would usually alert the system after the
flooding of the failed login requests and generates report for such requests.
4.4.1.10 Social Engineering Attacks

Social engineering attacks have become very common against enterprises. With program-
mers formulating perpetually astute techniques for tricking workers and people into giv-
ing over significant organization information, ventures must use due industriousness with
an end goal to stay two stages ahead of cyber criminals. These types of attacks commonly
include some type of mental control, tricking generally clueless clients or workers into
giving over classified or delicate information. Normally, social engineering includes e-mail
or other correspondence like asking to click a malicious link or to open an infected file.
It is important to analyze the severity levels of each of these threats and document it
before IoT deployment.
4.4.2 Security Requirements

The security requirements can be classified into three broad categories.
4.4.2.1 Data Security Requirements

A data security system is an arrangement to mitigate dangers while submitting to with
legitimate, statutory, inside, and legally binding created requests. Common strides to
building a system incorporate the meaning of control destinations; the evaluation and
recognizable proof of ways to deal with meet the targets, the determination of controls,
measurements, the foundation of benchmarks, and the arrangement of usage and testing
plans. The decision of controls is ordinarily relies on upon cost examination of various key
ways to deal with minimize the danger. The cost correlation normally differentiates the
expenses of various methodologies with the potential pick up money-related establish-
ment could understand as far as expanded availability, confidentiality, or integrity of the
information is concerned. Whatever be the approach, it should consider each and every
aspect of the following entries.
• There should be a standard policy or procedures

• Technology should be designed in a way that is compatible to all IoT products
• A typical data security system should support every gadget
4.4.2.2 System Security Requirements

Providing security nowadays has turned into an intense errand for all the business and
the distinctive associations. Security must be given to the clients and the vital informa-
tion to shield them from the vindictive and automatic leaks. Information is imperative for
each undertaking; it might be the custom records or licensed innovation. To keep the trust
between the IoT products and the consumers, it is vital to provide the security in the system.
4.4.2.3 Network Security Requirements

As the technology builds up, the utilization of Internet has turned out to be progressively
increasing. Furthermore, the use of various advancements turns out to be more. In the
meantime, security risk additionally turns out to be increasing and offers opportunity to
do more faults. Network attacks have become more common. When the network is more
targeted by the attackers, it is sometimes possible to lose data too. The Internet should be
the safest thing to do business. When the trust is broken, IoT does not have any meaning.
To avoid the network being targeted by the hackers, security policies should be taken that
safeguard the network from any such vulnerability.
4.4.3 Privacy Principles in IoT Technologies

There should be a mandate to let the IoT users know that kind of information that are
being collected from them. The users should rather have the flexibility of sharing limited
information. Before deploying the system in operational state, it is required to decide

the privacy principles. A proper analysis should be done on what kind of data should
be shared over the network. Proactive measures rather than reactive measures should
be taken. The security should be deployed as a part of IoT system development process.
The customers should be notified about their unexpected use of data. IoT organization
that employs this should ensure that the required privacy controls are installed perfectly.
A few innovative protections are usually talked about or executed: information minimi-
zation, encryption, access control, and advances giving the people upgraded control over
their own information, the way it is traded and secured. As in a legitimate examination of
the dangers and effects, one should not just concentrate on the technical aspects. Corporate
frameworks will have huge amount of data derived from the IoT sensors. Take the case of
organizations naturally gathering readings from client smart meters. They could parody
messages being sent from the meter to the service organization and send false information.
Yet, in the IoT that security capacity does not exist in a hefty portion of the gadgets that
will all of a sudden get to be associated. Security must be incorporated with the outline of
these gadgets and frameworks to make trust in both the equipment and trustworthiness
of the information. Further, the organization should collect the data about threat analysis
when the information is shared over the Internet. Application-specific privacy controls
should be provided based on the type of application. Companies should review their pri-
vacy policies regularly in order to cope up with the technology changes. A better trade-off
should be taken between security and privacy to achieve the objectives. Customers must
be provided with the glimpse of what kind of their data is shared with the third parties by
the IoT organizations as the data collected in IoT have a longer lifetime. This data protec-
tion should be available to secondary data also. For instance, if the data about when and
where your car moves and some other information like shopping habits; it becomes easy
for anyone to track you and do harm. Granular level data exemption should be provided to
the customers. Maintaining all the sensitive data becomes a very serious threat, as there is
a strong chance that the data could be viewed by anyone. IoT organization should ensure
that only authorized persons have the access in order to avoid mishandling data.
4.4.3.1 Privacy from the IoT Organization’s Perspective

The following principles help in achieving the privacy with respect to the organizations:
• Recognize the essential data streams in the associations and, particularly, the
arrangements of information that nourish center computational frameworks,
regardless of the possibility that information is not gathered or transmitted
electronically today. Wherever information gathering or information utilization
makes business esteem, you ought to hope to one day see associated (IoT) gadgets.
• Arrange the sorts of IoT gadgets expected and their benchmark administration
necessities, for instance, gadget discovery, remote configuration, inventory, and
programming update. Organize these data by timeframe.
• Characterize and organize the new threats of information leakage, particularly
new vectors that develop because of the fragmentation of working frameworks,
systems, and interfaces.
• Evaluate the danger of unapproved access to these gadgets. For instance, when the
medical devices are compromised, the negative impacts of this system should be
analyzed.
• Characterize the related security activities to be triggered. For example, the cir-
cumstances under which a gadget that is compromised would be removed the
associated system.
• Characterize your big data technique for IoT. By what method will you secure the
huge measure of business basic information that is delivered by the sensors in
these gadgets? Imagine a scenario in which monstrous measure of sensor infor-
mation exuding from a business basic gadget is traded off or spilled. Information
arranged security with element information relationship, investigation, and
insight is a center prerequisite for IoT.
• Create protection arrangements for sensor information. The expansion of sensors
will bring about more individual information being produced. The entrance to
and the security of this information will have numerous protection suggestions
with constrained direction from existing case law.
• Ensure these new associated gadgets against system interruptions and foreswearing
of-administration assaults. Endeavors have instruments to do this today; however,
now they will need to do it over a much more extensive arrangement of gadgets.
4.4.4 Privacy at the Granular Level

Securing an IoT foundation requires a thorough security inside and out system. Beginning
from securing information in the cloud, to ensuring information safe while in travel over
the general population web, and giving the capacity to safely procurement gadgets, every
layer assembles more noteworthy security confirmation in the general framework. Secure
these new associated gadgets against system interruptions and dissent of-administration
assaults. Ventures have systems to do this today, yet now they will need to do it over a
much more extensive arrangement of gadgets.
This security strategy can be produced and executed with dynamic cooperation of dif-
ferent players required with the assembling, advancement, and installations of IoT gadgets
and framework. For example, an IoT device manufacturer should make sure that following
activities are to be carried out:
• The equipment configuration ought to incorporate least elements required for

operation of the equipment and nothing more. A case is to incorporate USB ports
just required for the operation of the gadget. These extra components open the
gadget for undesirable assaults, which ought to be kept away from.
• There should be mechanism to identify physical altering of equipment, for exam-
ple, opening of gadget spread, expelling a part of the gadget, and so on. These alter
signs might be a piece of the information stream transferred to the cloud empow-
ering cautioning of these occasions to the administrators.
• Overhauling firmware amid lifetime of the gadget is unavoidable. Building gad-
gets with secure ways for redesigns and cryptographic certification of firmware
form will permit the gadget to be secured amid and after updates.
An IoT solution developer should ensure the following activities:
• Creating secure programming requires a ground-up thinking on security from

the initiation of the undertaking the way to its usage and testing. The decision of
platform and apparatuses is affected with this procedure.
• Open-source programming gives a chance to rapidly create solutions. At the

point when picking open-source programming, consider the movement level of
the group for every open-source segment. A dynamic group guarantees that pro-
gramming will be bolstered, issues will be found and tended to. On the other
hand, a dark and idle open-source programming will not be bolstered and issues
will most presumably not be found.
• A hefty portion of the product security imperfections exist at the limit of libraries
and APIs. Ensuring that all interfaces of segments being incorporated are secure
guarantees general security.
• As far as the IoT solution deployment is concerned, one has to follow the best
practices in the industry to make sure that IoT information is secured over the
Internet.
• IoT organizations may oblige equipment to be conveyed in unsecure areas, for
example, out in the open spaces or unsupervised regions. In such circumstances,
make sure that equipment sending is carefully designed to the greatest degree. On
the off chance that USB or different ports are accessible on the equipment, guaran-
tee that these are secured safely.
• Keep confirmation keys safe: amid arrangement, every gadget requires gadget IDs
and related verification keys created by the cloud administration. Keep these keys
physically safe even after its deployment. Any traded off key can be utilized by a
noxious gadget to take on the appearance of a current gadget.
The IoT solution operator should make sure that following best practices are to be made:
Stay up with the latest: Make sure that gadget OS and all gadget drivers are redesigned
to the most recent renditions. Windows 10, with programed updates turned on,
has stayed up with the latest by Microsoft, giving a protected working frame-
work to IoT gadgets. For other working frameworks, for example, Linux, staying
up with the latest guarantees they are additionally secured against malignant
assaults.
Secure against malignant movement: If the OS licenses, put the most recent hostile to
infection and against malware capacities on every gadget working framework.
This can alleviate most outside dangers. Most cutting edge working frameworks,
for example, Windows 10 IoT and Linux, can be ensured against this danger by
making fitting strides.
Review every now and again: Evaluating IoT base for security-related issues is the
key when reacting to security episodes. Most working frameworks, for example,
Windows 10, in occasion logging that ought to be looked into every now and again
to ensure that no security break has happened. Review data can be sent as a differ-
ent telemetry stream to the cloud benefit and then inspected.
Physically ensure the IoT base: the most exceedingly terrible security assaults against
IoT base are propelled utilizing physical access to gadgets. Ensuring against vin-
dictive utilization of USB ports and other physical access is a vital well-being and
security rehearses. Logging of physical access, for example, USB port utilization,
is critical to revealing any rupture that may have happened. Once again, Windows
10 empowers logging of these occasions.
Ensure cloud certifications: cloud validation accreditations utilized for arranging and
working an IoT sending are potentially the most effortless approach to obtain
entrance and trade off an IoT framework. Ensure the qualifications by changing
the secret key every now and again, and not utilizing these accreditations on open
machines.
4.5 Methods of Providing Security

Securing IoT is similar to a protocol and has to be implemented accordingly. Figure 4.6
provides an illustration of different layers of security that can be provided to an IoT sys-
tem. Apart from this there are five different steps of achieving perfect secrecy of an IoT
system. They are to be dealt in following sections.
4.5.1 Authenticity
When the gadget is connected to the system, it ought to confirm itself before accepting or
transmitting information. Profoundly implanted gadgets frequently do not have clients
sitting behind consoles, holding up to enter the certifications required to get to the system.
How, then, would we be able to guarantee that those gadgets are recognized accurately
before approval?
Pretty much as client verification permits a client to get to a corporate system in light of
client name and secret word, machine verification permits a gadget to get to a system in
light of a comparable set of accreditations put away in a protected stockpiling range.
4.5.2 Redesigns and Fixes

Once the gadget is in operation, it will begin accepting hot patches and programming
upgrades. Administrators need to take off patches, and gadgets need to verify them, in a
way that does not expend data transfer capacity or weaken the useful well-being of the
gadget. It is one thing when Microsoft sends upgrades to Windows clients and ties up
their tablets for 15 minutes. It is very important when thousands of gadgets in the field
are performing basic capacities, on the other hand benefits are subject to security patches
Data center/cloud
Core network
LOG network
Smart objects
Risk-based security Data protection and confidentiality
IP protection
Anti-temper and detection
FIGURE 4.6
Different layers of security.
to ensure against the inescapable helplessness that departures into nature. Programming
upgrades and security patches must be conveyed in a way that preserves the restricted
data-transfer capacity and irregular network of an implanted gadget and totally wipes out
the likelihood of trading off practical security.
4.5.3 Firewalling and IPS

The gadget additionally needs a firewall or profound bundle investigation ability to con-
trol movement that is foreordained to end at the gadget. Why is a host-based firewall or IPS
required if system based machines are set up? Profoundly installed gadgets have remark-
able conventions, unmistakable from big-business IT conventions. For example, smart
energy grid has its own arrangement of conventions representing how gadgets converse
with each other. That is, the reason business particular convention sifting and profound
parcel review capacities are expected to distinguish pernicious payloads covering up in
non-IT conventions. The gadget need not concern itself with sifting more elevated amount,
basic Internet movement—the system devices ought to deal with that—however, it needs
to channel the particular information bound to end on that gadget in a way that makes
ideal utilization of the restricted computational assets accessible. A firewall, in any case,
is not sufficient to ensure the IoT against programmers. That is on the grounds that control
is not the main issue—there is additional spying. Information encryption is required too.
4.5.4 Secure Booting

When the device is turned on, it is required to verify the software in the device with the
help of digital signatures. This has to be done with the help of cryptography. In addition to
this, the device has to be protected from different run-time and malicious threats.
4.5.5 Access Control

Next, various types of asset and access control are connected. Required or part based
access controls incorporated with the working framework restrain the benefits of gad-
get segments and applications so they get to just the assets they have to carry out their
employments. In the event that any part is bargained, access control guarantees that the
interloper has a negligible access to different parts of the framework as could be expected
under the circumstances. Gadget-based access control systems are undifferentiated from
system-based access control frameworks, for example, Microsoft Active Directory: even on
the off chance that somebody figured out how to take corporate accreditations to pick up
access to a system; traded off data would be constrained to just those regions of the system
approved by those specific certifications. The guideline of slightest benefit directs that
exclusive the negligible access required to perform a capacity should be approved keeping
in mind the end goal to minimize the viability of any rupture of security.
4.6 Building Safer IoT

Building a safer IoT is not just a single day’s task. A lot of parameters have to be considered
in order to provide the IoT system with a perfect security.
4.6.1 Cloud Security and the IoT

Securing the information in-transit from the edge of the system will likewise be the key.
Reportedly, medical device manufacturers and home automation are among the most
endangered industry segments, as far as cyber security is concerned. Encryption ought
to guarantee security and information honesty, yet associations will likewise require the
capacity to identify when these remote gadgets could be utilized as method for assaulting
their center system. Once more, it is impossible that this will be executed as something
besides a cloud administration.
As it is so often in modern IT, the cloud is likely to be the savior here. There are many orga-
nizations that set up individual identities and credentials for connected devices and revoke
them again to maintain the integrity of the system. The IoT will require the ability to seg-
ment the corporate network and to assume that many of the devices attached to it are at-best
vulnerable and at-worst actually a security risk (Gou et al., 2013). Yet, these administrations
will take years to develop—it could be close to the end of the prior decade they are working
at scale and with the required productivity level. In the meantime, expect IoT security—or
more likely the lack of it—to be making headlines for the next few years at least.
4.6.2 Security versus Efficiency

Security versus efficiency is a major problem in today’s IoT. It is misty right now about
what is the right answer for actualizing efforts to establish safety without affecting net-
work resources adversely. However, devices used in today’s technology cannot be secured
because of technical or business reasons. They are highly error-prone as it uses wireless
technology that is highly insecure in nature. Public key infrastructure (PKI) is a great
method to exchange secret key but in the case of IoT, this seems to be obsolete.
4.6.3 Defend against IP Spoofing

The following measures can be helpful to fight against IP spoofing:
• Applying authentication measures when exchanging keys between two parties.

• Prepare a list of authorized users to control the access. This will hinder making
Private IP addresses.
• Make different filtering techniques to control incoming or outgoing traffic. This
will make sure that a source is authenticated when incoming packets are approach-
ing, or it controls the outbound traffic.
• Routers should be configured not to accept the packets from the outside network.
Remember, these are the simple measures that you can follow to make sure that you are
free from IP spoofing. However, this spoofing needs an intelligent solution as it empha-
sizes the inbuilt TCP/IP design.
4.7 Security Issues in IoT Products

The specific security issues in IoT products can be addressed in multiple ways. Refer to
Figure 4.7 for a better understanding. Elements of the security solutions can be given in
the further sections.
Access Encryption
Updates Experience
FIGURE 4.7
Solution to IoT security issues.
4.7.1 Unauthorized Access

A standout amongst the most difficult security issues with brilliant gadgets is the strength
against presentation to physical attacks. IoT items are regularly left unattended and hence
are simple prey for a pernicious performing artist who can catch them and afterward
remove cryptography insider facts, interfere with or adjust their programming, or replace
these devices with preexploited ones under his control. Gadgets physically worked to
catch client login data at the source level provide reason for concern. Unauthorized access
could result in creating perceptible risks to physical safety, but IoT can be accessed from a
far as well.
4.7.2 Encryption
Another investigation by experts reveals that the attackers concentrate on the data that are
transmitted without any kind of encryption. This can lead to leaking sensitive information
like credentials, credit card information. So in order to protect the heterogeneous network,
it is necessary to encrypt the data whenever possible.
4.7.3 Updates and Patches

Associated gadgets should be upgraded all the time to stay safe at any rate to a great
degree unsophisticated threats. The danger of cyber attacks increments if patches are not
regularly updated much of the time. Disregarding this, it is demonstrated that close to
49% of organizations offer remote updates for their smart things. There can be many rea-
sons for this. The main concern to the firms involved in development of low end devices
may lack economical support to provide updates. Many IoT products are built with the
help of cheap rate chips. Then again, the reason could be that IoT sellers might not have
the specialized mastery to grow such upgrades. In some examples, the fundamental
issues is the absence of correspondence channels for the organization to remotely con-
vey the patches, in light of the fact that it is more helpful for these organizations to give
purchasers a chance to download and introduce them physically. Seen from the buyers’
perspective, be that as it may, this may not function admirably for them as some of them
experience challenges introducing the upgrades, or may not be at all mindful of their
presence in any case.
4.7.4 Lack of Experience

In the first place, IoT gadgets are frequently designed by consumer goods manufacturers.
On the contrary, some of the do not have the enough experience in providing the security
to the IoT products. To design a secured product, one has to have excessive skills incor-
porated with the ability to look foresee the future problems that might arise. The fewer
number of good security experts put a major obstacle in developing the secured products.
New entries do not have the experience to adapt to security issues in IoT.
4.8 Conclusion
Security at both the gadget and system levels is basic to the operation of IoT. The same
insight that empowers gadgets to perform their errands should likewise empower
them to perceive and neutralize dangers. Luckily, this does not require a progressive
approach, but instead an advancement of measures that have demonstrated effective in
IT systems, adjusted to the difficulties of IoT and to the limitations of associated gadgets.
Meanwhile, there are a couple of things you can do to secure your associated devices. In
the event that you are on a PC or cell phone, ensure that any accessible firewall and anti-
virus programming is actuated. Ensure you filter your framework routinely to recognize
and evacuate any malware or conceivable interruptions. A PC tainted with malware or
broken by a programer can be a starting point for assaults against the IoT hardware at your
home, or it might store passwords for IoT items that the programer can utilize. At last, keep
an eye out for phishing and social designing. Programers are exceptionally astute with
regards to sending messages and messages that request client names and passwords. Be
exceptionally suspicious. In the event that you get a telephone call requesting this secret
data, do not give it. Ensure you hang up and afterward call the quantity of the business
or association that the guest had guaranteed to speak to. Try not to utilize any telephone
number the guest may give. What is more, customers are gradually growing practical
insight. By consolidating rational strategies with state-of-the-art security technologies, we
can keep programers from turning our gadgets against us.
To shield buyers from potential information security breaks, organizations need to cre-
ate protection approaches that plainly detail how the information gathered from IoT items
will be utilized, and these strategies ought to be effectively available to shoppers. Everyone
is turning out to be more cognizant about where their information is being held, and an
association attempting to show purchasers what their information is being utilized for
will separate itself from the opposition. The IoT can possibly be huge for both purchasers
and organizations; however, security must be at the heart of each phase of the procedure.
We would not purchase a house that had no front entryway on it, so why might a shop-
per purchase an associated item with no security highlights set up? To rouse trust in the
IoT as it assembles pace, organizations need to guarantee they are supplying the lock and
additionally giving over the keys. Developing IoT solutions for security and protection are
promising. These incorporate making clients’ cell phones their security and protection key
that can affirm gadget matching, utilizing cryptography rather than a console and pass-
words, and protection saving individual information stockpiling frameworks so clients
control their private information shared crosswise over IoT frameworks. IoT can be made
secure, and client protection can be safeguarded if sellers, government, and undertakings
incorporate security with the IoT from the earliest starting point. The vision for the IoT is
to make our regular lives simpler and help achieving the effectiveness and profitability of
organizations and representatives. The information gathered will help us settle on more
intelligent choices. Be that as it may, this will likewise affect security desires. On the off
chance that information gathered by associated gadgets is compromised, it will undermine
trust in the IoT. We are as of now seeing shoppers place higher desires on organizations
and governments to defend their own data. Organizations ought to recognize the danger
level for their present introduction to the IoT and where it is going later on furthermore
consider the protection and security suggestions connected with the volume and sort of
information the IoT will create. It is genuinely a bold new world that guarantees numerous
energizing open doors. Trust is the establishment of the IoT and that should be supported
by security and protection. Furthermore, it is a discussion we as a whole need to begin
having now in the event that we are to profit from the associated world.
Bibliography
Baby monitors hacking issue description. Available: http://arstechnica.com/security/2015/09/9-baby-
monitors-wide-open-to-hacks-that-expose-users-most-private-moments/ (accessed September 28,
2016).
Basic IoT eavesdropping illustration. Available: http://www.infineon.com/cms/en/applications/
chip-card-security/internet-of-things-security/ (accessed September 28, 2016).
Car attacks description. Available: http://spectrum.ieee.org/telecom/security/how-to-build-a-
safer-internet-of-things (accessed September 28, 2016).
Cloud Security Characteristics description. Available: http://www.zdnet.com/article/internet-of-
things-finding-a-way-out-of-the-security-nightmare/ (accessed September 28, 2016).
Common Security Characteristics description. Available: http://resources.infosecinstitute.com/
security-challenges-in-the-internet-of-things-iot/ (accessed September 28, 2016).
Fabian, B. and O. Gunther. 2007. Distributed ONS and its impact on privacy. IEEE International
Conference on Communications, Glasgow, Scotland, pp. 1223–1228.
Gou, Q., L. Yan, Y. Liu, and Y. Li. 2013. Construction and Strategies in IoT Security System. Green
Computing and Communications (GreenCom), 2013 IEEE and Internet of Things (iThings/CPSCom),
IEEE International Conference on and IEEE Cyber, Physical and Social Computing, Beijing, China,
pp. 1129–1132.
Growth and Challenges of IoT systems description. Available: http://www.cisco.com/c/en/us/
about/security-center/secure-iot-proposed-framework.html (accessed September 28, 2016).
Mahalle, P. N., B. Anggorojati, N. R. Prasad, and R. Prasad. 2013. Identity authentication and capabil-
ity based access control (IACAC) for the internet of things. Journal of Cyber Security and Mobility,
1: 309–348.
Privacy and Security issue description. Available: http://iotforum.org/wp-content/uploads/2014/
09/D1.5-20130715-VERYFINAL.pdf (accessed September 28, 2016).
Zhao, G., X. Si, J. Wang, X. Long, and T. Hu. 2011. A novel mutual authentication scheme for internet
of things. International Conference on Modelling, Identification and Control (ICMIC), pp. 563–566.
5
Integration of Cloud Computing and IoT:
Opportunities and Challenges
Sara Ghanavati, Jemal Abawajy, and Davood Izadi
CONTENTS
5.1 Introduction ........................................................................................................................ 100
5.2 Motivation and Objectives................................................................................................ 101
5.3 Wireless Body Area Network........................................................................................... 101
5.3.1 Challenges in WBANs........................................................................................... 102
5.3.1.1 Node Characteristics .............................................................................. 102
5.3.1.2 Data Transmission Capability ............................................................... 102
5.3.1.3 Limitation of Resources ......................................................................... 103
5.3.1.4 WBANs and IoT Technology ................................................................. 103
5.4 The Internet of Things....................................................................................................... 103
5.4.1 Health-IoT ............................................................................................................... 104
5.4.2 IoT Health Care Services ....................................................................................... 105
5.4.2.1 Ambient-Assisted Living ....................................................................... 105
5.4.2.2 Adverse Drug Reaction .......................................................................... 105
5.4.2.3 Community Health Care ....................................................................... 106
5.4.3 IoT Health Care Challenges.................................................................................. 106
5.4.3.1 Storage Resources.................................................................................... 106
5.4.3.2 Computational Resources ...................................................................... 106
5.4.3.3 Communication Resources .................................................................... 107
5.4.3.4 IoT and Cloud Computing ..................................................................... 107
5.5 Cloud Computing .............................................................................................................. 107
5.5.1 Mobile Cloud Computing Technology ............................................................... 107
5.5.1.1 Advantages of Mobile Cloud Computing ........................................... 108
5.6 Cloud-WBAN-IoT Integration .......................................................................................... 109
5.6.1 Advantages of Cloud-WBAN-IoT Integration.................................................... 111
5.6.1.1 Communication ....................................................................................... 111
5.6.1.2 Storage ...................................................................................................... 111
5.6.1.3 Computation ............................................................................................ 111
5.6.2 Challenges of Cloud-WBAN-IoT.......................................................................... 112
5.6.2.1 Communications ..................................................................................... 112
5.6.2.2 Scalability and Flexibility ...................................................................... 112
5.6.2.3 Security and Privacy............................................................................... 112
5.6.2.4 Heterogeneity .......................................................................................... 113
5.6.2.5 Performance ............................................................................................. 113
5.7 Conclusion .......................................................................................................................... 113
References..................................................................................................................................... 114
99
5.1 Introduction
Wireless body area network (WBAN) provides the majority of hardware as an infrastruc-
ture support, through providing access to sensors and actuators. From their origin, WBANs
have been developed for specific monitoring purposes, for example, heart function of a
patient. In reality however, collecting only vital sign cannot be considered as a compre-
hensive health care monitoring system. Regularly, a comprehensive monitoring system for
telemedicine and ambient assisted living (AAL) needs to connect multiple medical devices
such as blood pressure meter, weighting meter, blood-glucose meter, and pulse oximeter
(Fei et al., 2010). Information about nurses, doctors, and also location of patients could also
be significantly important. In some situations, input–output devices and sensors such as
camera, microphone, speaker, can be extremely helpful in enhancing functionality of a
health monitoring system as patients can be remotely visited by their doctors. However,
WBANs are not capable to collect the information as biological sensors are very sophisti-
cated, sensitive with significant limitations.
To overcome the existing challenges, IoT technology can be applied to work in
parallel with WBANs in health care monitoring systems (Islam et al., 2015). IoT-based
services are ubiquitous and personalized and will speed up the transformation of
health care from career- to patient-centric (Klasnja & Pratt, 2012; Liu et al., 2011). In
the application, various medical devices include biomedical sensors, and diagnostic
devices can be considered as smart devices or objects representing core element of IoT
technology. The main purpose behind the innovation is to reduce costs, while enhanc-
ing quality of life. IoT provides also an efficient scheduling of limited resources by
ensuring their best use and service of more patients (Islam et al., 2015). IoT is characterized
by a very high heterogeneity of devices, technologies as well as procedures. Therefore,
ensuring about acceptance level of scalability, interoperability, reliability, efficiency,
availability, and security are considerably important in establishing IoT-based systems
(Islam et al., 2015).
Cloud Computing has the capability to resolve the majority of the aforementioned
problems (Ghanavati, 2015). Cloud Computing is in health care informatics that pro-
vides unlimited recourse of data that can be accessed anytime and anywhere in the
world (Rosenthal et al., 2010; Schweitzer, 2012). Health care systems require continuous
and systematic innovation in order to remain cost and time effective and efficient while
providing high-quality services (Rosenthal et al., 2010). Cloud Computing can ease data
processing collected from wireless sensors that need complicated computational tasks
to reveal patients’ health status (Fortino et al., 2014). In this case, real-time processing
of huge amounts of received data streams from WBAN and other related devices is a
memory- and energy-intensive task (Chen et al., 2011). Cloud infrastructure can help one
to run numerous processing tasks simultaneously in a real-time manner (Sultan, 2014).
Cloud Computing also provides additional features such as ease-of-access, ease-of-use,
and efficiency of deployment costs (Dash et al., 2010). However, integrating Cloud, WBAN
and IoT imposes several challenges such as security and privacy for health care applica-
tions that are currently receiving attentions by the research communities.
The remainder of this chapter is organized as follows: Section 5.2 briefly shows the moti-
vation and objective of the chapter. Section 5.3 introduces WBAN with advantages and
possible challenges. Section 5.4 displays IoT technology and the related IoT health care
services. Section 5.5 presents Cloud Computing and its related advantages. In Section 5.6,
integration of Cloud-WBAN-IoT is presented. Lastly, Section 5.7 concludes the chapter.
Integration of Cloud Computing and IoT 101
5.2 Motivation and Objectives

The motivation for this chapter comes from the fact that integrating the three Cloud
Computing, WBAN and IoT technologies can provide a high quality and scalable health
monitoring systems. Although there are many approaches that use sensor Cloud-based
monitoring systems, there is still a gap in the development of a comprehensive and
scalable common platform for patient monitoring systems (Löhr et al., 2010; Misra &
Chatterjee, 2014).
The main objectives of this chapter are to investigate the three technologies individually
and explore advantages as well as challenges in detail. Furthermore, the integration of the
three technologies is also studied in details.
5.3 Wireless Body Area Network

WBAN technology is a subfield of existing research in the field of WSNs and can be con-
sidered to be a specialization of biomedical engineering. WBANs consist of only less than
few dozen sensor nodes and are generally linked with the human body in which fewer
nodes are deployed than traditional WSNs. In WBANs, sensor nodes send their gathered
information to a base station (BS) known as coordinator. Due to the low transmission rate,
the network area of WBAN communication is normally within a few meters. Figure 5.1
shows a WBAN that is consisted of several body sensors such as Electrocardiogram (ECG),
blood pressure, Glucose, and motion sensor. For example, ECG (Ghanavati et al., 2016b) is
Positioning
ECG
Blood pressure
SPO2 and
Sensor node
motion sensor
PDA
Communication
Glucose link
sensor
Motion
sensor
FIGURE 5.1
Network size in a WBAN.
one of the most popular biomedical sensing systems these days. As has been investigated
in Nemati et al. (2012), Sullivan et al. (2007), ECG can monitor human heart activities
(Ghanavati et al., 2015). Another type of body sensor is blood pressure sensor that can
monitor flow and pressure in a blood vessel (Ghanavati, 2015). A glucose sensor, as it
has been extensively used in many projects (Li et al., 2015; Liao et al., 2012), can monitor
Glucose level in real time.
WBAN provides the majority of hardware as an infrastructure support, through pro-
viding access to sensors and actuators. From their origin, WBANs have been developed
for specific monitoring purposes for example heart function of a patient. That is due to
the reality, collecting only vital sign cannot be considered as a comprehensive health care
monitoring system. Regularly a comprehensive monitoring system for telemedicine needs
to connect multiple medical devices such as blood pressure meter, weighting meter, blood–
glucose meter, and pulse oximeter (Fei et al., 2010). Information about nurses, doctors,
and also location of patients could also be significantly important. Hence, ambulances can
be called by the monitoring system to pick up patients and somehow find their specific
doctors. In some situations, input–output devices and sensors such as camera, micro-
phone can be extremely helpful in enhancing functionality of a health monitoring system
as patients can remotely be visited by their doctors. However, WBANs are not capable
of collecting the information as biological sensors are very sophisticated, sensitive with
significant limitations.
5.3.1 Challenges in WBANs

In the following sections, some of the challenges in developing a comprehensive health
care monitoring system are discussed.
5.3.1.1 Node Characteristics

Small size of sensor nodes is usually preferred in most of the WBAN applications.
As patient comfort is an important factor in the applications, size of implantable or
wearable sensor nodes must be as small as possible. Apart from the size, accuracy of
data provided by sensor nodes is an extreme challenge in WBANs (Latré et al., 2011).
That is because the number of nodes is extremely limited in WBAN. Hence, nodes
need to be very accurate and sophisticated to reimburse the accuracy without data
redundancy (Li et al., 2006). For example, in ECG application that measures the activ-
ity of a human heart, sensors are attached close to the heart to sense information with
high accuracy (Latré, 2008). Therefore, we cannot expect the network to collect differ-
ent information such as location of patients, due to their unique features of WBAN
technology.
5.3.1.2 Data Transmission Capability

Due to the diversity of WBAN scenarios, data transmission parameters are different from
each application to another. The required values for data transmission parameters in dif-
ferent applications are already investigated in Ghanavati et al. (2015), Latré (2008), Latré
et al. (2011), Patel and Wang (2010).
As can be seen from the given information, data rates vary from simple with low rate to
more complicated data such as video streams. If any of WBANs applications use several of
these sensors together, data rate will reach over Mbps. In the case, as most of the current
existing low power radios are not able to pass such transmission rate, WBANs technology
is not a reliable paradigm to transfer different data such as sound and video of patients to
health centers. In fact, due to different data rate, congestion could be a possible significant
issue in such networks (Ghanavati, 2015).
5.3.1.3 Limitation of Resources

In WBANs, integrated batteries in sensor nodes provide the required energy for network
tasks. Due to the small size of sensor nodes in WBANs, batteries are also small and are able
to save limited amount of power. Mainly, there are three categories for energy consumption
in a WBAN: sensing information, wireless communications, and data processing (Latré,
2008). As have been investigated in Hanson et al. (2009), wireless communications such as
Wi-Fi, wireless USB, ZigBee, and Bluetooth consume higher amount of saved energy com-
pared with sensing part and processing circuits in a typical WBAN.
Battery usage is a critical issue in such networks. This is because for those sensors
implanted in the human body, changing or replacing the sensors is not feasible (Latré,
2008). Hence, expecting the sensor nodes with limited energy capacity to transfer large
amount of different information, such as GPS and voice data as well as their specific col-
lected data with high accuracy, is not realistic.
Apart from battery capacity, available memory, bandwidth, and processing capabilities
are among resource limitations of a WBAN that are because of very small size of the nodes.
Therefore, not all the collected information, such as voice and GPS data, can be processed.
These resource limitations cause a number of difficulties in implementing WBANs in a
comprehensive monitoring health care system (Ghanavati, 2015).
5.3.1.4 WBANs and IoT Technology

To overcome the described challenges, IoT technology can be applied to work in parallel
to WBANs in health care monitoring systems (Islam et al., 2015). WBANs can exist with-
out IoT however; IoT cannot exist without WBANs. That is because WBANs provide the
majority of hardware as an infrastructure support, through providing access to sensors
and actuators. From their origin, WBANs have been developed for specific monitoring
purposes, for example, heart function of a patient. In contrast, IoT is not focused on specific
applications as it would not be expected to collect specific types of sensor data.
5.4 The Internet of Things

In the late 1960s, Computer Network was invented, and communication between two com-
puters was made possible (Natalia & Victor, 2006). Then, TCP/IP stack was introduced in
the early 1980s. In the late 1980s, the World Wide Web (WWW) became available and made
the Internet very popular to be used commercially by 1991. That led people to become con-
nected to the Internet through different devices, including mobile devices, and form the
mobile-Internet (Associati, 2011). Finally, with the emergence of social networking, users
are now able to become connected through the Internet. The next evolution in the IoT is
to let the entire devices be able to communicate with each other via the Internet (INFSO,
2008). Figure 5.2 presents five typical phases in the evolution of the Internet.
Web Web Web

Web
FIGURE 5.2
Five phases in the evolution of the Internet.
Recently, IoT has become a significant technology in academia as well as industry. This
technology provides the opportunity of connecting the entire smart objects around us
to the Internet to communicate with each other and enables many applications in vari-
ous domains. Some objects have very limited capabilities of processing data, whereas
there is limitation in storing data. On the contrary, there are objects with larger memory
and higher processing capabilities and able to connect to the Internet intelligently. The
devices could be Internet-enabled computers and mobile phones, which already con-
nected 2.5 billion of them to the Internet. This rate is also increasing, and it is estimated
that it will reach 50 to 100 billion devices connected to the Internet by 2020 (Sundmaeker
et al., 2010). Thus, IoT technology is necessarily required to facilitate the communication
among the online devices to meet the required QoS in handling of massive number of
interactions.
The goal behind that is to reduce humans’ interactions to the devices and act to meet
people’s needs by bringing the capability of interacting autonomously to the monitor-
ing systems (Dohr et al., 2010b). This can be achieved by developing a proper architec-
ture based on applications (Alam et al., 2010; Patel et al., 2009). Application domain in IoT
generally can be mainly divided into industry environment such as Supply chain man-
agement (Chaves & Decker, 2010) and society categories (Sundmaeker et al., 2010) such
as Telecommunication, medical technology (Wang et al., 2011) that this chapter mainly
focuses on.
5.4.1 Health-IoT
In the near future, the current form of health care systems is expected to change from
the hospital-centric model to hospital–home-balanced in 2020 to the home-centric form of
health care in 2030 (Koop et al., 2008). To achieve that, IoT technology has been invented
to establish the in-home health care service. In fact, health care represents one of the most
attractive application areas for the IoT (Pang, 2013).
Health-IoT-based services are universal and personalized that enhance health care
data communication from career centric to patient centric (Klasnja & Pratt, 2012; Liu
et al., 2011). IoT has the potential to be applied in many medical applications such as
remote health monitoring that could include treatment and medication usage monitor-
ing at home by health care providers. In these applications, various medical devices such
as biomedical sensors and diagnostic devices are considered smart devices or objects
that play core roles in the IoT.
The main purpose behind the innovation is to reduce costs, while enhancing quality of
life. IoT paradigm can also efficiently schedule limited available resources to ensure their
best services in a real time (Islam et al., 2015).
5.4.2 IoT Health Care Services

IoT-based health care systems can be applied to variety of applications such as health mon-
itoring, supervising, and managing elderly patients as well as private health and fitness of
others. The services and applied procedures, however, require modifications and justifica-
tions to ensure properly function in the scenarios. These include different services such as
notification, resource sharing, and Internet services. For health care scenarios with major
connectivity with heterogeneous devices, cross-connectivity protocols and link protocols
are also required to be fully taken into account. Security, low-power and fast discovery of
devices and services can also be added to this list (Islam et al., 2015). In this section, popu-
lar IoT services are explained in detail.
5.4.2.1 Ambient-Assisted Living

An IoT platform equipped with artificial intelligence can address health care of aging and
incapacitated individuals, that is, called AAL. The main purpose of this service is to allow
elderly individuals living in their places in a convenient and safe manner. This service
ensures them a greater autonomy and providing them with a human-servant-like assis-
tance in the case of any problem.
Applying the central AAL paradigm over the IoT has been investigated in (Dohr
et al., 2010b). The authors explored combination of smart objects and health care ser-
vices to facilitate AAL. Such infrastructure enables communication between service
users such as elderly individuals, health care givers, and maybe family members.
For instance, a modular architecture for automation, security, control, and communication
is proposed for IoT-based AAL in Shahamabadi et al. (2013). This approach proposes a
framework for providing health care services to elderly and incapacitate individuals. The
authors used 6LoWPAN for active communications, and radio-frequency identifica-
tion (RFID) and near-field communications (NFCs) for passive communications. They
applied medical-based knowledge algorithms to detect issues faced by elderly people
and patients.
5.4.2.2 Adverse Drug Reaction

An adverse drug reaction (ADR) is an injury that might occur because of consuming a
single dose of a drug or a combination of two or more medicines. As this injury is not
a particular disease with a specific solution, a separate monitoring system is necessar-
ily required to be developed. An IoT-based ADR is proposed in Jara et al. (2010). In this
approach, authors developed a monitoring system by cooperating a pharmaceutical intel-
ligent information algorithm and sensor network. In this approach, the patient’s terminal
identifies the drug by applying barcode/NFC-enabled devices.
The collected information from the devices and patients’ information is processed
to ensure that the drug is compatible with their allergy profile and electronic health
record. The iMedPack has also been developed as part of the iMedBox to address the
ADR (Yang et al., 2014) by applying of RFID and controlled delamination material (CDM)
technologies.
5.4.2.3 Community Health Care

Community health care (CH) monitoring essentially comes with the idea of establishing a
network covering areas such as hospitals and rural communities. IoT paradigm can play a
backbone role in connecting several required networks to establish a cooperative network
structure.
A cooperative IoT platform for rural health care monitoring has been proposed in
Rohokale et al. (2011) to enhance energy efficiency. A community medical network has
also been proposed in You et al. (2011). This network applied WBANs to materialize CH.
The proposed platform shares the collected data from the networks to other medical
facilities. Thus, medical staff can have access to the information for making decisions.
In fact, the structure of a community medical network can be viewed as a virtual hospital
(Wang, Li et al., 2011).
5.4.3 IoT Health Care Challenges

To provide the best health care services, there are numbers of challenges to overcome.
In this section, we are addressing the challenges in detail.
5.4.3.1 Storage Resources

IoT involves a large amount of information as Things produce a considerable amount of
data. The data in different scenarios could be semistructured or even not-structured in
variety of sizes collected in different frequencies (Aguzzi et al., 2014). This collected infor-
mation then needs archiving and sharing, which cannot be accomplished by IoT paradigm
due to its capabilities (Rao et al., 2012). This integration apprehends a new convergence
scenario with newly arisen opportunities such as data aggregation (Fox et al., 2012) and
sharing with third parties (Zaslavsky et al., 2013).
Hence, a technology is required to offer the most convenient solution to deal with data
produced by IoT. It should capable of virtually offering unlimited, low cost, and on-demand
storage capacity to collect data (Rao et al., 2012). It should be capable of treating data in
a homogeneous manner through specified standards (Fox et al., 2012). Furthermore, it
should apply top-level security (Dash et al., 2010) and directly access to and visualize from
different locations (Rao et al., 2012).
5.4.3.2 Computational Resources

Due to the very limited processing resources of IoT paradigm, resources are not permit-
ted to process data on-site. Data processing is usually accomplished in powerful nodes in
which aggregation and realizing is possible (Rao et al., 2012). To address the issue, a third
party technology can be applied to provide unlimited real-time processing capabilities
and on-demand model to allow IoT processing be properly satisfied and enable analyses
of unprecedented complexity (Dash et al., 2010).
Another paradigm is also required to offer data-driven decision-making and predic-
tion algorithms to reduce risks and costs while enhancing performances with higher
energy efficiency (Zaslavsky et al., 2013). The performance needs to be real-time processing
(Rao et al., 2012), to implement scalable, collaborative, sensor-centric applications in com-
plex events (Rao et al., 2012).
5.4.3.3 Communication Resources

The main concept of IoT is enabling IP-enabled devices communicating through differ-
ent dedicated hardware. However, providing the capability for different communication
purposes can be considerably expensive. Therefore, a technology is required to offer an
effective and cheap solution to connect, track, and manage requests from anywhere at any
time using customized portals and built-in applications (Rao et al., 2012).
5.4.3.4 IoT and Cloud Computing

IoT is characterized by a very high heterogeneity of devices that are applying different
protocols to collect and communicate with each other. Thus, QoS parameters such as
scalability, interoperability, and reliability as well as security necessarily need to be fully
considered. However, the integration with the Cloud overcomes most of these concerns as
proved in Suciu et al. (2013). Cloud Computing also provides additional features such as
ease-of-access, ease-of-use, and efficiency of deployment costs (Dash et al., 2010).
5.5 Cloud Computing

The general concept of Cloud Computing refers to a new technique capable of deliver-
ing enterprise IT. Cloud-based services provide flexible deployment of applications to
adjust their resources to changes in client requirements. Cloud Computing provides vir-
tual machines (VMs) instances for servers, and other Internet-faced services. Thus, Cloud
Computing based models can be applied to any type of data centers regardless of its opera-
tional goals (Schweitzer, 2012).
One interesting application of Cloud Computing is in health care informatics, which
provides unlimited recourse of data, can be accessed anytime from anywhere in the world
(Ghanavati et al., 2016a). Health care systems require continuous and systematic innova-
tion in order to remain cost and time effective and efficient while providing high-quality
services (Rosenthal et al., 2010). Cloud Computing can ease data processing collected from
wireless sensors that need complicated computational tasks to reveal patients’ health sta-
tus (Fortino et al., 2014). In this case, real-time processing of huge amounts of received data
streams from WBAN and other related devices is a memory and energy-intensive task
(Chen et al., 2011). However, Cloud infrastructure can help one to run numerous process-
ing tasks simultaneously and in a real-time manner (Sultan, 2014).
Another challenging issue is managing collected data from the devices that include
WBANs as the data might be distributed in time and or space (Borradaile, 2013). Time dis-
tribution means occurrence of several activities in different times, whereas they have been
managed to reach the same goal. Space distribution refers to activities that take place in dif-
ferent locations, whereas they are managed by a same data network (Fortino et al., 2014). In
this case, a Cloud Computing solution could simplify the management of distributed data
by using methods such as data fusion in a scalable and low cost manner (Chatman, 2010).
5.5.1 Mobile Cloud Computing Technology

Mobile Cloud Computing (MCC) technology was invented not long after Cloud Computing
was introduced. The paradigm has been attracted attentions as a profitable business option
that reduces development and running cost of mobile applications (Ali, 2009).
Alternatively, MCC can be defined as a combination of mobile web and CC (Christensen,

2009), which is the most paradigm for stakeholders to access applications and services on
the Internet. MCC provides an infrastructure in which both the data storage and data
processing happen outside of the mobile device. The infrastructure moves the comput-
ing power and data storage into powerful and centralized platform located in Cloud. The
centralized platform and applications are then, accessed over wireless connection via not
just smartphones but mobile subscribers (Suciu et al., 2013). Thus, with this technology, the
mobile devices do not need a powerful configuration. The devices are not required high
CPUs speed and memory capacity either. That is because the entire complicated comput-
ing modules are processed in the Clouds.
5.5.1.1 Advantages of Mobile Cloud Computing

In this section, we explore the advantages of MCC in detail.
5.5.1.1.1 Extending Battery Lifetime

Efficiency of energy consumption is a significant concern in developing mobile devices.
Literature shows that there are a various number of developed approaches to enhance
CPU performances (Ghanavati, 2015) and to screen in an intelligent manner (Steinfeld
et al., 2015) to reduce power consumption. However, these solutions did not fully address
the issue as they require to investigate in the structure of mobile devices or new hardware,
which is not feasible for all mobile devices using in different systems. Then, the idea of
using offloading techniques is proposed (Ghanavati et al., 2016a). This avoids taking a long
application execution time on mobile devices and consequently enhance large amount of
power consumption.
For instance, (Barbera et al., 2013) evaluate the effectiveness of offloading techniques
such as applying MCC through several experiments. They proved remote application exe-
cutions that can save energy significantly up to 25%. Task migration and remote process-
ing is also another advantage of applying offload techniques. Mobile Cloud for Assisting
Healthcare (MoCAsH) (Hoang & Chen, 2010) is MCC-based information management sys-
tem that collects and utilizes vital sings. Users interact with the system via mobile agents
to the Cloud to communicate and manage the collected data.
5.5.1.1.2 Improving Data Storage Capacity and Processing Power

MCC is developed mainly with the purpose of ensuring mobile users are capable of stor-
ing and accessing a large data on the Cloud through wireless networks. Amazon Simple
Storage Service (S3) (Vartiainen & Väänänen-Vainio-Mattila, 2010) utilize a large stor-
age space in Clouds for authorized users. Apart from storage, running cost of compute-
intensive applications is also considerable. That is because of required long time and
large amount of energy especially when performed on the limited-resource devices. MCC
enhances the expected cost of running the applications. Amazon offers Elastic Compute
Cloud (EC2) that is developed to perform data computation on the stored data in S3. That
provides a number of advantages for users, including lower maintenance and operation
cost with higher utilization through virtualization. This type of computing model reduces
start-up time for new services as well as easier disaster recovery (Kumar & Lu, 2010).
5.5.1.1.3 Improving Reliability

Storing data and running applications on Clouds are an effective way to enhance data-
processing reliability. That is because of storing and backing up data and applications
on a number of computers or servers. This reduces the possibility and risk of data and
application lost on the mobile devices. MCC can also be developed with a fully com-
prehensive data security model for both service providers and users. For instance, the
Cloud is expected to remotely provide mobile users with security services such as mali-
cious code detection and authentication (Oberheide et al., 2008). Also, these Cloud-
based security services are capable of enhancing usage efficiency of the collected data
and records from different stakeholders to improve the effectiveness of the services
(Kumar & Lu, 2010).
5.6 Cloud-WBAN-IoT Integration

To develop a comprehensive monitoring health care system, Cloud Computing can be con-
sidered as the missing technology in the WBAN-IoT integration scenario. In fact, some
people believe that Cloud fills some gaps such as limited storage of WBAN-IoT integration
(Jeffery, 2014).
Health care applications usually generate a vast amount of data that require to be
managed efficiently. The collected information is also required to be analyzed to abstract
technical details while eliminating the need for expertise over the technology. In general,
the entire collected patients’ vital data via a WBAN and information from things (devices)
connected to the Internet, such as recovery, medication, logistics, finance, and even daily
activities, are managed, utilized, and processed through a Cloud Computing system
(Domingo, 2012).
The technology simplifies health care process and enhances QoS by improving the
cooperation among different entities. Moreover, the paradigm helps one effectively to
overcome existing challenges such as privacy and reliability of the applications (Nkosi &
Mekuria, 2010).
Figure 5.3 depicts an example of architecture for Cloud-WBAN-IoT for acquiring and
managing information including biomedical sensor data on Cloud. This monitors general
health situation of patients by collecting and forwarding requested information to Cloud
for further analysis, computational process, and storage. Cloud technology provides this
opportunity to adjust the system’s tasks and performance based on the users’ needs.
Specifically, the framework composes three subsystems that provide real-time monitoring
of patients. Figure 5.3 shows a general overview of the integrated framework:
In the following, we explain the key components of our proposed framework:
1. WBAN-IoT integration module: This module is the lowest layer of a health care
monitoring system composed of two main procedures described in the following:
a. Data collection phase: In this phase, patient’s vital signs such as blood pres-
sure, blood glucose, temperature, and others are monitored using body senor
nodes. There are adequate sensor nodes such as ECG (ElectroCardioGraphy)
and temperature sensor, and so on. These senor devices are able to sense, pro-
cess, and transmit health data (Abawajy et al., 2013; Zhang & Xiao, 2009). There
are also Things that can connect to the Internet to monitor patients’ health situ-
ations. Thus, information such as diet and drug consumptions of patients also
can be collected. In emergency situations, the patients can use specific cameras
WBAN-IoT Cloud-based subsystem End users
er y
Camera Qu Emergency
EEG Drug advice
Cloud Cloud
service platform
ECG Gateway Medical center
EMG
Cloud Database
Access infrastructure
point
Practitioner
Cloud services
Software-as-a-service (SaaS)
Food advice Financial Platform-as-a-service (PaaS)

service
Infrastructure-as-a-service (IaaS)
Medical
Body sensor
database
PDA
FIGURE 5.3
Cloud-WBAN-IoT integration for remote health care monitoring.
to be virtually visited by authorized doctors and discuss about their health

conditions.
b. Transmission phase: In this phase, the monitored data will be transferred via
wireless communication. The data are forwarded to the Cloud data centers for
monitoring and storage. There are several transportation protocols available
for WBAN such as Bluetooth over IEEE 802.15.1, ZigBee over IEEE 802.15.4,
UWB over IEEE 802.15.6 and radio-frequency identification (RFID) (Bhatia
et al., 2007).
2. Cloud-based module: The Cloud subsystem allows patients to submit their per-
sonal and medical information through a gateway. The medical records are acces-
sible by doctors and other health care givers as well as patients from a client-server
web service at any time. Authentication is needed for each time accessing Cloud.
This module provides the three types of Cloud services, namely SaaS, PaaS, and
IaaS. SaaS service offers various applications that are necessary for working with
health data. The PaaS includes tools such as operating system (OS), Database
Management System (DBMS), and virtualization as well as servers, storage, and
networking. Lastly, IaaS provides the physical infrastructure needed for funda-
mental computing recourses in a heterogeneous health care system. Cloud ser-
vice also provides a platform for development, testing, and execution of software
needed by both patients and medical staff. It suits to say that the Cloud provides
the standard interfaces for application integration.
3. End-user module (monitoring phase): This module offers patients with intelligent
monitoring services in a real-time manner. It consists of a variety of applications
that are used for analyzing-recorded health information. This module can also
have different terminals in different buildings such as emergency centers or family
members of patients, based on the service purposes and expectations. End-users

are also able to subscribe to the service to become updated automatically about the
recent transferred medical data and health situations.
5.6.1 Advantages of Cloud-WBAN-IoT Integration

In this section, we are investigating advantages of integrating Cloud Computing, WBAN,
and IoT.
5.6.1.1 Communication
The availability of high-speed networks of WBAN-IoT integration enables effective moni-
toring and control of Things, their coordination, their communications, and real-time access
to the produced data (Rao et al., 2012). In fact, WBAN-IoT is characterized by a very high
heterogeneity of devices, technologies, and protocols. Therefore, scalability, reliability, and
efficiency as well as availability and security need to take into account. The integration
with the Cloud resolves most of these challenges (Suciu et al., 2013). Furthermore, Cloud-
WBAN-IoT paradigm brings the advantage of enhancing data and application sharing.
The integration personalizes ubiquitous applications, whereas it reduces expenses of both
data collection and distribution (Rao et al., 2012). In this integration, Cloud offers an effec-
tive and cheap solution not just to connect any IP enabled devices from any location at any
time it provides opportunities to track and manage them by using customized portals and
built-in applications.
5.6.1.2 Storage
Integration of WBAN-IoT is basically identified by a large amount of information sources
from both biomedical sensors and Things that are required to work together in monitor-
ing systems. Thus, a large amount of data including nonstructured or semistructured are
expected to be produced and transferred (Aguzzi et al., 2014). In some cases, vast variety
types of data in different sizes and frequencies are collected and are required to be stored.
Therefore, a large-scale and long-lived storage is required.
To address the concern, Cloud Computing represents an import Cloud-WBAN-IoT driver
as it provides virtual low cost and on-demand storage capacities. In fact, Cloud Computing
generates new opportunities for data aggregation, integration, and sharing with third par-
ties (Rao et al., 2012). The data then can be directly accessed and visualized from any
places (Zaslavsky et al., 2013). Cloud also offers well-protected technology to protect data
by applying top-level security modules (Rao et al., 2012).
5.6.1.3 Computation
Data processing in the integration of WBAN-IoT needs to be properly satisfied for
performing real-time data analysis, implementing scalable and real time collab-
orative to manage complex events, and supporting task offloading for energy sav-
ing (Yao et al., 2013). However, the integration of WBAN-IoT has limited processing
and energy resources thus; complex and on-site data processing cannot be occurred.
Cloud Computing, however, offers virtually unlimited processing capabilities and an
on-demand usage model. This represents another important Cloud-WBAN-IoT driver
(Rao et al., 2012).
5.6.2 Challenges of Cloud-WBAN-IoT

Integrating Cloud, WBAN, and IoT imposes several challenges for health care applications
that are currently receiving attentions by research communities. In this section, we deeply
investigate the challenges.
5.6.2.1 Communications
Integration of Cloud-WBASN-IoT involves various communications among heterogeneous
devices with different protocols (Zaslavsky et al., 2013). For instance, smartphones that
usually capable of providing a wide range of mobile applications in different domains of
the integrations. This requires several heterogeneous network technologies continually
transferring data from Things such as wearable and implantable biosensors. Therefore,
dealing with this heterogeneity to manage source nodes including the Things and bio-
medical sensor nodes to enhance expected performance of the monitoring system is a
considerable challenge (Botta et al., 2008).
There are wireless network communication protocols, for example, Bluetooth (IEEE
802.15.1) or WiFi (IEEE 802.11a/b/p) technologies that are using for data transmissions.
However, there is still no agreement on the network architecture on managing the Things
in a uniform fashion of a heterogeneous scenario, whereas providing required perfor-
mance still represents an open issue (Botta et al., 2008).
Moreover, continuity in transmission of data increases the overall bandwidth consump-
tion dramatically whereas the current bandwidth limitation cannot support the increasing
trend (Xu et al., 2011). Thus, efficiency of bandwidth access and also its optimization is still
a significant challenge. In addition, in health care applications, providing and acceptable
rate of fault tolerant and reliable continues data transfer from Things to the Cloud is a chal-
lenge (Biswas et al., 2010).
5.6.2.2 Scalability and Flexibility

Cloud-WBAN-IoT requires efficient mechanisms to match collected data and tasks to
appropriate applications and services. As captured information about events and relevant
access services for different applications are distributed timely and geographically, col-
lection and dissemination becomes critical challenges. Therefore, providing flexible sub-
scription schemas and events management, while guaranteeing scalability with respect to
WBAN-IoT integration and users, is still considered an open issue (Le et al., 2010).
5.6.2.3 Security and Privacy

Critical monitoring health care applications requires a very high level of trust in services.
That is to provide a very high level of privacy by authorizing individuals to access specific
information.
In such applications, there is a high possibility of exposing several attacks such as
session riding, SQL injection, cross site scripting, and side-channel (Le et al., 2010).
Multitenancy can also compromise security and lead to sensitive information leakage
(Kapadia et al., 2011). Moreover, data integrity is an important factor that necessarily
needs to be provided from security and privacy-related aspects especially on outsourced
data (Liu et al., 2015).
Therefore, security and privacy of health care services is considered a challenge in

Cloud-WBAN-IoT to cope with different threats from hackers (Li et al., 2013).
5.6.2.4 Heterogeneity
Cloud services usually require resource integration to be efficiently customized for
particular purposes and applications. The requirements become significant in enhanc-
ing applications with multi-Cloud services (Grozev & Buyya, 2014). To achieve the
Cloud customizations and facilitate applications deployment particularly in health
care applications, there are significant challenges such as sophisticated information
selection and fusion mechanisms of heterogeneity to be taken into considerations
(Botta et al., 2016). In fact, a model in Cloud-WBAN-IoT in real-time data collection
process from heterogeneous devices needs to be developed to enhance decision-making
capabilities. Although research efforts have been investigated in this field, maximiz-
ing the intelligence in this context is also still an open challenge (Botta et al., 2016;
Pedersen et al., 2013).
5.6.2.5 Performance
Cloud-WBAN-IoT based health care applications usually require specific performance
and QoS requirements for different aspects such as communication and computation.
However, meeting the expected level of QoS in different scenarios may not be simply
achievable. Generally, obtaining acceptable network performance of the Cloud-WBAN-IoT
integration is a main challenge (Rao et al., 2012) for the applications. That is because health
care scenarios are real-time applications, and they necessarily need to provide services
with extremely high QoS performances. For example, in scenarios using mobility of data
resources, provisioning of data and services need to be performed with high reactivity
(Perera et al., 2014). However, timeliness may be heavily impacted by unpredictability
issues; real-time applications are mainly susceptible to performance challenges (Suciu
et al., 2013).
5.7 Conclusion
This study explored WBANs, IoT, and Cloud Computing technologies individually and
explore advantages as well as challenges in details. This study proved that merging the
three technologies can overcome many of the existing challenges in each individual para-
digm by explaining an example architecture for eHealth monitoring systems. In fact, we
justified that a WBAN-IoT-Cloud paradigm provides a comprehensive health care moni-
toring system with a number of enhancements such as location independency, reliability,
and acceptable security. In fact, it improves cooperation among different requests due to
the capability of processing different applications while sharing resources. However, there
are challenges such as required flexibility and privacy in processing heterogeneous data
need to be considered accurately.
References
Abawajy, J. H., Kelarev, A., & Chowdhury, M. (2013). Multistage approach for clustering and classifi-
cation of ECG data. Computer Methods and Programs in Biomedicine, 112(3), 720–730.
Aguzzi, S., Bradshaw, D., Canning, M., Cansfield, M., Carter, P., Cattaneo, G., & Stevens, R. (2014).
Definition of a research and innovation policy leveraging cloud computing and IoT combina-
tion. European Commission, Directorate-General of Communications Networks, Content & Technology,
Brussels, Belgium.
Alam, S., Chowdhury, M. M., & Noll, J. (2010). Senaas: An event-driven sensor virtualization approach for
internet of things cloud. Paper presented at the Networked Embedded Systems for Enterprise
Applications (NESEA), 2010 IEEE International Conference on.
Ali, M. (2009). Green cloud on the horizon. Paper presented at the IEEE International Conference on
Cloud Computing.
Associati, C. (2011). The evolution of internet of things. Focus. Milão, fev. Retrieved from http://
www.casaleggio.it/pubblicazioni/Focus internet of things v1.81%20-%20eng.pdf (accessed on
August 6, 2015).
Atzori, L., Iera, A., & Morabito, G. (2010). The internet of things: A survey. Computer Networks, 54(15),
2787–2805.
Barbera, M. V., Kosta, S., Mei, A., & Stefa, J. (2013). To offload or not to offload? The bandwidth and energy
costs of mobile cloud computing. Paper presented at the INFOCOM, 2013 Proceedings IEEE.
Bhatia, D., Estevez, L., & Rao, S. (2007). Energy efficient contextual sensing for elderly care. Paper pre-
sented at the Engineering in Medicine and Biology Society, 2007. 29th Annual International
Conference of the IEEE.
Biswas, J., Maniyeri, J., Gopalakrishnan, K., Shue, L., Phua, J. E., Palit, H. N., … Li, X. (2010). Processing
of wearable sensor data on the cloud-a step towards scaling of continuous monitoring of health and well-
being. Paper presented at the 2010 Annual International Conference of the IEEE Engineering in
Medicine and Biology.
Borradaile, G. J. (2013). Statistics of earth science data: Their distribution in time, space and orientation.
Berlin, Germany: Springer Science & Business Media.
Botta, A., de Donato, W., Persico, V., & Pescapé, A. (2016). Integration of cloud computing and inter-
net of things: A survey. Future Generation Computer Systems, 56, 684–700.
Botta, A., Pescapé, A., & Ventre, G. (2008). Quality of service statistics over heterogeneous networks:
Analysis and applications. European Journal of Operational Research, 191(3), 1075–1088.
Chatman, C. (2010). How cloud computing is changing the face of health care information technol-
ogy. Journal of Health Care Compliance, 12(3), 37–70.
Chaves, L. W. F., & Decker, C. (2010). A survey on organic smart labels for the internet-of-things. Paper
presented at the Networked Sensing Systems (INSS), 2010 Seventh International Conference
on, Kassel, Germany.
Chen, M., Gonzalez, S., Vasilakos, A., Cao, H., & Leung, V. C. (2011). Body area networks: A survey.
Mobile Networks and Applications, 16(2), 171–193.
Christensen, J. H. (2009). Using RESTful web-services and cloud computing to create next generation mobile
applications. Paper presented at the Proceedings of the 24th ACM SIGPLAN conference com-
panion on Object oriented programming systems languages and applications, Penn Plaza, NY.
Dash, S. K., Mohapatra, S., & Pattnaik, P. K. (2010). A survey on applications of wireless sensor net-
work using cloud computing. International Journal of Computer science & Engineering Technologies,
1(4), 50–55.
Dohr, A., Modre-Opsrian, R., Drobics, M., Hayn, D., & Schreier, G. (2010). The internet of things
for ambient assisted living. Paper presented at the 2010 Seventh International Conference on
Information Technology, IEEE.
Dohr, A., Modre-Osprian, R., Drobics, M., Hayn, D., & Schreier, G. (2010). The internet of things for
ambient assisted living. Information Technology: New Generations, 10, 804–809.
Domingo, M. C. (2012). An overview of the Internet of Things for people with disabilities. Journal of
Network and Computer Applications, 35(2), 584–596.
Fei, D.-Y., Zhao, X., Boanca, C., Hughes, E., Bai, O., Merrell, R., & Rafiq, A. (2010). A biomedical sensor
system for real-time monitoring of astronauts’ physiological parameters during extra-vehicular
activities. Computers in Biology and Medicine, 40(7), 635–642.
Fortino, G., Parisi, D., Pirrone, V., & Di Fatta, G. (2014). BodyCloud: A SaaS approach for community
body sensor networks. Future Generation Computer Systems, 35, 62–79.
Fox, G. C., Kamburugamuve, S., & Hartman, R. D. (2012). Architecture and measured characteristics of a
cloud based internet of things. Paper presented at International Conference on the Collaboration
Technologies and Systems (CTS), Denver, CO.
Ghanavati, S. (2015). Congestion control mechanism for sensor-cloud Infrastructure. Retrieved from
http://dro.deakin.edu.au/view/DU:30083703.
Ghanavati, S., Abawaji, J., & Izadi, D. (2015). A congestion control scheme based on fuzzy logic in wireless
body area networks. Paper presented at IEEE 14th International Symposium on the Network
Computing and Applications (NCA).
Ghanavati, S., Abawajy, J., & Izadi, D. (2016a). An alternative sensor cloud architecture for vital signs mon-
itoring. Paper presented at the Neural Networks (IJCNN), 2016 International Joint Conference
on, IEEE.
Ghanavati, S., Abawajy, J., & Izadi, D. (2016b). ECG rate control scheme in pervasive health care mon-
itoring system. Paper presented at the Fuzzy Systems (FUZZ-IEEE), 2016 IEEE International
Conference on.
Grozev, N., & Buyya, R. (2014). Inter-Cloud architectures and application brokering: Taxonomy and
survey. Software: Practice and Experience, 44(3), 369–390.
Hanson, M. A., Powell Jr, H. C., Barth, A. T., Ringgenberg, K., Calhoun, B. H., Aylor, J. H., & Lach, J.
(2009). Body area sensor networks: Challenges and opportunities. Computer, 42(1), 58.
Hoang, D. B., & Chen, L. (2010). Mobile cloud for assistive healthcare (MoCAsH). Paper presented at the
Services Computing Conference (APSCC), IEEE Asia-Pacific.
INFSO, E. (2008). Internet of Things in 2020: Roadmap for the Future. INFSO D, 4.
Islam, S. R., Kwak, D., Kabir, M. H., Hossain, M., & Kwak, K.-S. (2015). The internet of things for
health care: A comprehensive survey. IEEE Access, 3, 678–708.
Jara, A. J., Belchi, F. J., Alcolea, A. F., Santa, J., Zamora-Izquierdo, M. A., & Gómez-Skarmeta, A. F.
(2010). A Pharmaceutical Intelligent Information System to detect allergies and Adverse Drugs Reactions
based on internet of things. Paper presented at the Pervasive Computing and Communications
Workshops (PERCOM Workshops), 8th IEEE International Conference on.
Jeffery, K. (2014). Keynote: CLOUDs: A large virtualisation of small things. Paper presented at the 2nd
International Conference on Future Internet of Things and Cloud (FiCloud-2014).
Kapadia, A., Myers, S., Wang, X., & Fox, G. (2011). Toward securing sensor clouds. Paper presented at the
Collaboration Technologies and Systems (CTS), International Conference on, IEEE.
Klasnja, P., & Pratt, W. (2012). Healthcare in the pocket: Mapping the space of mobile-phone health
interventions. Journal of Biomedical Informatics, 45(1), 184–198.
Koop, E., Mosher, R., Kun, L., Geiling, J., Grigg, E., Long, S., & Rosen, J. M. (2008). Future delivery of
health care: Cybercare. Engineering in Medicine and Biology Magazine, IEEE, 27(6), 29–38.
Kumar, K., & Lu, Y.-H. (2010). Cloud computing for mobile users: Can offloading computation save
energy? Computer, 43(4), 51–56.
Latré, B. (2008). Reliable and energy efficient network protocols for wireless body area networks. Belgium:
Ghent University.
Latré, B., Braem, B., Moerman, I., Blondia, C., & Demeester, P. (2011). A survey on wireless body area
networks. Wireless Networks, 17(1), 1–18.
Le, X. H., Lee, S., True, P. T. H., Khattak, A. M., Han, M., Hung, D. V., Lee, & Y.-K. (2010). Secured WSN-
integrated cloud computing for u-life care. Paper presented at the Proceedings of the 7th IEEE
conference on Consumer communications and networking conference.
Li, B., Wang, Q., Yang, Y., & Wang, J. (2006). Optimal distribution of redundant sensor nodes for wire-
less sensor networks. Paper presented at the Industrial Informatics, 2006 IEEE International
Conference on.
Li, H., Guo, C.-Y., & Xu, C.-L. (2015). A highly sensitive non-enzymatic glucose sensor based on bime-
tallic Cu-Ag superstructures. Biosensors and Bioelectronics, 63, 339–346.
Li, W., Zhong, Y., Wang, X., & Cao, Y. (2013). Resource virtualization and service selection in cloud
logistics. Journal of Network and Computer Applications, 36(6), 1696–1704.
Liao, Y.-T., Yao, H., Lingley, A., Parviz, B., & Otis, B. P. (2012). A 3-CMOS glucose sensor for wireless
contact-lens tear glucose monitoring. IEEE Journal of Solid-State Circuits, 47(1), 335–344.
Liu, C., Yang, C., Zhang, X., & Chen, J. (2015). External integrity verification for outsourced big data
in cloud and IoT: A big picture. Future Generation Computer Systems, 49, 58–67.
Liu, C., Zhu, Q., Holroyd, K. A., & Seng, E. K. (2011). Status and trends of mobile-health applications
for iOS devices: A developer's perspective. Journal of Systems and Software, 84(11), 2022–2033.
Löhr, H., Sadeghi, A.-R., & Winandy, M. (2010). Securing the e-health cloud. Paper presented at the
Proceedings of the 1st ACM International Health Informatics Symposium, ACM: New York.
Misra, S., & Chatterjee, S. (2014). Social choice considerations in cloud-assisted WBAN architecture
for post-disaster healthcare: Data aggregation and channelization. Information Sciences, 284,
95–117.
Natalia, O., & Victor, O. (2006). Computer networks, principles, technologies and protocols for net-
work design. Hoboken, NJ: John Wiley & Sons.
Nemati, E., Deen, M. J., & Mondal, T. (2012). A wireless wearable ECG sensor for long-term applica-
tions. Communications Magazine, IEEE, 50(1), 36–43.
Nkosi, M., & Mekuria, F. (2010). Cloud computing for enhanced mobile health applications. Paper presented
at the Cloud Computing Technology and Science (CloudCom), IEEE Second International
Conference on.
Oberheide, J., Veeraraghavan, K., Cooke, E., Flinn, J., & Jahanian, F. (2008). Virtualized in-cloud secu-
rity services for mobile devices. Paper presented at the Proceedings of the First Workshop on
Virtualization in Mobile Computing, ACM: New York.
Pang, Z. (2013). Technologies and architectures of the Internet-of-Things (IoT) for health and well-being.
Patel, M., & Wang, J. (2010). Applications, challenges, and prospective in emerging body area net-
working technologies. Wireless Communications, IEEE, 17(1), 80–88.
Patel, P., Jardosh, S., Chaudhary, S., & Ranjan, P. (2009). Context aware middleware architecture for wire-
less sensor network. Paper presented at the Services Computing, IEEE International Conference
on.
Pedersen, T. B., Pedersen, D., & Riis, K. (2013). On-demand multidimensional data integration:
Toward a semantic foundation for cloud intelligence. The Journal of Supercomputing, 65(1),
217–257.
Perera, C., Zaslavsky, A., Christen, P., & Georgakopoulos, D. (2014). Sensing as a service model
for smart cities supported by internet of things. Transactions on Emerging Telecommunications
Technologies, 25(1), 81–93.
Rao, B., Saluia, P., Sharma, N., Mittal, A., & Sharma, S. (2012). Cloud computing for Internet of Things &
sensing based applications. Paper presented at the Sensing Technology (ICST), Sixth International
Conference on, IEEE.
Rohokale, V. M., Prasad, N. R., & Prasad, R. (2011). A cooperative Internet of Things (IoT) for rural
healthcare monitoring and control. Paper presented at the Wireless Communication, Vehicular
Technology, Information Theory and Aerospace & Electronic Systems Technology (Wireless
VITAE), 2nd International Conference on, IEEE.
Rosenthal, A., Mork, P., Li, M. H., Stanford, J., Koester, D., & Reynolds, P. (2010). Cloud computing: A
new business paradigm for biomedical information sharing. Journal of Biomedical Informatics,
43(2), 342–353.
Schweitzer, E. J. (2012). Reconciliation of the cloud computing model with US federal electronic
health record regulations. Journal of the American Medical Informatics Association, 19(2), 161–165.
Shahamabadi, M. S., Ali, B. B. M., Varahram, P., & Jara, A. J. (2013). A network mobility solution based
on 6LoWPAN hospital wireless sensor network (NEMO-HWSN). Paper presented at the Innovative
Mobile and Internet Services in Ubiquitous Computing (IMIS), Seventh International
Conference on.
Steinfeld, L., Ritt, M., Silveira, F., & Carro, L. (2015). Optimum design of a banked memory with
power management for wireless sensor networks. Wireless Networks, 21(1), 81–94.
Suciu, G., Vulpe, A., Halunga, S., Fratu, O., Todoran, G., & Suciu, V. (2013). Smart cities built on resil-
ient cloud computing and secure Internet of Things. Paper presented at the 19th International
Conference on Control Systems and Computer Science.
Sullivan, T. J., Deiss, S. R., & Cauwenberghs, G. (2007). A low-noise, non-contact EEG/ECG sensor. Paper
presented at the Biomedical Circuits and Systems Conference, IEEE.
Sultan, N. (2014). Making use of cloud computing for healthcare provision: Opportunities and chal-
lenges. International Journal of Information Management, 34(2), 177–184.
Sundmaeker, H., Guillemin, P., Friess, P., & Woelfflé, S. (2010). Vision and challenges for realising the
Internet of Things (Vol. 20). EUR-OP, Brussels, Germany.
Vartiainen, E., & Väänänen-Vainio-Mattila, K. (2010). User experience of mobile photo sharing in the
cloud. Paper presented at the Proceedings of the 9th International Conference on Mobile and
Ubiquitous Multimedia, New York.
Wang, W., Li, J., Wang, L., & Zhao, W. (2011). The internet of things for resident health information ser-
vice platform research. Paper presented at the Communication Technology and Application, IET
International Conference on, IEEE.
Wang, Y.-W., Yu, H.-L., & Li, Y. (2011). Notice of retraction internet of things technology applied in medi-
cal information. Paper presented at the Consumer Electronics, Communications and Networks
(CECNet), International Conference on, IEEE.
Xu, Y., Helal, S., & Scmalz, M. (2011). Optimizing push/pull envelopes for energy-efficient cloud-sensor
systems. Paper presented at the Proceedings of the 14th ACM international conference on
Modeling, analysis and simulation of wireless and mobile systems, Gainesville, FL.
Yang, G., Xie, L., Mäntysalo, M., Zhou, X., Pang, Z., Da Xu, L., Zheng, L.-R. (2014). A health-IoT plat-
form based on the integration of intelligent packaging, unobtrusive bio-sensor, and intelligent
medicine box. IEEE Transactions on Industrial Informatics, 10(4), 2180–2191.
Yao, D., Yu, C., Jin, H., & Zhou, J. (2013). Energy efficient task scheduling in mobile cloud computing.
Network and parallel computing (pp. 344–355). Springer: New York.
You, L., Liu, C., & Tong, S. (2011). Community medical network (cmn): Architecture and implementation.
Paper presented at the Mobile Congress (GMC), IEEE.
Zaslavsky, A., Perera, C., & Georgakopoulos, D. (2013). Sensing as a service and big data. arXiv pre-
print arXiv:1301.0159.
Zhang, Y., & Xiao, H. (2009). Bluetooth-based sensor networks for remotely monitoring the physi-
ological signals of a patient. Information Technology in Biomedicine, IEEE Transactions on, 13(6),
1040–1048.
Section II
Frameworks for the

Internet of Things
An Architectural Perspective
6
IoT Framework Based on WSN Infrastructure
with Different Topological Distributions
Marcel Stefan Wagner, Diogo Ferreira Lima Filho,

Miguel Arjona Ramírez, and Wagner Luiz Zucchi
CONTENTS
6.1 Introduction ........................................................................................................................ 121
6.2 General Sensor Description.............................................................................................. 122
6.2.1 Types of Sensors and Applications ..................................................................... 124
6.3 Node Interface Technologies ............................................................................................ 125
6.3.1 Connectivity Approach for IoT ............................................................................ 127
6.3.2 Data Sensing and Acquisition .............................................................................. 130
6.3.3 Data Transmission ................................................................................................. 133
6.4 Routing ................................................................................................................................ 134
6.4.1 Current Challenges in Routing ............................................................................ 134
6.4.2 Fundamentals of the TCNet Architecture ......................................................... 136
6.5 Data Management .............................................................................................................. 138
6.6 User Applications............................................................................................................... 142
6.7 Smart City with WSN ........................................................................................................ 144
6.8 Conclusion .......................................................................................................................... 147
References..................................................................................................................................... 148
6.1 Introduction
The internet of thinks (IoT) is based on the idea that devices, objects, and living beings have
interaction within the environment to achieve a particular purpose. These elements may
be distributed and connected across the network. The environment can comprehend dif-
ferent areas, such as terrestrial, air, space, underwater, or underground. The network can
be designed to work in different topologies and the network nodes can be fixed or mobile.
It is possible to provide an environment with monitoring and control using sensors and
actuators for the interactions amongst the different IoT network mechanisms. The use of
sensors helps the network to collect data about environment features. The use of actua-
tors serves to modify the environment or to carry out alterations in order to meet a main
purpose, according to the readings taken from the sensors and in accordance with the
network policy.
A wireless sensor network (WSN) consists basically of autonomous sensors distrib-
uted in an environment with specific addressing number for monitoring physical or
environmental conditions such as temperature, pressure, lightness, proximity, and
121
movement, among others. Another important activity of a node is the relaying or routing

of information received from other nodes.
Usually, the data collected by the sensors are sent over the network to a central loca-
tion, which is responsible for processing the information. The most modern networks have
nodes with high processing power and these nodes can operate with bidirectional com-
munication. In this scenario, they can also control sensor activity and manage devices
optimally.
As well as in many other branches of technology, initially the development of sensor net-
works was motivated by military applications regarding data collection. Currently, these
networks are also used for monitoring industrial processes, hospital management, com-
mercial issues, traceability, and other applications in various fields of activity. The nodes
may have different dimensions and because of the progress in nanotechnology, there are
chips or integrated circuits (ICs) of reduced dimensions, maintaining the same or provid-
ing increased capability processing.
6.2 General Sensor Description

Each IoT node normally is comprised by a power supply, which is a battery or an energy
harvesting module, an electronic circuit for interfacing sensors, microcontroller or micro-
processor and radio transmitter with an internal or external antenna. In Figure 6.1, the
blocks refer to the sensor internal modules for medium reading and actuation. From the
left to the right: analog-to-digital converter is responsible for the analog to digital conver-
sion (it can be represented by pulse width modulation) to a low-power microcontroller
or microprocessor. In Figure 6.1, around the microprocessor, there is a memory group
composed by random access memory/read-only memory for data processing and stor-
age, radio frequency (RF) transceiver for the chosen operating band as can be seen in
Section 6.4. In the reception (RX) path, the sensor node receives data from other sources to
perform some tasks in the network, such as setting an actuator to turn on a motor, valve,
or gate. It happens by a command from the microcontroller to provide the actuators with
digital or analog data.
Wireless sensor node

Memory
(RAM/ROM)
Sensor ADC
Microcontroller/
RF
microprocessor
transceiver
Actuator (ultra low power)
Actuator DAC
Node battery
FIGURE 6.1
Basic structure in blocks of a sensor node.
IoT Framework Based on WSN Infrastructure with Different Topological Distributions 123
Sensor node
APP
NET
BAT
QUE
MAC
PM PHY MOB
Air interface
to_CH pos_out
from_CH
FIGURE 6.2
Basic internal structure of a wireless sensor node.
As Sohraby et al. (2007) pointed, the cost of the sensors varies according to the applica-
tion and size, that is, it depends on the environment in which they will be used, includ-
ing climatic conditions. For instance, it may be important to know whether the system is
vulnerable to rain, if what the water level is, and whether the temperature has reached
a critical value. It should be considered for sensor nodes, resource constraints, such as
energy, memory, computational speed, and communication bandwidth (BW).
According to Dargie and Poellabauer (2010), the topology of sensor networks can vary
from a simple network with few static nodes till an advanced WSN with mesh distribu-
tion and multihop feature, including mobility. The propagation technique used among
the receivers, which influences the number of hops to the destination depending on the
application purpose, can be done by forwarding packets or broadcasting information to
network nodes.
In Figure 6.2, node mobility is assigned to the MOB module (related to the mobility) and
the backbone communication with the environment is assigned to PHY, which interfaces
with the channel (to_CH/from_CH) and outputs a position vector (pos_OUT). There is an
internal flow, in which APP corresponds to the application, NET to the network layer, QUE
to the data queue (if necessary it may perform part of the node internal processing), ACC
to network access layer, and PHY to the physical layer. Although in the internal structure
of the sensor, the BAT is related to energy storage unit and PM to the power management.
It is through the PHY layer that the node interacts with the network, generating the
transmission (TX) and RX of data. This layer is directly related to the RF transceiver block
of Figure 6.1.
The sensor functions are associated with the application and are suited to the type of
environment to be monitored, analyzed, and managed. Sensors use physical entities that
provide the electronic circuits with data collected from the environment. To provide it,
the sensors can use changes in resistance, capacitance, inductance, frequency, amplitude,
phase, and signal intensity.
The connectivity with sensors can be implemented using microcontrollers with TX

connection interfaces. These connections are essentially based on the communication
technology chosen, such as Ethernet, Fiber Optic, and Twisted Pairs for wired connec-
tions, and Infrared (IR), Bluetooth (BT), Wireless-Fidelity (Wi-Fi), ZigBee, radio-frequency
identification (RFID), near-field communication (NFC), LoRa, 3G, 4G, and 5G for wireless
communications.
6.2.1 Types of Sensors and Applications

The sensors can be classified into different groups for measurement or environment
reading according to the application, such as: electric current, potential, magnetic, radio,
humidity, fluid velocity, liquid flow, pressure, thermal, heat, temperature, gas, proximity,
optical, position, chemical, and others.
A general vision about possible types of sensors that can be used in the smart city sce-
narios is provided as follows:
• Speed: Speed is usually used for detecting speed of an object (or vehicle) and can be
implemented with different types of sensors, such as speedometers, wheel speed
sensors, ground speed, Doppler radar, and air speed indicators.
• Temperature: It gives the temperature measurement as an electrical signal in the form
of electrical voltage, which, in turn, is proportional to the temperature value. There
are different types of sensors to do it, classified as contact type and noncontact type,
with subclasses as mechanical (thermometer and bimetal) and electrical (thermistor,
thermostat, thermocouple, resistance thermometer, and silicon band gap).
• Presence: Passive IR (PIR) sensors are used for motion detection and are said to
be passive because they do not emit any energy or radiation for sensing motion.
These sensors are classified based on the angle over which they can detect motion
of the objects in the monitored environment, such as 120°, 180°, and 360°.
• Distance: The basic principle is similar to sonar, in which the reading of echoes
produced by a primary sound wave or RF is used to estimate the attributes of a
target, such as the distance of the object from the source. A transducer is used
for converting energy into ultrasound waves within bands above human hearing,
typically around 40 kHz.
• Humidity: It actuates in the measurement of humidity in the environment moni-
tored. They can convert humidity measures into electric voltage. Its measurement
is determined by the amount of water vapor present in a specific gas that can be
a mixture, such as air or a pure gas (nitrogen). Most commonly used units are:
Relative Humidity that is a relative measurement and a function of temperature,
Dew/Frost point as a function of the gas pressure but independent of the tempera-
ture (is also defined as an absolute humidity measurement), and parts per million
that corresponds to an absolute measurement unit.
• Gas and smoke: The detector, flammable and smoke/smother gas sensor is able to
detect combustible gases and smoke concentrations in the air. This sensor can
detect gases of different types, including smoke, liquefied petroleum gas, butane,
propane, methane, hydrogen, alcohol, and natural gas.
• Luminosity: Light Dependent Resistor (LDR) is a component whose resistance
varies with the intensity of the light. The more light falls on the component, the
lower the resistance is. These sensors are recommended for external use, which
through photocells recognize the decrease in light and trigger the system.
• Level: It operates as a power switch which can trigger keys, pumps, and lamps
or send a signal to a microcontroller. For liquids, they can be considered as level
switches in the type of magnetic float.
• Accelerometers: They can operate with different physical effects; therefore, they have
a wide range of acceleration values they are capable of measuring. These devices
are mainly used in positioning systems, tilt, and vibration sensors. A well-known
application of accelerometers is in mobile phone screens that adjust according to
the angle they make by means of g-forces or proper accelerations, which are accel-
erations due to reactions to the pull of gravity or other equivalent forces.
• Gyroscopes: These are devices consisting of a rotor suspended by a support formed
by two articulated circles. Its operation is based on the principle of inertia and
the rotating shaft has a memory effect, eliminating the need for a geographical
coordinate.
The use of the sensors described above can be made in conjunction and applied as parts
of the IoT nodes. They work as arms for data acquisition from the monitored environment,
to be allowed in microcontrollers or microprocessors for the decision-making process
(enabling or not the actuators).
6.3 Node Interface Technologies

To understand the nature and location of the interfaces between nodes, a short descrip-
tion about the technologies involved in IoT environments is presented ahead, based on
Wagner (2009) work. To visualize the differences among the various wireless TX technolo-
gies, Table 6.1 shows the application, coverage area, and operating frequency of the main
technologies in the telecommunications market.
BT is a wireless communication technology for short distances, designated to replace the
cables that connect many types of local devices, such as mobile phones, headsets, moni-
tors, audio, and medical equipment, among others. The energy–efficiency ratio makes it
ideal for use in devices that need to work from a small battery charge for long periods.
IR is a form of invisible electromagnetic radiation. The major advantages of IR are related
to its immunity to interference and the possibility of being transmitted and received with
low-cost and easy-to-operate devices.
The Wi-Fi term is a registered trademark of the Wi-Fi Alliance. The expression has
become a synonym for IEEE 802.11 technology, which allows the connection between dif-
ferent wireless devices. The range of a Wi-Fi network is variable, depending on the router
and the antenna used. This distance can vary from 100 m (indoor) up to 300 m (outdoor).
LoRa is a wireless technology developed to operate in the Low-Power Wide-Area
Networks (LPWANs). It seems to be an important tool for machine-to-machine (M2M)
and IoT applications, because the technology offers low power consumption, long range,
and secure data TX, so it also can be connected with existing cellular/wireless networks.
According to the LoRa Alliance, it is easy to plug in the LoRa to the existing cellular
infrastructure and also allow them to offer a solution for more low-power IoT applications.
TABLE 6.1
Comparison among Various Wireless Transmission Technologies
Technology Standard Application Coverage Area (m)a Frequency (Hz)a
RFID ISO/IEC 14443 and ISO/ WPAN 100 (active)/25 (passive) 13.56 MHz
IEC 18092
IR RECS-80, RC-5, CEA-931-A WPAN Up to 1.5 (targeted 35–41 kHz
and CEA-931-B direct)
NFC ISO/IEC 14443 A&B and WPAN 0.2 (contact) 13.56 MHz
JIS-X 6319-4
UWB IEEE 802.15.3a WPAN About 10 3.1–10.6 GHz
Bluetooth IEEE 802.15 WPAN 1 (class 3)/10 (class 2)/ 2.4 GHz
100 (class 1)
Wi-Fi IEEE 802.11a WLAN 25–100 (indoor) 5 GHz
Wi-Fi IEEE 802.11b WLAN 100–150 (indoor) 2.4 GHz
Wi-Fi IEEE 802.11g WLAN 100–150 (indoor) 2.4 GHz
Wi-Fi IEEE 802.11n WLAN 70 (indoor)/ 2.4 or 5 GHz
300 (outdoor)
ZigBee IEEE 802.15.4 WPAN/WLAN 10–100 0.8, 0.9, or 2.4 GHz
WiMax IEEE 802.16d WMAN 6400–9600 11 GHz
WiMax IEEE 802.16e Mobile/WMAN 1600–4800 2–6 GHz
LoRa Proprietary—Semtech LPWAN Greater than 15000 Sub-GHz band
(SX127X product family)
WCDMA 3GPP 3G WWAN 1600–8000 1.8, 1.9, or 2.1 GHz
CDMA2000 3GPP 3G WWAN 1600–8000 0.4, 0.8, or 0.9 GHz
1xEV-DO 3GPP 3G WWAN 1600–8000 1.7, 1.8, 1.9, or
2.1 GHz
LTE 3GPP 4G WWAN About 15000 0.7–2.6 GHz
MBWA IEEE 802.20 WWAN 4000–12000 3.5 GHz
a Data collected according to Brazil information and/or specification.
RFID is an automatic identification method that uses radio signals, retrieving and stor-
ing data remotely using devices called RFID tags. This tag generally is a small object that
can be placed on a person, animal, or things such as equipment, packaging, product, and
others. It contains silicon chips and antennas that allow transponders to answer radio sig-
nals sent by a transmitter (passive tags) and there are labels equipped with battery which
allows them to send the signal itself (active tags). It is an alternative to bar codes, enabling
product identification at a short distance from the scanner or self-off position.
NFC is a technology that allows the exchange of information wirelessly and securely
between compatible devices that are close to each other. Some examples are mobile phones,
tablets, badges, electronic ticket cards, and any other device that has an NFC chip inserted.
When the devices are sufficiently close, the communication is automatically established
without the need for additional configuration.
ZigBee corresponds to a set of specifications that are intended as wireless communica-
tion between electronic devices. It is based on features of low power operation, reasonable
data transfer rate, and low cost for implementation. This set of specifications can be corre-
lated with the Open System Interconnection (OSI) layers, as established by the IEEE 802.15.4
standard. The network was designed to connect small units to collect local data, together
with its control, using part of the radio spectrum. In Brazil, there is no need for licensing
this RF use; however, the devices must be approved by the National Telecommunications
Agency (Anatel) according with Brazil (2008).
The Inter-IC connection is used to communicate data with fewer wires, interfacing with
a lot of devices, such as displays, cameras, and others. This protocol works with Serial Data
and Serial Clock for bidirectional data transfer.
The 3G technology can be used to connect IoT nodes in a WAN manner, using high data
TX. With this connection scheme, data collected from end sensors can be delivered to data
centers or a respective node in a wide area, interfacing WSN with cellular networks. The
Wide-Band Code-Division Multiple Access is the air interface technology applied. For 4G,
two standards are attractive: WiMax and Long-Term Evolution. Both of them have high-
data transfer performance in the end-to-end TX. The interconnection between WSN with
other devices or WSN can be done using the approach WSN-cellular-end/WSN.
Global positioning system refers to a satellite service as the reference for location sys-
tems. It can operate with other services and devices to deliver a vast array of customized
applications, such as in agricultural machinery, security applications, and traceability.
5G technology is a promise for very high data transfer (around 20 Gbps) with connectiv-
ity among things, network, and cloud. This theme is yet a subject of extensive discussion
in the academic and market areas, providing scenarios for the future of communications
and IoT applications.
According with the range of presented technologies as possibility to IoT environments,
the technology chosen for describing in this work, at this section for a WSN scenario was
the ZigBee communication standard.
6.3.1 Connectivity Approach for IoT

Silva et al. (2014) pointed to the use of MICA motes sensors (MICAz) as a ZigBee use
solution, because of features as energy consumption, support to wireless communication,
and increased storage capacity. These motes are standardized by IEEE (2011) on the IEEE
802.15.4 at a 2.4 GHz and BW of approximately 250 kbps, ensuring low cost, low power,
and reliability.
Table 6.2 shows the description of the MICA motes sensor family features, linking the
information necessary to select the sensor to be used in the WSN. Tiny OS is an operat-
ing system aimed at events that was developed for use in sensor nodes. It was designed
to withstand intensive competition operations to access the channel that are common in
sensor networks and use few hardware resources. Against this background, it is possible
that a node not only acts on the environment phenomena detection, but also as a router,
actuator, and handle cases relating to the application, serving in this way, as a node for
IoT solutions.
Figure 6.3 shows the basic internal interaction of a generic node applied for the network
environment. This node is provided with cognitive processes as a solution proposed by
Wagner (2016) for IoT in the cognitive sensor networks scenario. Actually, nodes are pro-
vided with more processing power, higher speed, and storage, so its approach seems to be
a great solution for the management of IoT networks.
In the air interface of WSN, it should be considered the fading effect applied over the
signals transmitted for communication between nodes. In this scenario, for free space con-
dition, only part of the energy transmitted through the electromagnetic waves is captured
by the receiving antenna. This energy decreases with frequency and distance. This loss is
called loss in free space.
L = 32.5 + 20 log(d) + 20 log( f ) (6.1)
TABLE 6.2
MICA Motes Sensors Features
MICA Mote Sensors
Features MICA Mote MICA2 Mote MICAz Mote
Model MOT300–MPR300CA MICA2–MPR400CB MICAz–MPR2400CA
Radio transceiver TR1000 CC1000 CC2420–IEEE 802.15.4
Frequency 916 MHz 868/916 MHz, 433 MHz 2.4–2.48 GHz
(MPR410CB) or 315 MHz
(MPR420CB)
Transmission mode OOK (10 kbps) or 38.4 kbaud in Manchester 250 kbps in High Data Rate
rate ASK (115 kbps), with max Radio
50 kbps
Processor type Microcontroller Atmel Microcontroller Atmel Microcontroller Atmel
Atmega 103. AT90LS8535 Atmega128L, In Atmega128L. The radio
uses 0.75 mW, with 12 mA transmission 27 mA, MICAz (MPR2400), with
to transmission and reception 10 mA and in hardware security (AES-128).
1.8 mA to receipt Sleep <1 µA Uses 8 mA in Active mode
and less than 15 µA in Sleep
Communication Radio transceiver RF in Radio transceiver RF in Radio transceiver RF in 2.4
916 MHz, for 10 kbps. For 916 MHz. Transmission GHz. The external range is
transmission uses 36 mJ/s power between 0.01 and 75–100 m and for internal
and reception about 3.16 mW, in the range of purpose of 20–30 m
5.4 mJ/s–14.4 mJ/s. Range 5–91 m
approx. 20 m
Operational TinyOS. 178 bytes of TinyOS. 178 bytes of TinyOS. 178 bytes of memory
System memory memory
Memory 128 kB of programmable 128 kB of programmable 128 kB of programmable
memory, 512 kB SRAM for memory, 512 kB of flash memory, 512 kB of flash and
data and 32 kB for and 4 kB of EEPROM for 4 kB of EEPROM for
EEPROM configuration configuration.
Dimension 7.62 × 2.54 × 1.27 cm 5.71 × 3.17 × 0.63 cm. The 5.8 × 3.2 × 0.7 cm. The mote
mote weight is 18 g weight is 18 g
Enabled type of Light, temperature, Light, temperature, Light, temperature,
sensors accelerometer, noise, and accelerometer, noise, accelerometer, noise,
seismic magnetometer, humidity, magnetometer, humidity,
and barometric pressure barometric pressure, PH,
acoustic, magnetic, and other
MEMSIC sensors
Battery Utilizes 2 batteries AA 3V. Utilizes 2 batteries AA Utilizes 2 batteries AA
Can operate about working in 2.7–3.3V. Can working in 2.7–3.3 V. Can
approximately 1 year in operate about operate about approximately
low consume mode approximately 1 year in 1 year in low consume mode
low consume mode
Application WSN with routing capacity WSN with routing WSN with routing capacity in
in large scale networks capacity in large scale large scale networks (+1000
(+1000 nodes), Ad Hoc networks (+1000 nodes), nodes), Ad Hoc topology
topology covering and Ad Hoc topology covering, mesh networking
monitoring covering and monitoring and monitoring
Node WSN/CSN/IoT
1
2 Microprocessor Application
3 (processes decision) (User, M2M, ...)
4
Microcontroller (processing data)
Memory (RAM/ROM)
Radiofrequency
(TX/RX)
Actuator Sensor
FIGURE 6.3
Basic internal block interaction of a generic node applied for the network environment.
where d is the distance in kilometers and f the frequency in MHz.

For selective fading, the Rayleigh, Rice, Log-normal, and other approaches can be
applied. However, in general, environments for IoT implies in a multipath signal distribu-
tion; therefore, the Rayleigh fading is the approach chosen for explanation, because this
effect can possibly impair the connection between IoT nodes depending on the distance
of these end-to-end nodes or on the distance to the intermediate nodes in a multihop
communication.
As mentioned, the antenna’s range signal is directly linked to the fading effect in the TX
medium, that is, the greater the distance between the nodes, the higher the signal decrease
as the curve of the Rayleigh Probability Density Function, which is directly related to dis-
tance from the source node, representing:
R ∼ Rayleigh (σ) (6.2)
where σ is the scaling factor for the fading effect. The Rayleigh Probability Density
Function of random variable x is
 − x 2 
x  2
f ( x|σ ) = 2 e  2 σ  (6.3)
σ
where f(x|σ) is the function that can be understood as an amplitude feature of the signal
transmitted in the WSN environment with fading effect.
Figure 6.4 shows the sequence for 8σ simulations. It can be noted from the figure, that
the higher is the σ value, the softer becomes the curve of the function. The general use
approach in air interface for WSN is σ = 1; therefore, thinking in the IoT where the nodes
will collect data from the environments events or in a periodic data acquisition for moni-
toring, it is reasonable to establish a threshold for the RX warranty, determining here the
QoS for the service applied in delivering data from the source node to the destiny.
The generally applied topology for IoT scenarios include infrastructured distribution
(it indicates that all group of nodes are under a direct control of a unique coordination
function), Ad Hoc [also calling as independent mode, it indicates that devices can
communicate directly among them without the necessity for an access point (AP)], and
star (this topology generally is applied to LPWAN connectivity, where the devices are
connected directly with an AP).
1.4
Sigma = 0.5
Sigma = 1.0
Sigma = 1.5
1.2
PDF with amplitude signal of f(x,σ)
Sigma = 2.0
Sigma = 2.5
Sigma = 3.0
1 Sigma = 3.5
0.8
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7 8 9 10
x (m)
FIGURE 6.4
Probability density function (PDF) of Rayleigh functions.
FIGURE 6.5
A simple scheme representation of connection among nodes in an Ad Hoc topology scenario.
Figure 6.5 shows a simple connection scheme in an Ad Hoc topology scenario, where the
source (S) is in contact with the destination (D) node using a route determined by the range
of each node’s antenna, to the correct signal RX and re-TX between intermediate nodes.
Therefore, the data is wirelessly delivered and the network connection is remade when the
distance between nodes changes or if they lose their connection routes.
6.3.2 Data Sensing and Acquisition

The data acquisition is done by sensors present in the IoT nodes. According to Dias (2016),
the sensors are necessary for IoT systems to convert the features or conditions of an envi-
ronment, such as temperature, movement, and humidity in another element that can be
readable for microcontroller processing, that is, converting the measurements in digital
signals. These data will be carried to the application, in which it will be understood as
information.
Processing
task
Sensing Actuator
Knowledge
Environment
FIGURE 6.6
Node interaction when collecting data from the environment monitored.
A node is comprised by one or more sensors and transducers. The transducers are
responsible by the conversion of a primary energy form in another one. These devices can
transform any type of energy in electrical. The transducer can operate as an actuator in the
monitoring environment.
In Figure 6.6, it can be seen that a node collect data from the environment monitored,
and this cycle is executed in repetition according to the available sensing tools, such as the
types of sensors used by node on the desired application.
As mentioned in previous sections, the data acquisition is done by sensors measuring the
environment’s phenomena and converting these values into electrical signals to carry it for
the microcontroller’s inputs, that in turn, transform these receipted signals in the informa-
tion necessary to be, thus, forwarded to other network nodes or treated into the own node.
The physical medium corresponds to the environment where the sensors are inserted
for the management and control tasks. There are many types of sensors, such as capacitive,
inductive, and photoelectric, as described below.
The capacitive sensor (or proximity detector with capacitive effect) enables noncontact
detection and linear measuring for small displacements (it happens on the order of almost
zero to three centimeters with a resolution that can reach the nanometer). It uses capaci-
tance, which is the ability of a material to store electric charge (i.e., thinking in the elec-
tronics field; this principle is generally associated to the storage device called capacitor).
Figure 6.7 shows a parallel plate capacitor made with two conductive materials posi-
tioned in parallel and between these plates there is an insulating material called as dielec-
tric. The value of the mutual capacitance is proportional to the plate’s area and the dielectric
index material permittivity (property). It operates similarly to the capacitor; however, the
capacitance of the sensor varies according to the distance between sensor-reading surface
and material to be detected. So, there is a variation in capacitance according to the electri-
cal signal emitted by device.
The inductive sensor is responsible for conversion of a physical measurement to an elec-
tric signal that can be understood by a programmable logic controller as information. The
sensor has the ability to detect metallic objects at small distances. On account of lacking
moving parts, the inductive sensor possesses a long service life in relation to the mechani-
cal contacts sensors limit. In addition, they are very well-sealed, can work in dusty envi-
ronments (nonmetallic), in contact with liquids (i.e., they can be inserted in a tank filled
with water), and present great accuracy for small detection distance. Therefore, it can be
D
ie
le
ct
Conductive plate
ric
Conductive plate
FIGURE 6.7
LDR sensor and the characteristic curve of the component.
Inductive sensor
Metallic
target
Amplifier Schmitt trigger Oscillator Amplitude drops
FIGURE 6.8
A simple operation of an inductive sensor.
used in proximity repeatability measurements and can operate at continuous (DC) or

alternate (AC) output voltage level. In addition, the switching DC sensor can be found as
normally open and normally closed and must be chosen according to the needs of the IoT
project. In some DC models, transistors used in the amplifier circuit may be of PNP (posi-
tive potential is switched on) or NPN (connection should be made by negative potential)
type. Figure 6.8 presents a simple operation of an inductive sensor.
The photoelectric sensors are used in a lot of applications, including: control, embedded
electronics, security systems, industrial machinery, and medical equipment. The purpose
of a photoelectric sensor is to convert a light signal (light or shadow) into an electrical
signal that can be processed by an electronic circuit. LDR falls in the photo-sensors con-
ductive type and is also known as photo-resistor. In them, the resistance to electric current
flow depends on the intensity of light incident on a sensitive surface of cadmium sulphide
material. The typical assembly of this sensor is intended to maximize the sensitive area of
its electrodes, forming the structure shown in Figure 6.9.
As can be seen in the LDR sensor curve, the bigger the sensor, the greater is the ability to
control more intense currents and in the dark, the typical resistance can reach more than
1 MΩ.
The data collected is generally sent to a microcontroller/microprocessor to run the
embedded software (SW) and to attribute the referred task according to the data reading.
Dark night Average Bright

time sunny day sunlight
106
Resistance in Ω
LDR
symbol
Thin strip of
10 photoresistive
material
Dark 0.1 Illumination (Lux) 1000 Light (cadmium sulphide)
FIGURE 6.9
The characteristic curve of the LDR sensor.
There is a range of types of microcontrollers or microprocessors that can be applied in IoT

environments, such as Raspberry PI, Arduino, PIC, BeagleBone, Galileo, and Edison. The
most popular microcontroller family is Arduino and PIC. For the microprocessor, the most
used are Raspberry PI and Field Programmable Gate Array. They have become popular
because of their purchase facility and availability of source codes on the Internet. There
is a vast content available for those who are interested in these technologies, but some
knowledge is required about engineering, electronics, electricity, physics, and chemistry,
besides a working knowledge of computer languages such as C, C++, and Java, in order to
use most of them.
6.3.3 Data Transmission

The PHY layer is basically intended to be over wireless platform because of the emerging
necessity for IoT nodes to be mobile and versatile. As it happens with other technologies
for data sensing, the transceivers are the mechanisms that affect directly the TX. The basic
structure of the general model block applied to the communication between two devices
is shown in Figure 6.10.
RF signal based
on IoT technology
Transmitter
Receiver
chosen
(TX)
Data
(RX)
Data
DAC
ADC
Sensor node with TX Sensor node with TX
and RX modules and RX modules
ADC
RF signal based
on IoT technology
Transmitter
Receiver
chosen
Data
Data
(RX)
(TX)
DAC
FIGURE 6.10
Basic general model block structure for the communication between two wireless devices.
The TX is done by the frequency related to the technology applied in the IoT environ-
ment, which are the RF TX, and RX between nodes. In Figure 6.10, the oscillator module (~)
is responsible by the RF signal and the (►), (X), and (≈) modules are RF amplifiers and/or
drivers with the intermediate frequency; signal mixer according to the oscillator; and filter-
ing regarding to the low-pass, high-pass, band-pass, and/or band-stop filters, respectively.
The data is transmitted by the air interface after signal preparing stages mentioned as
an analog signal to be received by the receiver node, which treats the signal received in the
mentioned modules to finally turn the data able to be digitalized or used in analog man-
ner by the nodes.
It was considered the use of DAC and analog-to-digital converter, because of the micro-
controller/microprocessor modules utilization for the data management and possible
intelligence improvement in the system for the IoT purpose.
6.4 Routing
As mentioned in previous sections of this chapter, the activity of an IoT node is not limited
to the acquisition and actuation tasks, but it is also addressed with the advance in the appli-
cations and to the growing necessities. Thus the data routing among the devices is turning
each day more important. This section will treat about some routing schemes and protocols
that can be presented in the IoT scenarios, focusing in the newest Trellis Coded Network
(TCNet) approach.
6.4.1 Current Challenges in Routing

Routing and forwarding packets remain a challenge in today’s networks. In terms of the
Internet backbone, the goal is to develop a shared understanding that the large opera-
tors are facing regarding the scalability of Internet routing system. It is recognized that
the major contributing factors are the routing tables’ growth, constraints in the routers
technology and the limitations of Internet addressing architecture. Per year, the traffic vol-
ume is 50% (±5%) and the size of routing tables is around 15%–25%. Autonomous systems’
growth rates are around 10%.
Between January 2006 and January 2009, Border Gateway Protocol (BGP) prefix update
and withdrawal rates per day increased by a factor of 2.25 to 2.5, presenting an average
value of 2–3 up to 1000 peaks per second. The research efforts can be classified as follows:
• Incremental improvements in BGP leading to a BGP++.

• Development of a new class of Path-Vector based routing protocols.
• Searching for new routing paradigms.
The routing protocols target space is characterized by distributed and adaptive require-
ments and the main objective is to determine if there is more than one possible routing
scheme in that space besides BGP++ and Path-Vector.
On the other extreme of the Internet backbone lays the Ad Hoc and sensor networks.
Such networks tend to proliferate according to the deployment of smart objects based ser-
vices. They may be small or very large. In the latter case, the routing protocols also face the
scalability challenges. Both kinds of networks in general consist of elements with limited
resources in terms of processing power, storage space, and energy availability. The cur-
rent employed routing protocols as, for example, the Ad Hoc On-Demand Distance Vector
(AODV) suffers from a large signaling overload that entails large latencies. The situation is
worsened by node failures and by the hidden nodes phenomenon. In addition, considering
many of such networks may employ wireless TX; the spectral efficiency is also an issue.
To face the challenges presented by the described scenarios, geometric routing is being
considered as an interesting solution. Geometric routing is based on topological network
features that enable the implementation of greedy routing protocols.
For Papadimitriou and Ratajczak (2005), any planar three-connected graph can be
embedded in the plane in such a way that for any nodes s and t, there is a path from s to
t such that the Euclidean distance to t decreases monotonically along the path. As conse-
quence of this, would be that in any Ad Hoc network containing such a graph as a span-
ning subgraph, two-dimensional virtual coordinates for the nodes can be found for which
the method of purely greedy geographic routing is guaranteed to work. They present two
alternative versions of greedy routing on virtual coordinates that probably work. Using
Steinitz’s theorem, it was shown that any three-connected planar graph can be embedded
in three dimensions so that greedy routing works, albeit with a modified notion of dis-
tance. Experimental evidence that this scheme can be implemented effectively in practice
has been presented by them. They also present a simple, but robust version of greedy rout-
ing that works for any graph with a three-connected planar spanning subgraph.
Krioukov et al. (2010) consider that at the same time, the nodes in the network are not
aware of the global network structure, they analyzed if paths to specific destinations in
the network can be found without such global topology knowledge and how optimal these
paths can be. The salient feature of their model is that it allows one to study the efficiency
of such path finding without global knowledge, because the analyzed networks have
underlying geometry which enables greedy forwarding. As each node in the network has
its address, that is, coordinates in the underlying hyperbolic space, a node can compute the
distances between each of its neighbors in the network and the destination whose coordi-
nates are written in the information packet or encoded in the signal. Greedy forwarding
then accounts to forward the information to the node’s neighbor closest to the destination
in the hyperbolic space. As each node knows only its own address, the addresses of its
neighbors and the destination address of the packet. In this scenario, no node has any
global knowledge of the global network structure.
Comellas and Miralles (2011) propose a label-based routing for a family of scale-free,
modular, planar, and nonclustered graphs. They provide an optimal labeling and routing
algorithm for a family of scale-free, modular, and planar graphs with clustering zero.
Relevant properties of this family match those of some networks associated with tech-
nological and biological systems with a low clustering, including some electronic circuits
and protein networks. They suppose that the existence of an efficient routing protocol
for this graph model should help when designing communication algorithms in real net-
works and also in the understanding of their dynamic processes.
All the presented approaches assume that the shortest path from the source to the desti-
nation node is the best path. In the realm of software defined networks, the best path may
be defined considering a set of constraints, such as: cost, revenue, end-to-end delay, avail-
ability, packet loss rate, and so on. Based in the presented metrics, the use of the shortest
path as the only metric is clearly not sufficient. In addition, it is important to assume that
some metrics reflect technical performances whereas others are based on economic nature,
in which some of them are linear. In the case of nonlinear, the geometric routing is not pos-
sible unless the metric space can be transformed into an equivalent underlying hyperbolic
space in which the new source and destination nodes’ coordinates can be used to evaluate
the routing by means of a greedy algorithm. On the other hand, TCNet architecture was
proposed by Lima Filho (2015) and Lima Filho and Amazonas (2013, 2014) presenting a
different routing scheme to solve the pointed problems.
6.4.2 Fundamentals of the TCNet Architecture

The TCNet has been initially conceived for the scenario shown in Figure 6.11. This scenario
considers a WSN in which a sink node queries a set of sensor nodes in a predetermined
order. On the right side, the TCNet frame is presented.
In Figure 6.11, a node acts as an access point (AP) with internet protocol (IP) infrastructured
networks and also as a sink to the information generated by other sensor nodes. A multicast
scenario, where a sink node sends messages to a set of sensors and actuators, is a challenge
for WSNs, and depending on the situation, the message has to be forwarded by several inter-
mediate nodes before arriving at the desired destination. In such a case, each intermediate
node has to consult a routing table to discover the next hop. If the next hop is not available
in the table, a discovery procedure has to be initiated. The TCNet model changes this sce-
nario completely. The sink node wishing to interrogate a set of nodes in the network sends a
query qi in a specific order defined by some optimization criteria. Each node receiving a frame
undertakes the following actions:
• Recognizes that it is being queried.

• If the queried node concludes that it is an element of the route, it updates the
payload in the frame with the value of its sensed variable and rebroadcasts the
message.
The frame structure generated by the sink node, following the idea of the Multi-Protocol
Label Switching shim header, which consists of the fields WSN header|TCNet label|payload
data with the features described as follows:
• In the WSN header, a code word is loaded which enables metric calculation to be
used by the routing algorithm.
• In the TCNet label, a specific sequence defined by a QoS criterion is loaded and
transported along the route, allowing each node to recognize its position into the
route and is used to update the code carried by the WSN header.
• Payload data is the field that stores the updated data sensed by the sensors during
a query.
Internet
IP traffic “Query” - qi
WSN qi Sink/ WSN header TCNet label Payload data
qi AP
(hop) qi qi
(hop) qi
FIGURE 6.11
Illustration of a WSN in which a sink node queries a set of sensor nodes and in the right side the TCNet frame
is shown.
Kn(t)/outn(t)
State j
State i
Node j
Node i
FIGURE 6.12
Network node modeled by a state of a Finite State Machine.
The process ends when all nodes of interest have been visited, the payload field carries
up-to-date sensed data and the frame returns to the sink node.
The TCNet concept models the network as a digraph:
D = (V , E ) (6.4)
where V is the set of vertices vi representing the network nodes and E is the set of edges vi
and vj representing the physical or logical connections between the vertices vi and vj.
The model is based on finite automata or finite state machines (FSM) defined by a
cross function (kn/outn) where a sequence of input symbols {k n} generates a sequence
of output codes {outn} as shown in Figure 6.12. The k n(t) is the input symbol received
at time t corresponding to the outn(t) code output. It is also assumed that at time t, a
transition occurs at the FSM from state i to j. In TCNet, each state represents a network
node and the transition state indicates that the frame information must be sent from
node i to j.
A network node modeled as a state allows the application of FSM principles in order to
exploit the analogy between networks and state diagrams to define paths between nodes,
so as to enable each node to have full knowledge about the network by implementing a
path generator, based on mealy machine of low complexity with the use of Exclusive-OR
(XOR) gates and shift registers.
It is then possible to generate a specific route through a set of nodes, defined by a desired
optimization criteria, such as latency, packet loss, BW and cost, by shifting an input
sequence {kn} in the mealy machine and informing it in the frame as a TCNet label.
Using the described procedure, Figure 6.13 shows the route established by the protocol
label using {kn} = {1 1 0 0}. It can be observed that every node in the network is visited in
the order [(10), (11), (01), (00)], that is, at the end, the frame returns to the sink node with the
information collected from every other node.
The routing scheme helps in delivering data for network nodes. As described previously,
a great effort has been done to improve the knowledge routing mechanisms. There is a
wide field for network protocols optimization and should be considered that services and
human necessities change with time.
WSN
header 11 01 01 11
dHam (0)
00 00
00 00 Sink 00 n-00 00 Sink
11 11
11 (0) 11
10 10 0 n-10 n-10
00
00 10
10
(0)
01 01 01 01
1 n-01 n-01
01
01
(0)
10 10
11 11 n-11 n-11
Step 0 Step 1 Step 2 Step 3

(a) Node decoding trellis (b) Estimated route over the trellis
FIGURE 6.13
TCNet. (a) Decoding trellis and (b) Route established by the input sequence {kn} = {1 1 0 0}.
6.5 Data Management

One of the most important things to turn possible the IoT application in real environment
or to enable the smart city development is the utilization of intelligence in the system,
enabling the management optimization. Cognition and taking decision can be used to
form a framework of IoT intelligence.
The use of cognition for the network management and the decision-making process for
the actuators, not only in the way of reaction but aggregating learning to network nodes,
is the most important tool for the evolution of WSN into the IoT.
As mentioned, the sensors collect data from the environment and in the actual network
scenario, these data are sent to the microcontrollers or microprocessors for the decision
based on the predefined rules established by the embedded SW or specific command lines,
as a reaction form to the data collected, as, that is, if the lux measured in the environment
falls to a predefined threshold, the microcontroller send to the respective pin (of the own
IC) an information (binary 02 or 12) to turn on the light connected with this pin.
From the exposed, it can be seen that the actual mechanisms in the market yet do not
have the necessary intelligence aggregated to the devices (unless the most powerful smart-
phone, which have the condition to use a more elaborated embedded SW and the avail-
ability of a great number of sensors) to create intelligence in network.
Figure 6.14 presents the actual relation of the issues applied to the standardization struc-
ture of IoT on the left side, and the increased blocks for the management and machine
learning on the right side.
As pointed by Wagner et al. (2016), the cognitive processing allows the system to read just
itself and at the same time perform the network monitoring tasks for the data acquisition
and network management. This occur with the use of a control vector (VC) that is inter-
preted by the adaptive cognitive system (ACS) to realize the Fuzzy logic (FL) approach in
turn to deliver according to the defuzzification process output, the function to be adopted
IoT applications: It covers logistics, remote monitoring, alerting, remote control,
Machine learning: In this approach, it is important that machines can

in the network, as specific nodes with a higher buffer and a great data
maintenance, safety, security, optimization, and autonomous intelligence.
support the learning task to cover more upcoming services and turn
Network management: It comprises the use of management agents
available the real intelligence in the network, using here cognition.

Security: It covers encryption, authentication, access, identity, and
Systems integration: It covers design and implementation.

IoT analytics: It covers complex event processing, geospatial systems,
predictive analytics, and holistic analytics, such as IoT data, enterprise data,
social data, and environment data.
IoT development platforms: It covers

Storage platforms: It covers high
rapid development IDE, device/
scalability, NoSQL, and security.
middleware connection, and security.
Middleware: It covers workflow, Device connectivity: It covers device
storage.
trust.
business rules, cloud-enabled, and independence, device clouds, and
open standards. connectors.
Communication services: It covers cellular, satellite, cable, microwaves, and

private networks.
Local communications: It covers Ethernet, Wi-Fi, Bluetooth, RFID, access

points, gateways, and routers.
Devices: It covers platforms as Raspberry pi, Arduino and BeagleBone,

phones, and readers.
Components: It covers sensors, controllers, processors, communicators, GPS/

location, RFID, and I2C.
Generic vision regarding IoT framework Intelligence and cognition
FIGURE 6.14
Relation of the issues applied to the standardization structure of IoT and the increased blocks for the manage-
ment and machine learning of sensor nodes.
for some aspects readjust on the network nodes, which will direct the measures to be
adopted in the network and accomplish the learning process for the network robustness.
Possible metrics involved with the measurement and VC adoption are related with end-
to-end delay, throughput, latency, and energy consumption. The ACS architecture was
developed in order to seize control of WSN in multiple layers. For the ACS modeling, it
was defined the block diagram present in Figure 6.15 in which, the ACS module has inter-
action with the WSN, Cognitive Process or Module (CPMod) and Fuzzy blocks.
WSN
WSN
environment
ACS
Data and control
Graphics
Decision Decision limit

CPMod limit according to
threshold
Semantic Tables Data
engine management
Fuzzy logic Fuzzy rules

and inference
FIGURE 6.15
Adaptive cognitive system internal modules.
The Cognitive Network (CN) architecture, proposed by Thomas (2006) is considered

in ACS, in the logical interaction between the blocks. In Figure 6.15, at CN, the Software
Adaptable Network layer connects to the network medium, that is, the ACS performs the
WSN interface with CPMod. The cognitive process layer of the CN, which includes cogni-
tive elements in order to achieve the objectives outlined, is represented by CPMod block.
Cognition selfish layer, which has the end-to-end objective and concentrate the network
knowledge, is represented by ACS with the FL block.
In addition to the ACS representation as autonomous system along with the connections
shown, thinking on the specific architecture of WSN, Figure 6.16 shows how the ACS is
related to the different network layers, including FL and CPMod with availability to mak-
ing adjustments in different layers.
For the Link and Physical layers, Mitola III (2006) presented the cognitive radio, a tech-
nology that is being applied for different reasons in the reading of channels available,
because this technology enables the system to verify and optimize the use of radio spec-
trum. This radio system was idealized for the integration of substantial computational
intelligence, especially machine learning, computer vision, and natural language pro-
cessing in a Software-Defined Radio. Its application can be found in LoRa WAN radio
technologies.
As the chosen sensor node for the study was the MICAz (family MICA motes), then
the model of Figure 6.16a is more coherent because it specifically meets the IEEE 802.15.4
standard for attending the ZigBee technology. The model of 5 layers of Figure 6.16b is
also valid for the ACS, for the IEEE 802.11 standard. The CPMod is the intelligent part for
application in sensors, involving cognitive aspects to execute the decision-making process,
acting as required on specific aspects of sensors, which are related to modeling of memory,
type routing protocol, change processing queue, and it can apply energy saving.
Figure 6.17 shows the structure corresponding to a common sensing node with routing
capability, along with the insertion of a cognitive module responsible for performing cog-
nitive processes. In the scheme, the white blocks represent the traditional internal interface
node, the yellow blocks are the representation of the physical wireless environment and
green block represents the cognitive module. In CPMod, the solid red interfaces were ini-
tially implemented for processing, memory, and routing, and the dashed corresponds to
Task management plane
Mobility management plane

tem
Application layer sys N )
i t ive d C S Power management plane
n n
cog y, a
Network layer ive uzz em
apt od, F Application layer yst )
d
A M i v e s CSN
P i t
(C gn nd
MAC layer co zzy, a
ve
Transport layer
a pti d, Fu
Ad Mo
P
Physical layer (C
Network layer
Data link layer
Physical medium Physical layer

(Air interface)
(a) (b)
FIGURE 6.16
Layers stack with ACS multilayer approach. (a) 4 layers and (b) 5 layers.
Sensor application CPU model
Battery
Wireless Sensor model
ID RT agent agent Radio
model
Packet
Ad Hoc routing dispatcher
Sensor
Address PHY
resolution Link layer
protocol
Cognitive processor
Queue
module
MAC
802.11
Wireless propagation Wireless Sensor mobility Sensor actuator

model PHY model
Sensor Sensor node position Sensor propagation

channel tracker model
Wireless channel
FIGURE 6.17
Common sensing node with routing capability and cognitive processor module.
the part connected to the energy which provides the generation of four vectors analysis for
the rounds of analysis by the ACS.
From Wagner et al., (2015), the CPMod was created initially to operate according to the
IEEE 802.15.4 standard for mobile Ad Hoc WSN networks in wireless local area network
(WLAN), where ID corresponds to the identification of the node, the RT for the Routing
Table (regarding routing protocol), and Address Resolution Protocol.
The implementation of CPMod occurs in the application layer, but it does not change the
meaning and purpose of the process, that is, the realization of decision-making running
on cognitive module will also affect internal parameters of other layers, such as adaptation
in the processing queue of the packets, improving memory allocation, and modifications
in the routing protocols adopted to optimize the packet forwarding routes, directly influ-
encing the performance of the network. Here, it can be set up the use of TCNet protocol for
the data routing in the network.
The CPMod perform analysis on the best condition according to the steps and run the
data saving in another file as the values are defined in semantics block. The semantic block
is associated with the CPMod analysis meaning part, namely as the progress of the test
rounds on the system. CPMod uses fuzzy block to check the current network condition,
comparing the data collected from WSN with values present in semantics. With the sav-
ing of the data on the best conditions for the events according to the adopted semantic,
episodic memory is being formed and is the history of network building.
VC = {Energy, Route, Queue, Memory} (6.5)

From Equation 6.5, it can be seen that each detail is related with each possible metric
pointed before, as vectors in binary 02 or 12. So, that is, if the node receives VC = {0, 1, 0, 0},
it means that the routing protocol is subjected to be changed. In this approach, the com-
munication protocol passes not to be the reference for the network, but it turns a powerful
tool that can be changed and monitored to increase the WSN performance.
6.6 User Applications

A general vision about some cases of applying the WSN in the IoT environments and appli-
cations is done by the following:
• The use in the agriculture area with the utilization of drones: It implies in the
range area to be covered and the use of localization scheme, the standard recog-
nizing over the types of plantations, such as coffee, orange and soybeans, network
connection between drones, and with a central node and also, big data regarding
the data traffic to be in large scale.
• Access and presence in educational environments: The use of RFID tags for the control
of the students presence in the class and to enable them for the access to specific
environments, such as library, refectory, and gym room. For this application a
great database for the information storage and possible consulting is necessary.
It implies in security with high level, because the information about all students,
professors, and employees are available in the database, so a cryptography system
should be implemented here.
• The use of tags for monitoring and control animals: It implies in the use of RFID tags
for the animal identification. These tags should be implemented in a way that the
resistance to the climate would be higher than the normal situations, including the
sensor to be water-proof and withstand high temperatures. Database is important
here for the storage of animals’ information and availability for the researchers
consulting.
• Smart grids: They are in the real development area, because this issue directly
influences the citizens’ energy consumption. In this scenario, the use of a support
for reading data (if possible in the real-time manner) and to control the decisions
to be made over the stations, substations, and power lines is important. Big data
surely is applied here. The ZigBee and LoRa technologies can be considered here
for the nodes connection and monitoring.
• Structural Health Monitoring (SHM): According to Akhras (1997), SHM defines the
physical structures conditions of bridges, towers, cranes, and other large build-
ings. The measurements over a frame are used for computer models to analyze
state data structure and make forecasts and warnings. One of the biggest chal-
lenges of SHM is that it is impossible to draw conclusions from just one measure
about the use of only one sensor, which is well seen in the case of Figure 6.18
where a single sensor in a bridge collect data and no conclusions can be obtained
from the measures, because this sensor not cover all area and possibilities of
bridge disaster.
Command and control center
Steel/Fiber optic sensors
General scenario for smart bridges
Solar panel with wireless telemetry
Local control unit
Actuators and sensors
Steel/FRP jacket Attached and imbedded sensors Actuators and strain gauge sensor
FIGURE 6.18
Scenario of the structure of a bridge with smart sensors.
IoT Framework Based on WSN Infrastructure with Different Topological Distributions
143
The only feasible approach for SHM is to periodically measure a variety of physical
quantities of acceleration, stress, pressure, temperature, wind speed, chemical reactions,
and so on, which requires a considerable BW that can be solved taking into account fre-
quency sampling and analysis methods online and offline.
6.7 Smart City with WSN

Micali (2016) at the Intel Developer Forum 2016 (an event to presents the trends and fea-
tures of future of technology) has talked about virtual reality, robotics, visual intelligence,
drones, as well as the power of autonomous vehicles, and the importance of the 5G on
these scenarios. According to the event, over 50 billion things and devices must be con-
nected by 2020, in addition with more than 200 billion connected sensors, all of them gen-
erating huge volumes of data. So, for this need, high processing capabilities, analysis, and
storage in distributed network structure it will be necessary, in addition with permanent
connectivity between things, network, and cloud.
Thinking in the presented scenario, this section will provide some cutting edge pos-
sible scenarios for the WSN use as the base of IoT in Smart Cities, based on the Brazil
necessities, which can surely be applied in other places in the world. Below are described
the authors’ ideas concerning the IoT environments, taking into account the all connected
philosophy:
• Green line in the smart cities: The idea here is to set the priority on the streets and
avenues for ambulances, changing the aperture of the headlights, turning in this
way the ambulances traffic more efficient, and decreasing deaths in the way to
the hospitals. The principle applied here is that the ambulance contains a priority
tag, that in contact with headlights as sensor nodes, the timing of traffic lights is
changed. A control system can be applied for the monitoring and management of
vehicles, including tags for the ambulance drivers. For instance, semaphores net-
work will be created and can be clustered for connection between the lights and
also with vehicles.
• Interactivity of visually impaired or blind persons with traffic lights system: The idea in
this scenario is to enable visually impaired or blind people to more independent
life in the urban centers. The main goal is to enable cell phone or smartphone with
a SW for the user to receive audible information via the use of mobile devices.
These messages are sent according with the environments, such as roads, bus
stops, rail stations, airports, theatres, cinemas, galleries, events, shows, churches,
schools, shopping, and a lot of other places where the user can interact.
For the specific road scenario, semaphores can be connected as network nodes,
executing an interaction between the lights and the blind person. The semaphores
will interact with the users of the service receiving information about the traffic
lights timing, how many time for the crossing of the road, about cars situation,
and so on.
It is not only a beep or another type of code, but what is offered to the user is the
information as it is, or in other worlds, as audible information. The control can be
realized from the point of view of the network, determining if semaphores are out
of service and other related things. This project is patented by Wagner et al. (2013)
under the number PI 1105299-6 A2. Any queries about this issue can be treated
with the patent authors.
• Use of automobiles and motorcycles as network elements: Cars, trucks, buses, and
motorcycles, can represent majority traffic in the roads, avenues, and routes, so
the use of them for the Ad Hoc M2M connection network is incredible. With this
interaction between machines, a very large network can be implemented and it
can offer an enormous range for the network and reach increasing for nodes. Not
only vehicles can participate of the network, but semaphores (mobile elements in
connection with fixed elements), smartphone, notebooks, tablets, and electronic
devices in general.
This solution promises to deliver the all connect concept to the smart city (the
network coverage area extends to the edges of the city or between cities with the
use of highways). The 5G will be the great responsible for turn this dreaming city
in the real life possible.
• The use of camera system interconnected with vehicles for the recognizing of a stolen
vehicle and also to monitoring of the traffic in the roads, avenues, and routes: This sys-
tem works with a RFID in each camera point (or semaphore, traffic light, mast, or
possibly trees) can aggregate the management of the vehicles (such as the cars,
motorcycles, bicycles, buses, trucks, and other transport types), with tags inserted
into the vehicle structure (for nonviolation of vehicle data), turning the smart city
all connected for the government management (data can be collected from vehi-
cles for verification, such as if the vehicle is regular with payment of taxes, and
other emoluments). For this reason, is very important that the network and system
offer very high security mechanism, including cryptography and tags checking for
identifying tags RFID violations.
• Intelligent traffic system (ITS) in urban clusters: Several technologies are currently
applied to what is being called as ITS, ranging from images pattern recognition
until satellites, demanding complex infrastructures and large BW for data TX.
The scenario depicted in Figure 6.19 shows an example of an application in which
a vehicle can be tracked by transmitting its ID in broadcast mode to intelligent
nodes. The network nodes are associated to the states of a FSM that implements a
convolutional code using the TCNet protocol.
The sink or gateway nodes also need to be distributed along the city and the scenario
presented in Figure 6.20 illustrating how vehicles can be tracked by trellis clusters of
an ITS.
It is possible to change the configuration of the trellis changing the possible routes of
the TCNet. The left side of Figure 6.21 shows two different trellis clusters with the same
number of nodes but different number of routes. The first cluster has 8 direct paths
between nodes and its complexity is of 3 registers and 2 XORs. The second has 16 direct
paths and its complexity has increased only by 1 register and 1 XOR. This approach
suggests that the model is scalable and quite robust in the presence of failures.
Map
Traffic
Trellis cluster
Vehicles tracking
Trellis cluster
Trellis cluster
Sink/gateway
Node (ID record)

Slow Fast
Real time traffic Link
Route
Sink/gateway
Tracked vehicle/
transmit ID
FIGURE 6.19
Scenario of vehicles tracking systems by means of trellis clusters.
Trellis cluster
Trellis cluster
Trellis cluster
Sink/LERi
Sink/LERe
Route
Sink/LERi Trellis cluster Trellis cluster
Sink/LERe
Route
Sink/LER Record ID in buffer Node (stores ID) Tracked vehicle/
transmit ID
FIGURE 6.20
Illustration of vehicles tracked by trellis clusters.
Trellis cluster
000 000
Trellis cluster
Trellis cluster 010 010
Trellis cluster
100 100
110 110
FSM node 001 001

FSM node
011 011
101 101
111 111
(a) (b)
FIGURE 6.21
(a) Two different trellis clusters with the same number of nodes but different number of routes and
(b) Illustration of how a coverage area can be changed by changing the capacity of a trellis.
6.8 Conclusion
Considering the application of TCNet and AODV in fixed structures, where nodes move
slowly: for the TCNet performing a task, it is necessary to the node receive the obtained
sequence kn(t) from a primary or using a graphical means for the simple cases using trel-
lis inspection resources; in the case of AODV, the protocol performs the search of routing
independently, with the disadvantage of increased latency. In the case of fixed structures
where the nodes do not change the position, since the best routing table update, the AODV
efficiency is almost equal to TCNet.
The ACS system is viable in the point of view of the network because it can actuates to
improve the performance of the network related to the energy consumption of nodes, iden-
tifying the best network protocol to be used according to the network situation, the metric
applied to the end-to-end delay and latency, was enhanced because the flow of packets by
intermediate nodes is turned more controlled.
The cognitive approach can be inserted into an IoT device or for application in smart cit-
ies area, with the adaptation of the ACS/CPMod for operation in different nodes present in
the WSN and not only as a traffic analysis based on application layer.
The universe of application possibilities in the IoT environments is only limited by
our own thinking threshold and the connectivity is the key for the future develop-
ment in the world where the Internet domain supports the addressing for IoT elements
and the M2M, Machine-to-Living Beings and WSN-to-end/WSN are associated directly
with the platform for communication, thus aggregating intelligence, reliability, secu-
rity, and versatility with powerful smart devices usage, comprising a lot of variety of
sensors and actuators to be employed into the control, monitoring, and management of
networks.
References
Akhras, G. 1997. Smart structures and their applications in civil engineering. Civil Engineering.
Report CE97-2, Royal Military College (RMC), Kingston, Ontario, Canada. Available: http://
citeseerx.ist.psu.edu/viewdoc/download? doi=10.1.1.471.7279&rep=rep1&type=pdf.
Brazil. 2008. Presidência da República, Casa Civil, Subchefia para assuntos jurídicos. Law Decreto n.
6.654, 20/11/2008, that approve the Plano Geral de Outorgas de Serviço de Telecomunicações
prestado no regime público. Brasília/DF, Brazil. Available: http://www.planalto.gov.br/
ccivil_03/_Ato2007-2010/2008/Decreto/D6654.htm.
Comellas, F. and Miralles, A. 2011. Label-based routing for a family of scale-free, modular, planar
and unclustered graphs. Journal of Physics A: Mathematical and Theoretical. 44(20): 205102.
doi:10.1088/1751–8113/44/20/205102.
Dargie, W. and Poellabauer, C. 2010. Fundamentals of Wireless Sensor Networks: Theory and Practice.
Chichester, UK: John Wiley & Sons.
Dias, R. R. F. 2016. Internet das Coisas sem mistérios: uma nova inteligência para os negócios. HP RFID
Center of Excellence. São Paulo, Brazil: Netpress Books.
IEEE. 2011. Standard 802.15.4-standard for local and metropolitan area networks-Part 15.4: Low-rate
wireless personal area networks (LR-WPANs). IEEE Computer Society. 5 September. Available:
https://standards.ieee.org/about/get/802/802.15.html.
Krioukov, D. et al. 2010. Hyperbolic geometry of complex networks. Physical Review E, 82(3): 036106.
doi:10.1103/PhysRevE.82.036106.
Lima Filho, D. F. and Amazonas, J. R. 2013. TCNet: Trellis Coded Network-implementation of
QoS-aware routing protocols in WSNs. IEEE Latin-America Conference on Communications
(LATINCOM'2012). Cuenca, Ecuador, April.
Lima Filho, D. F. and Amazonas, J. R. 2014. A Trellis Coded Networks-based approach to solve
the hidden and exposed nodes problems in WSNs. Segundo Seminario Taller Latinoamericano de
Instrumentación Control y Telecomunicaciones (SICOTEL'2014). Available: http://www.gituq.edu.
co/memorias%20sicotel2014.pdf.
Lima Filho, D. F. 2015. Projeto e Implementação de um novo Algoritmo e Protocolo de Encaminha-
mento de Pacotes baseado em Códigos Convolucionais usando TCNet: Trellis Coded Network.
114 p. Doctoral Thesis. Sistemas Eletrônicos-Escola Politécnica, Universidade de São Paulo, São
Paulo, Brazil.
Micali, B. 2016. Preparem-se: o futuro previsto pela Intel na tecnologia traz revoluções. Tecmundo
in the Intel Developer Forum (IDF'2016). San Francisco, CA. Available: http://www.tecmundo.
com.br/intel/108519-preparem-futuro-previsto-intel-tecnologia-traz-revolucoes.htm.
Mitola III, J. 2006. Cognitive Radio Architecture: The Engineering Foundations of Radio XML. Hoboken,
NJ: John Wiley & Sons.
Papadimitriou, C. H. and Ratajczak, D. 2005. On a conjecture related to geometric routing. Journal
of Theoretical Computer Science-Algorithmic aspects of wireless sensor networks. 344(1): 3–14.
doi:10.1016/j.tcs.2005.06.022.
Silva, F. A. et al. 2014. Tecnologia de Nós sensores sem fio. Revista Controle e Instrumentação. UFMG.
Available: http://homepages.dcc.ufmg.br/~linnyer/bianossensores.htm.
Sohraby, K., Minoli, D., and Znati, T. 2007. Wireless Sensor Networks: Technology, Protocols, and
Applications. Hoboken, NJ: John Wiley & Sons.
Thomas, R. W. et al. 2006. Cognitive networks: Adaptation and learning to achieve end-to-end perfor-
mance objectives. Topics in Radio Communications-IEEE Communications Magazine. 44(12): 51–57.
Wagner, M. S. 2009. Influência de protocolos de segurança sobre o desempenho de redes UMTS.
141 p. Master Thesis. Sistemas Eletrônicos-Escola Politécnica, Universidade de São Paulo, São
Paulo, Brazil.
Wagner, M. S. 2016. Sistema cognitivo com tomada de decisão baseada em Lógica Fuzzy para aplicação
em ambientes de redes de sensores sem fio com múltiplos saltos. 237 p. Doctoral Thesis. Sistemas
Eletrônicos-Escola Politécnica, Universidade de São Paulo, São Paulo, Brazil.
Wagner, M. S., Ribas, C. A. M., and Oliveiras Junior, Z. D. 2013. PATENT number PI 1105299–6 A2.
Processo e equipamento para transmissão e recepção de informações auditivas para porta-
dores de deficiências visuais. Instituto Nacional da Propriedade Industrial (INPI). Revista da
Propriedade Industrial (RPI) 2231 de 08/10/2013, Seção I, p. 65, São Paulo, Brazil. Available:
http://revistas.inpi.gov.br/pdf/PATENTES2231.pdf.
Wagner, M. S., Ramírez, M. A., and Zucchi, W. L. 2015. Adaptive cognitive system applied on wire-
less sensor networks nodes decisions. IEEE International Conference on Communication Systems
and Network Technologies (CSNT'2015). Gwalior, India, April.
Wagner, M. S., Ramírez, M. A., and Zucchi, W. L. 2016. Adaptive cognitive system applied to WSN
decisions at nodes with a fuzzy logic approach. International Journal of Communication System
and Network Technology. 5(1). doi:10.18486/ijcsnt.2016.5.1.01.
7
Semantic Complex Service Composition
within an IoT Ecosystem
Grigorios Tzortzis, Charilaos Akasiadis, and Evaggelos Spyrou
CONTENTS
7.1 Introduction ........................................................................................................................ 151
7.2 Related Work ...................................................................................................................... 153
7.3 The SYNAISTHISI Platform ............................................................................................. 155
7.4 An Ontology for Smart Meeting Rooms ........................................................................ 157
7.4.1 Resource Model ...................................................................................................... 157
7.4.2 Service Model ......................................................................................................... 159
7.5 Service Composition.......................................................................................................... 160
7.5.1 Semantic Input/Output Matchmaking............................................................... 160
7.5.2 Manual Service Composition ............................................................................... 161
7.5.3 Semiautomatic Service Composition .................................................................. 162
7.6 Composition of Complex Applications in an IoT Ecosystem: A Use Case ................ 164
7.6.1 Available Services .................................................................................................. 164
7.6.2 Manual Service Composition Example .............................................................. 166
7.6.3 Semiautomatic Service Composition Example .................................................. 167
7.7 Discussion ........................................................................................................................... 169
7.8 Conclusion .......................................................................................................................... 170
Acknowledgment ........................................................................................................................ 171
References..................................................................................................................................... 171
7.1 Introduction
Recent technology advances in the fields of electronics, telecommunications, and
informatics have allowed everyday physical objects to be enhanced with the embodi-
ment of short-range and energy-efficient mobile transceivers. The extensive networking
of everyday objects has led to the so-called Internet of Things (IoT) (Atzori et al., 2010),
which has been considered by many to constitute the next industrial revolution (Gubbi
et al., 2013). IoT is expected to find numerous applications in many diverse and het-
erogeneous areas, including, but not limited to, industry, logistics, building and home
automation, smart cities and smart manufacturing. For further possible applications,
the reader is encouraged to study the works of Guo et al. (2013) and Bandyopadhyay
and Sen (2011).
It is typical for IoT ecosystems to incorporate a service-oriented architecture (SOA)
(Al-Fuqaha et al., 2015). In other words, each thing, whether hardware (e.g., a device)
151
or software (e.g., an algorithm), is exposed to the outer world as a web service, a self-
contained unit of functionality, which offers its services to other things. IoT services can
be categorized into three distinctive, yet interdependent, types: (1) services that capture
properties of the physical world and provide raw or slightly processed measurements
(sensing services); (2) services that process the acquired measurements and provide the
inferred results (processing services); (3) services that enable certain actions, based on
the results of the processing services (actuating services). On the following, these three
categories of services are referred to as S-, P-, and A-type services, respectively, or SPA
services as a whole.
Commonly, applications designed on top of an IoT ecosystem exploit the SOA paradigm
and involve distributed sensor networks at various scales that are interconnected with dis-
tributed processing modules (which autonomously process the measurements gathered by
the sensors) and actuation elements that are triggered according to the produced results.
Hence, the available SPA services can be combined to build complex applications that offer
some new desired functionality that none of the existing services is able to provide. This
procedure is referred to as service composition (Sheng et al., 2014). Key steps in composing a
service (thus building an application) are the discovery of suitable services, and their appro-
priate interconnection using an IoT-ready platform, such that a functional composition, is
ensured in which every service can be readily invoked (i.e., each service’s inputs and pre-
conditions are satisfied from the outputs and the effects of other services participating in
the composition).
The main focus of this chapter is to describe the procedure of developing complex appli-
cations within SYNAISTHISI (Pierris et al., 2015), an IoT-ready platform in which available
SPA services are semantically annotated using ontologies. In particular, a smart meeting
room ontology is utilized, which introduces classes and relations (properties) to model
the various service types encountered within the context of a meeting room that has been
augmented with several S-, P-, and A-type services. Moreover, it is shown that a developer
with knowledge of (1) the services required for the complex application and (2) their cor-
rect interconnection is able to exploit the semantic information in the service descriptions
(via SPARQL queries [Harris & Seaborne, 2013]) to manually compose a service. Also, a
service composition process is presented that utilizes semantics and allows developers
to compose services in semiautomatic fashion, alleviating the assumptions made by the
manual approach and considerably reducing the effort required by the developer and the
risk of human error.
In this chapter, a developer’s side perspective is adopted. Assume that there exists a set
of already developed services, and the developer needs to reuse some of them so as to
deliver new and more complex ones, as easily and as fast as possible. The following illus-
trate how a developer may search, discover, and interconnect available services using both
the manual and the semiautomatic composition methodology. As a test case to demon-
strate the potential of this framework, a person counting service is developed, implemented
within the context of a smart meeting room pilot scenario that exploits a subset of the
available sensing and processing services in the room.
The rest of this chapter is organized as follows: Section 7.2 provides an overview of
related work concerning service composition and service ontologies. Then, in Section 7.3,
the main characteristics of the SYNAISTHISI platform are outlined, along with the descrip-
tion of one of its pilot applications, namely the smart meeting room, in the context of which
service composition is performed. The proposed smart meeting room ontology is ana-
lyzed in Section 7.4, whereas the two service composition approaches follow in Section 7.5.
Semantic Complex Service Composition within an IoT Ecosystem 153
Section 7.6 presents a use case of constructing a person counting service by combining exist-
ing services available in the smart meeting room pilot. Section 7.7 discusses various inter-
esting features of the smart meeting room ontology and the two composition methods.
Finally, conclusions are drawn in Section 7.8.
7.2 Related Work

It is well known that orchestrating a collective functionality using highly heterogeneous
devices and modules exposed as web services is quite a difficult task. To obtain a compre-
hensive representation of web services and improve interoperability in the context of IoT
so that tasks such as service discovery and composition can be performed in a more effec-
tive and efficient manner, semantics are employed to enhance the description of services.
Several ontologies have been proposed for adding semantic content to web services, thus
transforming them into semantic web services (McIlraith et al., 2001). OWL-S (Martin et al.,
2004) and WSMO (Roman et al., 2005) are two renowned efforts, which provide highly
expressive models for annotating web services in heterogeneous domains. Rich conceptual
approaches, such as OWL-S and WSMO, demand considerable user effort to comprehend
them and appropriately annotate the services, whereas reasoning becomes a computation-
ally intensive process. Lately, a tendency has emerged to switch to lightweight semantic
models that focus only on the core semantics of web services, trading expressivity for
improved usability and complexity. SAWSDL (Kopecký et al., 2007) and WSMO-Lite (Vitvar
et al., 2008), which builds upon SAWSDL, are two such examples that allow semantic anno-
tations to be directly added to the WSDL description of services. Accordingly, for RESTful
services, hRESTS (Kopecký et al., 2008) and MicroWSMO (Roman et al., 2015), which is
based on hRESTS, have been developed, supporting the addition of semantics directly on
the HTML description of services. The Minimal Service Model (MSM) (Pedrinaci et al.,
2010) is another such approach that provides a common conceptual model for capturing
semantics of services, whether they are WSDL based or REST based. As noted by Pedrinaci
et al. (2010), MSM is largely compatible with OWL-S and WSMO service models. MSM has
been incorporated in the iServe platform (Pedrinaci et al., 2010), a service publication and
discovery platform.
The aforementioned ontologies constitute general purpose upper ontologies that include
high-level concepts that can be utilized across diverse domains. However, low-level con-
cepts specific to some particular domain are missing, and substantial effort is required
to define them. To tackle this issue, web service ontologies that are domain specific have
been exploited. Stavropoulos et al. (2012) proposed BOnSAI, a smart building ontology,
to facilitate the development of a smart ambient setting that provides automation and
energy savings. BOnSAI categorizes services/devices into sensors and actuators (i.e., S-
and A-type services) and defines classes and properties related to a smart building (e.g.,
lighting, air-condition, smart plugs, communication protocols used by devices, location of
devices, environmental parameters, such as temperature, measured [affected] by sensors
[actuators]). Past work of Akasiadis et al. (2015) has presented a preliminary version of an
ontology dedicated to smart meeting rooms that, in general, classifies services/devices
into sensors, processors and actuators (i.e., S-, P-, and A-type services) and subsequently
refines them by defining specific types of them in a class hierarchy, along with other
important aspects such as location and the *IOPEs of services. This ontology is extended
herein (Section 7.4) with additional concepts, to better describe the entities of smart meet-
ing rooms and improve the execution of service composition tasks.
Web services may be perceived as self-contained units of functionality that can be read-
ily published, discovered, and consumed using service-oriented architectures (SOA). To
accomplish the vision of IoT it is of paramount importance that apart from simply using
available services, users are able to easily combine them via service composition tech-
niques and produce more complex ones that better suit their needs. Service composition
approaches can be grouped into three main categories: manual, semiautomatic, and auto-
matic (Sheng et al., 2014). In manual composition, the user is responsible for every step of
the composition process. Appropriate services are manually discovered, and then their
interconnection must be defined. It may even be the responsibility of the user to bind the
web services according to the interconnection scheme. Although there is total freedom on
how to perform the composition, this advantage is compensated by the need for strong
programming skills (the user must be a developer) and the fact that the whole procedure
is time-consuming and error-prone.
In the semiautomatic case, the underlying system assists users by automating some parts of
the process (e.g., the discovery of suitable services, deciding whether two services can be con-
nected or not) and, at various steps during composition, presents them with possible choices
out of which they must select how to proceed. Liang et al. (2004) proposed an approach in
which the user first interacts with a query composer to select service categories that suit
his needs, thus forming a service request, then a service dependency graph that captures the
input and output dependencies of the available services is automatically constructed and an
AND-OR graph search algorithm is employed to extract multiple subgraphs that represent
possible compositions. These compositions are shown to the user, who must acknowledge
whether they are valid and satisfy his needs. Albreshne and Pasquier (2010) introduced a
system in which developers build generic process templates that describe the control flow of
possible composite services and end-users configure, via a graphical interface, those generic
templates according to their preferences. Based on the configured template, a service discov-
ery is executed to locate matching services and realize the composition.
Turning to automatic composition, the user typically sets his/her constraints and/
or preferences in a service request and is not allowed to intervene at any other part of
the process, which is executed in a fully automatic manner. Such methodologies are not
demanding, nor require specialized skills, and are ideal for everyday users. However,
as users have nocontrol on how the composition evolves, it may lead to a solution that
does not entirely fulfill their original intentions. PORSCE II (Hatzi et al., 2012) transforms
service composition to an AI planning problem that can be solved using off-the-shelf
PDDL-compliant planners. The initial state consists of the data that are provided to the
composition process (i.e., inputs), declared in a service request, whereas the goals are the
desired outputs of the composite service, which are also specified in the service request.
Actions for moving between states in the planning domain correspond to the available
web services. The applicability of a service (i.e., action) at a certain state depends on its
required inputs, whereas the resulting state is augmented with the outputs produced by
the service. PORSCE II offers a mechanism to replace a service from those involved in
the computed composite service if the user is unwilling to use it or the service has failed.
Rodríguez-Mier et al. (2016) presented a graph-based method for automatic service com-
position and implement it as part of the ComposIT system. A service request similar to
* IOPEs: inputs, outputs, preconditions and effects.

that of PORSCE II must be declared, containing the inputs available to the composite ser-
vice as well as the desired outputs it should produce, and a layered directed acyclic graph
of services is incrementally constructed. Each layer contains all those services, inputs of
which are provided by the outputs produced in previous layers and, therefore, are invok-
able at this layer. A heuristic search based on the A* algorithm is executed over the graph
to find the composition (i.e., a subgraph) with the minimum number of services that fulfill
the service request.
As a final note, all service composition approaches discussed here exploit semantic
descriptions of the individual services in order to avoid syntactic barriers and boost their
performance. However, this is performed in slightly different ways, according to the field
of application (see, e.g., the work of Sheng et al. [2014] that surveys recent work regarding
web service composition).
7.3 The SYNAISTHISI Platform

The platform presented herein has been developed within the context of the SYNAISTHISI
project,* the goal of which was to deliver energy efficient, secure, and effective applications,
as well as services to end users, aiming to minimize the environmental impact (services
would run only when needed), monetary costs (unreasonable energy consumption should
be prevented), user discomfort (room conditions should be comfortable for users), delays
and utilization of resources (network traffic should be minimized). In this section, a brief
description of the developed, IoT-ready, SYNAISTHISI platform (Pierris et al., 2015) is pro-
vided, along with an outline of how it enables the integration, interconnection, and coor-
dination of a large number of heterogeneous devices and algorithms, which are exposed as
semantic web services. The platform architecture is analytically discussed by Pierris et al.
(2015), and going into the architectural details is out of the scope of this chapter. In brief,
the platform supports communication channels that realize service interconnection, con-
trols the available resources, permits the deployment of custom applications, and provides
a set of tools to accommodate system administrators.
The available (semantic) services are registered into the service registry, implemented
by an RDF triplestore equipped with a semantic reasoner, and annotated using an
appropriate ontology, which shall be discussed in Section 7.4. Following the IoT para-
digm discussed in Section 7.1, they are divided into three distinct categories: (1) S-type
services corresponding to sensors that sense the physical world, (2) P-type services cor-
responding to processors (i.e., algorithms) that process the measurements of the S-type
services and/or the processed results of other P-type services, and (3) A-type services
corresponding to actuators that are used for the actuation of devices/signals based on
the acquired results. Seen from the services’ IOPEs perspective, S-type services pro-
duce outputs but do not have inputs (strictly speaking, they may only receive trigger
signals for their (de)activation), P-type services have both inputs and outputs, whereas
A-type services have inputs, and their output (actuation) is actually an effect that may be
observed. The three service types will be collectively referred to as SPA services hereafter.
Within the platform, services communicate by publishing messages and/or by subscrib-
ing to topics managed by a Message-oriented Middleware (MoM), which is a message
* http://iot.synaisthisi.iit.demokritos.gr
RGB Foreground
RGB
Background Publish
Camera Subtraction Subscribe
FIGURE 7.1
Two services interconnected with the SYNAISTHISI platform exchange messages. Solid arrows correspond to
messages published to topics, whereas dashed arrows depict the messages arriving to the subscribers of those
topics.
broker. Thus, all sensor measurements, processor results, and actuations are encapsulated
into messages and communicated via these topics. Once a message is published to a topic,
the MoM informs and delivers it to all clients that are subscribed to that topic. The MoM,
among others, supports the MQTT protocol (Banks & Gupta, 2014), main role of which is
to orchestrate the intercommunication among services in the aforementioned way. Each
service incorporates an MQTT client to be connected to the MoM. The MoM is respon-
sible to deliver the messages to subscribed clients. Scalability is facilitated, as new services
may be added with almost zero configuration. Moreover, the MQTT is an extremely light-
weight protocol, designed to require small code footprint and network bandwidth, while
also being available for many heterogeneous devices and programming languages. An
example of the interconnection of services using a MoM is depicted in Figure 7.1. A camera
is an S-type service that captures RGB video data from the physical world and publishes
them to a topic. A Background Subtraction module is a P-type service, which receives the
camera-generated data by subscribing at the same topic. Upon processing, it publishes its
output using another topic.
The SYNAISTHISI smart meeting room (Sfikas et al., 2015) is one of the pilot applications
of the SYNAISTHISI project and resulted upon the enhancement of a typical meeting room
with a set of SPA services that are integrated into the SYNAISTHISI platform. More spe-
cifically, it incorporates sensors and actuators, physically installed in the room, as well as
processing units running in the cloud. A Decision Maker (DM) service performs the main
actuations (implemented as A-type services), using measurements and results from S- and
P-type services. Upon recognizing specific simple events occurring within the room (e.g.,
temperature, humidity, luminosity, electricity consumption levels and occupancy), it trig-
gers actuations (e.g., controlling HVAC,* lights, the projector) aiming to minimize energy
losses, while maintaining certain comfort levels to improve the overall working environ-
ment. The smart meeting room scenario provided the motivation behind the ontology and
the service composition algorithms developed herein, although it must be made clear that
the service composition techniques are general and by no means limited to this scenario.
* HVAC: Heating, Ventilation, Air-Conditioning.

7.4 An Ontology for Smart Meeting Rooms

To enhance the efficacy of service discovery and composition toward realizing complex
applications, the SPA services of the SYNAISTHISI platform service registry are semanti-
cally enriched using ontologies. Specifically, a domain-specific ontology has been developed
that models the concepts related to smart meeting rooms (Section 7.3), in which all sensors,
processors, and actuators are exposed as web services.
The developed semantic model imports existing ontologies and reuses their knowledge.
The Internet of Things Architecture (IoT-A) ontology (De et al., 2012) includes high-level
concepts for describing key aspects of the IoT domain and forms its basis. Also, the popu-
lar Semantic Sensor Network (SSN) ontology is utilized to represent various features of
sensors and actuators (e.g., the measurement capabilities of a sensor), whereas the QU and
QUDT ontologies provide descriptions for physical quantities (e.g., temperature) and their
measuring units.
In what follows, the main classes and properties that make up the proposed ontology
are outlined, omitting secondary elements to avoid cluttering the presentation. A prelimi-
nary version of the ontology can be found in the work of Akasiadis et al. (2015), which is
here refined and extended with additional concepts.
The resource and service model of the IoT-A ontology are specialized by introducing
classes, subclasses, and properties related to the smart meeting room domain. In general,
the resource model is responsible for describing the characteristics of the actual device
(i.e., sensor or actuator) or processor, referred collectively to as resources hereafter, that is
hidden behind a web service, whereas the service model is responsible for describing the
characteristics of the web service (e.g., service input/output, service endpoint) that is used
to access a resource and expose its functionality to the outer world. Note that in the context
of this chapter, the notion of SPA services encapsulates both resource and service model
concepts.
7.4.1 Resource Model

The resource model, illustrated in Figure 7.2, consists of a core class, namely Resource,
that captures the notion of a resource. A hierarchy is defined to represent various types
of resources that may exist in a smart meeting room, many of which were not part of the
ontology of Akasiadis et al. (2015). The hierarchy includes specific types of sensors, for
example, temperature sensors, cameras and microphones; processors, for example, mod-
ules for subtracting the background of a scene; and actuators, for example, plug switches.
Class names are quite indicative of the resources’ functionality. A resource is equipped
with properties specifying its name (hasName), an ID (hasResourceID), some key-
words describing it (hasTag) and, importantly, the web service that exposes its function-
ality to the outer world (isExposedThroughService). The Location class defines the
location of a resource, while the Network class is used to describe the network interface
which makes the resource accessible through the web. Product-related information, for
example, the manufacturer’s name, can be provided using the Product concept. The
ontology also allows to associate a resource with its owner (ResourceOwner). Finally,
for sensing and actuating devices the SSN ontology concepts can be exploited using the
link to the SSN Device class.
158
CameraResource
MotionSensorResource
MotionDetectionResource BackgroundSubtractionResource
PlugMeterResource
LightSensorResource
SoundDirectionFacesResource HumiditySensorResource are
are PersonDetectionTrackingResource
TemperatureSensorResource
SoundDirectionResource ComplexEventRecognitionResource
are ImageProcessingResource
MicrophoneArrayResource
AudioProcessingResource SensorResource
AudioLevelResource
DecisionMakerResource
RecommendationSystemResource are
are ActuatorResource
FusionResource
ResourceOwner ProcessingUnitResource are are
are PowerONOFFActuatorResource
NetworkResource OnDeviceResource
Owner are
hasOwnerName hasOwnerID hasOwnerLink OnDeviceResource isHostedOn

(DataTypeProperty) (DataTypeProperty) (DataTypeProperty) Resource
are ssn : Device
xsd : string xsd : anyURI xsd : anyURI WirelessNetwork WiredNetwork
Resource
hasResourceID
hasOwner hasNetwork isExposedThroughService hasTag (DataTypeProperty) are
(ObjectProperty) hasName
hasProductInfo hasResourceLocation (ObjectProperty) (ObjectProperty) (DataTypeProperty)
(DataTypeProperty)
xsd : string Product Location Network sm : ResourceService xsd : string xsd : anyURI Network
ResourceOwner
hasModelName hasVersion hasImplementationLanguage hasManufacturer

(DataTypeProperty) (DataTypeProperty) (DataTypeProperty) (DataTypeProperty)
hasPort hasIPV4 hasIPV6 hasMACAddress
xsd : string xsd : string xsd : string xsd : string (DataTypeProperty) (DataTypeProperty) (DataTypeProperty) (DataTypeProperty)
xsd : unsignedInt xsd : string xsd : string xsd : string

Location
hasAltitude hasLatitude hasLongitude hasCity hasRoom hasPostcode

hasCountry hasBuilding hasGlobalLocation hasLocalLocation
(DataTypeProperty) (DataTypeProperty) (DataTypeProperty) (DataTypeProperty) (DataTypeProperty)
(DataTypeProperty) (DataTypeProperty) (DataTypeProperty) (DataTypeProperty)
Altitude xsd : string xsd : float xsd : string xsd : float xsd : string xsd : string xsd : anyURI xsd : anyURI xsd : string
FIGURE 7.2
Core classes and properties of the resource model. The sm and ssn namespaces refer to the service model and the SSN ontology, respectively.
7.4.2 Service Model

The service model, depicted in Figure 7.3, consists of a core class, namely Service, that
captures the notion of a web service and contains the necessary information for discov-
ering and invoking the service. Various service types are defined in a hierarchy, corre-
sponding to the aforementioned resource types. A service has a name (hasName), an ID
(hasServiceID), an owner (ServiceOwner), a link to the resource that is accessed using
the service (exposes) and an endpoint from where client applications can access the ser-
vice (ServiceEndpoint). The (physical) area that is affected when the service is invoked
can be determined via the hasServiceArea property. This property is particularly use-
ful for S-type services that declare the area that is observed by the sensor, and for A-type
services that declare the area that is affected by an actuation. Possible time constraints
on the availability of a service can be defined using the ServiceSchedule class. A very
important element in the semantic description of a service, especially for service composi-
tion tasks, is to accurately model its IOPEs. Here, this is accomplished through the sub-
classes of the Parameter class and by using the hasParameterType property to annotate
the IOPEs by providing the URI of the concepts (defined in some appropriate ontology)
that capture their meaning. Note that effects are considered as a special type of output
intended to describe the actuation of A-type services. Finally, it is possible to declare a
price for trading a service in a service marketplace (ServicePrice).
hasServiceOutput
(ObjectProperty) Output
endpointPath
rm : Resource (DataTypeProperty) xsd : string
hasServicePrice ServicePrice
endpointPort
xsd : unsignedInt
(DataTypeProperty)
exposes hasServiceEndpoint
(ObjectProperty) ServiceEndpoint
endpointHost
xsd : string
(DataTypeProperty)
rm : hasOwner
(ObjectProperty) ServiceOwner
endpointDescription
ResourceService xsd : string
(DataTypeProperty)
hasServiceArea
(ObjectProperty) ServiceArea
endpointProtocol
xsd : string
(DataTypeProperty)
hasEffect
(ObjectProperty) Effect
are Service
hasServiceParameter hasParameterType
Parameter xsd : anyURI
(ObjectProperty) (DataTypeProperty)
hasServiceSchedule
(ObjectProperty) ServiceShedule
Effect
hasName
xsd : string are
(DataTypeProperty)
ServiceOwner
are hasServiceID
xsd : anyURI Output Input Precondition
(DataTypeProperty)
rm : Owner hasPrecondition
Precondition are
(ObjectProperty)
hasServiceInput
Input Parameter
(ObjectProperty)
FIGURE 7.3
Core classes and properties of the service model. The rm namespace refers to the resource model. The hierarchy
of service types is not shown here, as it has a one-to-one correspondence to the resource type hierarchy.
7.5 Service Composition

The proliferation of the IoT relies heavily on the ability of users to readily combine the
available services within an IoT ecosystem and build complex applications that satisfy
their needs. On the following, two semantically-aware service composition methods that are
integrated into the SYNAISTHISI platform are presented; one falls under the manual com-
position paradigm while the other under the semi-automatic composition paradigm. In
its core, similar to Hatzi et al. (2012), Liang et al. (2004), and Rodríguez-Mier et al. (2016),
service composition is performed by matching-connecting the inputs of services to the out-
puts produced by other (preceding) services in the composition, so that all participating
services can be invoked, thus ensuring a functional composition. Note that preconditions
of services are not considered in this framework (similar to Liang et al. [2004]; Rodríguez-
Mier et al. [2016]) and effects are treated as a special case (i.e., subclass) of output pro-
duced by A-type services, therefore, hereafter, focus is entirely on the inputs and outputs
of services. On the following, the smart meeting room ontology (Section 7.4) is exploited to
obtain the semantic descriptions of SPA services, however the discussed approaches are
general, in the sense that they can be applied over any SPA services ontology.
7.5.1 Semantic Input/Output Matchmaking

In service composition an output of a service is regarded to match the input of another ser-
vice when the outputted data are compatible with the required input data and thus can be
readily consumed when inputted to the other service. To decide whether a match between
an input and an output exists, usually, the corresponding service descriptions are checked
to establish whether the input and the output reflect the same concept. In the most simple
case, two concepts are considered to be the same if they carry identical names, hence the
matching is done at a syntactical level (Sheng et al., 2014). However, as it is common in real
life, different names may be used to represent the same thing. Unavoidably in this case,
syntactic matchmaking is destined to fail. Semantics can greatly enhance the matchmak-
ing process by eliminating syntactic barriers through the identification of semantically
relevant (i.e., matching) concepts defined in some ontology.
An output concept A is considered to be semantically relevant to an input concept B
when one of the following hierarchical relationships (Paolucci et al., 2002) exists between
them: (1) exact(A, B) when the two concepts have the same URI or are equivalent in
terms of OWL equivalence, (2) plugin(A, B) when A is subsumed by (is a subclass of)
B, and (3) subsume(A, B) when A subsumes (is a superclass of) B. These hierarchical
relationships represent different degrees of semantic relaxation, in increasing order, and are
typical in the service composition literature (Hatzi et al., 2012; RodríguezMier et al., 2016).
The exact relationship implies that two concepts are the same and can be used inter-
changeably during composition. Plugin and subsume imply an approximate relationship
between two concepts and can be utilized to approximately match inputs to outputs, a
useful feature, especially when exact matches do not exist. Specifically, the plugin rela-
tionship means that the output providing service produces a more specific type of data,
than the input receiving service consumes. The subsume relationship suggests that the
provided output data are of a more general type than the one that is consumed in the input.
Hence, a composition containing a subsume match is not guaranteed to execute smoothly
during runtime, as there is a risk that a data-flow incompatibility may arise. Exact and plugin
matches avoid this pitfall and are thus a much safer option. In general, it is anticipated that
exact matches will be preferred over plugin matches, and plugin matches over subsume
matches. Hierarchical relationships are calculated by applying a semantic reasoner over
the ontology defining the concepts used to annotate the IOPEs of the SPA services (not the
smart meeting room ontology).
7.5.2 Manual Service Composition

The SYNAISTHISI platform is equipped with a manual service composition mechanism
(Akasiadis et al., 2015), in which the main steps for building a complex application are manu-
ally performed. Specifically, the necessary SPA services must be first discovered, and then
their interconnection (i.e., matching outputs to inputs) must be determined. The platform pro-
vides the tools to support the composition process (e.g., the service registry, a SPARQL query
[Harris & Seaborne, 2013] endpoint, the mechanism to state the interconnection scheme), but it
is not actively involved in the process per se. Obviously, manual composition is a quite daunt-
ing task (see also Section 7.2); thus, a developer is needed to execute it. From the developer’s
side, assume that there exists some prior knowledge regarding service composition; that is, what
kind of SPA services are required, and how they should be meaningfully interconnected.
To discover the services, the developer writes and issues through the platform SPARQL
queries conforming to the smart meeting room ontology, which describes the characteris-
tics of the desired services and subsequently recovers the matching services (retaining one
service for each part of the composition). By appropriately formulating the SPARQL que-
ries, the different hierarchical relationships between input/output concepts outlined in the
previous section can be exploited. As an example, assume that a developer has decided to
use in a composition a service that requires as input a video stream (declared using the
Video concept from some suitable external ontology where services’ IOPE-related seman-
tics are captured). The developer knows that such output is provided by camera services.
Therefore, he writes an appropriate SPARQL query adhering to the smart meeting room
ontology, shown in Chart 1.1, to discover these cameras. This query will return all camera
services that their output is either Video (exact match) or a subclass of Video (plugin
match), but not a super class of Video (subsume match), and the developer has to choose
one of them to include in the composite service.
Chart 1.1: Example developer-composed SPARQL query for discovering camera services
that output Video data. The sm and iope namespaces refer to the service model from
Section 7.4.2 and the external ontology where the Video concept is defined.
SELECT ?cam serv WHERE {

?cam serv a sm:CameraService.
?cam servsm:hasServiceOutput ?output.
{?outputsm:hasParameterTypeiope:Video.}
UNION
{?outputsm:hasParameterType ?type.
?type rdfs:subClassOfiope:Video.}
}
Having discovered the services to use, the next step is to formulate their interconnection
scheme. In particular, the inputs of each service participating in the composition must
be connected to compatible (according to the semantic hierarchical relationships) outputs
produced by some of the other participating services, to ensure all the involved services
can be invoked. Implicitly, this defines a partial ordering of the services, in which an output
providing service precedes those consuming the generated data through their inputs.
Remember that S-type services can be readily invoked as they do not have inputs (Section 7.3).
Hence, by utilizing one or more such services (recovered in the service discovery phase), it
is possible to establish a partial ordering in which all inputs are satisfied (S-type services
will be at the very beginning of this ordering). To describe the interconnection scheme the
developer assigns a unique identifier (a nonnegative integer number) to each service as a
whole, as well as to each of the service’s inputs/outputs. Then, a combined identifier con-
sisting of a service identifier and an output identifier is paired with a combined identifier
that consists of a service identifier and an input identifier; this reflects that this particular
output of the first service provides the data for the specific input of the second service. The
interconnection scheme is completed when such pairs of combined identifiers are declared
for all services’ inputs involved in the composition. The resulting scheme is submitted to
the platform that undertakes to enable the participating services and bring the required
communication channels between them into realization so that the composite service is
executed. Note that the developer is responsible for guaranteeing compatibility when pair-
ing inputs and outputs by respecting the semantic hierarchical relationships. The main
steps for manual service composition are summarized in Algorithm 1.
Algorithm 1 Manual Service Composition Steps

1: Decide what kind of SPA services are needed
2: for each kind of SPA service do // Service discovery.
3: Write a SPARQL query to retrieve matching services from the service registry
4: Select one of those services to be used in the composition
5: end for
6: for each selected SPA service do // Interconnection scheme.
7: for each input of the service do
8: Define a service-input combined identifier
9: Pair the combined identifier with a service-output combined identifier to
designate the origin of the data consumed in the input
10: end for
11: end for
12: Submit the interconnection scheme to the platform
7.5.3 Semiautomatic Service Composition

The manual service composition approach imposes some rather stringent assumptions that
require a considerable amount of effort on the developer’s side, high-level programming
skills, and a priori knowledge of the IoT ecosystem he operates in. Here, part of this effort is
transferred to the SYNAISTHISI platform that will guide the developer in building a complex
application, by proposing a semiautomatic service composition method that makes extensive
use of the semantic input/output matchmaking mechanism presented in Section 7.5.1.
Initially, the developer must specify a service request describing the desired functionality
of the composite service in terms of outputs that should be generated when executed. Each
output is declared using a suitable concept from an external ontology in which services’
IOPE-related semantics are captured. For each concept in the service request the platform
employs a distinct service discovery procedure that searches over the SPA services avail-
able in the service registry to locate those that produce an output (output concept) that
matches, hence satisfies, the particular service request concept and organizes them in a
matching services’ list. To populate this list, let us consider that two concepts match when
any of the three hierarchical relationships (exact, plugin, subsume) exists between them,
and a platform-generated SPARQL query conforming to the smart meeting room ontology
is issued to retrieve the corresponding services. An indicative SPARQL query formed by
a service discovery procedure that searches for services, the output of which matches, in
terms of either of the three hierarchical relationships; the Video concept determined in a
service request is illustrated in Chart 1.2. Services included in a matching list represent
alternative choices for satisfying a concept of the service request and are presented to the
developer who must select one to be incorporated into the composite service. Apparently,
a selection must be made for each service request concept. To ease selection, services in a
matching list are ordered according to the hierarchical relationship their output exhibits
w.r.t. the service request concept considered by the particular discovery process. Exact
matches precede plugin matches, which precede subsume matches, that is, the ordering is
done in increasing degree of semantic relaxation.
Chart 1.2: Example platform-generated SPARQL query for discovering SPA services
that output Video data. The sm and iope namespaces refer to the service model from
Section 7.4.2 and the external ontology where the Video concept is defined.
SELECT ?serv WHERE {

?serv a sm:Service.
?servsm:hasServiceOutput ?output.
{?outputsm:hasParameterTypeiope:Video.}
UNION
?type rdfs:subClassOfiope:Video.}
UNION
iope:Videordfs:subClassOf ?type.}
}
After selecting a service, it must be ensured that the service can be invoked, that is, all the
inputs of the service can be supplied with appropriate data, otherwise the composite service
will not execute. This is accomplished by retrieving the inputs (input concepts) of the ser-
vice and launching a separate service discovery procedure (analogous to the one described
above) for each input to recover those SPA services those SPA services that are available in the
service registry that produce an output (output concept) that matches the input; hence, they
provide suitable data for the input. Subsequently, for each input, the developer is prompted
to choose a service from the corresponding matching list that will be added to the compos-
ite service and a new round of service discovery and selection is initialized. This process
is repeated until there are no more SPA services participating in the composite service that
cannot yet be invoked. This way an interconnection scheme is semiautomatically formed
in the background and a partial ordering of the services is established. Note that S-type
services can be readily invoked as it do not have inputs (Section 7.3); thus, there is no need
to carry out service discovery if such a service is chosen. By progressively selecting S-type
services, it is possible to reach a state, in which all services are invokable and service compo-
sition terminates. After termination, the platform proceeds with realizing and executing the
composite service. Note that if, at any point, the service discovery procedure for an input
returns an empty matching list, the composition process fails, and no composite service is
created, as all available SPA services are deemed unsuitable.
Algorithm 2 Semi-Automatic Service Composition Procedure

Input: Dummy service SR, representing the service request
Output: Composite service CS, or
fail
1: Set Q = Ø// Queue structure with services as elements.

2: ENQUEUE(Q, SR)
3: repeat
4: S = DEQUEUE(Q)
5: for each input IS of service S do
6: L = SERVICE DISCOVERY(IS) // Matching services list.
7: if L == Ø then // If the list is empty.
8: return fail
9: end if
10: Prompt the developer to select a service SL from L
11: Add SL to CS to supply the data for input IS
12: if SL not an S-type service then
13: ENQUEUE(Q, SL)
14: end if
15: end for
16: until Q == Ø // Until all services can be invoked.
17: return CS
The semiautomatic service composition process is summarized in Algorithm 2, in which

the service request is represented as a dummy service without outputs, having the service
request concepts as inputs. A graphic example is depicted in Figure 7.4.
7.6 Composition of Complex Applications in an IoT Ecosystem: A Use Case

As has already been mentioned in Section 7.3, in the context of the SYNAISTHISI project,
a typical meeting room has been equipped with several sensors, interconnected through
the SYNAISTHISI platform, to provide S-type services. Moreover, a number of services
performing audiovisual analysis have been incorporated in order to provide cloud-based
P-type services. All these services have been semantically annotated using the proposed
smart meeting room ontology and form its instances. Following is a brief presentation of
some of these simple services, which are then utilized to compose a complex Person _
Counting service using both service composition approaches.
7.6.1 Available Services

The Person _ Tracking service is a P-type service that aims to continuously and
in real time detect and track people present in the smart meeting room. The algorithm
applied (Sgouropoulos et al., 2015) uses a blob- detection approach and is functioning
under the assumption that a moving blob at the foreground should belong to a human,
STEP 1: Matching list STEP 2: Developer STEP 3: Generated Repeat Until: S-type services are selected
I want a service of S(X) chooses S2 queries: S(A), S(B) process
Y S0 Y’ S’0
with output “X” S2 is an
X S1 A B
Specifies a service appropriate
service A B Result: Composed service...
request to X S2
discover services S0
with output “X” A* B* X S2
developer
A
X* Sk X S2 S’0
B
Platform generates an A* B*
S(X) ...senses the physical world and
appropriate query Let us assume that S2
X* SN I choose... produces output “X”
has inputs “A,” “B”
X* is an approx.
match of X
FIGURE 7.4
Semantic Complex Service Composition within an IoT Ecosystem
The process of semiautomatic service composition.

165
TABLE 7.1
S- and P-Type Services Utilized in Composing the Person _ Counting Service
Service Type Input(s) Output
Static_Camera S – RGB Video (RGB)
Stereo_Camera S – Depth Information (depth)
Dynamic_BG P RGB Video (RGB) Foreground (fg)
Person_Tracking P Foreground (fg), Depth Information (depth) Number of Persons (nop)
Person_Fusion P Number of Persons (nop) Fused Number of Persons (fnop)
as another moving entity is not expected to enter the meeting room. When a new per-
son enters the room, he/she is continuously tracked during his/her stay in the room.
The output of this service consists of the coordinates of the set of bounding boxes
that enclose the detected blobs, which are also enumerated; thus, their total number is
trivially calculated and outputted by the service. Accordingly, the number of tracked
objects is assumed to be the number of people present within the room.
However, in order for this P-type service to work, several other S- and P-type services
should interconnect and exchange information. More specifically, there exist several
RGB-D cameras that continuously capture video of the room’s interior. Each RGB-D cam-
era provides two services, one for RGB video and the other for depth. Thus, the available
S-type services are Static _ Camera and Stereo _ Camera, respectively.
The output of the Static _ Camera service is processed (i.e., is fed as input) by the
Dynamic _ BG service, a background subtraction module, which divides video content
into two parts based on motion: the background, which is discarded, and the foreground,
which consists the input (along with the depth dimension of the Stereo _ Camera ser-
vice) for the Person _ Tracking service that outputs the number of people present in
the foreground to the platform. The aforementioned services are summarized in Table 7.1.
As in general more than one camera may be used to surveil different parts of the
room, their outputs should be appropriately fused in order for the composed Person _
Counting service to provide an output that reflects the total number of people in the room.
Data fusion, in general, denotes the act of combining data from disparate and even het-
erogeneous sources, in order to obtain improved information compared to what is possible
when each source is used individually. To this goal, the implemented Person _ Fusion
service aims to combine measurements or raw data from heterogeneous or homogeneous
sensors. Specifically, Person _ Fusion is a P-type service, fed with the output of the
Person _ Tracking module produced by using two different cameras as input, that is,
from two different parts of the smart meeting room.
7.6.2 Manual Service Composition Example

Consider that the goal of the developer is to compose a Person _ Counting service, that
is, a service that estimates the number of the people present within the SYNAISTHISI
smart meeting room, by utilizing the two overlooking cameras. It is fair to assume that
the developer is familiar with basic computer vision tasks, thus is able to appropriately
interconnect two services upon observation of their inputs/outputs.
As has been already discussed, the room is fully covered by two RGB-D cameras (each
comprising of a Static_Camera and a Stereo_Camera service). Thus, the developer
first needs to discover these camera services. The developer is assumed to know that a
Person_Tracking service is needed and that it operates on the foreground of the scene
captured by a Static_Camera, and extracted by the Dynamic_BG service, whereas it
also requires, as input, the depth information of the scene produced by the corresponding
Stereo_Camera. He/she then interconnects the outputs of Dynamic_BG and Stereo_
Camera with the input of the Person_Tracking service. This process should be repeated
for both Stereo_Camera services. Also each Static_Camera must be interconnected
with the Dynamic_BG service to supply its input. Notice that the Person_Tracking ser-
vice is executed twice, once for each of the two cameras. The developer is aware that the
outputs of the two executions should be fused by the Person_Fusion service; therefore,
this service must be discovered and interconnected with the Person_Tracking service.
The Person_Fusion output corresponds to the desired output of the complex Person_
Counting service, that is, the total number of people in the room.
Service discovery is performed by issuing appropriate SPARQL queries to the smart
meeting room ontology via the SYNAISTHISI platform, as shown in Section 7.5.2. The
composed Person _ Counting service is illustrated in Figure 7.5. Note that the flow of
information between services should be considered before commencing the composition
process and that the developer is the only responsible for the correct interconnection of
the services.
7.6.3 Semiautomatic Service Composition Example

To present a use case of the semiautomatic service composition methodology, let us
revisit the previously described manual service composition example toward a Person_
Counting service. The goal of the developer is once again assumed to be the composition
of a service that counts the persons in the smart meeting room by utilizing the two cam-
eras. Here, the developer is assumed to be familiar with basic computer vision tasks (not
necessary as advanced as in the manual composition case), thus is able to understand the
flow of services suggested by the platform at each step and proceed to meaningful choices.
In Figure 7.6, the steps of the service composition process are illustrated. The process
begins with the developer’s intention to create a service with a specific functionality, that
P RGB S
fg
Dynamic BG
P Static camera
nop
depth
Person tracking S
Stereo camera
fnop P
Person fusion P RGB S
nop fg
P Dynamic BG Static camera
Person tracking depth S
Stereo camera
FIGURE 7.5
The composed Person _ Counting service using either manual or semiautomatic composition.
168
Step 1: Results of Step 2: Developer Step 3: Generated Step 4: Generated queries: Step 5: Generated
S(no_of_persons) chooses queries: query: S(RGB) × 2
I want a service S(no_of_persons) × 2 S(foreground) ×2 S(depth) × 2
that counts people person
person
fg person
fg stereo static
in the smart fusion is an fg static bg
nop nop RGB depth camera RGB camera
meeting room tracking appropriate tracking
depth depth
service ToF PTZ
single result dynamic RGB
Specifies a service nop nop fg RGB depth camera camera
person person bg
request to discover fnop fusion fnop Note: in some cases a
nop fusion
services with output nop single service is I choose I choose I choose
fnop: fused_no_of_persons discovered for a given
developer “no_of_persons” nop: no_of_persons dynamic bg stereo camera static camera
Note: the developer did output; the algorithm
fg: foreground not choose the exact advances automatically Note: the stereo Note: the static camera
The platform generates the
Note: fnop is a plugin match, but preferred the to the next step, camera is an S-type is an S-type service;
appropriate query
match of nop plugin match that is, without the service; no further no further queries
S(no_of_persons)
developer’s intervention queries required required
FIGURE 7.6
Semiautomatic service composition use case: Creating a person counting service. The depicted services’ IOPE concepts are assumed to be defined in an appropriate
external ontology.
is, he/she knows the service’s output. He/she expresses this intention via a service request,
and then two services are returned in the matching list. Even though the output of the
second service (namely Person _ Fusion) is not an exact match, but a plugin match,* his/
her intention to interconnect more than one cameras, along with his/her domain knowl-
edge, constitute the second service a preferable option over the first. The platform then
autoqueries, by generating suitable SPARQL queries of the form shown in Section 7.5.3, the
ontology and a single matching service, namely Person _ Tracking, that matches both
queries (one query for each input of the Person _ Fusion service) is returned. A further
autoquery allows the developer to select the next services, namely Dynamic _ BG and
Stereo _ Camera, thus satisfy the inputs of Person _ Tracking. Stereo _ Camera is
an S-type service, thus does not trigger another autoquery. Finally the process is termi-
nated when the developer selects another S-type service, namely a Static _ Camera to
supply the input of Dynamic _ BG. The composed service is the same as in the manual
composition example and is illustrated in Figure 7.5.
7.7 Discussion
This section discusses various aspects concerning the smart meeting room ontology and
the two service composition methods. Regarding the ontology, its most prominent char-
acteristics are (1) the utilization of existing and well-established semantic models; (2) the
ease of use in instantiating a smart meeting room, as the domain-specific nature of the
ontology means that not only high-level, but also low-level concepts are provided; and
(3) the ability to readily extend the ontology to include a new kind of SPA service, by add-
ing a new class in the resource-service type hierarchy. On the downside, as the ontology is
oriented toward modeling SPA services of smart meeting rooms, adaptation is necessary
so as to apply it on another domain.
Regarding the service composition, the exploitation of semantics allows one to overcome
syntactic barriers, which, otherwise, could lead to failures in producing the requested
services—for example, when exact matches do not exist, and the inputs and outputs of
services are approximately matched through plugin and subsume relationships. Also note
that both approaches currently operate over the smart meeting room ontology; however,
they are not limited to it. If SPARQL queries are appropriately rephrased, they can be read-
ily applied over any other ontology. The manual composition is targeted toward develop-
ers who possess the necessary programming skills in order to manually discover services
and interconnect them. A severely limiting factor in this kind of composition is the need
to know in advance which services are available in the registry so they can be discovered.
Such an assumption is not realistic in IoT ecosystems with millions of services that are
dynamically added and removed. Also the whole process is error-prone, as every step is
manually executed by the developer. In the semiautomatic framework the involvement
of the developer is limited to the basic tasks of defining a service request and selecting
services from platform-generated matching lists; hence, the whole process evolves with
minimum human intervention. Service discovery and interconnection is performed by the
platform, alleviating this way the need of a priori knowledge about the available services,
* This is just an assumption in order to showcase the semantic matchmaking mechanism.

and also acting as safeguard against human errors. Composing a service now becomes a
quite straightforward task, which can be even accomplished by an experienced user who
is not a developer but is skillful enough to comprehend the semantics surrounding the
service descriptions.
The semiautomatic composition algorithm guarantees that if a composite service is
created, all concepts included in the service request will be satisfied, and all services
combined in the composite service can be invoked. A drawback of the algorithm described
in Section 7.5.3 is the possibility of failing to produce a composite service, although the
necessary services are available. This is attributed to the fact that the composition process
immediately stops when an empty matching list is encountered after running the service
discovery procedure for one of the inputs of an already selected service. However, this
limitation is easily circumvented if, instead of stopping, the composition process prompts
the developer to replace the service previously selected with an alternative one contained
in the corresponding matching list. If no alternatives are available, it is possible to revert to
an earlier step of the composition process, replace the service at that step, and resume the
composition process from there.
7.8 Conclusion
Among the main visions of IoT is to seamlessly interconnect heterogeneous devices. This
chapter has built upon the SYNAISTHISI platform (Pierris et al., 2015) that follows the
service-oriented architecture (SOA) of IoT and exposes available resources as SPA seman-
tic web services. This chapter primarily focused on service composition, which constitutes
one of the most important and demanding tasks in unlocking the full potential of the
service-oriented architecture of IoT. The objective of service composition is to combine a
number of services, available in a service registry, in order to satisfy some user need when
none of the available services in the registry can do so. Key steps in composing a complex
service are the discovery of suitable services from the registry and their appropriate inter-
connection so as to ensure a functional composition, that is, one that every service in the
composition can be invoked, which, in general, translates to that all inputs and precondi-
tions of a service are provided/satisfied from the outputs and the effects of some other
service(s) that is (are) part of the composite service.
The way service composition can be performed in both manual and semiautomatic man-
ner within the SYNAISTHISI platform was demonstrated. The manual approach requires
considerable user intervention in every step of the composition process, both to discover
the services and to interconnect them, whereas in the semiautomated, the user only
specifies a service request and chooses services out of automatically populated match-
ing services’ lists presented to him, thus implicitly defining the service interconnection
scheme. To achieve this, an ontology was utilized in order to add semantic content to
the SPA services descriptions so that complete and accurate information regarding the
functionality offered by each service can be easily extracted and to permit the semantic
matchmaking between the inputs and the outputs of services. The ontology is domain-
specific, providing all necessary high- and low-level concepts to model a service-oriented
smart meeting room, similar to the one considered in the SYNAISTHISI project use cases.
However, the proposed service composition methods can operate over any ontology if
SPARQL queries are adapted accordingly.
Acknowledgment
This work was part of the “SYNAISTHISI” project results. The project was co-financed by the
Greek General Secretariat for R&T, Ministry of Culture, Education & RA and the European
RDF of the EC under the Operational Program “Competitiveness and Entrepreneurship”
(OPCE II), in the action of Development Grants For Research Institutions (KRIPIS).
References
Akasiadis, C., Tzortzis, G., Spyrou, E., & Spyropoulos, C. (2015). Developing complex services in an
IoT ecosystem. In IEEE World Forum on Internet of Things (WF-IoT) (pp. 52–56).
Al-Fuqaha, A. I., Guizani, M., Mohammadi, M., Aledhari, M., & Ayyash, M. (2015). Internet of
Things: A survey on enabling technologies, protocols, and applications. IEEE Communications
Surveys and Tutorials, 17(4), 2347–2376.
Albreshne, A., & Pasquier, J. (2010). Semantic-based semi-automatic web service composition.
Computer Department, Switzerland.
Atzori, L., Iera, A., & Morabito, G. (2010). The Internet of Things: A survey. Computer Networks, 54(15),
2787–2805.
Bandyopadhyay, D., & Sen, J. (2011). Internet of things: Applications and challenges in technology
and standardization. Wireless Personal Communications, 58(1), 49–69.
Banks, A., & Gupta, R. (2014). MQTT protocol specification. Retrieved from http://docs.oasis-open.
org/mqtt/mqtt/v3.1.1/mqtt-v3.1.1.html.
De, S., Elsaleh, T., Barnaghi, P. M., & Meissner, S. (2012). An internet of things platform for real-world
and digital objects. Scalable Computing: Practice and Experience, 13(1), 45–58.
Gubbi, J., Buyya, R., Marusic, S., & Palaniswami, M. (2013). Internet of things (IoT): A vision,
architectural elements, and future directions. Future Generation Computer Systems, 29(7),
1645–1660.
Guo, B., Zhang, D., Wang, Z., Yu, Z., & Zhou, X. (2013). Opportunistic IoT: Exploring the harmoni-
ous interaction between human and the internet of things. Journal of Network and Computer
Applications, 36(6), 1531–1539.
Harris, S., & Seaborne, A. (2013). SPARQL 1.1 query language, W3C Recommendation. Retrieved from
http://www.w3.org/TR/sparql11-query/.
Hatzi, O., Vrakas, D., Nikolaidou, M., Bassiliades, N., Anagnostopoulos, D., & Vlahavas, I. (2012). An
integrated approach to automated semantic web service composition through planning. IEEE
Transactions on Services Computing, 5(3), 319–332.
Kopecký, J., Gomadam, K., & Vitvar, T. (2008). hRESTS: An HTML microformat for describing restful
web services. In IEEE/WIC/ACM International Conference on Web Intelligence (WI) (pp. 619–625).
Kopecký, J., Vitvar, T., Bournez, C., & Farrell, J. (2007). SAWSDL: Semantic annotations for WSDL and
XML schema. IEEE Internet Computing, 11(6), 60–67.
Liang, Q., Chakarapani, L. N., Su, S. Y., Chikkamagalur, R. N., & Lam, H. (2004). A semiautomatic
approach to composite web services discovery, description and invocation. International Journal
of Web Services Research (IJWSR), 1(4), 64–89.
Martin, D., Burstein, M., Hobbs, J., Lassila, O., McDermott, D., McIlraith, S., … Sycara, K. (2004).
OWL-S: Semantic markup for web services, W3C Member Submission. http://www.w3.org/
Submission/OWL-S.
McIlraith, S. A., Son, T. C., & Zeng, H. (2001). Semantic web services. IEEE Intelligent Systems, 16(2),
46–53.
Paolucci, M., Kawamura, T., Payne, T., & Sycara, K. (2002). Semantic matching of web services capabil-
ities. In International Semantic Web Conference (ISWC) (pp. 333–347). Berlin, Germany: Springer.
Pedrinaci, C., Liu, D., Maleshkova, M., Lambert, D., Kopecky, J., & Domingue, J. (2010). iServe: A
linked services publishing platform. In Ontology Repositories and Editors for the Semantic Web
Workshop at the Extended Semantic Web Conference (ESWC) (pp. 71–82). Hersonissos, Greece.
Pierris, G., Kothris, D., Spyrou, E., & Spyropoulos, C. (2015). SYNAISTHISI: An enabling platform
for the current internet of things ecosystem. In Panhellenic Conference on Informatics (PCI)
(pp. 439–444). Athens, Greece: ACM.
Rodríguez-Mier, P., Pedrinaci, C., Lama, M., & Mucientes, M. (2016). An integrated semantic web
service discovery and composition framework. IEEE Transactions on Services Computing, 9(4),
537–550.
Roman, D., Keller, U., Lausen, H., de Bruijn, J., Lara, R., Stollberg, M., … Fensel, D. (2005). Web service
modeling ontology. Applied Ontology, 1(1), 77–106.
Roman, D., Kopecký, J., Vitvar, T., Domingue, J., & Fensel, D. (2015). WSMO-Lite and hRESTS:
Lightweight semantic annotations for web services and RESTful APIs. Journal of Web Semantics,
31, 39–58.
Sfikas, G., Akasiadis, C., & Spyrou, E. (2015). Creating a smart room using an IoT approach. In
Artificial Intelligence and Internet of Things (AI-IoT) Workshop, in conjunction with the 9th Hellenic
Conference on Artificial Intelligence (SETN), Thessaloniki, Greece.
Sgouropoulos, D., Spyrou, E., Siantikos, G., & Giannakopoulos, T. (2015). Counting and tracking
people in a smart room: An IoT approach. In IEEE International Workshop on Semantic and Social
Media Adaptation and Personalization (SMAP) (pp. 1–5).
Sheng, Q. Z., Qiao, X., Vasilakos, A. V., Szabo, C., Bourne, S., & Xu, X. (2014). Web services composi-
tion: A decade’s overview. Information Sciences, 280, 218–238.
Stavropoulos, T. G., Vrakas, D., Vlachava, D., & Bassiliades, N. (2012). BOnSAI: A smart building
ontology for ambient intelligence. In International Conference on Web Intelligence, Mining and
Semantics (WIMS). Craiova, Romania: ACM.
Vitvar, T., Kopecký, J., Viskova, J., & Fensel, D. (2008). WSMO-Lite Annotations for Web Services.
In S. Bechhofer, M. Hauswirth, J. Hoffmann, & M. Koubarakis (Eds.), The Semantic Web:
Research and Applications. ESWC 2008. Lecture Notes in Computer Science, Vol. 5021. Berlin,
Germany: Springer.
8
Enabling Smart Cities through IoT:
The ALMANAC Way
Dario Bonino, Maria Teresa Delgado, Claudio Pastrone, Maurizio Spirito,

Alexandre Alapetite, Thomas Gilbert, Mathias Axling, Matts Ahlsen,
Peter Rosengren, Marco Jahn, Raphael Ahrens, Otilia Werner-Kytölä,
and Jose Angel Carvajal Soto
CONTENTS
8.1 Introduction ........................................................................................................................ 173
8.2 Related Works..................................................................................................................... 174
8.3 The ALMANAC Platform ................................................................................................. 176
8.3.1 Architecture Overview ......................................................................................... 176
8.3.2 Data Flow ................................................................................................................ 177
8.3.3 Federation ............................................................................................................... 179
8.4 Application Scenarios and Real-World Pilots ................................................................ 180
8.4.1 Waste Management Domain ................................................................................ 181
8.4.1.1 Smart Waste Pilot in Turin..................................................................... 181
8.4.1.2 Deployment Activities ............................................................................ 181
8.4.1.3 Technical Details ..................................................................................... 182
8.4.1.4 Citizen Engagement ................................................................................ 184
8.4.2 Water Management Domain ................................................................................ 185
8.4.2.1 Mobile iOS Application: My Water ....................................................... 187
8.4.2.2 City Management and Communication Application ........................ 188
8.5 Results ................................................................................................................................. 190
8.5.1 Waste Management Domain ................................................................................ 190
8.5.1.1 Smart Waste Pilot in Turin..................................................................... 190
8.5.1.2 Citizen-Centric Application Usability ................................................. 193
8.5.2 Water Management Domain ................................................................................ 198
8.5.2.1 Discussion ................................................................................................ 200
8.6 Conclusion .......................................................................................................................... 200
References..................................................................................................................................... 201
8.1 Introduction
Smart cities advocate future environments where sensor pervasiveness, data delivery and
exchange, and information mash-up enable better support of a variety of aspects in every-
day life. As this vision matures, it addresses several application scenarios and adoption
perspectives, which call for scalable, pervasive, flexible, and replicable infrastructures.
173
This need is currently fostering new design efforts to deliver architectures and platforms
that cover the requirements on performance, reuse, and device and system interoperability
capabilities. The adoption of Internet of Things (IoT) as smart city enabler is challenging.
On one hand, IoT technologies are widely recognized as key components of successful
smart cities deployment. They, in fact, allow one to effectively tackle scenarios involving
multitudes of heterogeneous devices and subsystems interconnected through a variety of
networks, with different reliability, connectivity, and performance. On the other hand, city
life involves humans, businesses, and administrative entities that shall effectively cooper-
ate by means of the IoT technological infrastructure.
This chapter describes the smart city-enabling path pursued by the ALMANAC project
to develop a service delivery platform specifically targeted to smart cities, by integrat-
ing IoT and capillary networks with telecommunication networks. According to the evi-
dences emerging from the project, successful IoT adoption in smart city scenarios is better
achieved through the following:
1. Sound and informed design of an IoT technological platform tailored to the spe-
cific needs of smart cities.
2. Early involvement of smart city actors, for example, public administrations, social
communities, and utility services.
3. Explicit account of data ownership and access policies for data exchanged within entities.
4. Direct involvement of citizens.
In this chapter, technological choices, their underlying rationale, and the resulting
ALMANAC platform architecture are described in depth. Moreover, application scenarios
targeted by the project are discussed, and evidence from a real-world pilot is reported. For
each professional domain targeted by the project, namely water and waste management,
requirements gathered are presented and related to the technological solutions developed
by ALMANAC. Particular emphasis will be devoted to the pilot installed in Turin, in which
the ALMANAC platform has been deployed and tested. Finally, insights gained from the
pilot, outcomes of the real-world experimentation, and citizen feedback will be discussed
in relation to IoT technologies adoption in smart cities.
8.2 Related Works

Over the last years, the research community has reached a quite clear consensus about
the necessity and usefulness of making cities smarter through the use of innovative tech-
nologies. Many studies have been carried out on market potential (Sma, 2015), possible
architectures (e.g., I-architecture [Sma, 2012], IoT-A [Bauer et al., 2013]), deployments, and
roadmaps for real-world exploitation (Vermesan, 2009). Nevertheless, smart city is still a
very active research field, with solutions that are still experiencing difficulties demonstrat-
ing their full potential in a smart city context.
To address the challenges emerging in this domain, researchers from industry and
academia propose approaches and platforms supporting the smartification of exist-
ing cities. On the industrial side, big firms such as IBM (Fritz et al., 2012, Hogan
et al., 2011), HP, BOSCH, Siemens, and Cisco are pursuing smart city solutions to
Enabling Smart Cities through IoT: The ALMANAC Way 175
enable cities to become smarter by connecting city events and data, possibly exploiting
integrated cloud solutions.
On the research side, several approaches address different aspects of smart city design
and exploitation, from sustainable deployment and growth, to next-generation semantics
powered smart city platforms (SCPs). In their paper, Vilajosana et al. (2013) investigate the
underlying reasons for the slow adoption of smart cities and propose a procedure based on
big data exploitation through the so-called application programming interface (API) stores
concept. A viable approach for scaling business within city ecosystems is discussed and
currently available ICT technologies are described, exemplifying all findings by means
of a sustainable smart city application. Although this approach describes the motivations
for smart cities and establishes a general framework for their adoption, the proposed
ALMANAC framework is more oriented to the actual implementation of a federated SCP.
Bellini et al. (2014) propose a system for data ingestion and reconciliation of smart city
aspects. The system allows managing a large volume of static and dynamic data coming
from a variety of sources. These data are mapped to a smart-city ontology, called KM4City,*
and stored into an RDF† store in which it is available for applications via SPARQL‡ queries,
providing new services to the users via specific applications for public administration
and/or enterprises. Examples of potential uses of the knowledge base produced by the
Bellini et al. (2014) system are also offered, and they are accessible from the resource
description framework (RDF) store and related services. The framework and platform
architecture presented in this paper share several design principles with the approach of
KM4City, that is, semantic modeling of smart-city entities and big data approach to data
capture. However, while the former is more focused on ontology and dataset population,
ALMANAC is more concerned with federation of several instances and direct exploitation
of platform services by third-party application developers.
Several research projects are currently investigating aspects related to smart city design
and adoption both in Europe and the United States, among them:
• The Smart Santander project§ provides a city-scale experimental research facil-

ity in support of applications and services for a smart city. ALMANAC solutions
already integrate data managed by Smart Santander thanks to a strong collabora-
tion between the two projects.
• The Urban Water project,¶ installing smart meters in Almerı́a (Spain) to achieve
greater efficiencies in water management.
• The Open IoT project** defines an open-source cloud solution for the IoT, which
can be seen as a technological enabler for SCPs such as the one presented in this
chapter.
• The Mobosens U.S. research project†† provides citizens with a platform for col-
lecting and sharing environmental data, from stream quality to drinking water
safety.
* Knowledge Model for City.

† Resource Description Framework https://www.w3.org/RDF/, last visited September, 2016.
‡ Query Language for RDF https://www.w3.org/TR/rdf-sparql-query/, last visited September, 2016.
§ http://www.smartsantander.eu/, last visited March, 2015.
¶ http://urbanwater-ict.eu/, last visited March, 2015.
** http://openiot.eu/, last visited March, 2015.
†† http://nanobionics.mntl.illinois.edu/mobosens/, last visited March, 2015.
8.3 The ALMANAC Platform

The ALMANAC SCP collects, aggregates, and analyses real-time data from resources
(appliances, sensors and actuators, smart meters, etc.) deployed to implement smart cities.
Three specific applications, waste management, water supply, and citizens’ engagement,
have been selected for proof-of-concept implementation and evaluation.
8.3.1 Architecture Overview

Figure 8.1 depicts the logic view of the final ALMANAC architecture. It can be described
along two main axes, by vertical and horizontal divisions: the functional axis and the sub-
system axis. Although the functional axis is rendered as a stacked set of platform modules
each deserving specific documentation,* the subsystem axis (mainly horizontal) can be
exploited to gain a quick understanding of the main platform functions.
Subsystems composing the platform encompass from top to bottom:
• The cloud-based APIs, including all modules acting as service access point for exter-
nal, possibly third-party applications exploiting the ALMANAC platform. They
mostly include REST end points mapping to the offered platform functionalities.
Cloud-based APIs
Smart city resource Historical data Data fusion Provisioning Management

Live data API
library API API service API API API
Virtualization layer Security and privacy framework Network

VL policy
enforcement point
Linksmart global connect
Virtualization
layer core P2P backbone
Network Trust Security router/
manager manager manager backbones
Semantic representation Data management

framework framework Federated identity
Resource
catalog manager
Event broker
Metadata Data fusion DL policy
Smart city framework manager enforcement point
ontologies
Storage manager
Abstraction layer
SCRAL
SCRAL SCRAL device SCRAL policy

SCRAL core M2M
connectors managers enforcement point
Smart city resources
FIGURE 8.1
The ALMANAC logic architecture overview.
* Interested readers may refer to Bonino et al. (2015).

• The virtualization layer, which orchestrates the platform modules to fulfill Cloud
APIs requests, both within a single platform instance (PI) and among different
instances (federation). Moreover, it supports search, lookup, and addressing of
smart city resources and services.
• The semantic representation framework, which encompasses updated smart city mod-
els based on ontologies as well as a metadata management framework enabling
effective handling of machine understandable metadata on smart city resources.
• The data management framework, grouping all data-management modules that han-
dle effective storage, elaboration and fusion of live data, and that offer directory
services for resources registered in the platform.
• The abstraction layer, which provides technology independent, uniform access to
physical resources and maps both live and off-line metadata into shared, standard
representations that can be easily handled by upper platform modules.
• The security framework, which crosses all layers and provides services and utilities
to secure communication between platform modules, to check access permissions
and verify information disclosure on the basis of a policy mechanism. The frame-
work encompasses several enforcement end points, where application/user per-
missions are checked and accept/deny actions applied, and one federated identity
manager handling roles, permissions, and supporting end-to-end control on trans-
ferred data, both inside and outside of the platform.
• The networking subsystem, supporting effective communication between platform
modules, among different platforms (federation) and between the platform and
the outside world (e.g., through M2M communication).
8.3.2 Data Flow

The main information source in ALMANAC is the heterogeneous sensing network con-
stituting the smart city nervous system. Similar to capillaries in living beings, sensors
deployed in the city capture vital information on city processes and states (e.g., pollution
level, traffic, waste levels) and deliver it to the core elaboration center represented by the
ALMANAC platform. In this path, every single bit of information traverses a complex
set of modules, each performing a different operation, either on the data itself or on the
so-called metadata (i.e., information about delivered data). Raw values generated by
sensors are gathered by a capillary communication network, or directly by the smart
city resource abstraction layer (SCRAL) and undergo several checks and modifications
to transform each measurement point into meaningful and exploitable information. For
each sensor-generating data stream, a corresponding metadata description is gener-
ated, and kept up to date, exploiting standard representations (OGC SensorThings API*),
open data, and semantic models. The resulting metadata is attached to the data stream
and exploited to better characterize every single measure. For example, the raw tem-
perature value of 20°C measured by a DIY† sensor located in the center of Copenhagen
is enriched at SCRAL level with metadata about the source. This means that at SCRAL
level, the raw temperature value becomes an OGC Sensor Things API observation
* https://ogc-iot.github.io/ogc-iot-api/, last visited September, 2016.

† Do It Yourself.
SCRAL
Direct SmartBin
connection driver
ZigBee Event MQTT

Raw data Rich data
driver bus publisher
M2M
capillary M2M driver
Standard models,
External open data, domain ontologies
sensor-specific metadata
FIGURE 8.2
Data flow from the city to the ALMANAC lower layers.
referred to a well-defined property (e.g., the outside temperature), with a precise time
stamp, a standard unit of measure, and a geographical positioning in terms of latitude
and longitude. Attached to this bit of information, additional (meta-)data permits to
identify the sensor owner, for example, a private user, with medium access level, and
possibly data-quality indicators, for example, amateur-level sensor versus professional
measure system.
At the upper boundary of the SCRAL, data injected in the platform is no longer raw
and uncategorized. Instead, it is rich and expressed in a shared, standard, uniform, and
machine-understandable format, which can easily be handled by higher platform levels
where intelligent elaboration takes place (see Figure 8.2, which summarizes this first step).
Enriched data flow into the platform along several parallel paths (Figure 8.3). First,
data are captured by the platform storage manager (SM) which provides persistence
to gathered data. In parallel, the same data flow (selectively) into the data fusion man-
ager, which applies complex event-processing operators to the incoming data streams
and generates new (rich) data, either referred to same originating sensors (e.g., tempo-
ral aggregations, statistical properties) or pertaining to completely new sources, that
is, virtual sensors generated by combining together different data streams. New data
are in turn routed toward the SM for persistence. Although both the SM and the data
fusion manager handle live data, that is, measures plus metadata, the metadata infor-
mation flow is also routed to the resource catalog component, which is responsible for
maintaining a directory of currently connected data sources with its corresponding
metadata. Metadata can be further enriched by exploiting the inference mechanisms
provided by the metadata framework, which hosts domain ontologies exploited by the
ALMANAC platform.
At the highest architectural layer of the ALMANAC platform, rich data and metadata
are made available to end users, that is, applications using the ALMANAC Cloud APIs
and to other federated platforms. Data are either delivered through REST or Web Socket.
This layer can be seen as the end point of the field-to-platform information flow, and the
boundary at which the user-to-platform information flow begins.
Metadata
framework
Resource Metadata
catalog
Rich data
Storage
manager Historical rich data
SCRAL
Fused rich
data
Data fusion
Fused rich data
manager
Rich data
FIGURE 8.3
Data flow, from SCRAL to data management.
8.3.3 Federation
Every smart city is peculiar, with its own metering infrastructure, its own administrative
constraints, and its variegated set of actors and entities interacting together in everyday
city processes.
This scenario often leads to tight requirements on ICT systems, which are required
to support different—sometimes collaborative—entities, while having the ability to
enable independent operations of connected entities. Differently from several exist-
ing state-of-the-art approaches, which concentrate on technical requirements and only
consider social and legal interactions at a later design stage, ALMANAC tackles these
issues from the very early design phases. Interaction of entities involved in smart cit-
ies is addressed by means of the federation concept, an extension of the well-known
cloud federation idea (Rochwerger et al., 2009) to real-world city requirements and use
cases. Broadly speaking, the term cloud federation refers to interoperability of different
cloud systems as if they were one, with each single system maintaining a firm control
on shared information and functions. This concept can be extended quite well to smart
cities where different stakeholders need to work together as a single entity depending
on the current context and needs. The main challenge is addressing the fact that con-
text and stakeholders being part of a federation might change in time, and that their
exchanges are shaped and constrained by nontechnical boundaries set by policies and
regulations, which may also vary in time.
ALMANAC defines a framework in which multiple PIs can participate in federations
(Figure 8.4) offering cross-city, cross-nation, and cross-entity services. Such federations
are typically built around service-exchange agreements that consider the involved politi-
cal and legal issues. From a technical standpoint, they are deployed by establishing sets
Federation 2
Federation 1
ALMANAC P.I.
(Aarhus)
ALMANAC P.I.
(Turin municipality)
Federation 3
ALMANAC P.I.
(A.M.I.A.T.)
ALMANAC P.I.
(Copenhagen)
ALMANAC P.I.
(Brussels)
Federation 1—Turin and AMIAT (Waste service provider)
Federation 2—Turin, Copenhagen, and Aarhus
Federation 4
Federation 3—Aarhus and Copenhagen municipalities
Federation 4—Turin, Brussels, and Copenhagen municipalities
FIGURE 8.4
Multiple federations between ALMANAC platform instances.
of trust domains including two or more instances. PIs exchange data and distribute tasks
according to roles and privileges defined in the federation agreement (e.g., in a simplis-
tic on/off case, all users/platforms belonging to the federation are allowed to perform
tasks, whereas external platforms/users cannot exploit the federated services), and man-
aged through state-of-the-art solutions for federated identity and role management, for
example, based on the SAML* specification.
8.4 Application Scenarios and Real-World Pilots

Since the beginning of the ALMANAC project, a scenario-thinking approach has been
implemented to analyze and cocreate the selected application scenarios with relevant
stakeholders. Although the development of the ALMANAC SCP has been tailored to be
easily scalable to other cities and domains, the validation of ALMANAC results in the
waste and citizen-centric domains has been performed in the city of Turin, whereas activi-
ties regarding application of the ALMANAC platform in water management scenarios are
referred to a small pilot in Denmark.
* https://www.oasis-open.org/committees/tc_home.php?wg_abbrev=security, last visited September, 2016.

8.4.1 Waste Management Domain

In order to get a deeper insight into the waste domain in the Turin context, Turin’s
Municipality—which is one of the partners of the ALMANAC consortium—played an
important role in providing domain knowledge.
Thanks to the knowledge gained from the Environmental Department of the City of
Turin, the project learned that the waste management domain has mainly three protago-
nists: the citizens, the utility service operator (Amiat*), and the public administration. The
latter acts as both client and controller of the services provided by Amiat, while also behav-
ing as the interface between this and the general public, for example, by gathering sugges-
tions and complaints from the citizens concerning the services.
To include all relevant perspectives, these three main actors were engaged in a cocreation
process that puts in place ordinary workshop tools together with a set of ethnographic
tools, including field visits, scenario observations, and interviews with people from the
city environmental office and from the utility company.
From these activities, information related to the three main types of waste collection
methods in Turin was gathered: (a) street-based collection, based on waste containers
placed on the streets; (b) curbside collection, based on small waste bins located directly
inside households; and (c) underground ecological islands (UEIs), which are underground
containers with a large storage capacity (5 m3).
Regarding the quality of collected waste, street-based collection shows the lowest: recy-
clable waste is often of such poor quality it cannot be reused and ends up as unsorted
waste that is taken to an incineration plant or landfill. On the contrary, the quality of waste
collected through curbside collection is the highest. Citizens might feel a higher degree
of ownerships and pay more attention to the way waste is sorted, whereas it is easier for
the city to apply individual fines based on the quality of the collected waste. Nonetheless,
curbside collection is not always feasible due to space, aesthetic, and health reasons. UEIs
are not deemed to be as effective as the curbside collection (e.g., frequent abandoned waste
issues around the UEI); however, they are a good alternative to curbside collection both in
terms of the citizens feeling of ownership and in terms of collection optimization.
8.4.1.1 Smart Waste Pilot in Turin

From a technological point of view, the waste domain represents an interesting application
for the solutions developed by ALMANAC, especially for those considering federation of
IoT platforms operated by different owners—enabling data sharing and reuse of manage-
ment functionalities in a safe and controlled way—to optimize a service of common inter-
est. For these reasons, two UEIs were selected to deploy technology solutions considered
in ALMANAC. Data referred to waste collection in selected locations encompass both fill
level and weight measurements from collection trucks serving the UEIs.
The two locations selected for the pilot deployment are located in Turin city center, see
Figure 8.5, and they serve around 300 families and 90 businesses.
8.4.1.2 Deployment Activities

Deployment activities in the two selected UEIs started during August 2015. Fill-level
sensors and solar panels were installed in each waste container, and the two nonrecy-
clable waste containers were also equipped with radio-frequency identification (RFID)
* Azienda Multiservizi Igiene Ambientale Torino.
FIGURE 8.5
The sites selected for the ALMANAC project pilot.
access-control modules to grant exclusive access to the citizens involved in the experi-
mentation. During September 2015, the municipality proceeded to communicate to the
residents and businesses involved all relevant information. Users were informed about
the starting date of the trial and briefed on the new conferral mode for the nonrecyclable
waste that would include the use of a magnetic key to access the respective container.
During this time, Amiat validated the correct installation and corroborated the normal
functioning of fill-level sensors, solar panels, radio modules, and RFID-controlled access
modules, whereas the ALMANAC staff started the integration of the data generated by
these devices into the ALMANAC platform and reported any anomalies founded in the
collected data. On September 28th 2015, the experimentation officially started which is
foreseen to last until the end of 2016.
To assess the citizens’ perception of the experimentation activities, ALMANAC distrib-
uted a short questionnaire aimed at evaluating the perceived benefits after the end of the
trial period. The questionnaire was given to users involved in the experimentation and
was also available online.* A summary of the deployment-related activities described so far
is presented in Table 8.1.
8.4.1.3 Technical Details

Each site hosts UEI equipped with four waste containers† collecting the organic, glass,
paper, and not-recyclable fractions (Figure 8.6).
* https://it.surveymonkey.com/r/B2WJD3H.
† Funded by Amiat, the Turin Waste utility, and partly supported by Nord Engineering, a third party provider,
see http: //www.nordengineering.com/en/, last visited April 20, 2016.
TABLE 8.1
Summary of the ALMANAC Pilot Deployment Activities
Activity Date
UEIs equipment upgrade (sensors, solar-panels, and controlled access August 15–31, 2015
modules installation)
Informative letters sent to residents and businesses involved September 1–14, 2015
Key delivery to involved businesses September 1–26, 2015
Key delivery in city offices for involved residents September 14–26, 2015
Questionnaire to assess the perceived benefits of the deployment activities May 1, 2016
FIGURE 8.6
One of the UEIs involved in the ALMANAC pilot, located in Via Porta Palatina.
All containers were equipped with commercial-grade wireless fill-level sensors

provided by EcoTec* (specifications reported in Table 8.2) and kindly funded by Amiat.
Nonrecyclable containers carry an RFID-based access control system that allows waste
disposal from authorized citizens only. All sensors communicate with a central, legacy,
data collection service also provided by EcoTec. This setting reflects quite well the typi-
cal smart city scenario, where administrative constraints and commercial aspects are
strictly intertwined, often leading to deployment of heterogeneous data collection infra-
structures that shall be coordinated and operated as a single complex SCP.
The legacy data collection system gathering data from the pilot UEIs was integrated
into the ALMANAC platform thanks to a dedicated device manager, part of the platform
resource adaptation layer. Data collected every hour and at every waste disposal (for the
nonrecyclable fraction) are fed into the platform and stored for further processing and
* http://www.ecotecsolution.com/, last visited April 20, 2016.

TABLE 8.2
Specifications of Smart Waste Bins Deployed in the Pilot
Ecotec Smart Waste Bin
Power supply 12 V
#loggable disposal operation 250,000
#rfid keys 16,000
RFID reader 125 kHz or 13.56 MHz
Data transmission GSM/GPRS
Fill Level Sensor
Current rating <50 mA
Distance range 200–400 mm
Output range 4–20 mA
Distance resolution 4 mm
Operative temperature range −25°C + 70°C
elaboration. Collected data are available through the ALMANAC Cloud APIs, which fully
subsume the OGC SensorThings API SensingProfile,* thus offering standard means for
querying/retrieving specific data ranges.
8.4.1.4 Citizen Engagement

Citizen engagement in city processes is one of the crucial factors for successful deployment
of smart cities. ALMANAC sought citizen involvement from the beginning of the proj-
ect. Two main activities were carried respectively targeted at (a) attracting, engaging, and
keeping the pace with citizens living in the city quarter hosting the UEI pilot and (b) sup-
porting the whole citizenship in the tasks of duly disposing recycled waste.
8.4.1.4.1 Involvement in Pilot Activities

As part of the pilot deployment activities in Turin’s city center UEIs, the municipality together
with ALMANAC puts in place a campaign to engage citizens and promote more sustain-
able behaviors. General information regarding the ALMANAC project, the scope of the pilot
activities, and recycling educational material were distributed to the households and busi-
nesses involved in the experimentation. During the final phase of the experimentation, a
questionnaire was distributed to collect the perceived benefits and disadvantages of the UEIs
experimentation. Questionnaires were analyzed to assess the success of the experimentation
and to provide recommendations for a possible extension of the pilot to other UEIs in the city
in a near future. Involved citizens were invited to a final meeting where they were given the
opportunity to express themselves and provide further insights on the subject.
8.4.1.4.2 Recycling Support

In parallel with these activities, ALMANAC implemented a cocreation approach
(Figure 8.7) to design and develop a recycling support tool engaging the citizens of Turin.
The codesign activities have been carried out in collaboration with a selected community
from a temporary social housing project in the city of Turin† to mutually produce the
desired value outcome.
* http://ogc-iot.github.io/ogc-iot-api/, last visited April 29, 2016.

† http://www.sharing.to.it/index.php/en.html, last visited April, 2016.
FIGURE 8.7
Codesign workshop activities with a group of Turin citizens.
Interaction with the community consisted of a set of interviews and a one-off work-
shop, organized over a six-month timespan and aimed at designing an effectively engag-
ing recycling-support application. Although the heterogeneous and dynamic nature of
the sample did not allow for participation over a longer period of time, the quality of the
interactions compensated for fewer encounters and kept the community willingly and
happy to engage in the process.
Main findings from the codesign workshop included people facing difficulties in the
recycling process (e.g., what goes where?), thus becoming frustrated and developing bad
recycling habits. Also, the need to find a specific recycling bin emerged during the work-
shop, due to overfilled bins in the area or simply to dispose a specific packaging, for exam-
ple, a plastic bottle of water, while walking around in the city.
Needs and desires of the citizens concerning recycling and waste disposal were collected
and translated into high-level requirements for a recycling supporting mobile application
The WasteApp. The developed application has four (4) basic functions (Figure 8.8): (1) waste
collection calendar, (2) waste recycling guide extended with crowd-sourcing techniques for
harvesting waste classification from users and enabling direct bar code scanning, (3) waste
bin location map, and (4) drop-off centers map. The process of cocreation, the workshop
results, and the high-level requirements and related features of the mobile application
developed within ALMANAC are described in Bonino et al. (2016).
8.4.2 Water Management Domain

Water management is another interesting domain that can be exploited for benefiting the
whole smart city ecosystem. For the water utilities, the deployment of smart metering
devices (with flow meters) can bring huge economical and technical benefits, saving on
RELIABLE SMART SECURE

INTERNET OF THINGS FOR SMART CITIES
FIGURE 8.8
The WasteApp Home showing the four main features.
metering costs, and enabling advanced features such as real-time detection of the reverse
flow and tampering. Once the water management infrastructure is in place, smart meter-
ing devices can be used to provide citizens immediate feedback on consumption, alert
them about leakages in private water pipes, and so on. As a result, water management
applications can increase citizens’ awareness of smart city processes and promote moti-
vational schemes for sustainable behaviors through active engagement in individual and
community activities supported by new business models.
Water management applications developed in ALMANAC are linked to the require-
ments related to the consumption awareness scenario analyzed in the process. The areas
of interest are as follows:
• Showing user information about their own consumption, both historical and
current
• Aggregating results and showing consumption trends
• Sharing consumption data on social media
• Notification of leaks and other events.
<<component>> Cloud API

Data management framework (DMF)
Aggregated observations
<<component>>
<<component>>
VirtualizationLayer
DataFusionManager
Async query API
Persistent observations Raw observation
<<component>>
StorageManager
Sync query API
<<component>>
<<component>>
ResourceCatalog
Async thing registration SmartCity abstraction
layer (SCRAL)
FIGURE 8.9
ALMANAC Cloud APIs exposed by the virtualization layer and exploited to build the water management
applications.
Data flow from a WM Bus smart meter deployed in Aarhus to the ALMANAC SCP, and
then into the end-user mobile application. The applications show integrations with both
the SCP and external providers like Microsoft Azure and Apple apple push notification
service (APNS). Just as important as the actual applications are the libraries that have been
developed. These are now available not only as source code on GitHub but also as easy-to-
consume libraries that can be included in other applications.
The water management applications communicate with the open end points of the
ALMANAC SCP. This means that all communication is done through the virtualiza-
tion layer (Figure 8.9), which forward requests to all the relevant platform modules—and
enforces the security and policy frameworks configured within the ALMANAC SCP.
The water management application suite is composed of two separate software com-
ponents: (a) a mobile iOS application for citizens and water consumers, connected to the
ALMANAC SCP providing information about water consumption pertaining to the smart
meter installed in the user’s home; and (b) a city management and communication appli-
cation where city officials can push notifications to their citizens through the mobile app,
which is a desktop application designed to allow employees of the water utility or the city
to transmit notifications to users who live in the city.
8.4.2.1 Mobile iOS Application: My Water

My Water is an end-user application, which provides multiple views on the user’s water
consumption: it shows current consumption and compares it to the average from the last
30 days, and shows alerts and notifications. For normal use, the end user has to log-
in using the ALMANAC security and policy framework. This is done by loading an
FIGURE 8.10
Security and log-in in My Water.
external log-in page from within the application, which returns the necessary security
tokens to the application logic. The application can afterward communicate freely with
the ALMANAC SCP.
The security framework (Figure 8.10) is completely decoupled from the rest of the appli-
cation and is based on the techniques used by Google and Facebook. This allows the
application to seamlessly switch between security frameworks, without having to modify
core application logic.
Once the user is logged in, there are three tabs available to the user. One showing
current consumption, or for a specific day; another with graphs showing consump-
tion trends and more historical data; and finally one page showing historical alerts
(Figure 8.11).
The application also supports sharing personal consumption using the Facebook social
media. The user can choose to share today’s consumption with the public or just friends—
by clicking the thumbs up icon (Figure 8.12).
8.4.2.2 City Management and Communication Application

Through this desktop application, it is possible for a utility company to notify citizens of
water quality issues, for example, if a pipe has burst or if contaminants have been detected
somewhere in the water network infrastructure.
FIGURE 8.11
The My Water app.
FIGURE 8.12
Facebook-like button and feedback sharing page.
FIGURE 8.13
City notification service application.
The application can deliver notifications (Figure 8.13) to residents of specific target areas
or to the entire city, if needed.
8.5 Results
To better understand the impact of the ALMANAC Way of implementing a smart city ICT
infrastructure, the project consortium pursued several evaluation activities including: (a)
evaluation of the overall platform scalability, (b) evaluation of developed sample applica-
tions in terms of usability and willingness to adopt by end users, and (c) evaluation of
real-world deployment of the platform in terms of user satisfaction and perceived impact
on the user behavior.
Although full reporting of evidence and outcomes from experimentation activities is
slightly out of scope for this chapter, this section summarizes main results and relevant
insights gained during the experimentation of the ALMANAC approach.
8.5.1 Waste Management Domain

8.5.1.1 Smart Waste Pilot in Turin
To start understanding how citizens perceived pilot activities, a brief questionnaire (seven
questions) was designed and distributed among residents who were involved in the pilot.
The questionnaire was available in both paper form and online, focused on the waste
disposal habits, including assessment of collection sites neatness and organization, and
accessibility of waste containers. Questions are reported in Table 8.3. Readers must be
aware that translation from Italian to English might not retain all the nuances and careful
formulation of the survey questions.
TABLE 8.3
Turin Pilot Evaluation Questionnaire
Question Number Question Text
1 Do you think that using the electronic keys is easy?

2 Do you think controlled access to unsorted waste bins provides emptier containers?
3 Do you think that access control has contributed to a better waste differentiation?
4 During the trial, have you changed the way you sort waste at home?
5 Do you think the amount of abandoned waste has increased close to the underground
ecological islands?
6 Do you think that the installation of underground ecological islands with controlled
access would be an efficient way to improve the quality of waste collection?
7 Overall, is your opinion of the trial with the underground ecological islands positive?
The amount of questions in the questionnaire was purposely kept low with the goal
of reducing its complexity and obtrusiveness. Each question consisted of a statement for
which citizens were required to express their relative agreement using a 5-value Likert
scale ranging from strongly disagree to strongly agree.
8.5.1.1.1 Survey Outcomes

Out of 400 citizens and businesses who were engaged in the UEIs experimentation, only a
group of 30 people provided their opinions and perceptions through the questionnaire. This
could be related to the fact that no follow-up activities were taken by the partners deputed
to administer the survey (Turin municipality and Amiat) to promote a stronger citizen
engagement. Although results have low statistical significance due to poor coverage of the
pilot user base, they provide interesting insights on the user perception of pilot activities.
Several unexpected findings emerged from the questionnaire answers also fostered by the
availability of open comment spaces in all questions. Most participants took advantage of
this space to clarify their answers, often explaining motivations underlying their responses.
Only 2 out 30 persons submitted partially filled questionnaires, skipping Question 2 and 3,
respectively. In the following paragraphs, answers to the questions are discussed in detail
summarizing survey findings and discussing the corresponding outcomes.
Question 1: Question 1 aimed at understanding to which degree access control

is perceived as negative and annoying as feared by some of the most prudent
administrators. Survey answers demonstrate that there is no negative connotation
associated to access control and no particular complaints were registered about
the access key adoption. Overall, positive judgments covered 86% of the available
answers.
Question 2: Questions 2 and 3 aimed at verifying the assumption that forcing access
control on the nonrecyclable waste would trigger better recycling behaviors by
increasing the share of recycled materials. This design choice however did not
account for possible counter effects such as reduced quality of the other fractions
collected due to improper recycling or for possible increases of abandoned waste.
The results for this question show that 75% of the citizens found that the containers
for unsorted waste in UEI were emptier than traditional containers. The remaining
24% did not find any difference, possibly because visual inspection is physically
impossible in the UEI container, thus reducing the answers to pure perceptions.
Question 3: Due to the impossibility of performing visual inspection of the actual con-
tainer fill-level, answers to Question 3 are again purely perceptual. Nevertheless,
by looking at the question outcomes, citizens involved in the pilot feel that the
overall recycling efficiency has been improved by the access control measures: 70%
of them, in fact, provided a positive answer to the question, whereas only 30%
reported no changes.
Question 4: Although Questions 2 and 3 invited people to comment on other peo-
ple’s behavior and usage of the UEIs, Question 4 was related to changes in per-
sonal behaviors. In this case, the pilot seems to have forced some slight change in
the recycling behavior of interviewed people. Around 53% of answers reflected
changes in the way waste is sorted at home, which range from slight to significant
(only 13% reported disruptive changes). Nearly half of the respondents (47%) said
that they had not changed their behavior.
Question 5: Question 5 aimed at detecting if restraining the access to the nonrecyclable
container had a negative effect, for example, increase of the waste abandoned in
the proximity of the UEIs. Evidence from the survey demonstrates, in contrast with
many user comments, that there was no counter effect in terms of abandoned waste.
Quality of recycled materials could have decreased, but this was not captured by the
questionnaire. Out of the 30 respondents, 57% judged that abandoned waste actually
decreased during the experimentation, whereas 43% did not notice any change.
Question 6: Question 6 was aimed at understanding whether people think that
the installation of the UEIs with controlled access would be an efficient way to
improve the quality of waste collection. Almost 43% of the respondents think that
controlled access brings little or no changes in recycling behaviors, whereas the
majority (57%) thinks that it would actually be beneficial.
Question 7: Question 7 wanted to sum up citizens’ overall experience achieving over-
all positive results (80%), thus confirming the acceptance and viability of the pilot.
8.5.1.1.2 Outcomes from the Meeting with Public Authorities

As expected, citizens who attended the public meeting organized by Torino, with the
participation of representatives from istituto superiore mario boella (ISMB) and Amiat
provided feedback in line with the comments written in the questionnaires. Additionally,
they also provided some unrelated, yet interesting context information regarding noise
pollution due to glass disposal late at night, mainly from restaurants and other businesses
around. Among suggestions, the most recurring was enforcing by low glass disposal
during the daytime, to reduce noise annoyances during night hours. In line with this, citi-
zens proposed to extend access control to all waste containers, to reduce, or at least trace,
behaviors that might hamper the quietness of the neighborhood.
8.5.1.1.3 Discussion
Overall, the pilot in Turin was positively perceived by users and provided precious insights
to both city administrators and waste utility operators. A more structured approach at
surveying user perception together with follow-up techniques is needed to improve the
base of collaborative users willing to aid the city in identifying best ways to improve waste
management. Further investigations in this sense are needed, both to cover the several
dark points emerging from this survey and to better discern actual facts from feelings.
Extending the pilot to more locations in different city quarters would help in spotting and
compensating geographical and social biases. Moreover, it would provide a larger user set,
which together with proper follow-up techniques should provide statistical significance to
survey outcomes that at the time of writing are purely qualitative.
8.5.1.2 Citizen-Centric Application Usability

The citizen-centric application (namely WasteApp) was tested for usability, with a selected
group of users from the SHARING community. This preliminary test was aimed at spotting
the main usability issues in the prototype, thus allowing corrective actions to be taken before
final, long-term trials with users from SHARING and other connected communities in Turin.
To evaluate usability of the interactive prototype, four participants used the Citizen-
centric application in a controlled environment, performing six tasks each (Table 8.4).
Tasks were planned taking into account requirements and needs emerging from the
codesign workshop and were tailored to fit the ALMANAC waste scenario. The set of
tasks performed by the participants involved in the usability study reflect the typical
interactions we envision for the proposed application. Participants never met during the
evaluation. Their observations, together with data collected about their sessions with the
citizen-centric app, allowed us to perform a qualitative analysis of the application usabil-
ity, and helped to identify strong and weak points of the interface as well as to direct
future works and research activities.
Our analysis focuses on five basic questions about the usability of the citizen-centric
application:
• How easily do users understand the meaning and intended use of the four main
sections of the app?
• How quickly can users identify where to dispose of a specific item, that is, the type
of waste?
• How easily can users understand where to go for disposing waste of a given type?
• How easily users identify and understand the curbside collection calendar?
• How easily can users configure the app to send collection notifications?
8.5.1.2.1 User Sample

We recruited four participants for our user study (Table 8.5): two females and two males,
aged 31–55, residents of the SHARING community, who freely volunteered to partici-
pate in the study. Three out of four participants are Italian while one is originally from a
TABLE 8.4
Tasks Numbering and Its Descriptions
Task Number Task Description
T1 Where is the nearest paper/cardboard container located?

T2 Given a specific object that needs to be disposed of, identify
the right waste type using the citizen-centric application.
T3 When is glass and can waste being collected?
T4 Where should you dispose of your Microwave Oven?
T5 Which waste category does used tires belong to?
T6 Configure the citizen-centric application to generate a waste
collection notification at 7:00 a.m.
TABLE 8.5
Tasks Used for the Small-Scale Usability Study
Participant Age Range
P1 31–35
P2 51–55
P3 36–40
P4 31–35
TABLE 8.6
Job Positions of Users Participating in the Study
Participant Job Position
P1 Education
P2 Unemployed
P3 Computer science
P4 Business analysis
Middle-Eastern country. They all have a medium–high education level and quite diverse
job positions (Table 8.6).
Three out of four users do not work in technical fields and/or computer science. This
allows making an informed guess as to how the city population will react to the app, even
if the user sample considered in the study is too small to be of any statistical significance.
All users declared to regularly use smartphones in their daily activities, and one out of four
declared its adoption level as intensive. On average, they identified their frequency of opera-
tion around 11 to 20 times a day, thus a high degree of confidence with mobile application
usage and typical interaction patterns. This assumption is confirmed by the set of applica-
tions/purposes for which users declared to adopt smartphones (as reported in Table 8.7).
None of the users had previously used waste recycling support applications, thus con-
firming that no a priori bias toward some particular setup or function of the app occurred
during the tests.
8.5.1.2.2 Methodology
A within-subject design was employed where eah subject performed each task in random
order, to reduce ordering effects. Tasks involved different waste types and objects to avoid
any possible influence due to unwanted communication between users during the session
TABLE 8.7
Surveyed Users’ Smartphone Usage Patterns
Number of Users Exploiting a Smartphone
Application/Activity to Perform the Activity
E-mail/Web browsing 4
Messaging 3
Social networks 4
Maps and navigation 3
Phone calls 3
FIGURE 8.14
The usability test setup.
exchanges. Experiments were conducted in a controlled environment consisting of a sim-

ple single-room setup (Figure 8.14) with one moderator and two observers not interfering
with the test activities. In general, we followed recommendations for typical user studies
as reported in (Rubin, 2008).
8.5.1.2.3 Test Deployment

Every single-test session started with a short introduction of the ALMANAC project and a
brief description of the citizen-centric application.
After the introductory part, the user was allowed to freely experiment with the mobile
application for 2 minutes, with no suggestions and/or constraints from the moderator. After
this warm-up, each user was asked to fill in a pretest questionnaire aimed at gathering
first impressions and expectations for the citizen-centric application. This questionnaire,
in particular, investigated the user agreement on four different statements, using a 5-level
Likert scale for the allowed answer, which were as follows:
• I think the application is visually appealing

• I think the application is easy to use
• I think the application functions are clear and easy to identify
• I would like to use the application
The allowed answers were
• Completely disagree
• Disagree
• Neither agree nor disagree
• Agree
• Strongly agree
In the warm-up questionnaire, users were given space to provide feedback they might
want to share after this phase. Notice that the previous statements are translated from the
original Italian formulation; therefore, they just provide a means for the reader to under-
stand the posed questions while they do not necessarily reflect the meaning nuances and
the carefully formulated wording we used.
Each user was asked to perform six tasks (reported in Table 8.4), one at time. For four
tasks out of six, the participant was asked to use the think-aloud protocol to describe his/
her actions. User comments, suggestions, and complaints were collected by each observer,
both during the task execution and at the end of the test session. For each task, observers
recorded any unusual or unexpected behavior, and they kept track of time needed for task
completion.
At the end of this phase, users where required to fill in a posttest questionnaire with the
same questions of the warm-up questionnaire plus some additional statement to harvest
the most-liked functions and any free observation and/or feedback the user wanted to
provide.
8.5.1.2.4 Results
According to Nielsen’s Alertbox1*, we calculated the success rate of each participant as
the percentage of tasks that users were able to complete correctly, also giving credit for
partially completed tasks, that is, tasks completed with minor errors. We expected the
participants to easily grasp the app’s main features and to successfully accomplish all
tasks. Due to the limited size of the user sample and the small number of tasks to accom-
plish, quantitative results are not particularly significant; however, some evidence for
possible improvements did emerge. Table 8.8 reports the task success rates, where “S”
indicates a successful task (score of 100%), “P” a partial success (score of 50%), and “F” a
failed task (0%).
Although generally good, with an overall average success rate of 77%, results in terms of
success rate on single tasks were a little lower than expected. Three tasks were identified
as difficult. The participants had severe problems completing one of the tasks (T2—success
rate of only 37.5%) while in another, they displayed medium-level issues (T5). By analyz-
ing annotations taken by observers for tasks T2 and T5, we identified two main usability
issues in the citizen-centric prototype.
First, regarding task T2 where users were required to identify the waste category to
which a given object belonged, users were expected to exploit the barcode scanning feature
TABLE 8.8
Results of the Task Execution Test
Users/Tasks T1 T2 T3 T4 T5 T6
P1 S P S S P S
P2 S P S F F S
P3 S F S S S S
P4 S P S S P S
Average 100% 37.5% 100% 75% 50% 100%
Score
* http://www.useit.com/alertbox/20010218.html.
Too small
and difficult
to spot
FIGURE 8.15
The usability issue hampering T2 results.
of the citizen-centric application, with the aim of stressing the responsiveness of the applica-
tion for recycling support. Therefore, we assigned a full success (S) to waste identification
through barcode scanning and a partial success to waste identification by guide browsing
(P). Unfortunately, although useful and easy to understand, as confirmed by users in the
cooldown part of the test, the barcode scanning feature was not easy to detect (Figure 8.15).
The low success rate of task T5 is again due to a limitation in the user interface, which
did not sufficiently highlight the availability of the search by keyword feature of the recy-
cling guide (Figure 8.16). Half of the users ended up searching each waste category for
entries mentioning used tires. Moreover, one of the participants was not able to identify
the right guide entry and gave up the task after a few trials.
The T5 partial success had the effect of masking to some extent the severity of the
usability issue related to the search by keyword functionality. In fact, by inspecting the
waste recycling guide, it is clear that the search function is very difficult to spot as it is
hidden in a contextual menu. To mitigate this issue, common interaction patterns for
Android applications suggest positioning the search icon in the place currently occupied
by the barcode scanning feature. This modification is compatible with the one related to
the barcode scanning issue and can therefore be easily implemented in the next applica-
tion release.
The second issue relates to the absence of a filtering function on the waste bin map. Most
users were successful at finding the bins nearby. However, identifying bins for a selected
type of waste was cumbersome, as it was usually done by touching small icons on a map.
The main pattern we imagined was through the barcode scanning feature. In fact, this
feature offers disposal information and directions (map) to the nearest bin suitable for the
identified waste type. However, in most cases, this was not the pattern selected by users;
therefore, better interaction shall be supported, for example, by adding filter by waste-
type functionality to the map, and by making barcode scanning more prominent in the
interface.
Hidden in a
contextual
menu
FIGURE 8.16
Usability issue on the search by keyword functionality.
Other minor issues emerged from user observations gathered both during the task
execution and in the pre- and posttest questionnaires. Among these, it is worth citing
two issues that can easily be fixed:
1. On the EcoCenter location activity, it would be useful to have direct access to infor-
mation about accepted wastes, for example, by touching the EcoCenter icon, with
some additional function for getting the EcoCenter opening hours and contacts.
2. On the Nearby Waste bin activity, the availability of a filtering function would be
useful to ease the process of identifying bins for a specific waste type. Although
this function is already implemented in the prototype, for example, reachable
from the barcode scanning activity, it proved to be necessary in many cases thus
deserving an easier access interaction pattern.
8.5.1.2.5 Discussion
Comments verbally expressed by users where in general very positive. They appreciated the
notification ability and its ease of configuration (in fact, all users achieved 100% success in T6).
Moreover, they judged the application easy to use and supportive of their needs, even with
the issues emerging from the quantitative analysis discussed in the previous paragraphs.
8.5.2 Water Management Domain

A 10-question online survey was developed in order to get an impression of the poten-
tial end-user (private households) interest in the water management app. The survey was
promoted on the ALMANAC homepage, through the project partners’ contacts and on
social networks such as Twitter. We may therefore assume that many respondents were
from Danish households. The questions were a mix of statements where respondents had
to mark the degree of their agreement with the statement and multiple choices. Two ques-
tions were contextual to find out (1) how respondents paid for water (Q9) and (2) the num-
ber of members in their household (Q10). A total of 56 surveys were returned and the
overall results showed a positive interest in the water management app particularly with
respect to the leakage notification functionality. The results from the survey are analyzed
in more detail in the following paragraphs.
Nearly all respondents were interested in knowing more about their water consump-
tion. When comparing this with the knowledge from Question 9 (how they pay for water),
it appears that although most people (56%) get a specified water bill, they would still like
to know more. Question 1 was purposely designed as a general question to assess if there
is an end-user market for the water management app before moving onto questions that
are more specific. Thus, Question 2 specifies that water consumption information could be
made available via a mobile app. A total of 77% of the respondents were interested in an
app out of which more than half (54%) agreed or strongly agreed with the statement.
A slightly different pattern compared to Question 1 is noticed, as not all respondents
who want to know more would necessarily be interested in using a mobile app. In fact, one
of the comments to Question 3 was: I am not interested in the information if that means an extra
app on my phone (Table 8.9). Why does everything these days have to include an app?
A total of 13 respondents had made an entry in the Other field. Most respondents com-
mented that they would like a notification or indication of a possible water leak. Other sug-
gestions included tips on how to save water and detailed historical data: I would like to have
access to detailed historical data, so I can correlate water consumption with other data (e.g., shower
after exercise and dish-washing after cooking).
TABLE 8.9
Water Management Application Evaluation Questionnaire
Question Number Question Text
1 In general, I would like to know more about my household’s water consumption.

2 I am interested in a mobile app that gives me access to up-to-date information about my
household’s water consumption.
3 I would be interested in the following information consumption: current data, historical
data, notification, consumption comparison, bill estimation, water supply shortage
alert, other (more than one answer allowed).
4 I would be interested in receiving a notification if there is a public warning of water
contamination—tap/drinking water.
5 I would be interested in receiving a notification if there is a public warning of water
contamination—sea or lake water.
6 I would be interested in using the app for leakage notification at home.
7 How much would you be willing to pay to get the leakage notification system on the
app, if you saved between EUR 30–50 on home insurance policy? (Nothing, 1–5, 5–10,
10–15, more [in EUR])
8 I would be interested in using the app to connect with my social media accounts in order
to share my consumption data.
9 How do you pay for water consumption? (Included in my rent, flat annual rate, variable
annual rate, water provider bill)
10 My household consist of: 1, 2, 3, 4, 5, 6, or more persons.
Questions 4–6 inquired into the notification functionality offered by the ALMANAC
water management app. Questions 4 and 5 were related to notifications from the public
authorities regarding contamination of water and respondents expressed a high interest
in such a notification service.
Questions 6 and 7 were directly related to the value proposition in the Water Information
App. Not surprisingly, Question 6 showed a great interest in the leakage notification func-
tionality of the app and the answers to Q7 illustrate that respondents are willing to pay for
the leakage notification functionality.
The only statement, which got an overwhelming negative response, was Question 8 on
whether people were interested in using the app to share their water consumption on
social media. This clearly shows that the value of Water Management app is primarily as a
tool to monitor, control, and modify the individual household’s water consumption; shar-
ing water data with others is not perceived as having any value.
8.5.2.1 Discussion
Although the majority of respondents already receive a specified water bill, the over-
all results from the questionnaire revealed a substantial interest in being able to access
detailed current and historical data on the water consumption, as well as notifications of
possible water leakages and water contamination issues. Water bills are usually issued once
a year and can therefore not be used as an indicator (e.g., an unusual expensive bill), for
example, a water leakage in the household. Often leakages are not discovered until they
cause an actual pipe breakage resulting often in extensive water damages in the home. The
consequence of a water leakage is thus not simply one of an unusual high bill but also one
of expensive repair work.
8.6 Conclusion
Smart City ecosystem area is characterized by their complexity and variety of actors. Even
in the same city, different processes may be subject to different procedures, policies, and
legal frameworks. Introducing IoT technologies in such a context can enable higher pro-
cesses’ efficiency and better life quality for citizens in everyday life.
The presented ALMANAC SCP has demonstrated to provide support to decision pro-
cesses in both day-to-day and long-term city management, while implementing intelligent
control of a large number of smart city resources.
Introducing IoT technologies in the waste domain have provided deeper understanding
of waste generation dynamics, both for the municipality and for the utility service opera-
tor. IoT technologies can facilitate the optimization, planning, and control of the waste
collecting and cleaning system, generating operational savings and promoting innovative
and more sustainable practices in the field. Additionally, IoT technologies in the water
management domain could provide further insights to citizens, water utility companies,
and the city municipality. Real-time information linked to consumption habits and water
contamination are of great importance to these actors, for economic, health, and secu-
rity reasons. Overall, citizens are inclined to accept the use of innovative technologies to
improve everyday processes in their city and their lives.
ALMANAC has paved the way for introducing new technologies and approaches in
real-life settings. Results obtained during the project lifespan should be further explored
within the City of Turin and in other European cities to fully achieve the potential impact
of IoT technologies in smart city ecosystems, always keeping in mind that citizens should
be at the center of these.
References
Bauer, M., Boussard, M., Bui, N., and Salinas, A. (2013). Final architectural reference model for the
IoT v3.0. Technical report, Internet of Things Architecture IoT-A.
Bellini, P., Benigni, M., Billero, R., Nesi, P., and Rauch, N. (2014). Km4city ontology building vs
data harvesting and cleaning for smart-city services. Journal of Visual Languages & Computing,
25(6):827–839.
Bonino, D., Delgado, M. T., Alapetite, A., Gilbert, T., Axling, M., Udsen, H., Soto, J. A. C., and
Spirito, M. (2015). Almanac: Internet of things for smart cities. In Future Internet of Things and
Cloud (FiCloud), 2015 3rd International Conference on, pp. 309–316. IEEE: Rome, Italy.
Bonino, D., Delgado, M. T., Pastrone, C., and Spirito, M. (2016). Wasteapp: Smarter waste recycling
for smart citizens. In International Multidisciplinary Conference on Computer and Energy Science–
Splitech 2016.
Fritz, P., Kehoe, M., and Kwan, J. (2012). IBM smarter city solutions on cloud. White paper, IBM.
Hogan, J., Meegan, J., Parmer, R., Narayan, V., and Schloss, R. J. (2011). Using standards to enable the
transformation to smarter cities. IBM Journal of Research and Development, 55(12):1–10.
Rochwerger, B., Breitgand, D., Levy, E., Galis, A., Nagin, K., Llorente, I. M., Montero, R. et al. (2009).
The reservoir model and architecture for open federated cloud computing. IBM Journal of
Research and Development, 53(4):535–545.
Rubin, J. and Chisnell, D. (2008). Handbook of Usability Testing: How to Plan, Design, and Conduct
Effective Tests, 2nd ed. Hoboken, NJ: Wiley.
Sma, (2012). ICT architecture supporting daily life in three smart cities. Technical report, Smart
Cities EU Project.
Sma, (2015). Smart cities market by smart home, intelligent building automation, energy manage-
ment, smart healthcare, smart education, smart water, smart transportation, smart security, &
by services–Worldwide market forecasts and analysis (2014–2019). Technical report, marketsand-
markets.com.
Vermesan, O., Harrison, M., Vogt, H., Kalaboukas, K., Tomasella, M., Wouters, K., Gusmeroli, S.,
and Haller, S. (2009). The internet of things–strategic research roadmap, Cluster of European
Research Projects on the Internet of Things.
Vilajosana, I., Llosa, J., Martinez, B., Domingo-Prieto, M., Angles, A., and Vilajosana, X. (2013).
Bootstrapping smart cities through a self-sustainable model based on big data flows.
Communications Magazine, IEEE, 51(6):128–134.
9
NFC in IoT-Based Payment Architecture
Alak Majumder, Shirsha Ghosh, Joyeeta Goswami, and Bidyut K. Bhattacharyya
CONTENTS
9.1 Introduction ........................................................................................................................ 203
9.1.1 Reader/Writer Mode ............................................................................................. 204
9.1.2 Peer-to-Peer Mode.................................................................................................. 204
9.1.3 Card Emulation Mode ........................................................................................... 205
9.1.4 NDEF Stack Description ....................................................................................... 206
9.1.5 Security in NFC...................................................................................................... 206
9.2 Basic Overview of IoT Universe ....................................................................................... 208
9.3 Use Cases of NFC in IoT Universe................................................................................... 210
9.4 NFC-Based Payment Systems .......................................................................................... 213
9.5 Common Threats and Vulnerabilities of NFC-Based Payment System
and the Process to Eliminate Them ................................................................................ 217
9.6 Conclusion .......................................................................................................................... 218
References..................................................................................................................................... 219
9.1 Introduction
In the era of ubiquitous computing, many wireless technologies have emerged to serve
several types of communication needs. Near-field communication or NFC is a short-
range wireless communication protocol, which is a successor of popular Radio-Frequency
Identification or RFID technology that was patented by Charles Walton (1983). NFC was
standardized as an ISC/IEC standard in 2003 and works in a very short range using
13.56 MHz frequency that makes it perfect for contactless payments. It extends the ISO
14443 RFID standard. Since its inception, NFC is on the limelight because of its versatility,
inherent security, and interoperable nature. The basic mechanism of NFC is based on the
Faraday’s Law of Induction in which current is flown between two devices by creating
a magnetic field. In the case of NFC, the reader or the active device creates a magnetic
field using its antenna coil. Then, the passive devices present in that field get energy from
the magnetic field and modify the properties of the incoming frequency. When the active
device gets information about the passive device, a very small amount of AC current
emerges from the sinusoidal waves. The AC current is converted into DC inside the tag
chip using the rectifier and the passive device also generates carrier wave to pass infor-
mation to the active device.
Currently NFC is involved in three major devices: Smartphones, NFC readers, and
NFC tags. Based on the different working and application areas, NFC has three different
203
operating modes known as reader/writer mode, peer-to-peer mode, and card emulation
mode. NXP semiconductors have documented all these modes in a document entitled
“NFC Everywhere” (2015a).
9.1.1 Reader/Writer Mode

In this mode, the communication settles between a passive NFC tag and an active NFC
device such as NFC enabled smartphone or NFC terminal. The main purpose of this
communication is two:
• To read prestored data from the tags. The involving steps are as follows:
• Read request: The user requests data by active device such as mobile to the NFC
tags that may be installed into various places.
• Data transfer: The prestored data are then transferred to the active device from
the passive tag to the reader.
• Processing by the reader device: After getting data from the passive tag, the reader
processes the data for various functions.
• To write values to the NFC tags or writer mode. In this case if some data are pre-
stored in the tag, then the writer overwrites or updates those data by the latest.
The steps are as follows:
• Write request: First user makes a request to write data to the NFC tag.
• Data transfer: Then the NFC tag endorses the success of the operation.
This operating mode has data rate up to 106 kbps. The main applications of this mode
are smart posters, ticketing, accessing Internet or any media files, and so on. Figure 9.1
represents the mode.
9.1.2 Peer-to-Peer Mode

This operating mode is about the communication between two NFC active devices. This
operating mode takes the advantages of two communication technology: NFCIP-1 (Near-
Field Communication Interface and Protocol), LLCP (Logical Link Control Protocol), and
SNEP (Simple NDEF Exchange protocol). In NFCIP-1 mode, both NFC devices must have
own power source to generate magnetic fields. And in the LLCP communication, both
NFC passive device (Tag)

NFC active device
(Smartphone)
FIGURE 9.1
NFC reader/writer mode.
NFC in IoT-Based Payment Architecture 205
NFC active device NFC active device

(Smartphone) (Smartphone)
FIGURE 9.2
NFC peer-to-peer mode.
the devices stay identical. SNEP allows the exchange of NDEF messages analogous to tag
operation specifications. In peer-to-peer mode, first data are exchanged between two active
devices then using those data, several tasks are executed. A 424 kbps data rate is provided
by this operating mode. This mode has several important tasks including exchanging data,
peer-to-peer money transfer, pairing devices, and so on. Figure 9.2 shows the mode.
9.1.3 Card Emulation Mode

Card emulation mode provides a way to use an NFC enabled mobile as a substitute of con-
tactless smart card in which the mobile devices can store different credit card information
with a much secure way. The working steps of this operating mode include the following:
• Service request: The user establishes an NFC communication with an NFC reader
and makes a request to any service provider in which the reader transfer the data
collected’ from the mobile phone to the service provider. Figure 9.3 shows the mode.
• Backend service: The service provider will run this service in backend after receiving
the required data.
• Service usage: Finally the service provider processes the data and provides a service
to the user.
NFC active device Point of sale terminal

(Smartphone) (POS)
(Using as emulated card)
FIGURE 9.3
NFC card emulation mode.
The major application fields of this operating mode are Payment, Ticketing, Identity ser-
vice, Smart environment, and so on with the advantage that it removes the physical objects
and cash money and gives higher security.
9.1.4 NDEF Stack Description

The format by which data are exchanged between two NFC devices is known as NFC
Data Exchange Format or NDEF. NFC Forum specified all the technical specification in
their official documentation (2006) on NDEF. It is a binary message format in which every
message contains several records, and each record consists of a payload of size 232-1 octet.
Each record has header fields and a payload field, where header field contains five flags:
MB (Message Begin): Points the first record of an NDEF message.

ME (Message End): Points the last record of an NDEF message.
CF (Chunk Flag): Defines the continuation of payloads from one record to its next record.
SR (Short Record): Specifies the size of the payload length that is the number of octets
present in the payload.
IL (ID Length Present): Presence of the optional ID field and its corresponding length
field.
The NFC Protocol Stack is shown in Figure 9.4.
9.1.5 Security in NFC

As NFC is widely used for payment-related applications, it is very important to take care of
the security measures. Due to NFC’s very small distance communication range, it is very
difficult for an attacker to intersect the communication medium, but it is very important
to increase the security as high as possible to eliminate the chance of data theft. NFC uses
hardware secure elements that provide security at its level best.
Typically, the secure element comes included with the NFC active element such as
Smartphone. Several libraries are also written in the software stack to combine NFC with
several role players such as Trusted Service Manager (TSM), Mobile Network Operator
(MNO), and POS. The communication is performed over a secure channel using sym-
metric or asymmetric cryptographic keys.
There are several types of secure elements available. They are as follows:
Universal Integrated Circuit Card (UICC): They are mandatory for GSM applications
and necessary about the network. They are coded on Java and Global Platform that
allow 3rd party developer to run their applications using UICC. The main host
controller uses Single Wire Protocol (SWP) to communicate with UICC.
Embedded Secure Element (eSE): This type of secure element is separate hardware
that is embedded in the mobile handset by the manufacturer. These will verify
the legitimacy of all NFC-based Transfer. Once embedded in the hardware, they
can only be removed by the manufacturers.
MicroSD: Several 3rd parties other than the smartphone manufacturers, and Mobile
Network operators, use the MIcroSD as an alternative to UICC and eSE.
Trusted Execution Environment (TEE): It is a part of main processor core that can be
used to store and process sensitive information.
Application layer NFC forum NDEF messages
NFC forum
Tag types Type 1 tag Type 2 tag Type 4 tag Type 3 tag
formatted tag
SLE66CL
Product example MIFARE MIFARE MIFARE
Topaz Felica NFC device
STD 1K/4K ultralight DESfire
16RF
NFC in IoT-Based Payment Architecture
Applicative protocol ISO/IEC 7816-4 (APDUs)
Topaz MIFARE MIFARE ISO 14443 ISO 14443 LLCP

Protocol ultralight
protocol protocol protocol A-4 B-4 (peer-to-peer)
Initialization Felica/JIS X
No anti ISO 14443 6319-4 and ISO
anti-collision ISO 14443 A-3/ISO 18092
collision B-3 18092
product activation
ISO 14443 ISO 18092

RF ISO 14443 A-2/ISO 18092
B-2
Physical ISO 14443

ISO 14443 A-2/ISO 18092
characteristics B-1
FIGURE 9.4
NDEF stack.
207
9.2 Basic Overview of IoT Universe

The concept of IoT-based application began from the inception of RFID technology. The
primary focus of IoT development was to devise some electronics equipment that will
boost the use of RFID. But soon when the need of connectivity among several peripherals
was raised, researchers started to explore the use cases of other wireless technologies. Soon
IoT became a much broader scope for research, and it continued to be bigger ever since.
As given by International Telecommunication Union (2005), the main motto of IoT
development is very simple: “from anytime, anyplace connectivity for anyone, we will
now have connectivity for anything.” The vision is that there will be a world where things
can communicate with Internet and provide important feedback about their environment
that will rather improve the quality of human life. For proper implementation of the motto
of IoT, seamless integration between several wireless technologies is very much essential.
Actualization of the IoT concept into the real world is possible through the integration of
several enabling technologies.
Though several technologies are integrated in the development in IoT, RFID remains
the main driver technology. RFID tag is basically a microchip with an antenna used
for transmitting the tag ID and receiving the reader signal. They are assigned a unique
identifier. The tags are attached with different objects and using the unique identifier,
different objects can be identified, and mapped into a virtual environment (Atzori et al.,
2010).
Another critical technology in IoT universe is wireless sensor network or WSN. In
Wireless Sensor Network, several sensor nodes are connected in a distributed or central-
ized and multihop fashion. The sensor data are shared between the different nodes that
thereafter collected by the main server for data manipulation. WSN mainly contains the
following elements:
1. Hardware: Comprises sensor interface, A/D Converter, transceiver, and power supply.
2. Communication stack: Designing the software stack for convenient topology, suitable
routing and MAC Layer.
3. Middleware: A software infrastructure to combine the WSN hardware, communi-
cation stacks, cyber infrastructure, and application.
4. Secure data aggregation: It extends the lifetime of the network and ensures reliable
data collection.
The recent trend is to integrating the sensor technologies with the RFID chip. This allows
much more capability of sensor nodes. This system is very much useful in medical sys-
tem, in which microsensors can be fitted into patients body and get essential medical
data from there.
Atzori et al. (2010) compared the RFID system, WSN, and RFID Sensor Network (RSN)
in their literature as shown in Table 9.1.
For an IoT system to be useful, it should be scalable. Means addition or reduction of
sensor hardware should not affect the performance of the system. To make the system
scalable, Uniform Resource Name (URN) system is widely used in IoT system. Several
replicas of resources are made by the URN that is then accessed by URL (Gubbi et al.,
2013). All the different sensor data are acquired by the central node are made addressable
by URN and thereafter made accessible to the web server by URL.
TABLE 9.1
Comparison between RFID system, WSN, and RFID Sensor Network (RSN)
Processing Sensing Communication Range Power Lifetime Size Standard
RFID No No Asymmetric 10 m Harvested Indefinite Very Small ISO18000

WSN Yes Yes Peer-to-Peer 100 m Battery <3 years Small ISO 802.15.4
RSN Yes Yes Asymmetric 3 m Harvested Indefinite Small None
The gathered sensor data are passed to a web server through internet. As the data
gathered from the sensor nodes are enormous, it needs to be manipulated. Researchers
are developing several algorithms using Artificial Intelligence, Neural Network, Genetic
Algorithm, and so on for automatic response from the gathered data. These algorithms are
executed on a centralized infrastructure to get analytics from the collected data. Nowadays,
IT giants such as Google, Microsoft, Amazon, and so on are providing cloud-based server
and tools for developers. These services are gaining popularity because of their reliability,
uptime guaranty, and many other features for the developers.
After processing the data, the statistics of the result are sent to the users from the web
server. For this process, mobile Computing plays a crucial rule. Currently developers
and researchers prefer to use mobile based applications for data visualization because
of huge number of smartphone users in market. Using a mobile application, important
information can be visualized anywhere at any time. Smartphone also offers touch
screen inputs that are very intuitive to get input from the user.
IoT has very useful practical application in our daily life. With more and more research
and development on the technology, different application areas are coming up regularly.
GSM Association came up with some application idea in their IoT report on July 2014.
They are given below:
IoT for Smart Cities:

Dim-based smart streetlights on ambient conditions to save energy.
Smart Display board for passenger information.
Self-driving vehicles to increase safety and reduce CO2 emission.
Smart traffic lights.
Smart camera to reduce crime and to provide quick response for emergency cases.
Smart payment system for insurance, Toll or road charges.
IoT for Health:
Connected devices to get important health parameters and updating medical reports.
Doctors can see that the treatment is properly followed or not.
Patients can take proper medicines in proper time.
Family members can ensure proper care taking.
IoT for Education:
Students carry all the materials in Smart devices.
Results are updated in real time.
Using connected device students and teachers can work collaboratively.
Self-guiding education helps adults to fill up their skill gaps.
IoT for Productivity:

Video conferencing for long distance communication.
Real time data can be shown in any meeting.
Reducing the product delivery time in e-commerce business by delivering the
product from the closest inventory.
Prototyping using 3D printer.
9.3 Use Cases of NFC in IoT Universe

As discussed earlier, NFC is successor of RFID technology, which is the main role player
in IoT development. NFC has the same capability as RFID but with increased security and
features. In an article written by Paula Hunter (2016) of NFC Forum, she pointed out four
major points for which IoT needs NFC. The points are as follows:
1. Using NFC, any unpowered device can also be connected with network. Just by
taping an NFC tag, useful information can be gathered about an object.
2. Using NFC, the user can choose his own way of connection. NFC provides the
simplest and intuitive way to take an action by the user.
3. Using NFC, the handshaking procedure for connection of devices can be elimi-
nated just by using a simple tap.
4. Using NFC, the chance of eavesdropping is reduced to a great extent.
IHS Technology predicts that in 2018, 1.2 billion smartphones will be shipped with NFC
embedded on them. This provides a great opportunity to developed connected devices
with least amount of hardware required. Presently only using an NFC-enabled phones
many IoT-based applications are developed. Some of the applications include Home
Automation, Smart Meters, Healthcare, Home appliance control, Consumer Electronics,
Cloud-based applications, and so on. In the whitepaper published in June 2016, NFC
Forum (2016) pointed out some very useful application of NFC in Smart Home such
as Water Metering, Appliance Controlling, Appliance Servicing, Ambient Setting, and
so on. They also pointed out that “one of the main challenges for providers of smart
home devices is providing a unified device commissioning flow that is independent of
the underlying communication framework.” NFC simplifies the device commissioning
process by a single tap-and-go feature, which is completely independent of underlying
communication protocol and based on standardized mechanisms. Another important
factor for smart home is to connect and pair with Wi-Fi. Due to the alliance between NFC
Forum and Bluetooth SIG, NFC can speed up Bluetooth pairing process by eliminating
the need of time consuming processes like device discovery and device pairing. NFC
forum has also alliance with Wi-Fi. Just by tapping an NFC tag that is glued with Wi-Fi
router, any smartphone gets all the information such as SSID and Passcode and configure
itself to connect with the Wi-Fi network. When an unpowered device with an NFC tag
is tapped on the NFC module, the network parameters are written on the tag. Later, this
information is used to connect the device to the network. NFC also helps for Smart home
access control. The users can share their smartcard keys using smartphones with their
relatives without meeting them personally. The keys then can be used with NFC to open
any kind of electronic lock or smartcard based locks.
NTAG I2C Chip is another low-cost NFC module from NXP Semiconductor (2014). It
provides ideal solution in the places where it is very difficult to embed any display or
remote controlling features. NTAG I2C Chip works like a bridge between the appliance
and users NFC-enabled mobile to make the mobile an extended display and remote
controller for the appliance. The system reduces the cost due to its much simpler
design and lease number of fancy hardware and also increases the security by elimi-
nating the possibility of unauthorized access. This feature is quite useful in different
applications such as Fitness Tracking Wearable, Home Appliances, and Thermostat.
It also provides different features such as Self-Serve maintenance, Administrative
access to Electrical Equipment, Firmware update, Bluetooth and Wi-Fi pairing, Home
Automation Commissioning, and so on.
Texus Instruments also launched NFC Dynamic Transponder RF430CL330H. It provides
the same features as NXP NTAG I2C Chip but has two inbuilt communication protocol:
I2C and SPI.
Researchers are trying to take advantage of the features of NFC and use this technol-
ogy in different fields. NFC can be a vital role player in in healthcare application. Some
researches on the use of NFC in healthcare are given below. Freudenthal et al. (2007a)
implanted 13.56 MHz frequency based RFID tag TI—TagitTM transponder, on animal
tissues and proved the usefulness of using passive NFC to communicate with the medi-
cal devices embedded in human cadaver. Morak et al. (2007b) used NFC technology
for real-time monitoring of different health parameters such as blood pressure, body
weight, and so on. They can be monitored just by tapping NFC-enabled smartphone in
the mobile device. Researchers came up with a system, by which using NFC-enabled
phones, patients with impaired fine motor skills can get meaningful information. They
have fabricated a smart poster. The patients will get information just using simple tap
in the smart poster. The system also has a physician’s dashboard, in which using a web-
portal, doctors can monitor all the patient related information in real time (Prinz et al.,
2011). In 2008 Deutsche Bahn, Vodafone, German rail authority, Deutsche Telekom along
with some support from industry jointly launched an NFC-based ticketing system in
Germany known as Touch&Travel. The system covers trains, metros busses, and a ferry.
In the same year, in San Francisco, contactless payment Solution provides ViVOtech
along with some other companies such as Sprint and First Data, enabled Bay Area Rapid
Transit (BART) using NFC. In another literature, researchers used NFC to schedule
appointment in the hospital (Sankarananrayanan and Wani, 2014). The system can also
prioritize appointments based on patient’s physical condition. Other than health care,
NFC can be also used in On the Go ticketing services. In the research papers by Nasution
et al. the researchers have developed a prototype of a ticketing system based on NFC
enabled android mobile (Nasution et al., 2012). In a whitepaper published by NFC Forum
(2011), the authors have pointed out the usefulness of NFC-based ticketing over conven-
tional Paper- or PC-based ticketing. NFC-based mobile provides lot more facility and
robustness than traditional procedures such as getting information about timetables,
up-to-date weather report before travel, location map for navigation, discounted travel
fares, next bus arrival time, and so on. The consumer gets many benefits such as getting
the ticket electronically without the need of physically going in the office to collect the
tickets, less chance of lost or theft, easy sharing, and so on. Developed countries like
London, Paris, and Madrid already have implemented automatic fare collection from
the NFC-enabled mobile phone.
Educational institutes are also relying on IoT-based system to increase the relation-
ship between the student and the management, in which NFC can play an important
role. In the literature, Miraz et al. (2009) pointed out the efficiency of NFC to enhance the
teaching administrative and payment-related services. The researchers also developed
smart posters for different departments and faculties. Just by tapping any NFC-enabled
mobile, all the information regarding the concerned department or faculty can be viewed
instantly. Bueno-Delgado et al. (2012b) present a preliminary survey on smart univer-
sity about the actual acceptance to the users. They have also devised some mechanism
for different academic applications such as attendance monitoring and payment collec-
tion using NFC technology. The U.S. Department of Transportation also included NXP’s
RFID tagging technology for vehicle-to-vehicle and vehicle-to-infrastructure technology
(Bodden, 2016a).
For any NFC-based project both NFC reader and Writer device are essential. If some-
body is good at Android programming, then Android-based mobile phone has the capa-
bility of handling all types of NFC communication modes, that is, reader/writer mode,
Host Card Emulation Mode, and peer-to-peer mode. The detailed guideline can be found
on Android’s official guideline on Android’s official website by Google. Researchers have
also published many books that provide step-by-step guideline for NFC-based Android
Application development. Like, Beginning NFC: Near-Field Communication with Arduino,
Android, and PhoneGap by Brian Jepson, Don Coleman, and Tom Igoe, Professional NFC
Application Development for Android by Busra Ozdenizci, Kerem Ok, and Vedat Coskun;
Near-Field Communication with Android Cookbook by Vitor Subtil and many more. The books
also provide many readymade examples to start with. By using a Simple Android based
mobile phone and NFC tags, anyone sitting at their home with a computer can start build-
ing NFC-based projects for different domains such as Smart poster, Access Control, and
so on. NFC tags are also very cheap (Starts from 5$ at Amazon), so huge amount of cost
involvement is also not required.
Hardware-based NFC projects required some basic skillset in Microcontroller program-
ming. For any beginner, one of the best hardware platforms is Arduino. It is very popular
because it is completely Open Source, huge documentation, cross platform IDE and has a
large community. Anyone can visit Arduino’s official website (https://www.arduino.cc/)
to start with. All the documentation, hardware details, tutorials are readily available in
the website. Raspberry-Pi is another very popular hardware development platform based
on ARM processor. It is also an open source and has a large community base. For high
performance computing, Beaglebone is mostly preferred. These are ARM Cortex-A8 based
development board with 2 GB onboard Flash memory. Both Raspberry-Pi and Beaglebone
has onboard Ethernet connectivity option.
For NFC communication, the most popular IC is PN532 fabricated by NXP. Many com-
panies such as Adafruit INC, Sparkfun, and Seeedstudio have manufactured readymade
NFC module based on PN532 IC. This module is very easy to use with any Microcontroller
board. They either use Serial Communication or Serial Peripheral Interface (SPI) protocol
communicating with the host microcontroller. There are also other NFC transceiver ICs
such as TRF7970A by Texas instrument and ST95HF by STMicroelectronics, and so on.
Though NFC is widely used for different application, Payment remains the main use of
NFC. As the need of market is increasing rapidly, the Payment-related technologies are
growing exponentially. In the next part, the usefulness of NFC in payment architecture in
discussed in detail.
9.4 NFC-Based Payment Systems

Due to its very less communication range, NFC is preferred for payment-related needs.
In 1995, Seoul Bus Transport first implemented contactless payment technology. In 2007,
Barclay card first embedded NFC in its payment card to support contactless payment.
Since then, NFC development took a huge leap, and contactless transaction is increasing
every day. Many researchers are exploring the use cases of NFC for different payment-
related issues. Figure 9.5 shows a typical NFC-based payment system architecture.
NFC provides several advantages over other payment options like
1. Ease of Use: NFC is very easy and intuitive to use. Users can pay just by using a
single tap. NFC can be also integrated with mobile wallet based app that provides
more convenience to customers in digital payment domain.
2. Versatility: As NFC is embedded on Smartphone, it can be used I a wide range of
application such as POS and peer-to-peer Payments, Ticketing, Reward system
access control, and so on. So NFC is one of the most versatile technologies of
present day.
3. Security: NFC provides hardware level security using inbuilt/external secure
element. NFC-based payment cards are most secure. And as NFC is contactless,
retailers do not get physical card information. And as NFC can be integrated with
digital payment easily, NFC-based transactions can be made more secure using
digital cryptography.
Google integrated NFC in its Nexus-S device in 2011, and launched their first Digital Wallet
known as Google Wallet that uses Host Card Emulation. After that due to a tremendous
development in Mobile Computing platform, different Mobile wallets are coming up in
the market, most of which are based on NFC or Contactless technology.
POS provider
Retailer
Mobile handset Mobile network

Customer Merchant
manufacturer operator
Secure element Trusted service Contactless payment

Acquirer
manufacturer manager application provider
Brand
FIGURE 9.5
NFC card emulation mode.
Monteiro et al. (2012) have made a mobile application, which can transfer mobile phone
credit between two phones using NFC and Bluetooth. In the research paper (Mainetti
et al., 2012), the authors have devised a mechanism, by which all the credit card infor-
mation are securely stored in the mobile device using the secure element. Apple and
Samsung both released their payment systems in 2014 and 2015, respectively. Both of them
are highly secure using Biometric Authentication and Tokenization. Apple Pay can only
work on NFC-enabled Point of Sale terminals, in which Samsung pay works in traditional
Magnetic Stripe based Point of Sale terminals using a technology called Magnetic Secure
Transmission (Wallner, 2014). Google also launched Android pay in 2015 using Biometric
authentication and Tokenization.
Researchers are predicting that the pervasive computing is field of next generation com-
puting in which IoT plays a vital role. So soon people started to develop payment solutions
for IoT-based products. The areas such as Public Transport ticketing, Smart University,
Mobile Commerce, Logistics, peer-to-peer payments, and so on need a flawless, robust,
and scalable payment architecture. NFC can be a crucial element in this context. The main
advantage of NFC is its very less power consumption and very small antenna size, which
makes it perfect to be mounted on several things that need to be connected within the IoT
network.
In the article “How to Integrate Payments in IoT Devices?” (2015), it is found that sev-
eral fintech startups are focusing on IoT-based payments to provide innovative payment
solution to the consumer using NFC, Mobile Apps, Sensors, and so on. One of the most
payment companies, PayPal has made developed payment app solely for IoT development,
which is featured on Samsung’s wearable devices. Pebble is another very popular wear-
able. Different companies are also launching payment app based on Pebble. Semiconductor
companies are also embedding NFC functionalities in the core of their processor. For exam-
ple, Qualcomm is embedding NXP’s NFC and Embedded Secure Element in some of their
Snapdragon-based processors. Intel Corporation & Ingenico Group are jointly launching
a tablet that will serve the purpose of EMV and NFC payments. Card Payment Network
Processing companies such as VISA, American Express are also investing on IoT-based
product for payment.
When Apple launched Apple Pay in 2014, they also introduced Apple Pay for their wear-
able Apple Watch. Using Apple Watch users can have a seamless payment functionality
across contactless point of sale terminals.
NFC is also introduced in Automated Vending machine. USA technologies introduced
NFC-based contactless payment on their vending machine in about 100,000 locations
across the United States.
In the article “Gemalto and Visa roll out contactless payment wristbands at Eurovision,”
Security giant Gemalto also launched contactless payment wristband that can be used
at any VISA contactless terminal around the globe (Bodden, 2016b). In another article by
the same author, it is found that Chinese Ecommerce giant Alibaba, in partnership with
SAIC Motor Corp, has introduced internet car. In this car, the driver will be able to make
a purchase using company’s mobile payment platform Alipay (Bodden, 2016c). In 2016,
prototype of an NFC-based digital card is invented by Alak et al. that serves all kinds of
Payment needs as well as Multiple ID Card Virtualization. The card also uses Tokenization
and Biometric Security (Majumder et al., 2016). This card is based on NFC peer-to-peer
communication mode. Researchers have shown a very interesting way to make a transac-
tion without using any smartphone. One digital card sends the transaction details to other
TABLE 9.2
Different Stack Holders in NFC Ecosystem
Stake Holder Role
NFC Chip They design the silicon chip for NFC communication. They may also include secure
Manufacturer element in the NFC chip, for example, NXP, TI, and so on
Secure Element They are behind the designing of secure elements by which every transaction is validated.
Manufacturer The secure element may be hardware or Cloud based service. For example, NXP and
Apple manufactures secure element chip whereas Google provides Cloud-based Secure
element solution for their Google Wallet service
Mobile Phone They manufactures the mobile handsets for consumer and integrate NFC payment solution
Manufacturer into their hardware and software design, for example, Samsung, Apple, and so on
Trusted Service One of the most important role players. As Payment services are of heterogeneous
Manager (TSM) structure, it is very important to bring all different users, mobile networks, handset
manufacturers under a common platform so that the system works seamlessly. TSM
plays a very crucial role by combining all under one place
Mobile OS They provide Operating system to run over the mobile hardware, such as Google’s
Provider Android, Apple’s iOS, Microsoft’s Windows, and so on
Application Application developers are behind the top notch application for a mobile device.
Developers The applications should be intuitive, easy to use, and should provide great security
Card using NFC peer-to-peer communication protocol. These details are then forwarded
to cloud server using GSM and a message server. Server processes the data and makes
the transaction happen. The card can also be used in any POS terminal using Host Card
Emulation feature, though it is not required as because one card can alone work as a POS
terminal. The digital card is also biometric protected which increases the security of the
system.
As discussed in the previous section, NFC may be a very vital role player for On the Go
ticketing services. It provides several useful features than traditional ticketing services.
Using automated machines, tickets can be purchased and also validated. User now has
the freedom to get the tickets as per their convenience and can also share tickets over
Internet to their friend or relatives. All the tickets are purchased from a central database.
The administrators can easily get the statistics from the central database. Based on user
demands, personalized advertising can be directly sent to the user’s mobile.
In any NFC-based payment system, there are several role players. For proper functioning
of the payment network, it is very essential to perform their role efficiently. Ghosh et al.
(2015) discussed the role of the stake holders in their literature. The stake holders for an
NFC-based payment network are listed in Table 9.2.
Figure 9.6 describes the whole NFC ecosystem. For Development of IoT-based product,
first consideration is that the module must have connectivity peripheral inbuilt. The device
should be small such that it can be mounted anything to get information from it. Other
than Arduino, Raspberry-Pi and Beagle bone, any companies are manufacturing develop-
ment board customized for IoT-based application only. Some of them are listed below:
• Panstamp: These are very small development boards based on Atmega328p MCU
and provides RF connectivity via Texas Instruments CC1101 RF interface.
• RFduino: These are fingertip sized mini development boards based on ARM
Cortex-M0 processor and provide on-board Bluetooth connectivity.
NFC chip
manufacturer
Secure
Application element
developers manufacturer
NFC
ecosystem
Mobile OS Mobile phone

provider manufacturer
Trusted
service
manager
(TSM)
FIGURE 9.6
Stack holder in NFC ecosystem.
• Pinoccio: These are microcontroller development boards based on Atmel

ATmega256RFR2. They provide the provision to use them for mesh networking as
well as web connectivity.
• mbed: These are based on ARM Cortex M3 processor and provide Ethernet
connectivity.
• Wi-Go Module: These modules are based on Freescale KL25Z (ARM Cortex M0+)
processor. They also have many sensors inbuilt on the board such as Accelerometer,
Magnetometer, Altimeter, and Ambient Light Sensor. These modules provide
Wi-Fi connectivity and many more.
All these modules are perfect to kick off any IoT-based project. For any NFC-based project,
it has to be verified that whether the board has the provision for either serial connection
or SPI connection. For NFC peer-to-peer communication mode, it is necessary to use SPI
protocol due to constraints in the software stack. To use the peer-to-peer communication,
two Protocols have to be followed. One is SNEP, and the other is LLCP. The description of
both can be found from the NFC Forum’s website. Researchers and NFC Module manufac-
turing companies have written LLCP and NDEF software stack for popular development
environment such as Arduino and Raspberry-Pi. With minor modifications, they can be
run in any environment.
For processing and visualizing the data from different nodes, one can use the standard
web development languages such as HTML, PHP, and JavaScript. But nowadays, there are
many readymade solutions available, which are efficient and reliable, such as Thing speak,
IoT Toolkit, Berg Cloud, and Electric Imp. They are also very easy to learn and implement
in the real world.
9.5 Common Threats and Vulnerabilities of NFC-Based

Payment System and the Process to Eliminate Them
The fact that criminals are trying to hack mobile phones for fraudulent transaction in
unquestionable. As the technology is becoming harder to crack, the attacks are also becom-
ing more refined. To make any payment system more sustainable, it is very important to
understand the current and possible future threats and vulnerabilities and devise a full
proof mechanism to reduce down the chance of them.
In a Whitepaper by Mobeyforum (2013), they pointed out the present vulnerabilities of
NFC-based payment system. They are as follows:
• Untrusted Communication Channel: As it is impossible to authenticate both the

POS and NFC devices in an unprotected channel, there is a possibility of Identity
Hijacking.
• Denial of Service in the Physical Layer of Communication: If a huge amount of data
can be sent from any device using same 13.56 MHz frequency, then it may create a
Denial of Service at the communication channel.
• Denial of Service in Payment Cards: An attacker can also hack the POS terminal
causing it to exceed the limit of failed transaction.
• No Encryption in Communications: As there is no encryption in the channel between
the NFC device and the POS, with proper hardware any attacker can intercept in
the communication.
The common threats in NFC-based payment systems are listed below:
• Eavesdropping: With proper antenna if any attacker intercepts the message in the
channel between the NFC device and the POS terminal, he may get sensitive
financial information. But the attacker’s antenna should be very near to the NFC
device for successful eavesdropping.
• Data Corruption: Using same frequency antenna, at the time of data transmission,
attacker may insert or modify the data sent over NFC.
• Phishing: If the NFC tags are replaced by any attacker, then instead of getting
useful information, users may be forced to visit malicious websites. The user
may also download malicious software or virus in their device.
• Card Cloning and Skimming: Another common mode of attack in contactless sys-
tems. Using a hidden electronic device called skimmer, attackers try to get the card
information.
Researches are going on to secure the NFC-based transactions. Here are some common
practices recommended by NFC Forum.
• Eavesdropping: As NFC has very short communication range (<3 cm), it is very
difficult to plot an eavesdropping attack. To make it even more secure, it is better
to use any standard encryption protocol to secure the channel between the POS
and NFC device.
• Data Modification: NFC devices have the ability to check for RF signals in the chan-
nel. So if a device is programmed to check for an unknown RF signal, it can sense
data modification attack. This attack can be also reduced by securing the commu-
nication channel.
• Man in the Middle: Same as eavesdropping, the chance of this type of attack is also
very less. Man in the middle attack can be reduced by using Half Duplex commu-
nication mode. It means, only one device will transmit data one time. The devices
may also detect any disturbance in the channel.
It can be seen that from the above discussion, the main target or objective of designing
security measure for NFC devices is securing the channel. Researchers are developing new
encryption algorithms based on hash Function. In the literature (Chang et al., 2011c), the
researchers have developed a method by using XOR operation and one-way hash functions
using which many drawbacks of RFID communication such as Replay attack, DoS attack,
and so on are addressed. The authors also have theoretically proved their claims. Sun and
Zhao (2012) found some loop holes in the previous literature and proposed a new scheme,
by which the loop holes of the previous paper are properly overcome. Zhuang et al. (2014)
improvised previous algorithm and applied for NFC Ecosystem by optimizing the power
consumption, reducing the errors, and optimizing the storage usage. Urien (2014) pro-
poses a new model to use NFC P2P model for Internet of Things. This model is basically a
TLS security layer written over NFC Forum’s LLCP which provides data integrity.
9.6 Conclusion
One of the oldest relationships that exist from the evolution of the earth is between the
Payer and Payee. The concept of exchanging goods came into picture with the need basic
commodities in daily life. Soon it turned to the concept of payment that provides high
priced goods such as precious metals, gems, and so on. With the steady development
of technologies and human needs, the payment methods changed enormously. With
the progress of the society and technology, people changed the means of payment. In the
recent years, with the adaption of POS payments, Contactless payments, Cloud-based
payments, the growth of digital money transfer is increasing exponentially. Continuous
progress in technologies and quality of life, has made a strong impression on the devel-
opment on payment techniques. After the evolution of NFC technology, contactless
payment has recently got a huge attention because of its short-range conducive nature.
Again, as mobile computing took a great leap due to enormous development of smart-
phone platform, companies such as Google, Samsung, and Apple have embedded NFC
in smartphone to provide On the Go payment eliminating the need of payment cards. The
digital payments are becoming more secure with the invention of technologies such as
biometric security, 3D secure algorithm, Tokenization, and so on. In all cases, the custom-
ers opted for a solution, which is secure and more than that, which is convenient for the
user. Big IT giants such as Apple, Google, and Samsung are continuously trying to be a
market leader in the digital payment domain, by embedding more features in their flag-
ship smartphones. People are also slowly adapting them in countries such as the United
States, United Kingdom, and so on. For example, 20% of users that uses iPhone have tried
Apple Pay in their daily life. Though Developing countries are still far behind from the
first world country, they are gradually adapting digital payment means as services such
as, Mobile Wallets, Electronic Fund transfer, and so on.
In this chapter, we have mainly focused on the pros and cons of NFC technology and
why we must choose it for IoT universe, mainly in e-payment infrastructure.
References
A Security Analysis of NFC Implementation in the Mobile Proximity Payments Environment (2013).
Available: http://www.mobeyforum.org/w/wp-content/uploads/NFCSecurityAnalysis_
Part1_FINAL1.pdf
Amit, How to Integrate Payments in IoT Devices? (2015). Avaliable: https://letstalkpayments.
com/how-to-integrate-payments-in-iot-devices/
Atzori, L., Iera, A., and Morabito, G. (2010). The internet of things: A survey. Computer networks,
54(15), 2787–2805.
Bodden, R. (2016a). US Smart City Challenge adds NXP’s RFID tagging solutions NFC World.
Available: https://www.nfcworld.com/2016/05/18/344832/us-smart-city-challenge-adds-
nxps-rfid-tagging-solutions/
Bodden, R. (2016b). Gemalto and Visa roll out contactless payment wristbands at Eurovision.
Available: https://www.nfcworld.com/2016/05/10/344604/gemalto-visa-roll-contactless-
payment-wristbands-eurovision/
Bodden, R. (2016c). Alibaba Unveils Connected Car for Alipay Mobile Payments in China. Available:
https://www.nfcworld.com/2016/07/19/346300/alibaba-unveils-connected-car-alipay-
mobile-payments/
Bueno-Delgado, M. V., PavÓn-Marino, P., De-Gea-García, A., and DolÓn-García, A. (2012b). The Smart
University experience: A NFC-based ubiquitous environment, Sixth International Conference on
Innovative Mobile and Internet Services in Ubiquitous Computing (IMIS), 799–804.
Chang, Y. F., Lin, S. C., and Chang, P. Y. (2011c). A location-privacy-protected RFID authentication
scheme. IEEE International Conference on Communications (ICC), pp. 1–4.
Freudenthal, E., Herrera, D., Kautz, F., Natividad, C., Ogrey, A., Sipla, J., Sosa, A., Betancourt, C.,
and Estevez, L. (2007a). November. Suitability of NFC for medical device communica-
tion and power delivery. Engineering in Medicine and Biology Workshop, 2007 IEEE Dallas,
pp. 51–54.
Ghosh, S., Goswami, J., Kumar, A., and Majumder, A. (2015). Issues in NFC as a form of contactless
communication: A comprehensive survey. In Smart Technologies and Management for Computing,
Communication, Controls, Energy and Materials (ICSTM), pp. 245–252.
Gubbi, J., Buyya, R., Marusic, S., and Palaniswami, M. (2013). Internet of Things (IoT): A vision,
architectural elements, and future directions. Future Generation Computer Systems, 29(7),
1645–1660.
Hunter, P. (2016). IoT + NFC: Four reasons why IoT needs NFC. Available: http://internetofthingsagenda.
techtarget.com/blog/IoT-Agenda/IoT-NFC-Four-reasons-why-IoT-needs-NFC
Mainetti, L., PatronoI, L., and Vergallo, R. (2012). IDA-Pay: An innovative micro-payment system
based on NFC technology for Android mobile devices, 20th International Conference on Software,
Telecommunications and Computer Networks (SoftCOM), pp. 1–6.
Majumder, A., Bhattacharyya, B. K., Ghosh, S., and Goswami, J. (2016). A digital card serving identity
and payment purpose, filed as Indian Patent at IPO Kolkata, Application No. 201631004666.
Miraz, G. M., Ruiz, I. L., and Nieto, M. Á. G. (2009). How NFC can be used for the compliance of
European higher education area guidelines in European universities. First International Workshop
on Near Field Communication, pp. 3–8.
Monteiro, D. M., Rodrigues, J. J. P. C., and Lloret, J. (2012). A secure NFC application for credit transfer
among mobile phones. International Conference on Computer, Information and Telecommunication
Systems (CITS), pp. 1–5.
Morak, J., Kollmann, A., Hayn, D., Kastner, P., Humer, G., and Schreier, G. (2007b). Improving tele-
monitoring of heart failure patients with NFC technology. In Proceedings of the Fifth IASTED
International Conference: Biomedical Engineering, pp. 258–261.
Nasution, S. M., Husni, E. M., and Wuryandari, A. I. (2012a). Prototype of train ticketing application
using near field communication (NFC) technology on android device. International Conference
on System Engineering and Technology (ICSET), pp. 1–6.
NFC Everywhere (2016). Available: www.nxp.com/documents/brochure/939775017634.pdf
NFC Forum (2016). Simplifying IoT: Connecting, commissioning, and controlling with near field
communication (NFC) NFC Makes the Smart Home a Reality. Available: http://nfc-forum.
org/wp-content/uploads/2016/06/NFC_Forum_IoT_White_Paper_-v05.pdf
NFC Forum (2011). NFC in public transport. At http://nfc-forum.org/wp-content/uploads/2013/12/
NFC-in-Public-Transport.pdf
NXP Semiconductors (2014). NFC for embedded applications: Your critical link for the internet of
things, Available: https://cache.nxp.com/documents/brochure/75017587.pdf
Prinz, A., Menschner, P., Altmann, M., and Leimeister, J. M. (2011). inSERT—An NFC-based self
reporting questionnaire for patients with impaired fine motor skills. 3rd International Workshop
on Near Field Communication (NFC), pp. 26–31.
Sankarananrayanan, S. and Wani, S. M. A. (2014). NFC enabled intelligent hospital appointment and
medication scheduling. 2nd International Conference on Information and Communication Technology
(ICoICT), pp. 24–29.
Sun, X. and Zhao, Z. (2012). An improved location-privacy-protected RFID authentication scheme.
Journal of Hangzhou Dianzi University, 32(5), 57–60.
Texus Instrument NFC in the Internet of Things. https://e2e.ti.com/blogs_/b/connecting_
wirelessly/archive/2013/09/17/nfc-in-the-internet-of-things-iot
The Internet of Things (2005). ITU Internet Reports. Avaliable: https://www.itu.int/net/wsis/tunis/
newsroom/stats/The-Internet-of-Things-2005.pdf
Urien, P. (2014). LLCPS: A new secure model for internet of things services based on the NFC P2P
model. IEEE Ninth International Conference on Intelligent Sensors, Sensor Networks and Information
Processing (ISSNIP), pp. 1–6.
Wallner, G. (2014). System and method for a baseband nearfield magnetic stripe data transmitter.
U.S. Patent 8,628,012.
Walton, C. A. (1983). U.S. Patent No. 4,384,288. Washington, DC: U.S. Patent and Trademark Office.
Zhuang, Z. J., Zhang, J., and Geng, W. D. (2014). Analysis and optimization to an NFC security authen-
tication algorithm based on hash functions. International Conference on Wireless Communication
and Sensor Network (WCSN), pp. 240–245.
10
Trust Management in IoT
Avani Sharma, Pilli Emmanuel Shubhakar, and Arka Prokash Mazumdar
CONTENTS
10.1 Introduction ...................................................................................................................... 221
10.1.1 Internet of Things: Introduction, Characteristics, and Challenges ............222
10.1.2 Security, Privacy, and Trust in IoT ................................................................... 224
10.1.3 Motivation ........................................................................................................... 224
10.2 Trust: Properties and Management...............................................................................225
10.2.1 Properties of Trust .............................................................................................225
10.2.2 Management of Trust ........................................................................................ 227
10.3 Trust-Management System ............................................................................................. 227
10.3.1 Information Gathering ...................................................................................... 227
10.3.2 Trust Computation............................................................................................. 230
10.3.3 Trust Dissemination .......................................................................................... 233
10.3.4 Update and Maintenance ................................................................................. 233
10.4 Attacks against Trust Management ..............................................................................234
10.4.1 Attacks Based on Reputation Management ...................................................234
10.4.2 Attacks Based on Identity Management ........................................................ 237
10.5 Conclusion ........................................................................................................................ 237
References..................................................................................................................................... 238
10.1 Introduction
The advent of new technologies and innovation in networking enabled communication to

be wireless, pervasive, and ubiquitous across the globe. A new generation of World Wide
Web, that is, Web 3.0 (Hendler, 2009), is bringing the intelligent web to reality by providing
Internet-based services using the semantic web, microformats, machine learning, and so
on. With an increasing number of Internet users, research is being focused on developing
new technologies that can fulfill user’s demands with their satisfaction level. This has
led to the evolution of Internet of Things (IoT) (Atzori et al., 2010) which is an emerg-
ing paradigm that facilitates communication between uniquely identifiable heterogeneous
physical objects seamlessly through the Internet. These physical objects, called IoT devices,
have embedded intelligence and decision-making power which makes them smart enough
to communicate with each other to get data, information, or services. Deployment of IoT
for real-world application suffers various challenges such as addressing the huge number
of devices, analyzing heterogeneous information, open communication channel, and so
on (Borgia, 2014; Stankovic, 2014). Security is another critical issue that adversely affects
221
the performance of IoT applications (Sicari et al., 2015). Characteristics of IoT environment
such as wireless channel, resource constraint devices, limited bandwidth, Internet con-
nectivity, and so on make it vulnerable toward various kinds of attacks. Managing trust
is an important aspect of security in uncertain IoT environment, in which devices may
behave maliciously with time, to interrupt the services and performance. A trusted envi-
ronment, in which devices can share, process, and access information without bothering
about security and privacy, is needed to effectively utilize IoT services. This chapter is
intended to introduce the readers about the importance of trust while building a secure
IoT system. Describing the major characteristics and challenges for IoT environment, this
section gives a glance on the problem of Privacy, Security, and Trust in IoT environment.
Section 10.2 presents detail description about trust covering various properties of trust
and its management. In the next Section 10.3, various phases associated with the trust-
management system (TMS) are explained and commonly used computation models for
the trust evaluation are discussed. Finally, Section 10.4 introduces various attacks which
are possible to be launched against TMS. These attacks continue to thwart performance of
IoT even after embedding trust-management solutions.
10.1.1 Internet of Things: Introduction, Characteristics, and Challenges

IoT was first introduced by Kevin Ashton of Massachusetts Institute of Technology at
Auto-ID Center in the year of 1999 with the integration of radio-frequency identification
with sensor technology. Today, IoT has become the foremost technology to achieve a smart
vision of the world. According to Gartner’s Hype Cycle 2015, IoT took place on the top of
emerging IT technologies. Cisco Internet Business Solutions Group forecasts connectivity
of 50 billion devices to the Internet by the end of 2020 (Evans, 2011).
Various organizations define IoT but no standard definition of IoT exists in literature that
can be used globally. Cisco (Evans, 2011) states that “IoT is simply the point in time when
more things or objects were connected to the Internet than people.” According to ITU
vision of IoT (Pena and Ismael, 2005), “From anytime, anyplace connectivity for anyone, we
will now have connectivity for anything.” In support of connecting various heterogeneous
devices to the Internet protocol (IP) based network, IPSO (IP for smart Objects) Alliance
forum (Dunkels and Vasseur, 2008) was formed in September 2008. The forum brings the
importance of wise adaptation in IP together with IEEE 802.15.4 for full deployment of
IoT across the globe. Collaborative efforts have been made by Internet Engineering Task
Force and IEEE 802.15.4 working group to standardized IoT. They provide a paradigm shift
from traditional IP network-based protocol stack to the smart protocol stack with power
efficiency that can run on low-power IoT devices. Figure 10.1 presents layered architecture
of IoT with various enhanced communication protocols and standards.
The major driving force behind the success of IoT is the integration of various tech-
nologies such as wireless sensor networks, radio-frequency identification, IPv6 &
enhanced communication protocol, cloud computing, nanotechnology, identification
and tracking (Tan and Koo, 2014). Some of the important characteristics of IoT network
inherited from the above technologies are heterogeneous and light-weight devices, low
power consumption, limited computation capacity, limited memory, battery operated,
embedded intelligence, decision making, and distributed computing. Deploying IoT
for real world applications cope up various challenges that need to be addressed for
effective realization of IoT (Borgia, 2014; Stankovic, 2014). Some of the major chal-
lenges are as follows:
Trust Management in IoT 223
Application layer CoAP MQTT HTTP
DTLS
Transport layer UDP
Network layer RPL IPv6/IPv4
6LoWPAN
Data link layer
IEEE 802.15.4
Bluetooth 4.0
Wi-Fi
NFC
Physical layer
FIGURE 10.1
Layered architecture for Internet of Things.
• Addressing and identification: IoT is a collection of large number of devices. To cope

up with the existing IP-based Internet functionality, a larger address space is
required for IoT devices which uniquely identify these devices. IPv6 overcomes
this issue but present challenge is how to get these addresses and how to integrate
addresses with the existing solution is a challenge in IoT.
• Architecture: A generic architecture to permit easy connectivity, control, and com-
munication between heterogeneous devices are required. Designing such a stan-
dard architecture with heterogeneous devices is a nontrivial task.
• Creating knowledge: Huge amount of data is generated by heterogeneous IoT
devices as they are continuously sensing information. Building knowledge from
this uncertain data to create wisdom is a key issue.
• Robustness: Dynamic behavior of IoT environment makes it susceptible toward
asynchronous activities like clock synchronization, which is undesirable. A robust
design which overcomes issue of reconfiguration with the dynamically changing
situation is the need of ours.
• Security: Collaboration of different technologies in IoT environment makes it vul-
nerable toward various security issues insured with underlying technologies. An
unauthenticated or malicious device in the system can compromise system perfor-
mance by launching various types of attacks.
• Openness: Most of the communication between IoT devices is performed wire-
lessly. Openness in IoT refers to the freedom of device to configure itself with
network infrastructure. IoT devices do not need standard protocols stacks to join
the network. This openness introduces uncertainty in the system.
• Interoperability: Supporting operability between heterogeneous devices is a major
task that should be handled by IoT infrastructure carefully. Cross platform design
of protocol is required to support heterogeneous communication.
10.1.2 Security, Privacy, and Trust in IoT

Security is a measure to define system’s reliability, robustness, and effectiveness.
It is defined as the “quality of remaining secure from the risk” (Merriam-Webster,
2007). For a system to be secure, five security measures are defined: Confidentiality,
Integrity, Availability, Authentication, and Authorization (ISO, 2005). One important
aspect of security is privacy (Sicari et al., 2015). Privacy deals with data protection and
confidentiality of user’s information. In open environment like IoT, where informa-
tion access can be done using the Internet, the probability of the disclosure of a user’s
personal information among the others is higher than traditional information systems.
Privacy solutions are generally concerned with whom and when information must be
released or disclosed. Mechanisms like cryptography and access control are used to deal
with the problem of security and privacy. These mechanisms are termed Hard Security
by Rasmusson and Janssen (1996) in which a security policy is always assumed to exist
for authorization of entities in the system and to give them right for accessing the infor-
mation. However, existing security mechanisms do not ensure the reliability of a sys-
tem in the presence of malicious activities performed by an insider entity due to its
behavioral change. An authenticated entity can be compromised to disrupt the system
performance. Even with an embedding security solution, a system can behave in an
untrustworthy manner with respect to certain context.
Trust (Eder et al., 2013; Josang, 2007) solves this problem by maintaining legitimacy
between the entities by analyzing their behavior with respect to particular context. It
defines a measure of confidence about the expected behavior of entities. The concept of
Trust comes under the category of Soft Security (Rasmusson and Janssen, 1996). Mechanisms
to implement soft security follow ethical norms of a system for assessing the behavior.
They basically deal with maintenance of reputation that defines general perception about
behavior of an entity seen globally in a system.
In IoT, traditional solutions for security and privacy such as cryptography, digital
signature, key management, cannot be directly applied because of resource constraint
nature of IoT devices. Involvement of different protocol stacks, technologies, and
standards need a unified approach that can be embedded with a heterogeneous
environment. Moreover, scalability is essentially refined by a huge number of
interconnected heterogeneous devices. There is a need to embed advanced security
mechanism like trust and reputation (Yan et al., 2014) that can capture dynamic
behavior of a system.
10.1.3 Motivation
IoT facilitates various applications where information or services can be accessed ubiq-
uitously through the Internet. A lot of work has been proposed to optimize and use these
applications in the best possible way. However, vulnerability assessment, which is the
foremost aspect that should have been considered for these latest technologies, is still
far from the required progress. Malicious devices across the globe are able to compro-
mise the security primitives of IoT. Behavior-based analysis of devices is required that
can predict the device performance over the time. A trustworthy system is needed to
prevent from unwanted activities conducted by malicious devices. Trust is an important
aspect of security which encompasses security and privacy with goodness, strength,
ability, and availability. Trust management provides analysis on the dynamic behav-
ior of devices considering their past behavior, reputation, or recommendation received
from the other’s opinion. After analyzing the importance of trust management in IoT,
following conclusions can be drawn:
• Trust management facilitates reliable data fusion and mining, qualified service in
heterogeneous IoT environment.
• Trust management is required in various decision-making processes like intru-
sion detection, management of keys, authentication, and so on.
• In uncertain conditions, where a device can leave or join the network at any
moment of time, trust management is useful to make the decision with the limited
amount of knowledge available.
• Trust management helps users to overcome the risk engaged in acceptance and
consumption of IoT services.
• Trust management accelerates access control by isolating misbehaving devices in
the system.
10.2 Trust: Properties and Management

The concept of trust is not confined to a single domain. Trust is a multidomain, multidi-
mensional concept having importance in every aspect of life like social science, economics,
philosophy, psychology, and so on. (McKnight and Chervany, 1996). Trust defines reliance
about the behavior of an entity. Cook (2003) defines trust as the degree of subjective belief
about the behaviors of a particular entity. Various definitions for trust have been proposed
in the literature (Castelfranchi and Falcone, 2000; Cook, 2003; Gambetta, 2000). These are:
Trust is an attitude that we have toward people whom we hope will be trustworthy,
where trustworthiness is a property, not an attitude (Castelfranchi and Falcone, 2000).
Trust basically is a mental state, a complex mental attitude of an agent ‘x’ towards another
agent ‘y’ about the behaviour/action ‘a’ relevant for the result (goal) ‘g’ (Cook, 2003).
Trust (or, symmetrically, distrust) is a particular level of the subjective probability with
which an agent assesses that another agent or group of agents will perform a particular
action, both before he can monitor such action (or independently of his capacity ever to
be able to monitor it) and in a context in which it affects his own action (Gambetta, 2000).
Wanita et al. (2013) present importance of trust in a social scenario which derives various
facts about the theory of trust to be used in domain of communication and networking.
IoT is all about making ubiquitous network of physical devices where different devices can
communicate with others to share information or to request services from other devices
anywhere and at any time. Trust between the devices is required to create a reliable
environment where various devices can share their thoughts, opinions, and experiences
without being bothered about the privacy of their information.
10.2.1 Properties of Trust

Adaptation of trust in IoT environment should fulfill certain properties. These properties,
given in Figure 10.2, define the necessary condition on trust management (Cho et al., 2011;
Lopez et al., 2010; Yu et al., 2007).
Nontransitive
Dynamic Context-
dependent
TRUST
Subjective Asymmetric
Adaptive
FIGURE 10.2
Properties of trust.
• Asymmetric: In heterogeneous IoT environment, in which each device with differ-

ent capacity (computation power and energy) plays different roles, trust remains
asymmetric (Rasmusson and Janssen, 1996; Yan et al., 2014). A device A, trusting
another device B, may not receive the same level of trust from B. For example, a
provider of a service can trust on requester of service but the same may not be true
because opinion, perception, and expectation of requester may vary.
• Adaptive: For a dynamic environment like IoT where devices can quit or join the
network at any moment of time, trust should follow adaptability with the envi-
ronment. To capture dynamicity of IoT environment, trust must be expressed
in continuous form with regular update in its value so that uncertainty in the
network can be mapped to current state (Wanita et al., 2013).
• Dynamic: Value of trust may be changed according to a new observation. It may
increase or decrease with time as experienced by the devices. With each interac-
tion, opinions of devices may change, which results in change in trust value for
the device.
• Context dependent: Scope of trust is context specific. For example, a device A,
trusting on other device B which is providing weather information, will not have
the same degree of trust on the other device C providing temperature or pressure
information (Lopez et al., 2010; Yu et al., 2007).
• Nontransitive: Trust follows nontransitivity. The principle of nontransitivity says
that if a device A trusts on a device B and the device B trusts on another device C,
then it is not necessary that the device A will trust the device C (Cho et al., 2011;
Lopez et al., 2010).
• Subjective: Trust is subjective with respect to the context in which it is applied
(Lopez et al., 2010; Yan et al., 2014). Opinions of different trustors (evaluator of trust)
may have the different level of trust on the subject’s behavior. Trustworthiness of
a subject not only depends on its behavior but also on the perception of trustors
about subject’s behavior.
10.2.2 Management of Trust

Trust management integrates individual’s views about behavior of an entity with the
global opinions seen by the other members of a system. These global opinions are called
reputation. Blaze et al. (1996) were first toward bringing the concept of trust for security
computation. He introduced the term Trust Management and highlighted its importance
as a separate component for building secure systems. According to Blaze, management of
trust in a system opens the way to interpret and specify security policies, relationships,
and credentials. For IoT, the importance of trust management is not confined to a single
layer of communication stack (Hoffman et al., 2009; Lacuesta et al., 2012; Yan et al., 2014).
All the three layers: physical, network, and application are subjected to achieve security
goals while embedding trust management. These goals are as follows:
• Physical layer/device layer: At the physical layer, efficient and scalable identity man-
agement (IM) for the devices is required in IoT because of handling numerous
heterogeneous devices. These identities can be any real-world identifier or may be
assigned by the users. The objective of trust management is to prevent the system
from outcomes of acquiring faked or multiple identities.
• Network layer: At the network layer, solutions for reliable data fusion, transmission,
mining, and communication are required. The objective of trust management is to
analyze and process huge amount of sensor data generated from heterogeneous
devices for the purpose of receiving accuracy and reliability.
• Application layer: At the application level, ensuring quality of services of the
requested information is a major concern with the expectation Only me, only now
and only here. Trust management should ensure that requested services should
precisely deliver to the right person at a right time without losing the quality.
Moreover, preserving the privacy of a user’s information according to the expecta-
tion and policies of user should be there.
10.3 Trust-Management System

TMS can be represented as a framework of multiple processes which iteratively evaluate,
disseminate, and update trust score of an entity on the basis of information gathered to
maintain trustworthiness between the entities. Definition of the entity may vary accord-
ing to application scenario in which they are used. In the case of IoT, an entity is an end
device who is involved in the process of communication with the embedded decision-
making capacity. Various components are associated with management of trust like expe-
rience, knowledge, reputation, recommendation, and so on. Bringing trust to the domain
of computation is closely related to the term reputation (Eder et al., 2013). Although Trust
represents subjective view about the trustworthiness of an entity, reputation defines general
perception about the behavior of an entity computed globally (Dobelt et al., 2012). The
generic structure of the TMS involves four phases as shown in Figure 10.3.
10.3.1 Information Gathering

The first phase of TMS deals with defining preliminary requirement to compute trust in a
system and acts as input to the process of trust computation. For example, what parameters
1. Information gathering
4. Update and maintenance
Start * Trust parameters
* Time driven
* Source of information
* Event driven
* Type of information
2. Trust computation
3. Trust dissemination
* Statistical models End
* Probabilistic models * Centralized
* Fuzzy models * Distributed
* Flow models
* Machine learning
FIGURE 10.3
Phases involved in the designing of trust-management system.
should be chosen to define trust in a system, how the information is collected, and how to
represent the information in computational form?
Strategies adopted for TMS may vary according to the application scenario. In a social
network, trust can be built by measuring the degree of friendship or community of interest
between the members of a group. Similarly, in some applications like banking or finance,
honesty and fairness play an important role in building trust (Wanita et al., 2013). These
qualitative attributes are called trust parameters which can be represented in quantitative
forms like binary, discrete, or continuous (Ruan and Durresi, 2016). Before evaluating trust
score for an entity, one should be aware of about what trust parameters should be chosen
and what mathematical representation should be used to maintain this information in the
computational form. To understand the process of trust management, we imply trustor, as
an entity who is an evaluator of trust, and trustee, as an entity for which trust is calculated.
1. Trust parameters: Figure 10.4 depicts various trust parameters that can be used to
compute trust. Three essential attributes that are used to derive these parameters
are knowledge, experience, and reputation (Ruan and Durresi, 2016). Reputation
about an entity can be maintained based on the knowledge and experience about
the behavior of that entity. General parameters which are involved in the process
of defining trust are as follows:
• Honesty refers to belief or loyalty of trustee toward trustor.
• Behaviors defines different patterns of individual with each interaction.
• Attitude represents positive and negative views of an entity.
• Friendship represents state of intimacy with someone.
• Fairness represents quality of treating each entity equally in a reasonable way.
• Benevolence defines desire of an entity for being good or kind toward others.
Reputation
Benevolence
Behavior Attitude
Honesty Trust Col
Fairness Friendship
Cooperativeness
Experience Knowledge
FIGURE 10.4
Trust parameters.
• Cocooperativeness represents compliance of trustee with trustor in a social

application.
• Community of interest represents similar characteristics or capacity of entities in
same community.
2. Source of information: Collecting raw information for trust computation is a major
task in designing of any TMS. There are two measures on the basis of which infor-
mation can be obtained (Hoffman et al., 2009; Sabater and Sierra, 2005). These are
as follows:
• Direct measure: A trustor can rely on own observations or findings about the
behavior of an entity, like analyzing the number of successful or failed transac-
tions, resource utilization for each transaction, and so on. Such type of infor-
mation is interpreted as first-hand information.
• Indirect measure: Information received from third party or obtained through
transitive relation is called second-hand information or recommendation.
Recommendation is the most common way to utilize second-hand informa-
tion where reputation or trust evaluation of an entity depends on global view
observed by the other entities in the system.
3. Scaling of information: Information collected for trust computation must be repre-
sented in a form in which computation can be carried out. Various numeric for-
mats, such as continuous, discrete, and intervals, exist that can be used to represent
gathered information in mathematical computation form (Ruan and Durresi, 2016;
Wanita et al., 2013). Presenting the information as a discrete value can take any of
the following two forms: binary discrete and multinomial discrete. In binary dis-
crete form, information is presented in the binary form of 0’s and 1’s. Binary values
are easy to understand and can easily represent behavior of an entity by using
0 for distrust and 1 for trust. Multinomial discrete values are used when trustor
does not have binary opinion about the behavior of an entity. To scale more trust-
worthiness of an entity, trustor can use different scales of measurement like High
Trust, Trust, Low Trust, Distrust, and so on. This type of representation provides
more accurate information toward the degree of trust of an entity having more
opinions. Another way to represent trust information is to use continuous values
which are generally used in the situations where probability density functions
(PDFs) are used for input distribution. One more possible representation is to use
interval value, in which information is maintained in the form of intervals which
are used by most fuzzy models.
10.3.2 Trust Computation

The process of trust computation deals with the selection of an appropriate computational
model for trustworthiness evaluation of entities. Various trust computation models exist in
literature that can be used to build TMS for IoT. These models take input from the previous
phase according to the requirement of the application. On the basis of the approach adopted
for output calculation, computation models can be classified as deterministic and probabilis-
tic (Hoffman et al., 2009). In deterministic approach, an output of any computation process
can be determined having knowledge of input parameters. Such kinds of approaches are
suitable where size of the system is small and computation depends on the accuracy of
individual views instead of considering global perspective. Performing simple weighted
sum or average of the feedback ratings falls under this category. In probabilistic approach,
input variables used to derive output possess random values. Computation is performed
over the random set of values instead of having discrete input values. These are also called
stochastic algorithms where predictive outputs have certain error bounds. Error bounds
define approximation in predicted values with an acceptable level of error. Bayesian model
(Josang, 2001; Josang and Ismail, 2002) and Belief model (Josang, 2001) are the most com-
monly used probabilistic models in the domain of trust computation. Some models are
designed based on the machine learning algorithms to capture dynamic behavior of enti-
ties. Hidden Markov Model (HMM) (Anjum et al., 2015) and Artificial Neural Network
(ANN) (Zong et al., 2009) are the examples that use machine learning as a tool to compute
trust. The other models discussed in the literature to compute trust are fuzzy models
(Chen et al., 2011; Manchala, 1998; Mahalle et al., 2013; Sabater and Sierra, 2002), graph-
theory-based models (Theodorakopoulos and Baras, 2006) and flow models (Levien, 2009;
Page et al., 1999).
• Simple statistical models: The easiest and straight-forward form of computing trust
of an entity is to perform simple mathematics over the feedback provided for each
transaction like sum of ratings (positive and negative rating) and averaging the
rating to measure the degree of trustworthiness. eBay’s reputation management
(Resnick and Zeckhauser, 2002) deploys such approach to maintaining trust over
its service providers and customers. Other commercial sites that use such type
of computation are Epinion (Massa and Avesani, 2005) and Amazon (Massa and
Avesani, 2007). The model efficiently computes trust score; however, the systems
deploying such model lacks in maintaining the global reputation of the entities.
• Probabilistic models: Instead of taking the direct discrete value of input, these models
take the probability of occurrence of an event. Though various algorithms consider
probability as a measure to derive trust score, most commonly used probabilistic
models which use probability distribution functions are as follows:
• Bayesian models: In this, the concept of probability and evidence is considered
to compute trust score (Josang and Ismail, 2002). Binary values with beta PDF
(Falls, 1974) are taken as input for the model. These binary ratings, depict-
ing positive and negative behavior of entities, are used to upgrade trust
score. To calculate the posterior probability of an event, Baye’s rule is used
in which prior probability of occurrence of an event derives predicted poste-
rior probability. Representation of trust score with beta PDF considers tuple of
two parameters, α and β, in which α denotes positive behavior and β denotes
negative behavior of an entity. The beta PDF, represented by beta(p|α,β), can be
expressed in the form of gamma function, Γ, as
Γ(α + β) α −1
beta(p α ,β) = p (1 − p)β −1 Where 0 ≤ p ≤ 1, α , β > 0
Γ(α)Γ(β)
The expectation value of beta distribution, which is used as trust score, is given as
α
E(p) =
α+β
Using Bayesian model in a system provides effective trust or reputation com-

putation as compared with the simple statistical models because of the infer-
ence of prior information in posterior behavior prediction. However, Bayesian
model does not give any information about the selection of prior distribution.
Moreover, these models consider constant probability distribution over each
transaction which restricts their efficiency for deploying dynamic behavior of
the system. To overcome this problem, some algorithms use decay factor or forget-
ting factor while building trust score (Bao and Chen, 2012; Bao et al., 2013; Chen
et al., 2015). Decay factor is a kind of scaling variable which is used to give more
weight to recent information as compared with the previous behavior seen.
• Belief model: These types of models too are related to the theory of probabil-
ity, but summation of probabilities seen with the all possible outcomes of an
event does not necessarily give 1. Residual probability is treated as uncertainty
that can be anything depending on the application scenario. For example, one
basic idea given by Josang (2001) is to model trust with the integration of three
components: belief, disbelief, and uncertainty. A tuple of four parameters (b, d, u,
and a) is used to derive opinion about an entity where b represents belief, d rep-
resents disbelief, and u represents uncertainty with the summation b + d + u = 1. a
is called relative atomicity which is used to derive probability expectation value
for the opinions as b + au. A well-known example of reasoning with uncertainty
is Dempster–Shafer Theory (Shafer, 1992).
• Fuzzy models: In Fuzzy models, computation of trustworthiness or reputation is
done using the fuzzy set of values instead of having deterministic values. Input
variables in such type of model have randomly assigned value (Chen et al., 2011;
Mahalle et al., 2013). A membership function is used with the observable vari-
ables to describe degree of trust in the interval ⟨0, 1⟩. Instead of computing trust
using simple statistical calculations, fuzzy logic uses set of rules incorporated
with IF-Else conditions. These rules are defined to deal with uncertainty in the
network and to perform reasoning with fuzzy measurements. The schemes which
fall under this category are reputation model for gregarious societies (REGRET)
(Sabater and Sierra, 2002) and the reputation system given by Manchala (1998).
• Machine learning-based models: Machine-learning algorithms can be used to ana-
lyze future behavior of entities based on analyzing their previous transactions.
Most common machine-learning algorithms to capture the dynamic behavior

of an entity are HMM (Anjum et al., 2015) and ANN (Zong et al., 2009). HMM
predicts dynamic behavior of an entity using the concept of states. Each inter-
action of trustor with trustee is associated with some contextual information
such as number of successful transactions, number of failure transactions, and
delay in receiving services. The output of interaction defines levels of trust rat-
ing where each level represents the state of an entity for that transaction. For a
HMM (λ), trustor can infer next state of trustee, L, using a sequence of observa-
tion Fm + 1 = F0 F1F2...Fm with the probability function, P(S = L|Fm + 1). A forward
algorithm is applied to recursively derive this probability. Finally, the trustor
is able to decide whether to interact with the target entity having the highest
probability based on trust rating. In ANN based algorithms, learning is applied
to map input vector into output values. These algorithms can work with any
nonlinear relationships between input and output by applying iterative train-
ing to data sets.
Linear discriminate analysis (LDA) and decision tree (Liu et al., 2014) are other
two most common machine learning-based data analytics techniques which can
be used to derive trustee’s future behavior based on historical data of past inter-
action. These two techniques are application dependent, in which interaction
between trustor and trustee is described by a set of features and two classes, suc-
cessful and unsuccessful, define trustworthiness of entities. In LDA, a trustor after
classifying the historical data into Successful and unsuccessful performs LDA on
two groups to obtain a linear classifier. This classifier is then used to estimate clas-
sification of potential interaction in successful group. In decision tree algorithms,
first a tree is constructed using training data. After that, classification algorithms
are applied to the potential transaction by starting from the root of the tree and
moving down until a leaf node is found.
• Flow models: In these models, trust or reputation is computed iteratively
considering the long chain of transitive relationship between the entities.
Computation of trust score considers opinions of other entities in the system.
Incoming flow and outgoing flow concentrated on an entity is used to update
true score of that entity. For example, ranking about an entity can be deter-
mined by considering how many of others referring that entity. Application
of these models can be seen in Google’s page rank (Page et al., 1999) and
reputation scheme given by Levien (2009). In Google’s page rank, hyperlinks
toward the page rise its ranking and hyperlink from the page leads to decrease
its rank on the web. Another example is EigenTrust model (Kamvar et al.,
2003) used with peer-to-peer networks. In EigenTrust, repeated and iterative
aggregation and multiplication of trust score derived along with transitive
chain is used to make final trust score.
• Graph theory models: In Graph Theory, centrality and distance are generally
used measures to calculate degree of trust for an entity. Centrality defines
importance of an entity by considering in-degree (transaction received) and
out-degree (transaction initiated) on that entity. Distance between two entities
defines connectivity between them. Most popular methods for graph-based
trust measure are Google page rank (Page et al., 1999) and EigenTrust (Kamvar
et al., 2003).
• Discrete models: Sometime it is better to use human’s general perception instead of

using any existing computational model to measure degree of trustworthiness of
an entity. The situation arises in the systems unable to handle complex informa-
tion lacking with computation resources. Some heuristic approach can be used to
compute trust score similar to look-up table (Carbone et al., 2003; Liau et al., 2003).
Disadvantage of using such approaches is that they do not present generalized
structure for trust management.
10.3.3 Trust Dissemination

Communicating trust information between the entities of a system can opt following two
approaches to design the architecture of a system (Hoffman et al., 2009; Josang, 2001):
• Centralized
• Distributed
1. Centralized architecture: In this architecture, a central entity is chosen, who takes

in charge of all the computation, storage, and decision making. Information
about the behavior of entities is collected by those entities who have some previ-
ous interaction with the evaluating entity. A central authority is chosen which
takes responsibility of maintaining reputation of all the entities and to make them
available for public use. To know the trust score of a device, trustor requests cen-
tralized authority which, in turn, will do all the needful and return trust score
about the device to the requesting entity. In this approach, trust computation has
global opinions instead of having bias value. However, the approach suffers from
potential single point of failure. Most common application scenario which imple-
ments this architecture is wireless sensor networks in which sink node act as base
station to perform all the centralized computation and decisions on the behalf of
other nodes in the network.
2. Distributed architecture: In this type of architecture, there is no central entity for com-
puting and managing trust scores. Instead, each entity contributes toward the pro-
cess of information exchange and stores opinions about past behavior. A well-suited
environment for distributed architecture is peer-to-peer networks in which every
node plays the role of both client and server. In the case of IoT, where devices are able
to perform computation and decision making on their own without the need of any
centralized server, use of distributed architecture is more appropriate compared
with the centralized architecture. The advantage of using distributed approach is
that no single point of failure affects trust management. At the same time, memory
requirement and computation cost per entity increases. Moreover, such approaches
have higher convergence time compared with the centralized architecture.
10.3.4 Update and Maintenance

Final stage of TMS is the maintenance of trustworthiness of entities in the network.
To maintain recent behavior of the entities, their trust score must be updated and
maintained with time. Specially, in distributed and dynamic scenario similar to IoT, in
which a device’s behavior is not constant throughout the time, computed trust score
needs to be updated. Two basic approaches toward updating trust scores are as follows
(Guo and Chen, 2015):
1. Event driven: Trust of a device can be updated upon encountering an event. In such
a scenario, each activity is considered as an event and trust value is updated on
each activity. For example, wherever a trustor receives a service from trustee, the
trustor may update previously computed trust score according to current behav-
ior seen.
2. Time driven: Trust of a device can be updated periodically. Counter-based strate-
gies are adopted to implement such kind of approach. Here, trust score is updated
with each time unit instead of waiting for occurrence of an event.
10.4 Attacks against Trust Management

Trust management comprises of managing and building the reputation of entities
locally (direct measure) and globally (indirect measure) in the system. In open systems
similar to IoT in which devices can join and leave the network at any moment of time,
maintaining trust and reputation between heterogeneous devices is a challenging task.
Even after embedding trust management strategies, the system can be compromised by
insider adversaries. These adversaries can launch the attack either for self-benefit or to
perform malicious activity (Koutrouli and Tsalgatidou, 2012). A selfish attacker subverts
the system performance with the intension of increasing own benefit, whereas a mali-
cious attacker targets other devices to degrade the system performance. These attackers
can work individually or may collude with other devices in the system (Marmol and
Perez, 2009). In later category, attacking device forms a group with the other devices to
perform a malicious activity and boost the effect of attack on the system performance.
One major aspect that affects the process of trust management is IM. Identity belong-
ing to an entity (device) plays an important role while building the reputation of that
entity. These identities can be any real-world identity or can be defined by the users.
Pseudonym in assigning identities allows attackers to bear multiple identities. If a sys-
tem is compromised with faked identities, any trust management strategy can be vul-
nerable toward attacks. Therefore, a system is assumed to embed efficient IM scheme
for managing authentication between heterogeneous identities. We classify the attacks
(Figure 10.5) into two categories where an attacker can disrupt either reputation man-
agement or IM system.
10.4.1 Attacks Based on Reputation Management

Effectiveness of reputation management depends on the information gathered about the
behavior of devices using their past history and recommendations received from the other
devices in the system. An attacker can behave in inconsistent manner or may provide the
dishonest recommendation about a device to maintain false reputation. By showing con-
flicting behavior, an attacker can spoil the process of information gathering which in turn
results in maintaining spurious trust score about the devices. Such type of behavior makes
system prone toward Denial of Service attack where system resources can be depleted
Attacks
Reputation Identity
management management
On–Off Sybil
Discriminative Whitewashing
behavior
Repudiation
Self-promoting
Bad-mouthing
Slandering
Ballot-stuffing
FIGURE 10.5
Attacks classification against trust management.
because of processing and storing false information about the devices. Attacks that come
under this category are as follows:
1. On–off attack: In this attack, the attacker switches its behavior between
good and bad by alternatively providing good or bad services (Koutrouli
and Tsalgatidou, 2012). Here, the attacker shows inconsistency in the time
domain. As the reputation goes down during the course of bad behavior,
the attacker regains its reputation by providing good services. This alter-
native behavior makes the presence of attacker undetected and difficult to
update and maintain effective reputation timely. The attack is also called
traitors, because of the sudden change in the behavior of an attacker. Most
of he TMS are vulnerable toward on–off attack, because of considering the
only recent behavior of devices instead of combining past behavior with the
current one comprehensively.
Solutions to deal with on–off attack consider decay factor or forgetting factor
while computing trust score (Bao et al., 2013; Chen et al., 2015). This decay
factor is used to give more weight for recent information over the past infor-
mation. Another approach is to consider bad behavior in the form of distrust
along with the trust while building the reputation of a device because only a few
bad transactions can drastically affect reputation over the consistent long term
good transactions (Sun et al., 2008).
2. Discriminative behavior attack: In this attack, the attacker shows inconsistency in
user domain instead of targeting time domain (Josang et al., 2007). An attacker
may show good behavior toward one group of devices and may misbehave with
another small set of devices. Such discrimination leads to increase reputation of
attacker significantly by the group who experienced good behavior compared
with decrease in reputation by the group who experienced bad behavior. For
example, a service provider can provide good services to most of the requesters
and can target one or few requestor with bad services. While building reputa-
tion, recommendation received with the targeted requesters (who experienced
bad services) has less or no significance as compared with the recommendation
received from the large pool of users.
One possible way to deal with these situations is to consider direct measure
only for trust computation instead of using global recommendation values. Using
global recommendation increases chances of introducing false reputation values
in building trust.
3. Self-promoting attack: Here, an attacker tries to increase one’s own reputation by
falsely reporting positive feedbacks or rating (Josang, 2001). This attack infects the
systems in which trust computation uses simple mathematics over the positive
feedback received. An attacker can be an individual device or may be the group
of malicious devices who collude to fabricate reputation value. Main targets for
such kind of attacks are the systems which do not implement any authentication
scheme to distinguish between legitimate and fabricated feedbacks. Systems, lack-
ing with proper authentication schemes, are susceptible toward acquiring faked
identities (Sybil attack). Having multiple identities, an attacker can significantly
promote own importance.
Mitigating solutions for these attacks include efficient authentication scheme
with a proof of accountability and rate limiter in acquiring multiple identities
(Hoffman et al., 2009).
4. Slandering attack: In slandering attacks, the attacker or group of attackers
provide a false recommendation about other devices in the system to maintain
false reputation about them (Sun et al., 2008; Josang et al., 2007). Here, attackers
are the recommenders who are involved in process of deriving global reputation
of devices. Targets of these attacks are the systems where trust computation is
carried out using indirect measure. The attack exploits the transitive property of
TMS where trust is computed on the behalf of recommendations received from
the third party. Recommenders can show slandering behavior in the following
manner:
• Bad-mouthing attack: Attacker has bad mouth about the victim device and pro-
vides dishonest recommendations for the legitimated victim device to reduce
its reputation. By reducing other’s repudiation, attacker increases own chances
to be selected as an honest entity in the system.
• Ballot-stuffing attack: Meaning of ballot stuffing is to provide rating more than
the required legitimated number. In this, malicious recommender colludes
with other dishonest devices in the system and increases their reputation by
providing good recommendation about them.
Methods to alleviate the effect of slandering attacks include statistical analysis on the rec-
ommendations received from multiple recommenders such as similarity-based filtering.
Moreover, instead of limiting number of recommenders, aggregate the recommendations
from a large pool so that the effect of false recommendation can be diminished (Hoffman
et al., 2009; Sun et al., 2008).
10.4.2 Attacks Based on Identity Management

Identity scheme adopted for any system deals with registration and authentication of enti-
ties in the system. Attackers can exploit identities either during registration or during the
authentication process to spoil system performance. Attacks under these categories are:
1. Sybil attack: Attacker can associate with multiple identities. These identities take the
blame of being malicious in the system (Hoffman et al., 2009; Marmol and Perez,
2009). With these fake identities, an attacker can perform differently each time
reputation is generated or provided. This different behavior creates anonymity in
computing reputation value. Sybil attack is a form of the initiator to launch subse-
quent attacks on reputation management. For example, attacker having fabricated
identity can degrade reputation of another device to increase own reputation that
results in self-promotion attack. Moreover, providing negative information about
an entity results in bad-mouthing attack.
2. New comer/whitewashing attack: In such a kind of attack, attacker subverts the
system performance for some duration and then after knowing about degrada-
tion in owns reputation value, whitewashes its bad history by disappearing from
the system (Koutrouli and Tsalgatidou, 2012). The attacker can rejoin the system
having a new identity to take a fresh start. Attacker behaves as a new comer in
the system with a different identity which allows regaining the same reputation
even after abusing the system with the previous identity.
3. Repudiation attack: An attacker, providing or receiving recommendations, may deny
involvement in such kind of information (Koutrouli and Tsalgatidou, 2012). The
attack is against authentication scheme in which the sender of bogus information
remains unidentifiable because there is no prove of blaming a device in the network.
To defend against identity-related attacks, a strong authentication scheme is required

where recommenders are imposed with proof of their legitimacy while providing recom-
mendations about the devices. Use of digital signature, unique identity, and restriction in
identity generation can thwart the process of acquiring multiple identities (PKI Page; Jurca
and Faltings 2003).
10.5 Conclusion
Embedding security solutions in IoT environment is of prime concern to get reliable
services. Despite using traditional security mechanisms, there is a need to propose some
advanced solutions that can be integrated with resource constraint IoT devices while sup-
porting heterogeneity. Trust management is a major aspect toward mitigation of secure
issues. Importance of trust and its management for IoT applications to achieve the goals of
security and privacy has been presented in this chapter. This chapter intends to describe
the design of a TMS while introducing the various computational models that can be used
to evaluate trust in IoT. Vulnerability assessment of the process of trust management is to
be considered while building robust TMS. This chapter familiarizes the readers with the
various kinds of attacks that can be launched to compromise the management of trust.
References
Anjum, B., Rajangam, M., Perros, H., and Fan, W. (2015). Filtering unfair users: A hidden Markov
model approach. Information Systems Security and Privacy, pp. 147–154.
Atzori, L., Iera, A., and Morabito, G. (2010). The Internet of things: A survey. Computer Networks,
54(15), 2787–2805.
Bao, F. and Chen, I. R. (2012). Dynamic trust management for Internet of things applications.
Proceedings of International Workshop on Self-Aware Internet of Things, pp. 1–6.
Bao, F., Chen, I. R., and Guo, J. (2013). Scalable, adaptive and survivable trust management for com-
munity of interest based Internet of Things systems. International Symposium on Autonomous
Decentralized Systems (ISADS), pp. 1–7.
Blaze, M., Feigenbaum, J., and Lacy, J. (1996). Decentralized trust management. IEEE Symposium on
Security and Privacy, pp. 164–173.
Borgia, E. (2014). The Internet of things vision: Key features, applications and open issues. Computer
Communications, 54, 1–31.
Carbone, M., Nielsen, M., and Sassone, V. (2003). A formal model for trust in dynamic networks.
BRICS Report Series, 10(4), 54–61.
Castelfranchi, C. and Falcone, R. (2000). Trust is much more than subjective probability: Mental com-
ponents and sources of trust. IEEE Hawaii International Conference on System Sciences, pp. 1–10.
Chen, D., Chang, G., Sun, D., Li, J., Jia, J., and Wang, X. (2011). TRM-IoT: A trust management model
based on fuzzy reputation for Internet of things. Computer Science and Information Systems, 8(4),
1207–1228.
Chen, I. R., Bao, F., and Guo, J. (2015). Trust-based service management for social Internet of things
systems. IEEE Transactions on Dependable and Secure Computing, 13(6), 684–696.
Cho, J. H., Swami, A., and Chen, R. (2011). A survey on trust management for mobile ad hoc net-
works. IEEE Communications Surveys & Tutorials, 13(4), 562–583.
Cook, S. (2003). Trust in Society. Russell Sage Foundation Series on Trust. Retrieved from https://www.
russellsage.org/.
Dobelt, S., Busch, M., and Hochleitner, C. (2012). Defining, understanding, explaining TRUST within
the uTRUSTit project. White Paper.
Dunkels, A. and Vasseur, J. P. (2008). IP for smart objects. Internet Protocol for Smart Objects (IPSO)
Alliance, White Paper.
Eder, T., Nachtmann, D., and Schreckling, D. (2013). Trust and reputation in the Internet of things.
Retrieved from https://web.sec.uni-passau.de/projects/compose/papers.
Evans, D. (2011). The Internet of things, How the next evolution of the Internet is changing every-
thing. Cisco Internet Business Solutions Group, pp. 4–11.
Falls, L. W. (1974). The beta distribution: A statistical model for world cloud cover. Journal of
Geophysical Research, 79(9), 1261–1264.
Gambetta, D. (2000). Can we trust trust? Trust: Making and Breaking Cooperative Relations, Oxford, UK:
University of Oxford, pp. 213–237.
Guo, J. and Chen, R. (2015). A classification of trust computation models for service-oriented Internet
of things systems. IEEE International Conference on Services Computing (SCC), pp. 324–331.
Hendler, J. (2009). Web 3.0 emerging. Computer, 42(1), 111–113.
Hoffman, K., Zage, D., and Nita-Rotaru, C. (2009). A survey of attack and defence techniques for
reputation systems. ACM Computing Surveys (CSUR), 42(1), 1.
ISO (2005). Code of practice for information security management. ISO/IEC IS17799-Information
Technology.
Josang, A. (2001). A logic for uncertain probabilities. International Journal of Uncertainty, Fuzziness and
Knowledge-Based Systems, 9(3), 279–311.
Josang, A. (2007). Trust and reputation systems. Foundations of Security Analysis and Design IV, Berlin
Heidelberg: Springer, pp. 209–245.
Josang, A. and Ismail R. (2002). The beta reputation system. Proceedings of the 15th Bled Electronic
Commerce Conference, Slovenia, Balkans, pp. 2502–2511.
Josang, A., Ismail, R., and Boyd, C. (2007). A survey of trust and reputation systems for online service
provision. Decision Support Systems, 43(2), 618–644.
Jurca, R. and Faltings, B. (2003). An incentive compatible reputation mechanism. Proceedings of the
IEEE Conference on ECommerce, pp. 285–292.
Kamvar, S. D., Schlosser, M. T., and Molina, H. G. (2003). The EigenTrust algorithm for reputation
management in P2P networks. International World Wide Web Conference, pp. 640–651.
Koutrouli, E. and Tsalgatidou, A. (2012). Taxonomy of attacks and defence mechanisms in P2P repu-
tation systems-Lessons for reputation system designers. Computer Science Review, 6(2), 47–70.
Lacuesta, R., Navarro, G., Cetina, C., Penalver, L., and Lloret, J. (2012). Internet of things: Where to be
is to trust. EURASIP Journal on Wireless Communications and Networking, 1, 1–16.
Levien, R. (2009). Attack resistant trust metrics. Computing with Social Trust, London: Springer,
pp. 121–132.
Liau, C. Y., Zhou, X., Bressan, S., and Tan, K. L. (2003). Efficient distributed reputation scheme for
peer-to-peer systems. International Conference Human Society at Internet, Berlin, Germany:
Springer, pp. 54–63.
Liu, X., Datta, A., and Lim, E. P. (2014). Computational Trust Models and Machine Learning. Boca Raton,
FL: CRC Press.
Lopez, J., Roman, R., Agudo, I., and Fernandez-Gago, C. (2010). Trust management systems for wire-
less sensor networks: Best practices. Computer Communications, 33(9), 1086–1093.
Mahalle, P. N., Thakre, P. A., Prasad, N. R., and Prasad, R. (2013). A fuzzy approach to trust based
access control in Internet of things. International Conference on Wireless Communications,
Vehicular Technology, Information Theory and Aerospace and Electronic Systems (VITAE), IEEE,
pp. 1–5.
Manchala, D. W. (1998). Trust metrics, models and protocols for electronic commerce transac-
tions. Proceedings of the 18th International Conference on Distributed Computing Systems, IEEE,
pp. 312–321.
Marmol, F. G. and Perez, G. M. (2009). Security threats scenarios in trust and reputation models for
distributed systems. Computers & Security, 28(7), 545–556.
Massa, P. and Avesani, P. (2005). Controversial users demand local trust metrics: An experimen-
tal study on epinions.com community. Proceedings of the 20th National Conference on Artificial
Intelligence, pp. 121–126.
Massa, P. and Avesani, P. (2007). Trust-aware recommender systems. Proceedings of Conference on
Recommender Systems, pp. 17–24.
McKnight, D. H. and Chervany, N. L. (1996). The Meanings of Trust. Minneapolis, MN: University of
Minnesota.
Merriam-Webster (2007). Retrieved from http://www.m-w.com/.
Page, L., Brin, S., Motwani, R., and Winograd, T. (1999). The PageRank citation ranking: Bringing
order to the web. Technical Report, Stanford Digital Library Technologies Project.
Pena, L. and Ismael. (2005). Internet report 2005: The Internet of things. ITU.
PKI Page, An Overview of Public Key Infrastructures, Retrieved from http://www.techotopia.com/
index.php/An_Overview_of_Public_Key_Infrastructures_%28PKI%29.
Rasmusson, L. and Janssen, S. (1996). Simulated social control for secure Internet commerce.
Catherine Meadows, Proceedings of the New Security Paradigms Workshop.
Resnick, P. and Zeckhauser, R. (2002). Trust among strangers in Internet transactions: empirical
analysis of eBay’s reputation system. The Economics of the Internet and ECommerce, Advances in
Applied Microeconomics, 11(2), 23–25.
Ruan, Y. and Durresi, A. (2016). A survey of trust management systems for online social communities–
Trust modeling, trust inference and attacks. Knowledge-Based Systems, 106, 150–163.
Sabater, J. and Sierra, C. (2002). Social ReGreT, a reputation model based on social relations. SIGecom
Exchanges, 3(1), 44–56.
Sabater, J. and Sierra, C. (2005). Review on computational trust and reputation models. Artificial
Intelligence Review, 24(1), 33–60.
Shafer, G. (1992). Dempster–Shafer theory. Encyclopedia of Artificial Intelligence, 330–331.
Sicari, S., Rizzardi, A., Grieco, L. A., and Porisini, A. (2015). Security, privacy and trust in Internet of
things: The road ahead. Computer Networks, 76, 146–164.
Stankovic, J. A. (2014). Research directions for the Internet of things. IEEE Internet of Things Journal,
1(1), 3–9.
Sun, Y., Han, Z., and Liu, K. R. (2008). Defence of trust management vulnerabilities in distributed
networks. IEEE Communications Magazine, 46(2), 112–119.
Tan, J. and Koo, S. G. (2014). A survey of technologies in Internet of things. IEEE International
Conference on Distributed Computing in Sensor Systems (DCOSS), pp. 269–274.
Theodorakopoulos, G. and Baras, J. S. (2006). On trust models and trust evaluation metrics for ad hoc
networks. IEEE Journal on Selected Areas in Communications, 24(2), 318–328.
Wanita, S., Surya, N., and Cecile, P. (2013). A survey of trust in social networks. ACM Computing
Survey, 45(4), 1–33.
Yan, Z., Zhang, P., and Vasilakos, A. V. (2014). A survey on trust management for Internet of things.
Journal of Network and Computer Applications, 42, 120–134.
Yu, H., Shen, Z., Miao, C., Leung, C., and Niyato, D. (2007). A survey of trust and reputation
management systems in wireless communications. Proceedings of the IEEE, 98(10), 1755–1772.
Zong, B., Xu, F., Jiao, J., and Lv, J. (2009). A broker-assisting trust and reputation system based on
artificial neural network. IEEE International Conference on Systems, Man and Cybernetics (SMC),
pp. 4710–4715.
Section III
Interdisciplinary Aspects
of the Internet of Things
11
Highway Safety of Smart Cities
Sándor Szénási
CONTENTS
11.1 Introduction ...................................................................................................................... 244
11.2 Accident Prevention......................................................................................................... 245
11.2.1 Available Services ............................................................................................... 245
11.2.1.1 Intelligent Navigation Systems ........................................................ 245
11.2.1.2 Warning Messages ............................................................................. 245
11.2.1.3 Voice Recognition ...............................................................................246
11.2.1.4 Monitoring Vehicle Status .................................................................246
11.2.1.5 eCall ......................................................................................................246
11.2.1.6 Self-Driving Cars ................................................................................246
11.2.1.7 Additional Services ............................................................................ 247
11.2.2 Internet Access in Vehicle.................................................................................. 247
11.2.2.1 Using Cellular Network ....................................................................247
11.2.2.2 Using Wi-Fi Networks .......................................................................247
11.2.2.3 Communication between Vehicles ...................................................248
11.2.2.4 Intervehicle Communication ............................................................248
11.2.2.5 IoT Design in Intervehicle Communication ................................... 249
11.2.3 Risks ..................................................................................................................... 250
11.2.3.1 Personal Data Theft ............................................................................250
11.2.3.2 Hacking................................................................................................ 250
11.3 Recording Road Accident Data ...................................................................................... 250
11.3.1 Main Steps of Scene Investigation Work ......................................................... 250
11.3.2 Accident Location Identification....................................................................... 251
11.3.2.1 Using Global Positioning System Technology ...............................251
11.3.2.2 Using IoT to Gain More Accurate Data ........................................... 252
11.3.3 Taking Photos about the Accident Scene ........................................................ 253
11.3.3.1 Using IoT Services to Retouch Photos .............................................254
11.3.4 Recording Additional Accident Data............................................................... 255
11.3.5 Sending Traffic Message Channel Messages .................................................. 255
11.3.5.1 Sending Traffic Message Channel Using a Smart Device ............ 261
11.4 Analyzing Road Accident Data...................................................................................... 261
11.4.1 Advantages of the New IoT Devices ................................................................ 261
11.4.1.1 Using the Data from IoT Sensors ..................................................... 261
11.4.1.2 Sensor and Data Types ...................................................................... 262
11.4.2 Traditional Method to Find Accident Hotspots ............................................. 262
11.4.2.1 Sliding Window Method ................................................................... 262
11.4.2.2 Adaptation for Two-Dimensional Coordinates ............................. 264
243
11.4.3 Data-Mining Approach-Finding Accident Hotspots..................................... 265

11.4.3.1 Clustering Algorithms....................................................................... 265
11.4.3.2 Density-Based Clustering of Applications with
Noise Algorithm .................................................................................265
11.4.3.3 Details of the Density-Based Clustering of Applications
with Noise Method ............................................................................ 265
11.4.3.4 Calculating Accident Density ........................................................... 267
11.4.4 Analysis of Hotspot Candidates ....................................................................... 268
11.4.4.1 Using Information Gathered by IoT Devices..................................268
11.5 Conclusion......................................................................................................................... 269
References..................................................................................................................................... 269
11.1 Introduction
The main area of road safety engineering is the development of methods used to prevent
road users from being killed or seriously injured. The goal of the eSafe initiative (declared
by the European Union in November 2002) is to halve the number of public road accidents.
The target of a similar project (called the ITS Ten-Year-Plan and announced in January
2002) in the United States is attempting to decrease the number of public road accidents
by 15%.
Beyond several other factors (changes in legislation, construction or reconstruction of
road objects, etc.), an important part of these projects is to utilize the benefits that come
from new tools that have appeared in the last few decades (GPS, the Internet, wireless
communication, cheap sensors, etc.). The public road network has fairly large physical
dimensions, so the direct connection between the installed devices is impracticable. That
is the point at which the Internet of Things (IoT) paradigm can help by creating a loosely
coupled, decentralized network of nodes.
These nodes can be the following:
• Sensors embedded into the surface of the road, traffic signs, and so on
• Sensors built into vehicles
• Devices in vehicles to communicate with the driver
• Applications in the cloud to collect analyze and distribute data
Although most of the technical requirements are fulfilled, there are not any publicly avail-
able, well-functioning, massive interconnections between these devices that take advan-
tages of every opportunity. Even the required standards are missing to start building these
networks tomorrow. But there are several smaller projects and pilot applications already in
use. Further integration of these or a new system based on the experience from their use
may bring about the expected results.
This chapter contains three main sections regarding the fields where the IoT will become
essential in the near future:
• Accident prevention: IoT devices in vehicles and traffic control objects will help
decrease the number of accidents.
Highway Safety of Smart Cities 245
• Accident data recording: When an accident has already occurred, road accident scene
investigators have to gather and record all information about the crash, vehicles,
persons, environment, and so on; the IoT paradigm is widely used in this process.
• Accident analysis: A substantial portion of road safety engineers’ work is the deep
analysis of accidents; the information given by IoT devices makes this work more
comfortable and more accurate.
11.2 Accident Prevention

According to the IoT design, various nodes (wireless sensors, RFID sensors, satellite net-
works, and other intelligent technologies) can be integrated into a new heterogeneous
network to bring many benefits to highway users. Predicting dangerous situations and
decreasing the number of accidents is one of the major goals of these developments; how-
ever, there are several additional benefits, including reducing road congestions, managing
traffic, decreasing environmental pollution, and so on.
11.2.1 Available Services

In the near future, there should be a large network containing all road-safety-oriented
objects (sensors, road signs, etc.) and vehicles controlled by intelligent services in the cloud
(He et al. 2014). Although the technological conditions are given, only some independent
applications currently exist (Bennakhi and Safar 2016). But these show that it is worth
investing in these ideas; most of the existing systems are easy to use and—directly or
indirectly—help one to reduce the number of road accidents.
11.2.1.1 Intelligent Navigation Systems

It is obvious that using printed maps is obsolete; GPS-based navigation systems mostly
do a better job in route planning and turn-by-turn navigation. But we are already one
step ahead. It is possible to use online traffic data to avoid traffic jams and roadworks.
Navigation systems based on this philosophy are one of the best examples of the IoT
design. Millions of sensors (traffic counters, cameras, etc.) and the vehicles themselves
send a lot of data to the cloud.
These are just small information crumbs (the speed of a given vehicle, etc.) and use-
less alone, but collecting and deep analyzing these together gives a very clear picture of
the traffic situation. This knowledge makes it possible to create more appropriate route
and travel time estimations. As the communication is of two-way, in return for their
information, clients are informed about any unexpected events submitted by other users
(accident, roadworks, bad weather, etc.) that can help avoiding accidents.
11.2.1.2 Warning Messages

Unexpected situations are the most dangerous situations. Driving in bad weather condi-
tions (ice, snowy road surface, heavy rain, fog, etc.) is not an issue if the driver can prepare
for it. There are various sensors to detect any of those dangerous situations. Using the IoT
paradigm, these sensors can send a signal directly to the approaching vehicles, or more
likely to the cloud. The onboard system of the car can access all data concerning it and
warn the driver to be careful.
11.2.1.3 Voice Recognition

Driving any vehicle in traffic requires full attention from the driver. However, even the
most careful driver occasionally has to watch elsewhere (the gauges of the car, the enter-
tainment system, etc.). From a road safety perspective, it would be better to minimize these
moments; voice recognition can partially solve this problem.
Nowadays, this service is available only in the premium segment because of its high
resource demand. But these services are also available without high onboard computation
power using cloud services. Voice recognition is just an example; there are several comfort
or safety issues easily solved by appropriate cloud applications. IoT design helps using this
heterogeneous architecture.
11.2.1.4 Monitoring Vehicle Status

Modern cars are fully equipped with electronic sensors. The control unit of the onboard
system needs a lot of information from different components to efficiently coordinate their
operation. Using the IoT paradigm, it is possible to send some essential information back
to the car manufacturer. They have the necessary experience and computation power to
analyze these and send back a warning message if it is necessary to avoid further danger-
ous technical problems.
11.2.1.5 eCall
Although the main topic of this section is the prevention of accidents, the eCall system is
worth mentioning. It is activated after a crash, but it can help one to decrease the sever-
ity of an unavoidable accident. The eCall system is intended to bring rapid assistance
to any participants in an accident anywhere in the EU. The initiative aims to deploy a
small device in all vehicles that will automatically dial 112 in the event of a serious road
accident and send some information (GPS coordinates, etc.) to local emergency services.
The eCall system is not Internet-based, but in the case of an already-available continuous
online connection, the development of an IoT-based similar service is expected (Jankó and
Szénási 2013).
11.2.1.6 Self-Driving Cars

Robotics is one of the fastest growing businesses today (Takács, és mtsai 2016). At the time
of writing this chapter, the development of self-driving cars is one of the most popular
research areas (Atzori et al. 2010). Today, these projects are in an early stage; however,
some results are very promising. Google’s driverless cars have already driven millions of
kilometers.
At the same time, some manufacturers, such as Tesla Motors, have already enabled the
self-driving feature in their cars. The combination of IoT and self-driven cars may be the
pathway to a world without any road accidents (Fazekas and Gáspár 2015). But nowadays,
it is just a possibility in the distant future.
11.2.1.7 Additional Services

There are various additional services based on IoT that are loosely connected to the road
safety area (Kanoh 2013; Ashokkumar et al. 2015). Every application that decreases the
traveling time (and consequently the number of vehicles on the road) or the frustration of
the driver can indirectly help one to decrease collisions. A lot of research papers present
results about automatic parking systems, and so on.
11.2.2 Internet Access in Vehicle

As the name suggests, IoT cannot exist without continuous Internet access. It is obvious
that we can provide an online connection for static objects in urban or interurban areas,
but it is more difficult to do the same for a moving vehicle. There are several protocols
to solve this, usually based on the cellular phone network or on local Wi-Fi networks
(Chen 2012).
11.2.2.1 Using Cellular Network

The cellular network infrastructure allows Internet access for users from the 1990s. The
main advance of this method is that the coverage is especially good in industrial coun-
tries, and the access is relatively cheap. In the last decade, these networks have prepared
to serve broadband requests, making mobile Internet access usable for IoT purposes. The
speed of a common 3G network is enough for the necessary communication. Movement
of clients does not cause any problems; the protocols used are able to switch between
relay stations if necessary. Although there is a speed limitation, it does not affect the
everyday usage.
To connect a vehicle to the cellular network, one of the following options is used:
• Built-in connectivity: In this case, a cellular communication module is integrated

into the onboard system of the car. The Internet connection relies on this built-
in module. It has several advantages (there is no limit for the size of the radio
antenna, and this module can be an integral part of the onboard entertainment
system), but it is expensive. Therefore, this is used by premium brands only.
• Brought-in connectivity: In this case, the driver or any of the passengers share the
Internet connection of his or her smartphone with the car. MirrorLink technology
standardizes the protocol for this two-way connection. Persons in the car have
access to the entertainment system of the car by their smart devices, whereas the
car gains immediate access to the Internet.
11.2.2.2 Using Wi-Fi Networks

Using Wi-Fi networks may be a cheaper solution, and it is also available in some smart cit-
ies. Although the efficient range is significantly smaller in this case, but, in built-up areas,
the density of access points is quite high. Recent research shows that it is possible to access
the Internet inside a moving vehicle within the coverage of Wi-Fi hotspots.
This method is called drive-thru Internet. It has several limitations (the coverage of Wi-Fi
hotspots, the time required to make a connection to a Wi-Fi hotspot for high-speed vehi-
cles), but there are several ideas to solve these. For example, a vehicle connected to a Wi-Fi
hotspot can act as an access point to share that connection with the nearby clients.
11.2.2.3 Communication between Vehicles

Intelligent Transport Systems (ITS) try to solve the issues caused by the high traffic rate of
our century. Several large projects begun in Europe, Japan, and the United States since the
mid-1980s focus on the following areas (Tsugawa 2005):
• Traffic safety: To decrease the number of accidents (or at least decrease the number
of killed persons).
• Efficient transportation: To decrease the overall traveling time, and consequently
the environmental footprint, fuel cost, and others.
Although the goals to be achieved are very different, the requested tools are quite identical.
Both kinds of projects need a lot of sensors collecting data about vehicles, traffic, and the
environment; and one of the key points is the effective communication between the nodes.
It is usual to distinguish the following communication subtypes (Lu et al. 2014):
• Intervehicle communication: The main communication form between two partici-

pants in traffic (also often referred to as vehicle-to-vehicle communication); this can
be an interaction between drivers or at a lower level between onboard computers.
• Road-to-vehicle communication: The communication between some passive nodes
(like temperature sensors) and participants in traffic.
11.2.2.4 Intervehicle Communication

The requirement for intervehicle communication is not new. There were some early proj-
ects in the 1980s in Japan (such as the Association of Electronic Technology for Automobile
Traffic and Driving) to share information between cars. It became a heavily researched
area in Europe and the United States to make systems improving car-to-car communica-
tion to relay incidents or emergency information from a proceeding car or just to send
location data to nearby participants to avoid accidents caused by low visibility.
Several wireless communication protocols are known, but the intervehicle communica-
tion has some very specific requirements:
• The nodes of the network are not stationary. Individual cars are following different
routes; therefore, the network of nearby vehicles is constantly changing.
• The speed of the vehicles is an important factor. Fast-moving nodes have not got
enough time to execute slow interconnection procedures.
• Sensors are connected to the main battery of the car charged by the electric system;
therefore, there is no energy constraint (Figure 11.1).
There were several attempts to solve this issue using various wireless techniques, such as
Bluetooth (two of the constraints of this technology are the limited number of connecting
devices and the limited communication distance, etc.), radio-frequency identification (the
two major challenges are the large power loss at some locations and the unreliable data
transfer among simultaneous transmissions), ultra-wide band (UWB) radio technology,
and so on.
FIGURE 11.1
Example for intervehicle communication.
11.2.2.5 IoT Design in Intervehicle Communication

The rapid spread of wireless Internet opens a whole new path in this area. Instead of the
direct connection of vehicles, each car connects to some application in the cloud individu-
ally. Vehicles can send and receive all corresponding data to and from the cloud via the
appropriate services.
The only constraint is the necessary bandwidth on both sides. On the vehicle side, the
limiting factor is the speed of the wireless network; it may be not enough fast to enable
this two-way data stream. On the server side, the biggest challenge is the high number
of clients. In the case of thousands or millions of cars, it can be very difficult to store and
process all of the input data.
Using this IoT design has several advantages.
• The direct connection between devices becomes unnecessary, avoiding the previ-
ously mentioned issues. Each vehicle has to connect only to the cloud and keep
this connection alive.
• In the case of multiple neighboring vehicles, it is not necessary to send the same
message multiple times. It is enough to send it to the service, and the service will
distribute it to the affected clients.
• Clients have the ability to get messages from faraway locations. For example, mes-
sages from all nodes affecting the planned route can help avoid further accidents.
Some applications have been already developed and widely used on the basis of the previ-
ously presented architecture. For example, Waze is a very popular geographical naviga-
tion application developed by Waze Mobile (purchased by Google in 2013) that won the
Best Overall Mobile App award in 2013. It differs from other navigation application soft-
ware in that it is community driven. Drivers can report incidents (such as accidents, traffic
jams, and police traps), and other drivers receive these events.
Thanks to the IoT design, this is the biggest step forward since the appearance of online
maps on smart devices. This step turns the navigation application into an interactive tool
that is no longer based on a static map but instead is able to instantly adapt the planned
route to current real-world events and to warn the driver to be careful in the case of dan-
gerous places.
11.2.3 Risks
Every new technology has some disadvantages beyond the advantages. This is true for IoT
devices; it is worth considering the potential risks of using online devices. The two major
risks are personal data theft and the potential for the onboard systems to be hacked.
11.2.3.1 Personal Data Theft

Data from sensors providing weather-related information or general traffic information is
usually publically available for all clients. The accessibility of data from devices owned by
road operators, such as camera systems, is more confusing. Nowadays, there are several
publically accessible camera systems (webcams); however, it is possible to track vehicles
using these online video streams.
Data gathered and submitted by vehicles of private persons are more sensitive. No one
wants to let everyone know the details of his or her trips (source, destination, GPS location,
speed, etc.). The information itself is valuable for the community, but the default attitude is
to hide the identity of the source. Most available applications promise that they use anony-
mous data and do not store any unnecessary information about the drivers.
11.2.3.2 Hacking
The more complex a computer system is, the more vulnerable it is to attack. In the world of
the Internet, computer hacking is a permanent real threat. In the case of wireless networks,
this becomes extremely dangerous, because the client does not have any effective physical
method of prevention. Moving vehicles connected to the Internet with a two-way connec-
tion can be very risky in the near future. Hacking the system of a self-driving car can lead
to a tragic end.
11.3 Recording Road Accident Data

Although there are several new techniques to prevent highway accidents, it is a fact that
road accidents happened, happen, and will happen. It is how we should accept the world,
and we have the methodology to handle these accidents. It is important to determine the
reasons for each accident, because they can lead to some kind of potential solution (Yang
et al. 2013; Elmaghraby and Losavio 2014). Theoretically, it is not possible to eliminate all
further accidents, simply because of the exorbitant cost and general inconvenience. But it
is necessary to do our best.
11.3.1 Main Steps of Scene Investigation Work

It is hard to determine the cause of an accident, but it is usually one—or some combination—
of the following:
• Bad driving habits

• Bad weather conditions
• Vehicle error
• Engineering of roads
In accident analysis, we have to keep in mind all these factors. The most important is the
last one, because traffic engineers can easily and quickly change the environment (main-
tain/build/rebuild objects, etc.). On the contrary, it is very hard and time-consuming to
change general driving habits.
Nevertheless, it is very important to record as much data as possible about the crash.
It is important to understand that most of these data are available on the accident scene
only. There is no way to reproduce the circumstances of the accident later. Therefore, it is
very important for accident scene investigators to have the appropriate tools to efficiently
record all accident data (Wan et al. 2014). This requires instruments for the following steps:
• Accident location identification.

• Taking some photos of the environment.
• Recording all additional data about the accident.
• Submitting all information to the cloud.
Traditionally, accident scene investigators had to use a special form to record all accident
data. This led to several errors: it was very hard to determine the exact location of the
accident (they had to measure the distance from the nearest milestone), and it was usual to
skip some fields of the form or to write in contradictory data (e.g., the accident type was a
head-on collision, but only one participant was recorded).
The developments of the last decade significantly changed the methodology to be fol-
lowed. New hand-held devices allow the investigators to solve all the mentioned problems
using a single device with higher accuracy than has ever been available. The IOTs’ point of
view helps one to merge and utilize all the available tools.
11.3.2 Accident Location Identification

The location of the accident is the most crucial data for accident analysis. First, to analyze
the environment, we need a precise location to check the road signs, road surface quality,
and so on. In the case of an error of dozens of meters, the result of the analysis can be very
different and misleading. Second, the location is very important for the hotspot search
algorithms, and these are the first steps of most higher level accident analysis procedures.
11.3.2.1 Using Global Positioning System Technology

Global Positioning System (GPS) technology allows us to determine accident location in an
appropriate and quick way as never experienced before. Using traditional handheld GPS
devices, the coordinates can be read from the screen after only a few minutes. Although it
seems very unambiguous, this leads to serious issues in practice.
• There are several coordinate systems for coding a location on the surface of the
Earth. Police officers sometimes accidentally change the settings of the device
and use another format. For example, the following values are all coding the same
location: N43°38′19.39″, 43°38′19.39″N, 43 38 19.39, 43.63871944444445, and so on.
• The other problem is that numerical coordinates are meaningless for users (except
some well-trained experts). It is easy to find the difference between a right and a
mistyped street name (Neuannsreet versus Neumann street), but it is easy to errone-
ously copy the coordinates from the GPS device to the paper form (and later, from
the paper to the personal computer of the officer).
11.3.2.2 Using IoT to Gain More Accurate Data

A smart device with an Internet connection is the perfect solution for this problem. It is
possible to develop software that is able to show the actual position of the device based on
the built-in GPS receiver and other sensors (available Wi-Fi hotspots).
Using the Internet, the smart device is able to upload this location data to the server
without any manual work. This makes the entire process error-free, meaning that there is
no chance to make any typos in the coordinate pair.
However, it is worth keeping in mind that the precision of GPS technology is not always
satisfactory. Especially inside built-up areas, the deviation between the real location and
the supplied coordinates can be 10–30 m. This can lead to unacceptable results such as
coding the accident to the neighboring street.
That is the other point in which the online connection can help the investigators work.
It is possible to use online map services to show the accident location. It is possible to
display the following data (Figure 11.2):
• The GPS coordinate given by the smart device.

• The corresponding topographical identification (road number and road section or
street name and house number).
• A small map centered to the position given by the coordinates.
For a human, it is hard to directly check the validity of GPS coordinates; but thanks
to online maps, that is no longer necessary. Based on the given street name + house
(a) (b) (c)

FIGURE 11.2
Screen captures from the presented Android-based data recording application: (a) show current position; (b) enter
general information about the accident (outside or inside built-up area, etc.); and (c) give the traditional location
definition (road name and section).
FIGURE 11.3
Screen capture from the presented location checker application.
number pair and the displayed map, it is easy to decide whether the GPS coordinate is
correct or not.
After uploading the coordinates to the server, there are more advanced tools to ana-
lyze data location. For example, Figure 11.3 shows the position given by the GPS and the
position according to the given road number and section. This program can check the
uploaded data and instantly warn the investigator via the online connection to recheck the
given data if the distance between the two positions is too high.
11.3.3 Taking Photos about the Accident Scene

It is ordinary to take photos for documentation purposes in every field. In the accident
scenes, there are some special requirements:
• The photos must contain every detail of the scene environment (road signs, vehi-
cles, other objects, road surface, etc.).
• Depending on the legislation, these photos cannot contain any personal data.
There must not be the face of any person or any special identification objects (such
as registration plate of the participants’ cars).
• The environmental conditions are often very unsuitable for taking photos. In
other fields of photography, it is possible to choose the correct time for shooting;
but in the case of accident investigations, there is no time to wait. The investiga-
tor must be as fast as possible to minimize the traffic jam caused by the crash and
roadworks.
Using analog cameras is outdated, and nowadays it is common to use digital ones. These
are very easy to use, flexible, and give the opportunity to check the photos taken on the
scene (and shoot more if necessary) (Rövid et al. 2016).
Advanced smartphones are able to take pictures in high resolution. Maybe, these are not
applicable for advanced artwork, but the result is totally satisfactory for these purposes.
The software given to scene investigators has a function to take a photo and upload it to
the cloud instantly.
This is very convenient compared with the traditional method (taking a photo, sending
it to the laboratory, cataloging the final pictures, etc.). The uploaded photos are instantly
attached to the appropriate accident, have location geocodes, and if the smartphone makes
it available (with a built-in magnetic sensor), also contains the shooting direction data.
11.3.3.1 Using IoT Services to Retouch Photos

The role of the cloud is much more than just storing uploaded photos. It opens a connection
between the scene investigator and his or her colleagues in the headquarters. Someone can
check and evaluate all uploaded photos and ask for corrections if they are necessary.
It is also important to note that speed is a very important factor (Kertész and Vámossy
2015). The accident scene must be kept untouched until the investigation ends; therefore,
the investigator has to take all the photos as fast as possible. On the other side, it is obvious
that there is not any other chance to take these pictures.
All necessary facts must be recorded and checked before the investigator leaves the
scene. The mentioned automatic two-way communication enables this (Figure 11.4).
Some other necessary operations must be taken with the photos. It is also true for normal
photos and aerial images (Kiss 2014; Molnár et al. 2016). It is necessary to blur all parts of
the image that make it possible to identify persons or vehicles. This is very tedious and
tiresome work for humans, but it is easily automated (e.g., it is well known that Google
Street View has similar algorithms to blur some parts of the taken images). This process is
very resource intensive (Windisch and Kozlovszky 2015); therefore, it is hard to execute on
smart devices. The IoT point of view can help in this problem. It is possible to upload the
images to the cloud and use some service to perform the necessary modifications.
This methodology is also usable in the case of taking photos under unsuitable circum-
stances. Contrary to general photography, it is common to shoot pictures at night or in
FIGURE 11.4
Retouched accident photo.
wrong environmental conditions (because these are usual accident reasons). It is also possi-
ble to use some cloud services to improve the image. It is a big advance that this is available
on the scene because in the case of failure, the investigator can take some additional photos.
11.3.4 Recording Additional Accident Data

Beyond the location and photos, there are several other pieces of information to collect at
the accident scene. One of them in particular is especially important: the estimated time
of the crash. But there is a lot of additional data to find from the following data sources:
• The accident itself

• The vehicles
• The persons
• The road and environment
Table 11.1 contains the main attributes of a typical road accident (Table 11.1).
Using the IoT, all of these data can be instantly uploaded from the accident scene. On the
server side, there are several systems and intelligent algorithms to check and evaluate the
uploaded data. These can send back a warning message if there is any missing or errone-
ous information. This instant response is very useful, because the investigator can check
all warnings at the scene and modify the uploaded data (Figures 11.5 through 11.8).
11.3.5 Sending Traffic Message Channel Messages

Traffic Message Channel (TMC) is a technology for distributing information about
unexpected events in the public road network. The main topology is many to many:
messages from multiple sources received by the Traffic Information Center (TIC) and
forwarded to all receivers. Obviously, the permissions of the users are strictly regu-
lated. Only some privileged sources can send messages (road accident scene investiga-
tors are typically in this group), but the broadcasted messages are usually available for
everyone.
TABLE 11.1
Main Attributes of Road Accidents
Level Attribute Example
Accident Time 2016.06.12. 22:20

Location LAT 47°48′22″, LON 18°45′38″
Nature of accident Collision between cars and bicycle
Vehicles Category Motor-car
Brand Opel
Persons Age 23
Injury Light
Vehicle ID 1
Road and environment Road surface Wet
Weather Rainy
Visibility Below normal
FIGURE 11.5
Select accidents using given filtering rules.
(a)
FIGURE 11.6
Check the details of a given road accident: (a) accident data. (Continued)
(b)
(c)
FIGURE 11.6 (Continued)

Check the details of a given road accident: (b) vehicle and personal data, (c) road and traffic data.
(d)
(e)
FIGURE 11.6 (Continued)

Check the details of a given road accident: (d) map preview, and (e) notes/photos and maintenance tab.
FIGURE 11.7
Visualize the accidents of a given road interval.
FIGURE 11.8
General statistics of a given road interval.
Traffic incidents are binary-coded in TMC messages. The main data fields of a message
are the following:
• Event code—one of the codes from the predetermined code table

• Location code—in this case, the location of the accident crash site
• Expected duration—how much time is needed to clean up the area
• Other additional data—not used in this case
It was a good idea to assign codes for events and not to send plain text. Using this meth-
odology, it is easy to translate the message code to the language of any user. Moreover, the
database of events is directly available for statistical queries.
The event itself takes 11 bits from the message. This limits the number of available events
to 211, but that is absolutely enough to send the most frequent occurrences. These mes-
sage codes cover a variety of events. Some of the codes describe an unexpected and short
duration event (like car accidents), whereas some of them describe the opposite (e.g., road
construction work; Table 11.2).
TABLE 11.2
Road Accident Related TMC Codes
TMC Code Description
857 Unprotected accident area(s)

858 Danger of unprotected accident area(s)
12 Accident(s), traffic being directed around accident area
141 All accidents cleared, no problems to report
201 Accident(s)
202 Serious accident(s)
203 Multivehicle accident (involving Q vehicles)
204 Accident involving (a/Q) heavy lorr(y/ies)
205 Accident(s) involving hazardous materials
206 Fuel spillage accident(s)
207 Chemical spillage accident(s)
208 Vehicles slowing to look at accident(s)
209 Accident(s) in the opposing lanes
215 Accident(s). Stationary traffic
222 Accident(s). Queuing traffic
228 Accident(s). Danger of queuing traffic
229 Accident(s). Slow traffic
240 Road closed due to accident(s)
241 Accident(s). Right lane blocked
242 Accident(s). Center lane blocked
243 Accident(s). Left lane blocked
244 Accident(s). Hard shoulder blocked
245 Accident(s). Two lanes blocked
246 Accident(s). Three lanes blocked
247 Accident. Delays (Q)
248 Accident. Delays expected
249 Accident. Long delays (Q)
250 Vehicles slowing to look at accident(s). Stationary traffic
11.3.5.1 Sending Traffic Message Channel Using a Smart Device

At the accident scene (after some more important safety procedures), the investigator should
send a TMC message to the TIC. Using the Internet, this can be done using his smart device.
It depends on the officer when to send this message. If it is important, it is possible to send the
message out before any other operation. Another way is to input some of the previously men-
tioned accident data (location of the accident, nature of the accident, etc.), and based on this
information, the software on the smart device can determine the appropriate message code.
It is very auspicious that the scene investigator can directly handle these TMC messages
from the scene. First, he or she is an expert; therefore, he or she can send a reliable and
trustworthy opinion about the actual situation. The TMC message contains a field about
the expected duration of the event. It is just an approximation; the scene investigator has to
guess how much time he or she needs to collect all necessary information and perform the
requested cleaning operations.
Of course, it is possible to manually stop the event. The investigator has to remain on the
crash site until all the previous operations have been completed. At the end, he or she is able
to send a closing TMC message, indicating to all vehicles that the road is opened for traffic.
11.4 Analyzing Road Accident Data

11.4.1 Advantages of the New IoT Devices
As previously mentioned, the reason or reasons for an accident usually belong to one or
more of the following categories (Paridel et al. 2011):
• Bad driving habits

• Vehicle errors
• Engineering of roads
The main task of road safety experts is to find the critical areas of the public road networks,
analyze these, propose some possible improvements, and estimate the expected positive
effects.
11.4.1.1 Using the Data from IoT Sensors

Traditionally, experts have to use one database containing accident data (Gellerman et al.
2016). This database is controlled by the scene investigators; they were the source of all
available accident information.
In addition to the resource-intensive human work, this leads to some inaccuracy. A lot of
attributes are estimated by the scene investigators several hours after the accident.
Nowadays, it is possible to install several kinds of sensors recording various data (Perera
et al. 2014). These are usually cheap devices, and it is possible to use a lot of them in vehicles,
traffic objects, and so on. However, this new possibility raises some new questions. First,
how can we store this enormous amount of data? And second, how can we process this data?
Fortunately, data storage devices have become more and more cheap, and it is very easy to
record all available data in a convenient way using cloud technology. Processing this amount of
data is a more complex problem, but there are several data-mining algorithms for this purpose.
11.4.1.2 Sensor and Data Types

Here are some examples of how embedded devices can assist in data gathering.
• Traffic measurement sensors: There are several ways to measure the traffic at a given
section of the highway. There are camera systems with this additional feature, and
there are special target hardware developments for this purpose (sensors built into
the road surface using induction to count the number of vehicles). These embed-
ded devices can send all recorded data to the cloud.
• Speed measuring sensors: These sensors are able to measure the speed of a given
vehicle. It is possible to upload all recorded data to a central warehouse.
• Analyzing weather conditions: The weather has a very big impact on road accidents.
However, it is very rare that the weather is the direct cause of an accident; but
inclement weather conditions can change a semiaccident situation into a real
crash. It is possible to measure the following using various sensors:
• Temperature
• Light conditions
• Atmospheric pressure and so on
• Camera systems: The biggest help for analyzing the reasons is a short
video about the accident. Nowadays, it is possible to install camera systems in
dangerous places, and thanks to advanced compression algorithms and fast
wireless Internet connections, it is possible to upload all recorded data to the
cloud.
• Sensors inside vehicles: It is also possible to gather several kinds of information from
the built-in vehicle sensors. For example, the speed of the vehicle can be inferred
by the built-in GPS receiver or the smartphone of the driver (Zarronbashar and
Mahmud 2010; Isen et al. 2013; White et al. 2011).
It is possible using multiple data sources to determine the appropriate value of a given
parameter. In the case of accidents, the most important information is the location.
Using multiple sources (a smart device used by the accident scene investigator, a GPS
device built into the vehicle, camera images, etc.) makes it possible to have a more accu-
rate result.
11.4.2 Traditional Method to Find Accident Hotspots

Accident hotspots are the area of the public road network in which the density of acci-
dents is higher than expected. As previously mentioned, the new devices make it avail-
able to assign GPS coordinates to accident locations (Reshma and Sharif 2012). This has
several advancements (see Section 11.4.1) but also raises some interesting questions
(Figure 11.9).
11.4.2.1 Sliding Window Method

Traditional hotspot searching algorithms (Sorensen and Pedersen 2007; Montella 2010;
Costa et al. 2015) use road number + road section based geocoding. It allows the usage
FIGURE 11.9
Hotspot in built-up areas.
FIGURE 11.10
Traditional one-dimensional sliding-window method.
of some basic methods such as the well-known sliding window technique (Lee and
Lee 2013; Szénási and Csiba 2014). This method contains the following consecutive
steps:
• Select all accidents on the examined road (this method is not applicable for analyz-
ing accidents occurring on multiple roads).
• Sort these accidents by the road section number.
• Slide a window with a given size (e.g., 300 m) over the road.
• Count the number of accidents inside the window according to its actual position.
• If the number of accidents is higher than a given threshold value, the position can
be considered as a hotspot candidate (Figure 11.10).
This method has several advantages:
• It is easy to use.
• There are several already-existing applications.
• The results are easily readable for humans.
But it has some disadvantages in the case of built-up areas:
• Using the given technique, it is important to know the distance between two
accidents. Outside urban areas, when we use road number + road section pairs,
it is easy to calculate. The distance between two accidents equals the difference
between the two road section values.
• But in the case of built-up areas, the street name and house number pair is the
common identification method. There is not any simple way to calculate the dis-
tance between two accidents based on these.
• Public roads constitute a network of road sections and junctions. These differ in
their geometric characteristics and traffic technological design. Within built-up
areas, a significant amount of accidents can be linked to junctions. In such cases,
based on the decision of the scene investigator, it differs to which street the acci-
dent will be allocated.
• In the case of larger junctions, it is possible to have up to four or five different
street names appearing at the same junction (and sometimes the junction also has
a square name). Thereby, these accidents in the database appear under different
street names, and this makes it harder to find the hotspots.
11.4.2.2 Adaptation for Two-Dimensional Coordinates

There are several papers about the sliding window technique and its adaptations (fixed
window size, flexible window size, using several special weighting factors, etc.). It was
widely used in the past, but the appearance of GPS coordinates mean that it will soon be
out-dated. All geocoded information from the embedded sensors uses GPS coordinates.
The appropriate format (WGS’84, EOV, etc.) does not really matter because it is easy to
convert from one to the other.
But all of these use a significantly different approach compared with the road number +
road section identification. These are all some kind of two-dimensional spatial coordi-
nates, and it is not possible to use the sliding window method in this case (it is possible to
use a two-dimensional window instead of the one-dimensional one, but it is not widely
popular; Figure 11.11).
FIGURE 11.11
Two-dimensional sliding-window method.
11.4.3 Data-Mining Approach-Finding Accident Hotspots

It is necessary to develop new hotspot searching algorithms based on the newly available
data sources. GPS coordinates give unequivocal and accurate locations, but it is necessary
to use novel spatial methods instead of the old-fashioned sliding window.
11.4.3.1 Clustering Algorithms

One of the most promising methods is from the field of data-mining. Data-mining is a
special database application used to discover hidden correlations in data. As a multidis-
ciplinary field, it combines fields such as database management, statistics, and artificial
intelligence. The main advantage of this methodology is that it tries to automate certain
processes and expert activities helping in processing a large amount of data, which requires
disproportionately large efforts by experts. The main difference between traditional statis-
tical operations and data-mining is that in the latter case, it is possible to find correlations
between the entities inside an existing database, which are not already known.
A special field of data-mining concerns clustering algorithms. Clustering means that we
have to find groups (also known as clusters) of items as follows: All items inside a cluster
are similar to each other (according to a prespecified criterion), whereas objects outside the
cluster (or in different clusters) are quite different from each other (according to the same
criterion).
In the case of accident analysis, the requested output is the groups of accidents, which
are similar to each other (according to location or another accident attribute). As in the case
of similar accidents, we can assume that the reason was the same, it is possible to eliminate
this factor in order to significantly decrease the number of future accidents.
11.4.3.2 Density-Based Clustering of Applications with Noise Algorithm

There are several well-known clustering algorithms. One of these, the Density-Based Spatial
Clustering of Applications with Noise (DBSCAN), seems ideal for accident hotspot search-
ing (Szénási and Jankó 2016). The purpose of this algorithm is to find areas in the multidi-
mensional search space where the density of elements is higher than a given threshold. One
of the advantages of this method is that it is able to work effectively in a noisy environment.
In the case of accidents, the search space is the examined country/city/etc. The elements
inside this search space are the occurred accidents. The position of a given element in the
search space accords to the location of the corresponding accident. Hotspots are areas of
the search space where the density of accidents is higher than a given threshold. Every
accident not belonging to any hotspot is considered noise.
Another advancement of the DBSCAN algorithm is the capability to find clusters of any
shape, in contrast to the two-dimensional sliding window method, which always uses a
rectangular-shaped window. In the case of junctions, irregularly shaped areas can lead to
significantly higher accident density.
11.4.3.3 Details of the Density-Based Clustering of Applications with Noise Method

The main input parameters of the DBSCAN algorithm are the following:
• ε: A radius type variable (meters)

• MinPTS: Minimal number of accidents within a cluster (number of accidents)
To understand the mechanism of the DBSCAN algorithm, it is necessary to introduce

some definitions.
• ε-environment: Space radius of ε of an element (it is necessary to define a distance

concept).
• Internal element: There are at least MinPTS number of elements in the
ε-environment of this element.
• Directly dense accessibility: One element is directly dense accessible from another if
it is in the ε-environment of it.
• Dense reachability: One element is accessible from another through a chain of
directly densely accessible ones.
• Dense connection: Two elements are densely connected if there is any element from
which both of them are densely reachable.
• Density-based cluster: A domain of densely connected elements that shows maxi-
mum accessibility of density.
Based on these definitions, the goal of the DBSCAN algorithm is to find domains of acci-
dents in which all elements are densely connected, and no further expansion is possible
(the number and the size of clusters are maximal).
It is also necessary to define the concept of distance. In the case of accidents identified by
GPS coordinates, this can be considered to be simply the Euclidean distance between the
two locations. However, it is possible to use more sophisticated methods.
The main steps of the DBSCAN algorithm are the followings (Figure 11.12):
1. Select a starting point (any point of the search space).

2. If there are not at least MinPTS accidents in the ε-environment of this point, it is
possible to skip the remaining steps and select another starting point.
3. If there are at least MinPTS accidents, it is worth starting the further cluster analy-
sis. This is an internal point so that we can create a new cluster-candidate contain-
ing this and the other elements in its ε-environment.
4. The ε-radius of these new elements is also investigated, and if any of the existing ele-
ments meet the necessary growing conditions, they are added to the cluster candidate.
FIGURE 11.12
DBSCAN method to find accident hotspots.
5. The search continues using these new points with Step 4.

6. If it is unable to select a new point to continue the search, the recursion finishes.
Based on the goal of the algorithm, there are several variations of the previous algorithm:
• In the simplest variation, the presented algorithm will find clusters, in which the
number of accidents is higher than a given threshold.
• It is also possible to use some weighting factors. The outcome of the accident is
known, and it usually belongs to one of the following classes: fatal accident, acci-
dent with serious injury, accident with light injury, accident without personal
injury. For example, here is a possible weighting factor configuration:
• Fatal accident: 10
• Accident with serious injury: 3
• Accident with light injury: 1
• Only property damage: 0.5
11.4.3.4 Calculating Accident Density

The presented variations are accounting for only the number of accidents. But in the case of
hotspot identification, we are looking for areas with high accident density. Density means
the number of accidents divided by the bounding area. If the accidents are located on mul-
tiple roads, it is necessary to use a spatial area definition based on the shape spanned by
the accidents.
To find the convex shape with the minimal area containing all accidents (all of them
have to be inside the shape or in the contour/edge of the shape). If this polygon is
known, it is possible to calculate its area using the Shoelace formula (also known as the
Gauss area formula) that requires only the appropriate sequence of the two-dimensional
edges. The outer points of the polygon must be in a given order (e.g., in a clockwise
direction).
The production of this sequence is very resource-intensive; therefore, the DBSCAN algo-
rithm maintains this automatically when a new accident is inserted into the cluster.
• The first two accidents are automatically inserted into the sequence.
• The third accident is also inserted into the sequence, but if necessary the algo-
rithm reorders the points to ensure their clockwise order.
• If the fourth or any further element is outside the already-existing polygon, then
it is necessary to expand it and to insert the accident into the contour sequence.
• If the fourth or any further element is inside the already-existing polygon, there is
nothing to do. This will change the accident density, but not the area of the mini-
mal convex bounding polygon (Figure 11.13).
At some point, the DBSCAN algorithm will not be able to find any new accidents to expand
the actual cluster. It calculates the area of the minimal convex bounding polygon and the
accident density of the cluster. If this density is higher than a previously given threshold,
then this is an interesting area for further processing. If not, then it restarts the DBSCAN
algorithm from another starting point.
New point
FIGURE 11.13
Area calculation method.
11.4.4 Analysis of Hotspot Candidates

The result of the data-mining algorithm is a set of clusters. We usually call these hotspot
candidates. The word candidate makes it clear that these are only accidents grouped
together using some kind of mathematical method. It is possible that these are totally inde-
pendent accidents without any common attributes; in this case, this is not a real hotspot,
but just an unfortunate distribution of accidents. Because of this, all hotspot candidates
require deeper analysis.
11.4.4.1 Using Information Gathered by IoT Devices

That is the point in which all data recorded by the IoT devices can help in understanding
the situation:
• Traffic measuring sensors: The above-described method gives only a raw analysis of
accidental data. The result is the areas of the road network in which the density
of accidents is high. The expected number of accidents in road sections with high
traffic is higher than in road sections with low traffic.
Therefore, it is worth taking into consideration the traffic information of the
scene (in the most simplified case, as a divisor).
• Temperature sensors: It is possible to check the recorded temperature data. The
average temperature on a daily/hourly basis is already known. Based on this, the
expected distribution of temperatures at the time of accidents is also available.
If it significantly differs from the actual distribution, it is worth finding out the
reasons. For example,
• If most of the accidents occurred in low-temperature environments, it is worth
checking the surface of the road in cold situations.
• If most of the accidents occurred in high-temperature environments, it is worth
checking the surface of the road in hot situations.
• Weather sensors: These sensors can give us additional information about the
weather conditions at the time of the crash. For example, the atmospheric pres-
sure, rain, snow, sunset, and so on. It is also worth checking if there is any uncom-
mon pattern in these data series. For example,
• If most of the accidents occurred in rainy situation, it is worth checking the
road surface. Maybe, it becomes unusually slippery when it is wet.
• If most of the accidents occurred in sunny weather, it is worth checking the
routing of the road. Maybe, the sun shines directly into drivers’ eyes.
• If most of the accidents occurred on snowy days, it is worth considering limit-

ing the speed or performing other preventive actions.
• Speedometers: The speed of the participant vehicles can be very important in the
analysis phase. Speeding is one of the most frequent reasons for an accident. On
the one hand, it is one of the things which is really hard to influence (although
there are some methods, such as speed limit road signs, fixed installation of
speedometers, etc.). But on the other hand, it is worth knowing that the origin
of the problem is bad driving habits and not road engineering issues.
• Camera images: If there are videos of the crash events, these can help one to under-
stand the reasons for the accident. It is worth watching these videos one by one. If
there are any common patterns among the occurred accidents, it is a real hotspot,
and it is necessary to take preventive actions.
11.5 Conclusion
As can be seen, IoT can help in several points of the data analysis process. It makes it avail-
able using accurate location information about accidents, which leads to usable clustering
methods. It also can help in the next phase of the analysis project, the deep investigation,
in which the main goal is to uncover the common patterns among accidents.
The IoTs paradigm provides two additional advancements for experts:
• The data collection process becomes comfortable and automated. For example, in
the past, it was possible to count the number of cars by using inductive devices
built into the road surface; but these devices have only a very limited storage, and
someone has to manually read out and clear the recorded data weekly.
• Using IoT, it is not necessary to use local storage; devices can upload all data
instantly to the cloud. It is able to read all recorded information only seconds later.
It is more important for traffic control applications, but it also helps the work of
accident analysts.
• Using the new devices also makes the gathered information even more precise.
Embedded systems allow for the determination of vehicle speed and direction in a
very accurate way. The distribution of camera systems gives essential information
about traffic situations.
References
Ashokkumar, K., B. Sam, R. Arshadprabhu, and Britto. 2015. Cloud based intelligent transport sys-
tem. 2nd International Symposium on Big Data and Cloud Computing, pp. 58–63.
Atzori, L., A. Iera, and G. Morabito. 2010. The Internet of things: A survey. Computer Networks 54:
2787–2805.
Bennakhi, A., and M. Safar. 2016. Ambient technology in vehicles: The benefits and risks. The 6th
International Symposium on Frontiers in Ambient and Mobile Systems, pp. 1056–1063.
Chen, Y.-K. 2012. Challenges and opportunities of Internet of things. 17th Asia and South Pacific
Design Automation Conference. IEEE, Sydney, pp. 383–388.
Costa, S. D., X. Qu, and P. M. Parajuli. 2015. A crash severity-based black spot identification model.
Journal of Transportation Safety and Security 7 (3): 268–277.
Elmaghraby, A. S., and M. M. Losavio. 2014. Cyber security challenges in smart cities: Safety, secu-
rity and privacy. Journal of Advanced Research 5: 491–497.
Fazekas, Z., and P. Gáspár. 2015. Computerized recognition of traffic signs. Acta Polytechnica Hungarica
12 (5): 35–50.
Gellerman, H., E. Svanberg, and Y. Barnard. 2016. Data sharing of transport research data.
Transportation Research Procedia 14: 2227–2236.
He, W., G. Yan, and L. D. Xu. 2014. Developing Vehicular Data Cloud Services in the IoT Environment.
IEEE Transactions on Industrial Informatics 10 (2): 1587–1595.
Isen, L., A. Shibu, and M. S. Saran. 2013. Evaluation and treatment of accident black spots using
geographic information. International Journal of Innovative Research in Science, Engineering and
Technology 2 (8): 3865–3873.
Jankó, D., and S. Szénási. 2013. Expected impacts of the implementation of ITS on congested
periods regarding the Hungarian national road network. Moving Toward an Integrated Europe
Pan-European eCALL Service in the HeERO Countries. Budapest, pp. 72–77.
Kanoh, H. 2013. Development of MMRS (Mind Map and Relief System), an information sharing
system for children’s safety. Procedia Computer Science (22): 762–771.
Kertész, G., and Z. Vámossy. 2015. Current challenges in multi-view computer vision. 10th Jubilee
IEEE International Symposium on Applied Computational Intelligence and Informatics. Timisora,
Romania.
Kiss, D. 2014. Eigenvector based segmentation methods of high resolution aerial images for pre-
cision agriculture. 5th ICEEE-2014 International Conference: Global Environmental Change and
Population Health: Progress and Challenges. Budapest, Hungary, pp. 155–162.
Lee, S., and Y. Lee. 2013. Calculation method for sliding-window length: A traffic accident frequency
case study. Easter Asia Society for Transportation Studies 9: 1–13.
Lu, N., N. Cheng, N. Zhang, X. Shen, and J. W. Mark. 2014. Connected vehicles: Solutions and chal-
lenges. IEEE Internet of Things Journal 1 (4): 289–299.
Molnár, A., D. Stojcsics, and I. Lovas. 2016. Precision agricultural and game damage analysis applica-
tion for unmanned aerial vehicles. International Journal of Applied Mathematics and Informatics
10: 38–43.
Montella, A. 2010. A comparative analysis of hotspot identification methods. Accident Analysis &
Prevention 42 (2): 571–581.
Paridel, K. et al. 2011 Teamwork on the Road: Efficient Collaboration in VANETs with Context-
based Grouping, Procedia Computer Science (5): 48–57.
Perera, C., A. Zaslavsky, P. Christen, and D. Georgakopoulos. 2014. Sensing as a service model
for smart cities supported by Internet of things. Transactions on Emerging Telecommunications
Technologies 25 (2): 81–93.
Reshma, E. K., and S. U. Sharif. 2012. Prioritization of accident black spots using GIS. International
Journal of Emerging Technology and Advanced Engineering 2 (2): 117–122.
Rövid, A., Z. Vámossy, and S. Z. Sergyán. 2016. Thermal image processing approaches for secu-
rity monitoring applications. In S. László Nádai, S. József Padányi (Eds.), Critical Infrastructure
Protection Research: Results of the First Critical Infrastructure Protection Research Project in Hungary,
pp. 163–175. Switzerland: Springer International.
Sorensen, M., and S. K. Pedersen. 2007. Injury severity based black spot identification. Annual
Transport Conference. Aalbor, pp. 1–10.
Szénási, S., and P. Csiba. 2014. Clustering algorithm in order to find accident black spots identified By
GPS coordinates. 14th GeoConference on Informatics, Geoinformatics and Remote Sensing. Albena,
pp. 497–503.
Szénási, S., and D. Jankó. 2006. Black spot treatment system using a “hunting for irregular pattern”
process and a safety knowledge-base. On safe roads in the XXI Century. Budapest, Hungary,
pp. 100–105.
Szénási, S., and D. Jankó. 2016. A method to identify black spot candidates in built-up areas. Journal
of Transportation Safety & Security 9 (1): 1–25.
Takács, Á., D. Á. Nagy, I. Rudas, and T. Haidegger. 2016. Origins of surgical robotics: From space to
the operating room. Acta Polytechnica Hungarica 13 (1): 13–30.
Tsugawa, S. 2005. Issues and recent trends in vehicle safety communication systems. IATSS Research
29 (1): 7–15.
Wan, J., D. Zhang, S. Zhao, L. T. Yang, and J. Lloret. 2014. Context-aware vehicular cyber-physical
systems with cloud support: Architecture, challenges, and solutions. IEEE Communications
Magazine 52 (8): 106–113.
White, J., C. Thompson, H. Turner, B. Dougherty, D. C. Schmidt, and Wreckwatch. 2011. Automatic
traffic accident detection and notification with smartphones. Mobile Networks and Applications
16 (3): 285–303.
Windisch, G., and M. Kozlovszky. 2015. Parallel image sharpness measure for supercomputing envi-
ronment. 13th International Symposium on Applied Machine Intelligence and Informatics. Slovakia,
Herlany: IEEE, pp. 283–288.
Yang, L., S. H. Yang, and L. Plotnick. 2013. How the Internet of things technology enhances emer-
gency response operations. Technological Forecasting and Social Change 80 (9): 1854–1867.
Zarronbashar, E., and A. R. Mahmud. 2010. Intelligent GIS-based road accident analysis and real-
time monitoring system. International Journal of Engineering 2 (1): 1–7.
12
Connected Bicycles: Potential Research
Opportunities in Wireless Sensor Network
Sadik Kamel Gharghan, Rosdiadee Nordin, Nor Fadzilah Abdullah,

and Kelechi Anabi
CONTENTS
12.1 Introduction ...................................................................................................................... 273
12.2 Problem Background ....................................................................................................... 275
12.3 Toward Battery-Free Internet of Bicycle ....................................................................... 277
12.3.1 Sleep/Wake Strategy ..........................................................................................277
12.3.2 Data Mitigation ...................................................................................................278
12.3.3 Radio Optimization ............................................................................................279
12.3.4 Routing Protocols ...............................................................................................279
12.3.5 Energy Harvesting ............................................................................................. 279
12.4 Why Connected Bicycles? ............................................................................................... 280
12.4.1 Smart City Driven ...............................................................................................280
12.4.2 Intelligent Transport System Concept .............................................................281
12.4.3 Economic Value Chain .......................................................................................281
12.4.4 Cyclist Safety Driven ..........................................................................................282
12.4.5 Technology Driven ............................................................................................. 282
12.5 Potential Opportunities on the Internet of Bicycles .................................................... 283
12.5.1 Long Range, Low-Rate, Low-Power, Low-Cost Internet of Bicycle
Wireless Backhaul .............................................................................................. 283
12.5.2 High-Rate 5G Backhaul .....................................................................................284
12.5.3 Energy-Efficient Transmission Algorithm.......................................................284
12.5.4 Energy-Harvesting Radios ................................................................................285
12.5.5 Big Data Analytics ..............................................................................................285
12.5.6 Network Security Challenge............................................................................. 286
12.6 Conclusion......................................................................................................................... 286
References..................................................................................................................................... 287
12.1 Introduction
Wireless sensor networks (WSNs) can be used to monitor sports activities as suggested
in several recent research works. In the area of high performance cycling, cyclist per-
formance can be considered as one of the most important application. BWSNs (bicycle
273
WSNs) have been identified as a technology candidate that is suitable for the mobility
model, energy model, and real-time monitoring of a cyclist. A few key WSN technolo-
gies that have been utilized are Bluetooth, ZigBee, Wi-Fi, and Advanced and Adaptive
Network Technology (ANT). For example, in the work of Balbinot et al. (2014) and Casas
et al. (2016), the Bluetooth wireless protocol was used to monitor the three-dimensional
forces applied on the bicycle crankset. The monitoring system was achieved based on
a data logger. Similar research work adopted ZigBee wireless protocol to monitor the
location of a cyclist on a cycle track (Shin et al., 2013), bicycle biomechanical parameters
(Olieman et al., 2012), bicycle location estimation and radio reachability (Hayashi et al.,
2012), and knee and ankle angles monitoring (Marin-Perianu et al., 2013). The ANT pro-
prietary wireless protocol was used by Baca et al. (2010) to monitor the biomechanical
and physiological parameters of the bicycle and cyclist. By utilizing the infrastructure of
the mobile and Internet networks, the cyclist parameters can be transmitted to a remote
location via a framework system that consists of wireless protocol and mobile phone
device. Although Wi-Fi technology consumed more power and larger in size compared
to Bluetooth, ZigBee, and ANT, it was also used to monitor the cyclist performance in
some research work (Yu et al., 2009).
In terms of available commercial products, the Schoberer Rad Messtechnik device can
be considered as the most widely used to monitor the professional cyclist parameters
such as power, cadence, speed, heart rate, and temperature (Bini et al., 2011).
Previous research works and commercial products mostly focus on methods of
measuring cycling performance and transferring these parameters from the bicycle
sensor nodes to the monitoring device that is fixed on the bicycle handlebar as shown
in Figure 12.1. With the advancement of sensors technology, wireless communication
technologies and cloud computing, the BWSN is expected to join the Internet of Things
(IoT) hype.
Heart rate
sensor node
Speed/cadence
sensor node Head unit
(monitoring
device)
Torque sensor node
FIGURE 12.1
The general configuration of the BWSN with related sensor nodes.
Connected Bicycles: Potential Research Opportunities in Wireless Sensor Network 275
12.2 Problem Background

High performance cycling requires a high fitness level with sufficient training program
and teamwork. Cycling is a highly competitive sport, where the margin between victory
and defeat is in the order of milliseconds. The biomechanical and physiological param-
eters of the bicycle and cyclist can be used to develop and refine the cyclists and team tactic
and training programs to become more competitive in international competitions. These
parameters can be measured and wirelessly transmitted to the coach or remote location
that can be used to analyze the performance and progress of the cyclist. In addition, the
measured parameters can be transmitted to the monitoring device that mounted on the
handlebar of the bicycle to display on the liquid crystal display to be seen in real time by
the cyclist.
BWSNs are unlike other types of WSNs. Several limitations can be highlighted as
below:
1. Limited energy resources: All sensor nodes are portable on the bicycle and use the
battery as the main power source. Recharging or replacing batteries when the
bicycle in competition case is impossible, especially in the long distance event.
Accordingly, the ANT and ZigBee wireless protocols that consume less energy
during operation can be used. Besides that, energy-efficient techniques are an
essential demand.
2. Limitation on size, weight, and memory size of sensor nodes: The sensor node needs to
be small as possible to reduce the weight and size. In addition, the mathematical
relationship between the measured parameters of the bicycle needs to be merged
into one sensor node. A task-division technique through a coordinator node can
also be used to avoid exhaustive calculation in sensor nodes.
3. Inaccurate distance estimation: Radio communication is extremely sensitive to the
location of the bicycle on the cycle track or road due to the changing distance and
varying channel conditions between the cyclist and the coach. For efficient energy
utilization, the RF-transmitted power needs to be modified based on distance or
signal-to-noise-ratio (SNR) estimation.
4. Difficulty to use energy harvesting as an alternative to the battery source: The harvest-
ing apparatus adds extra weight and the biker is faced with rolling resistance or
aerodynamic resistance.
Of the limitations stated above, it can be seen that energy is the most critical source in
BWSNs. Therefore, reducing power consumption and extending battery life are manda-
tory because bicycle sensor nodes use battery power as a primary energy source. Several
schemes of power-reduction techniques toward prolonging the battery life of BWSNs will
be presented in Section 12.3.
Wireless connectivity supports wearable equipment that allows for unobtrusive and
noninvasive monitoring. The development of wireless technologies has helped many
practical applications. In sports application, WSNs allow the gathering and use of data,
possibly in real time, and potentially performing the comparison or combination of the
athletes’ performance. As stated in the previous section, Bluetooth, ZigBee, ANT, and
TABLE 12.1
Comparison of Wireless Technologies Used in Bicycle Monitoring
Wireless Technology
Parameters Bluetooth (IEEE 802.15.1) ZigBee/IEEE 802.15.4 ANT
Frequency band(s) 2.4 GHz 868/915 MHz, 2.4 GHz 2.4 GHz
Modulation type GFSK BPSK and OQPSK GFSK (Zhang et al., 2013)
Data rate 1 Mbpsa (Wong et al., 2013); 20h, 40i, and 250 kbpsj 250 kbps, 1 and 2 Mbps
1.2 Mbpsb (Georgakakis et al., (Cavallari et al., 2014)
2011); 3 Mbpsc (Decuir, 2014);
24 Mbpsd (Georgakakis et al.,
2011)
Output power 0e, 4f, and 20 dBmg (Liu et al., (−25 to 0) dBm −18, −12, −6, and 0 dBm
2012) (Kohvakka et al., 2010)
Transmission range 10–100 m (Abbasi et al., 2014) (10–100) m 30 m (Caballero et al.,
2011)
Power consumption (40–100) mW 1 mW 0.183 mW
Battery life (coin cell) (1–7) days (4–6) months 3+ years (Adibi, 2012)
Latency <10 sec (20–30) ms ≈ Zero (Maharjan et al.,
2014)
Spreading type FHSS DSSS Adaptive isochronous
(Gasparrini et al., 2013)
Number of devices in 7 (active)/255 (total) 216 232
network (Padgette et al., 2012)
Number of RF channels 79 Oneh, 10i, 16j (Mraz, 125 (Zhan & Yu, 2013)
Cervenka, Komosny, &
Simek, 2013)
Network topology Ad-hoc, point-to-point, and Ad-hoc, peer-to-peer, Broadcast ad-hoc
star mesh, and star peer-to-peer, mesh,
and star (Rawat
et al., 2014)
a Bluetooth LE, b Bluetooth 1.2, c Bluetooth 2.0 + EDR, d Bluetooth 3.0 and Bluetooth 4.0, e Class 3, f Class 2,
g Class 1, h for 868 MHz, i for 915 MHz, j for 2.4 GHz.
Wi-Fi have been considered for BWSNs. However, Wi-Fi technology is unsuitable for
BWSN due to the high power consumption and large node size; thus, it is not considered
in the next sections.
The main dissimilarities between the Bluetooth, ZigBee, and ANT wireless protocols
used in bicycle monitoring are summarized in Table 12.1 (Hassan, 2012; Khssibi et al., 2013;
Li & Zhuang, 2012; Rault et al., 2014). Bluetooth and ZigBee protocols are based on an
IEEE 802 family of standards while ANT is a proprietary standard. Particularly, Classic
Bluetooth, ZigBee, and ANT use a low data rate. Bluetooth and ZigBee are designed for a
wireless personal area network communication distance of nearly 100 m, whereas ANT
has shorter transmission distance of approximately 30 m (Casamassima et al., 2013). The
RF-transmitted power level can be modified for all three wireless technologies. As seen
in the table, the power consumption is the lowest power for ANT, followed by ZigBee and
Bluetooth.
ZigBee protocol more typically is used in WSNs or control applications that have a lower
requirement on data rate and complexity (Amzucu et al., 2014) as compared to Bluetooth
protocol. In addition, ZigBee can support 65,535 nodes in the network, with simple pro-
cedures for removal or addition of nodes to the network, supports mesh configuration
such that the failure of one node does not bring down the rest of the network, and lower
power consumption. ZigBee/IEEE 802.15.4 has a data rate of 250 kbps (Abbasi et al., 2014;
Wang, 2013) compared to ANT that caters higher data rate options of 250 kbps, 1 and 2
Mbps. Therefore, ANT is capable of transmitting over the air for a shorter duration com-
pared to ZigBee for the same amount of data. ANT has been designed to stay in an ultra-
low power mode (sleep mode) for long periods, quick wake up, speedily transmit for
150 µs, and return to sleep mode. In terms of latency, the ANT offers near zero latency
(Smith, 2011; Tabish et al., 2013), ZigBee latency is below 30 ms, and Bluetooth has the
highest latency of up to 10 s. Latency is a very crucial problem for race-cycling applica-
tions because the cyclist parameters must be transmitted and monitored in real time.
ZigBee employs direct sequence spread spectrum (Nagarajan & Dhanasekaran, 2013),
whereas Bluetooth uses frequency-hopping spread spectrum (Langhammer & Kays, 2012)
schemes, to ensure the security of the wireless channel. However, the ANT employs an
adaptive isochronous (Gasparrini et al., 2013) network technology to ensure coexistence
with other wireless protocols.
From the table, it is clear that the classic Bluetooth wireless protocol is inappropriate for
wireless monitoring of sports cycling performance due to the limitation of communication
range, power consumption, and number of nodes in the network. ANT is more appropri-
ate when the power consumption is a critical demand, whereas ZigBee is more suitable in
terms of communication distance. In addition, latency is worse for Bluetooth and ZigBee
compared to ANT.
12.3 Toward Battery-Free Internet of Bicycle

This section provides a taxonomy view of several energy savings and harvesting
approaches specific to the bicycle wireless sensor nodes as seen in Figure 12.2. Advantages
and limitations of each technique will be highlighted. At the end of the section, several
techniques related to battery-free Internet of Bicycle (IoB) will be presented.
12.3.1 Sleep/Wake Strategy

The sensor nodes radio can be put in sleep mode with the sleep/wake strategy to reduce
the power consumption of WSNs. In sleep mode, no data communication occurs. The
sensor nodes wake to collect and transmit their data to a portion of the time and then
return to sleep mode to conserve energy. The sleep/wake strategy can be performed with
duty cycling (Anchora et al., 2014; Carrano et al., 2014), medium-access control pro-
tocols (Chen et al., 2014; Jing Liu et al., 2015; Khanafer et al., 2014), topology protocol
(Karasabun et al., 2013; Zebbane et al., 2015), and passive wake-up radios (Ba et al.,
2013; Farris et al., 2016). In the duty cycling method, sensor nodes were alternating
between active and sleep modes, according to network activity. A low duty cycle saves
a huge amount of energy relative to the traditional operation in the order of dozens
of milliamperes but significantly increases communications latency (Rault et al., 2014).
Therefore, a trade-off between energy consumption and latency should be taken into
consideration when designing a WSN.
Energy-efficient techniques
Sleep/wake Radio Energy- Energy-

Data mitigation
strategy optimization efficient routing harvesting
Data gathering
Duty-cycling TPC Sink mobility Solar energy
(RCD)
Data driven Multi-path Electromagnetic

MAC protocols Cognitive radio
(combine sensors) routing radiation
Modulation Cluster Magnetic resonant

Topology control Data rate coupling
scheme architecture
Passive wake-up Data Cooperative Relay node

Wind energy
radio compression communication position
Directional Mechanical
Network coding Routing metric
antennas vibration
FIGURE 12.2
Energy-efficient techniques in WSN.
A number of research works have proposed effective duty cycle settings to enhance the
power consumption of WSNs (Anchora et al., 2014; de Paz Alberola & Pesch, 2012; Rasouli
et al., 2014). Topology control cannot be taken into account in BWSN because the existence
of redundant sensor nodes added extra weight to a bicycle, increases hardware size and
complexity, and increases aerodynamic resistance. In addition, this method is expensive,
specifically when the sensor nodes are deployed in a vast area. Also, the passive wake-up
radios method is inappropriate for BWSN because it has short communication distance of
3 to 6.7 m.
12.3.2 Data Mitigation

Data mitigation is another type of power consumption solutions for WSNs. This method
aims to reduce the amount of transmitted data from source nodes to destination node.
Data mitigation can be achieved with data gathering (Dhasian & Balasubramanian,
2013; Talele et al., 2015), data rate, data-driven techniques (Mesin et al., 2014), net-
work coding (Qin et al., 2014; Rout & Ghosh, 2013), and data compression (Incebacak
et al., 2015; Srisooksai et al., 2012). Among the data mitigation types, the data gather-
ing and data rate were used to minimize the power consumption of the bicycle sensor
and mobile nodes for track cycling application has been proposed in our research works
(Gharghan et al., 2014, 2015, 2016). For data gathering, the reduction is accomplished by
using the proposed redundancy and converged data algorithm for speed/cadence sen-
sor node (Gharghan et al., 2014) and the average torque data for the bicycle torque sensor
node (Gharghan et al., 2015). Moreover, by taking advantage of the data link between bicycle
speed and cadence measurement method, the bicycle speed and cadence sensor nodes are
merged into one node. Consequently, the transmitted data volume is minimized. Thus, the
power consumption, cost, size, and complexity of the bicycle speed/cadence sensor node
will also be reduced. The data rate of BWSN wireless protocol can be adjusted by bicycle
mobile nodes to reduce their power consumption based on accurate distance estimation
between the coach and the bicycle on the track. High data rate translates to lower receiver
sensitivity and lower communication distance. Therefore, a trade-off between data rate
and communication range is necessary.
12.3.3 Radio Optimization

Previous research works have shown that most of the power are dissipated in the RF
components of WSNs compared to data processing units, such as microcontrollers and
microprocessors (Mesin et al., 2014). Radio optimization schemes were used to reduce
the power consumption of the RF components of sensor nodes using schemes such as
transmission power control (TPC) (Cotuk et al., 2014; Ramakrishnan & Krishna, 2014;
Yan et al., 2013), adaptive modulation (Rosas & Oberli, 2012; Sendra et al., 2011), cogni-
tive radio (Mansoor et al., 2015; Naeem et al., 2013), directional antennas (Karapistoli
et al., 2009; Khatiwada & Moh, 2015), and cooperative communication (Dong et al., 2015;
Garcia et al., 2013; Jung et al., 2011). TPC scheme is considered for cycling application
in Gharghan et al. (2016), where the RF-transmitted power of the ANT mobile node was
modified based on accurate distance estimation between the cyclist and the coach. A
combination of two power-reduction schemes (i.e., sleep/wake and TPC) was presented in
Castagnetti et al. (2014) and Dai et al. (2012).
Note that adaptive modulation schemes cannot be used in the cycling sensor nodes based
on 2.4 GHz ZigBee and ANT wireless protocols because only single modulation scheme
is supported, specifically offset quadrature phase-shift keying (Setiawan et al., 2015) and
Gaussian frequency-shift keying (Zhang et al., 2013) for ZigBee and ANT, respectively. It is
also difficult to implement cooperative communication in BWSN because the installation
of adjacent nodes in a bicycle equates to extra weight. In addition, the directional anten-
nas approach cannot be employed in cycling application because of the adopted wire-
less protocols (i.e., ZigBee and ANT) have been designed with the use of omnidirectional
antennas.
12.3.4 Routing Protocols

Routing protocols in WSNs are unlike other wireless networks due to the limited energy
resources of sensor nodes, data are gathered from several nodes to a sink node and a single
global address cannot be adopted in the random positioning of the sensor nodes (Arora
et al., 2016). Due to the fact that sensor nodes are commonly battery powered (Hayes & Ali,
2016), the routing protocols in BWSNs are mainly designed to improve power consump-
tion, prolong the network lifetime, use of cluster architecture, and sink mobility (Rault
et al., 2014) that is suitable when the cyclists are moving in group.
12.3.5 Energy Harvesting

A major limitation of sensor nodes is limited battery capacity. An alternative method
that has been used to address the problem of limited nodes lifespan is to employ
energy harvesting. Numerous technologies have been developed to allow sensor
nodes to harvest energy from different sources such as solar (Yongtai et al., 2012),
electromagnetic radiation (Cortés-Sánchez et al., 2014), magnetic resonant coupling
(Li et al., 2015), piezoelectric microgenerators (Nechibvute et al., 2012), mechanical
vibration (Yang et al., 2012), and kinetic and wind energy (Sudevalayam & Kulkarni,
2011). Compared with conventional sensor nodes, rechargeable sensor nodes can
work continuously, ideally for an infinite lifetime. The ambient energy can be con-
verted to electrical energy and can be employed directly to supply the sensor nodes or
can be stored and used later on. The energy of the sensor nodes needs to be estimated
to evaluate the battery-charging cycle. Hence, the sensor nodes vital parameters can
be optimized such as duty cycling, sampling rate, and TPC to adjust their power
consumption consistent with the periodicity and amount of the harvesting source.
The energy harvesting mechanisms can be operated alongside batteries of the sensor
nodes. For example, sensor node using solar energy to charge the batteries can work
extensively during the daytime, whereas the sensor node can use their batteries based
on power-reduction techniques such as sleep mode (i.e., duty cycle) in the night to
save energy. Furthermore, a sensor node can enter higher sleep periods (i.e., low duty
cycle) and low transmission power when the batteries of the sensor node have low
residual energy (Nintanavongsa et al., 2013). Recent developments in wireless power
transfer (WPT) are expected to significantly increase the lifetime of WSNs and make
them continuously operational, as the WPT techniques can be employed to transmit
power between transmitter and receiver devices without any contact between them.
Many research works used WPT to charge the sensor nodes for different applications
such as medical sensor nodes for implantable devices, unmanned aerial vehicles or
drones, and WSNs (Griffin & Detweiler, 2012; Kong & Ku, 2016; Nintanavongsa et al.,
2013; Xie et al., 2013; Yoshida et al., 2013).
Energy harvesting is not suitable for use in race cycling because the harvesting equip-
ment adds extra weight. However, it is possible for use in day-to-day cycling.
12.4 Why Connected Bicycles?

The market value of data in the age of the IoT was estimated at US$19 trillion per year
by the CEO of Cisco in 2014 (Ravindranath, 2014), of which $4.6 trillion was public sec-
tor savings because cities are working better through interconnectivity. This section
explores several aspects of IoT end-user application in the bicycle sector herein referred
as IoBs.
12.4.1 Smart City Driven

Cycling contributes significantly to the concept of smart city initiatives by preventing con-
gestion because cyclist takes up less space than parked cars. Cycling also contributes to
the health of the society and promotes the concept of green energy. Therefore, it is no sur-
prise that there has been an increase in the number of people using a bicycle as their pre-
ferred mode of transportation (Wegman et al., 2012). Studies conducted in Monon Trail in
Indianapolis concluded with the report that homes closer to urban bicycle greenways are
11% more marketable compared to homes without such facility (Fischer et al., 2010). As
part of the Connected Streets project, London-based Future Cities Catapult has come up
with innovative technologies that could improve cyclists’ route choices, wayfinding and
safety on the urban road, by using the IoT to complement existing cycling infrastructure.
These technologies that are currently in the prototype stage, include (1) a handlebar-mounted
device to measure air quality and suggest cleaner roads, (2) a route-rectifying attachment for
bike-sharing schemes, and (3) a helmet visor that offers alternative directions utilizing the
idea of Head-Up Displays to overlay real-time data that provide a visual nudge to the user,
offering safer directions based on the immediate streetscape around them, an adaptation to
the Google Glass.
12.4.2 Intelligent Transport System Concept

The intelligent transport system involves the incorporation of artificial intelligence unit
resting on the upper layer of wireless sensor application protocol stack. In reality, many
projects have been initiated under the intelligent transport system concept driven by
wireless sensors. SafeSpot, a project of INFRASENS, deploys cameras, laser scanners,
and RFID-systems to detect dangerous situations. The advanced driver assistance sys-
tems use the in-vehicle sensor to increase the safety of cyclist. The Dutch InnoCross
project aiming to decrease the number of accidents involving the cyclist by decreas-
ing the number of red light violence of the cyclists with the aid of green light assis-
tance (Thielen et al., 2012). With the rise of driverless or autonomous cars, streets will be
made safer and air quality experienced by cyclists will also be improved. In addition,
with advanced sensor technology and a networked traffic system, it is possible that in
the future all mobility devices will be connected, eliminating traffic accidents almost
completely (Mayne, 2016a).
In recent years, there have a lot of hype on electronic bicycle, or commonly known as
e-bike. The time has come to integrate the e-bike industry into the smart mobility plat-
forms. Using the capability of location-tracking capability, bike share is a particularly
powerful tool as an information source about people’s mobility habits. The cycling data
collection could be used by the city for transport planning and infrastructure design.
Furthermore, it could also be used as a marketing avenue for cities and businesses. For
example, the mobility data could be used to plan targeted advertising to a rider individual-
ized journeys, transmitted to the bike via a smartphone or to a docking station where the
ride ends (Mayne, 2016b).
12.4.3 Economic Value Chain

The connected bike will open up a host of revenue opportunities for entrepreneurs
wanting to create floating e-bike-rental schemes in cities. A similar service, but for scoot-
ers, has been introduced in Paris to decongest the city and boost the tourism industry,
called Cityscoot (Hayashi et al., 2012). With Cityscoot, you can locate an available electric
scooter via the Cityscoot smartphone app, and book the scooter for an emission-free
drive around the city whenever you want, without any commitment and the downsides
of owning them.
In May 2013, Citi Bike (Citi Bike NYC), a privately owned public bicycle-sharing system
in New York City has been launched. To date, there are 8000 bikes and 500 stations across
Manhattan, Brooklyn, Queens, and Jersey City, which marks as the largest bike share
program in the United States. As of March 2016, Citi Bike has a total of 163,865 annual
subscribers, with an average of 27,287 rides per day in 2015. In March 2014, the VeloCitta
project (VeloCittà) was launched, where five participating European cities (Krakow,
London, Szeged, Padua, and Burgos) are looking to develop a new approach to the organi-
zational and financial aspect of the bike-sharing systems. On further analysis, the direct
and indirect spill-over effects accruing from bicycling industry are $46.9 billion on food,
transportation, lodging, gift, and entertainment (Flusche, 2012). Another possible business
opportunity is hosting organized rides and races events, which will draw thousands of
people to the host communities and help boost the local economy.
12.4.4 Cyclist Safety Driven

Worldwide, more than 1.2 million people lose their lives in road accidents annually (WHO,
2015). According to U.S. Department of Transport study (USDOT, 2010), bicycle and pedes-
trian crashes accounted for more than 15% of traffic fatalities. As modern urban plan-
ning concepts emphasize and encourage nonmotorized modes of transportation, there
is an important need for innovative and proactive methods of analyzing and improving
cyclists’ safety and flow efficiency. A cyclist is generally considered as vulnerable road
users and exposed to both fatal and reduced injury (Thielen et al., 2012). To mitigate or at
least reduce the occurrence of these accidents, several candidate solutions have emerged. In
Goldhammer et al. (2012), cooperative sensor network has been proposed to reduce cyclist
accident at road intersections. The idea behind their proposals is to reduce the cyclist acci-
dent probability by increasing their field of view that creates a bird’s eye view of their
surroundings. Besides that, there is considerable research conducted on cyclist safety
focused on analyzing collision data and factors that contribute to the collision. These fac-
tors include attributes such as bicycle helmet usage, cyclist–vehicle-driving characteristics
(e.g., speed limit and gap acceptance), conflicts at different road topologies, and separated
cycle paths infrastructure design (Dill et al., 2012).
An inventive invisible cyclist safety helmet, called the Hövding, has been invented by
students at the University of Lund in Sweden (Hövding). It is an airbag helmet that is
actually a collar, worn around the neck, which inflates instantly during a crash using an
algorithm that can distinguish normal cycling from accidents.
12.4.5 Technology Driven

Many cyclists use heart rate monitors strapped to their chests or wrists, and even the
lactate threshold work on the calf to measure the level of lactic acid builds up in the blood-
stream due to intensive exercise, causing discomfort and forcing the athlete to slow down.
Such devices require the rider to look down to view their data on different displays, or
take their hands off the bike. These smart wearables can present a safety issue, particu-
larly when riding in a close pack of cyclists, or navigating crowds. The Solos smart cycling
glasses (Solos) have been used for U.S. track cyclists training for the 2016 Olympic Games,
which displays the real-time performance metrics culled from the athletes’ bike sensors,
heart rate monitors, and other self-tracking devices and provides micro-display of these
at the right-hand corner of the cyclist’s glass. This enables riders to view their critical stats
without taking their eyes off the road.
The smart pedal by Connected Cycle (Smart Pedal) is an add-on to any normal bike,
which comes with GPS, GPRS, Bluetooth, Accelerometer, and a backup battery that
recharges itself through the bike electric circuit. The pedal automatically records the
speed, route, incline, and calories burned on each bike trip. These statistics are sent to the
cloud and made available to users through the Connected Cycle application available on
smartphones. It provides a simple solution to bike theft, where the bike owner is instantly
notified via smartphone when the bike is moved. The location capability not only prohibits
theft but also allows the owner to locate where they last parked their bike. BitLock (Bitlock)
and Skylock (Skylock) are examples of a smart keyless lock for a bicycle. Another stand-
alone bicycle add-on that has been invented is the Bluetooth and GPS-enabled Helios
(Helios bars), a set of aluminum handlebars that integrates headlight and blinker system
for bicycles.
12.5 Potential Opportunities on the Internet of Bicycles

Without any ambiguity, the goal of IoB is the enablement of seamless integration of any
potential bicycle sensor device into the Internet, thus creating a platform for new forms
of machine-to-machine communication and machine-to-human communication. The
implementation of IoB is expected to face some challenges as it may require substantial
remodification of the protocol stack from the physical layer up to data representation and
service composition.
12.5.1 Long Range, Low-Rate, Low-Power, Low-Cost Internet

of Bicycle Wireless Backhaul
The three wireless network topologies capable of simulating the IoT concept are the cel-
lular, IEEE 802.15 standards, and Low-Power Wide-Area Networks (LPWANs). At the
moment, tested IoT applications are based on multihop short-range IEEE 802.15.4 stan-
dards that include: ZigBee and Bluetooth (Gubbi, Buyya, Marusic, & Palaniswami, 2013).
Fortunately, this is same technology driving the BWSN. These technologies offered the
needed degrees of freedom to implement many IoT user applications as they operate in
the license-exempt radio spectrum, thrives on very low power consumption, and suitable
for miniaturized devices, which are fundamental requirements for wireless sensor-driven
IoT devices. However, ZigBee and Bluetooth suffer from poor coverage in the range of 10 to
100 m line-of-sight that limits their application especially in scenarios that require urban-
wide coverage similar to the IoB concept.
On the other hand, the traditional cellular technology specifically, Third-Generation
Partnership Project, which drives the universal mobile telecommunications system
(UMTS) and long term evolution (LTE) wireless standards, was not originally designed
to support IoT because of the huge number of devices requiring connectivity (Centenaro
et al., 2015). The IoT traffic will be sporadic involving sending short packages and requires
lots of signaling overhead. To illustrate this point, take for instance, when bicycles are
fitted with road intersection accident warning systems, packets are transmitted only at
the road intersection and when there is an impending danger. In essence, BWSN cannot
be supported by cellular topology. Nonetheless, there are concerted efforts to revamp the
2G/GSM to cater for the expected IoT traffic through Cellular IoT architecture (Centenaro
et al., 2015).
A temporary solution may lie in the so-called LPWANs. The LPWANs networks oper-
ate in the sub-GHz, unlicensed frequency bands using mesh or star topologies. A unique
feature of LPWANs is the presence of collector node referred as the gateway to other
collocated LPWANs. There are several LPWANs standards and proprietary technologies
such as TV White Space technology IEEE 802.22 wireless rural area network (WRAN)
(Fitch et al., 2011), LoRa (Long Range) (Centenaro et al., 2015), SIGFOX™ and Ingenu™.
The actual description of these technologies is beyond the scope of this chapter. In sum-
mary, for low requirement of IoT data rates, the wireless backbone for IoB should meet the
listed criteria:
• Operating frequency bands: 784, 868, 915 MHz

• Transmission range: 10 to 15 km in rural areas, and 2 to 5 km in urban areas
• Topology support: For large mesh networking or star clusters
• Quality of service (QoS) Support for low latency devices
12.5.2 High-Rate 5G Backhaul

Industry experts believe that 5G will hit the market by the beginning of 2020. However to
date, the decision on 5G spectrum band license has yet to be allocated. One of the goals of
5G is to cater for the rising aggregate spectrum demand driven by the IoT concept. Some
IoT applications will require an aggregated data rate in order of Gbps that can only be met
by the proposed 5G networks attributes (Mitola et al., 2014):
• Combines multiple physical layer (PHY), medium-access control, radio access

network (RAN), mobility management IP core technologies involving multiple
bands with different propagation characteristics, and access rules.
• Heterogeneous coexistence involving networks with transmission power, access
scheme, cell topology consisting picocell, femtocell, microcell, and macrocell.
• Interference mitigation and management techniques with QoS constraints.
12.5.3 Energy-Efficient Transmission Algorithm

In the IoB concept, hundreds or even thousands of sensing nodes might be involved using
data fusion rules. Hence, there is a need to exploit large numbers of these sensing nodes
to create a cheap and energy-efficient but high-quality network. Therefore, protocols
must be designed capable of achieving fault tolerance when confronted with node failure
while minimizing energy consumption. Inspired by these problems, several efficient rout-
ing algorithms have been proposed such as Low-Energy Adaptive Clustering Hierarchy
(Arora et al., 2016). The protocol provides localized coordination and control of cluster
setup and operation, randomized rotation of cluster head selection, and data compression
capabilities. Although performance wise, this protocol is efficient, however, no specific
standard was attached to this protocol.
Exploiting this limitation, several energy-efficient algorithms have been proposed for
bicycle sensor transmission. Baechler (2015) has proposed an energy-efficient low-power
algorithm based on either ANT/ANT+, ZigBee, or Bluetooth standard. The proposal
includes an adaptive power-reduction algorithm to reduce the power consumption at the
speed/cadence sensor bicycle node. Furthermore, the algorithm is capable of switching
between redundancy and converged data algorithm and the sleep/wake algorithm for
torque sensor node power consumption as presented in Section 12.3. Several other works
have utilized XBee and ANT wireless protocol (Dementyev et al., 2013). Both wireless proto-
cols considered the low power consumption, low cost and weight, small size with accept-
able communication range, and above all, they operate in the license-exempt 2.4 GHz
industrial, scientific, and medical (ISM) band. Of recent, an improved version of the
Bluetooth technology known as Bluetooth Smart or Bluetooth Low Energy has emerged
(Khssibi et al., 2013). Though several energy-efficient algorithms have been proposed for
bicycle sensor, there is a need to explore the following in the context of IoB:
• Energy-efficient algorithm based on distance and traffic

• Low complexity cluster head selection
• Optimal cluster size
• Admission control policies
12.5.4 Energy-Harvesting Radios

Energy-harvesting techniques are an integral component of the modern sensor network to
prolong the wireless sensor transmitter lifetime. The details on recent research on energy-
harvesting radios have been presented in Section 12.3.5. However, the question remains
on the most effective and optimal approach among the listed approaches to power the IoB
radios.
The OpenBike (OpenBike) project mentioned in Section 12.5.5 is an example of
energy-harvesting effort by converting pedal power to electric power and subsequently,
charging the battery. Nonetheless, some research gaps still exist on energy-harvesting
protocol design:
• Designing a protocol that exploits intermittent and random energy arrival

• Optimal energy-scheduling algorithm taking cognizance of energy harvest
constraint
• Specific algorithm categorizing source of the energy power source
• Adaptive and response algorithm capable of selecting/switching to different
sources of energy
12.5.5 Big Data Analytics

Each of the IoB drivers discussed in the previous section has the capacity to generate small
data which will snowball into big data. Big data is simply making intelligence out of struc-
tured and unstructured dataset and has been increasingly adopted in the many sectors
such as: health, sciences, engineering, education, and the society at large (Hashem et al.,
2015). Not only the big data techniques possess the capacity to assist vulnerable road users
in automating and computerizing their schedules, but they also provide reduced infra-
structure maintenance cost, efficient management, and user access. There are many advan-
tages of exploring the big data techniques in IoB concept, which include: extrapolating
hidden/latent information as well as pattern recognition using any of the statistical tools.
IoB is expected to fit into the four Vs of big data, which are volume, variety, velocity, and
value (Ding et al., 2014). Wireless sensor lacks the memory and computing power to ana-
lyze data and makes an informed decision. Therefore, IoB will be inclined toward cloud
computing. The advantages of using cloud computing schemes are the elimination of
expensive hardware, a complex algorithm, dedicated hardware space, and a reduction in
CAPital EXpenditure (CAPEX) and OPerating EXpenditure (OPEX). However, before these
performances gain can be achieved, there is a need to address some of the basic challenges
of big data in the context of IoB data capture, storage, search, analysis, and virtualization.
12.5.6 Network Security Challenge

As with other ICT-based system, security challenges must be overcome for IoB concept
to be accepted. Security measures ensure the interactions between bicycle to bicycle or
human operators to bicycle over the wireless link are conducted in a secured platform
devoid of foreseeable security challenges. Several security mechanisms that must be inte-
grated in IoT in general and adopted by the IoB have already been enumerated by the
research community (Heer et al., 2011; Roman et al., 2013) such as:
• Protocol and network security characterized by the cryptographic algorithm for

secured end-to-end communication.
• A simple and yet complex identity management scheme embedded in a highly
integrated authentication framework driven by a user-friendly authorization
module.
• Privacy policy to address concerns bordering on profiling, stealing, and loss of
control must be designed. Anonymity and transparency are the key attributes of
the privacy modules that must be upheld.
• Robust and fault tolerance are essential for IoB concept as a cyclist cannot afford
to have a system that is not resistance to a fault.
• Trust and governance framework encompass machine learning and human con-
trol in the IoB concept. Although trust creates a platform for entities to deal with
future actions of collaborating devices, governance invokes the human mecha-
nism ability to punish, exclude, and readmit users.
Highlighting the possible security framework is important, it is equally important to

expose some of the tactics that will be deployed to execute the security threats stated below:
• Denial of service (DoS): This is the most widely anticipated techniques to launch
security attack in a wireless system. DoS can target IoB system by jamming their
communication channels. Thus, rendering the system inaccessible.
• Physical damage: Although DoS is technical involving the use of highly sophisti-
cated tools, physical damage denotes a condition in which the attacker has direct
access to the IoB hardware.
• Eavesdropper: Denotes incubation stage in which the attacker in a passive mode is
focusing on monitoring, analyzing, and observing the data flow chain in the system.
• Node capture: This is the advanced form of eavesdropping in which the attacker
decides to wirelessly and remotely control the IoB system.
12.6 Conclusion
IoBs is expected to be one of the things to capitalize from the IoT. The benefit of IoB has
improved rider safety, greener transportation option to the urban community, improve
health management of the cyclist, and new economies of scale opportunities. The IoB tech-
nologies compose of wireless communication and sensor technology, backhaul network,
cloud computing, and big data processing. Energy has been identified as one of the most
critical resources toward enabling IoB due to the requirement of low bicycle weight, mobil-
ity support, low cost, and low system complexity. However, battery technology still has not
reached maturity. One of the grand challenges toward IoT is a lifetime of the sensors. An
alternative way is to exploit another form of surrounding energy, known as energy harvest-
ing. The taxonomy view on energy harvesting techniques has been presented. Ultimately, the
future aim is to achieve battery-free supply. Besides that, several energy-efficient approaches
have been identified. A prototype has been developed in this work as an example of a case
study based on sleep/wake strategy and radio optimization. Finally, latest development and
potential research topics in IoB have also been highlighted in this chapter.
References
Abbasi, A. Z., Islam, N., & Shaikh, Z. A. (2014). A review of wireless sensors and networks’ appli-
cations in agriculture. Computer Standards & Interfaces, 36(2), 263–270.
Adibi, S. (2012). Link technologies and BlackBerry mobile health (mHealth) solutions: A review.
IEEE Transactions on Information Technology in Biomedicine, 16(4), 586–597.
Amzucu, D. M., Li, H., & Fledderus, E. (2014). Indoor radio propagation and interference in 2.4 GHz
wireless sensor networks: Measurements and analysis. Wireless Personal Communications, 76(2),
245–269.
Anchora, L., Capone, A., Mighali, V., Patrono, L., & Simone, F. (2014). A novel MAC scheduler
to minimize the energy consumption in a Wireless Sensor Network. Ad Hoc Networks, 16,
88–104.
Arora, V. K., Sharma, V., & Sachdeva, M. (2016). A survey on LEACH and other’s routing proto-
cols in wireless sensor network. Optik-International Journal for Light and Electron Optics, 127(16),
6590–6600.
Ba, H., Demirkol, I., & Heinzelman, W. (2013). Passive wake-up radios: From devices to applications.
Ad Hoc Networks, 11(8), 2605–2621.
Baca, A., Kornfeind, P., Preuschl, E., Bichler, S., Tampier, M., & Novatchkov, H. (2010). A server-based
mobile coaching system. Sensors, 10(12), 10640–10662.
Baechler, H. (2015). Sensor apparatus and method for determining pedalling cadence and travelling
speed of a bicycle: Google Patents number: US9075076B2, BMC Trading AG: Switzerland.
Balbinot, A., Milani, C., & Nascimento, J. S. B. (2014). A new crank arm-based load cell for the 3D
analysis of the force applied by a cyclist. Sensors, 14(12), 22921–22939.
Bini, R. R., Hume, P. A., & Cerviri, A. (2011). A comparison of cycling SRM crank and strain gauge
instrumented pedal measures of peak torque, crank angle at peak torque and power output.
Procedia Engineering, 13, 56–61.
Bitlock. Smart keyless bike lock. Retrieved from https://bitlock.co.
Buratti, C., Conti, A., Dardari, D., & Verdone, R. (2009). An overview on wireless sensor networks
technology and evolution. Sensors, 9(9), 6869–6896.
Caballero, I., Sáez, J. V., & Zapirain, B. G. (2011, June). Review and new proposals for zigbee applications
in healthcare and home automation. Paper presented at the International Workshop on Ambient
Assisted Living, Torremolinos-Málaga, Spain.
Carrano, R. C., Passos, D., Magalhaes, L. C., & Albuquerque, C. V. (2014). Survey and taxonomy
of duty cycling mechanisms in wireless sensor networks. IEEE Communications Surveys &
Tutorials, 16(1), 181–194.
Casamassima, F., Farella, E., & Benini, L. (2013). Power saving policies for multipurpose WBAN. Paper
presented at the 23rd International Workshop on Power and Timing Modeling, Optimization
and Simulation (PATMOS), Karlsruhe, Germany p. 83–90.
Casas, O. V., Dalazen, R., & Balbinot, A. (2016). 3D load cell for measure force in a bicycle crank.
Measurement, 93, 189–201.
Castagnetti, A., Pegatoquet, A., Le, T. N., & Auguin, M. (2014). A joint duty-cycle and transmission
power management for energy harvesting WSN. IEEE Transactions On Industrial Informatics,
10(2), 928–936.
Cavallari, R., Martelli, F., Rosini, R., Buratti, C., & Verdone, R. (2014). A survey on wireless body area
networks: Technologies and design challenges. IEEE Communications Surveys & Tutorials, 16(3),
1635–1657. doi:10.1109/SURV.2014.012214.00007
Centenaro, M., Vangelista, L., Zanella, A., & Zorzi, M. (2015). Long-range communications in unli-
censed bands: The rising stars in the IoT and smart city scenarios. arXiv preprint arXiv:1510.00620.
23(5): p. 60–67.
Chen, K., Ma, M., Cheng, E., Yuan, F., & Su, W. (2014). A survey on MAC protocols for underwater
wireless sensor networks. IEEE Communications Surveys & Tutorials, 16(3), 1433–1447.
Citi Bike NYC The Bike-Share Planning Guide. Retrieved from www.citibikenyc.com.
Cortés-Sánchez, J., Velázquez-Ramírez, A., Lucas-Bravo, A., Rivero-Angeles, M. E., & Salinas-Reyes,
V. A. (2014). On the use of electromagnetic waves as means of power supply in wireless sensor
networks. EURASIP Journal on Wireless Communications and Networking, 2014(1), 1–10.
Cotuk, H., Bicakci, K., Tavli, B., & Uzun, E. (2014). The impact of transmission power control
strategies on lifetime of wireless sensor networks. IEEE Transactions on Computers, 63(11),
2866–2879.
Dai, G., Qiu, J., Liu, P., Lin, B., & Zhang, S. (2012). Remaining energy-level-based transmission
power control for energy-harvesting WSNs. International Journal of Distributed Sensor Networks,
2012, 1–12.
de Paz Alberola, R., & Pesch, D. (2012). Duty cycle learning algorithm (DCLA) for IEEE 802.15.4
beacon-enabled wireless sensor networks. Ad Hoc Networks, 10(4), 664–679. doi:http://dx.doi.
org/10.1016/j.adhoc.2011.06.006.
Decuir, J. (2014). Introducing Bluetooth smart: Part II: Applications and updates. IEEE Consumer
Electronics Magazine, 3(2), 25–29. doi:10.1109/MCE.2013.2297617.
Dementyev, A., Hodges, S., Taylor, S., & Smith, J. (2013, April). Power consumption analysis of Bluetooth
Low Energy, ZigBee and ANT sensor nodes in a cyclic sleep scenario. Paper presented at the IEEE
International on Wireless Symposium (IWS), Beijing, China.
Dhasian, H. R., & Balasubramanian, P. (2013). Survey of data aggregation techniques using soft
computing in wireless sensor networks. IET Information Security, 7(4), 336–342.
Dill, J., Monsere, C. M., & McNeil, N. (2012). Evaluation of bike boxes at signalized intersections.
Accident Analysis & Prevention, 44(1), 126–134.
Ding, G., Wu, Q., Wang, J., & Yao, Y.-D. (2014). Big spectrum data: The new resource for cognitive
wireless networking. arXiv preprint arXiv:1404.6508, p. 1–13.
Dong, L., Tao, H., Doherty, W., & Young, M. (2015). A sleep scheduling mechanism with PSO col-
laborative evolution for wireless sensor networks. International Journal of Distributed Sensor
Networks, 2015, 12, 517250.
Farris, I., Militano, L., Iera, A., Molinaro, A., & Spinella, S. C. (2016). Tag-based cooperative data gath-
ering and energy recharging in wide area RFID sensor networks. Ad Hoc Networks, 36, 214–228.
Fischer, E. L., Rousseau, G. K., Turner, S. M., Blais, E. J., Engelhart, C. L., Henderson, D. R., … Tobias,
P. A. (2010). Pedestrian and bicyclist safety and mobility in Europe. Transportation Research Board
(TRB): Washington, DC.
Fitch, M., Nekovee, M., Kawade, S., Briggs, K., & MacKenzie, R. (2011). Wireless service provision in
TV white space with cognitive radio technology: A telecom operator’s perspective and experi-
ence. IEEE Communications Magazine, 49(3), 64–73.
Flusche, D. (2012). Bicycling means business: The economic benefits of bicycle infrastructure. The
National Academies of Sciences, Engineering, and Medicine, 17, 28.
Garcia, M., Sendra, S., Lloret, J., & Canovas, A. (2013). Saving energy and improving communications
using cooperative group-based wireless sensor networks. Telecommunication Systems, 52(4),
2489–2502.
Gasparrini, S., Gambi, E., & Spinsante, S. (2013, October). Evaluation and possible improvements of the
ANT protocol for home heart monitoring applications. Paper presented at the IEEE International
Workshop on Measurements and Networking Proceedings (M&N), Naples, Italy.
Georgakakis, E., Nikolidakis, S., Vergados, D., & Douligeris, C. (2011). An analysis of Bluetooth,
zigbee and Bluetooth low energy and their use in WBANs. In J. Lin & K. Nikita (Eds.), Wireless
Mobile Communication and Healthcare (Vol. 55, pp. 168–175). Berlin, Germany: Springer.
Gharghan, S. K., Nordin, R., & Ismail, M. (2014). Energy-efficient ZigBee-based wireless sensor net-
work for track bicycle performance monitoring. Sensors, 14(8), 15573–15592.
Gharghan, S. K., Nordin, R., & Ismail, M. (2015). An ultra-low power wireless sensor network for
bicycle torque performance measurements. Sensors, 15(5), 11741.
Gharghan, S. K., Nordin, R., & Ismail, M. (2016). Energy efficiency of ultra-low-power bicycle wire-
less sensor networks based on a combination of power reduction techniques. Journal of Sensors,
2016, 21. doi:10.1155/2016/7314207.
Goldhammer, M., Strigel, E., Meissner, D., Brunsmann, U., Doll, K., & Dietmayer, K. (2012,
September). Cooperative multi sensor network for traffic safety applications at intersections. Paper
presented at the 15th International IEEE Conference on Intelligent Transportation Systems,
Anchorage, AK.
Griffin, B., & Detweiler, C. (2012, May). Resonant wireless power transfer to ground sensors from a UAV.
Paper presented at the IEEE International Conference on Robotics and Automation (ICRA),
Saint Paul, MN.
Gu, L., & Stankovic, J. A. (2005). Radio-triggered wake-up for wireless sensor networks. Real-Time
Systems, 29(2–3), 157–182.
Gubbi, J., Buyya, R., Marusic, S., & Palaniswami, M. (2013). Internet of Things (IoT): A vision, archi-
tectural elements, and future directions. Future Generation Computer Systems, 29(7), 1645–1660.
Hashem, I. A. T., Yaqoob, I., Anuar, N. B., Mokhtar, S., Gani, A., & Khan, S. U. (2015). The rise of “big
data” on cloud computing: Review and open research issues. Information Systems, 47, 98–115.
Hassan, M. A. A. (2012, June). A review of wireless technology usage for mobile robot controller. Paper
presented at the Proceeding of the International Conference on System Engineering and
Modeling (ICSEM’12), Singapore.
Hayashi, H., Kagami, O., & Harada, M. (2012, June). Results of field trials with wide-area ubiquitous
network. Paper presented at the IEEE MTT-S International Microwave Symposium Digest
(MTT), Montreal, QC.
Hayes, T., & Ali, F. (2016). Robust ad-hoc sensor routing (RASeR) protocol for mobile wireless sensor
networks. Ad Hoc Networks, 50, 128–144.
Heer, T., et al. Security Challenges in the IP-based Internet of Things. Wireless Personal Communications,
2011. 61(3): p. 527–542.
Helios bars. Transform any bike into a smart bike. Retrieved from www.kickstarter.com/projects/
kennygibbs/helios-bars-transform-any-bike-into-a-smart-bike.
Hövding. Airbag for urban cyclists. Retrieved from www.hovding.com.
Incebacak, D., Zilan, R., Tavli, B., Barcelo-Ordinas, J. M., & Garcia-Vidal, J. (2015). Optimal data
compression for lifetime maximization in wireless sensor networks operating in stealth mode.
Ad Hoc Networks, 24, 134–147.
Jung, J. W., Wang, W., & Ingram, M. A. (2011, March). Cooperative transmission range extension for duty
cycle-limited wireless sensor networks. Paper presented at the IEEE 2nd International Conference
on Wireless Communication, Vehicular Technology, Information Theory and Aerospace &
Electronic Systems Technology (Wireless VITAE), Chennai, India.
Karapistoli, E., Gragopoulos, I., Tsetsinas, I., & Pavlidou, F.-N. (2009). A MAC protocol for low-rate
UWB wireless sensor networks using directional antennas. Computer Networks, 53(7), 961–972.
Karasabun, E., Korpeoglu, I., & Aykanat, C. (2013). Active node determination for correlated data
gathering in wireless sensor networks. Computer Networks, 57(5), 1124–1138.
Kaseva, V. A., Kohvakka, M., Kuorilehto, M., Hännikäinen, M., & Hämäläinen, T. D. (2008). A wire-
less sensor network for RF-based indoor localization. EURASIP Journal on Advances in Signal
Processing, 2008(1), 1–27.
Khanafer, M., Guennoun, M., & Mouftah, H. T. (2014). A survey of beacon-enabled IEEE 802.15.4
MAC protocols in wireless sensor networks. IEEE Communications Surveys & Tutorials, 16(2),
856–876.
Khatiwada, B., & Moh, S. (2015). A novel multi-channel MAC protocol for directional antennas in ad
hoc networks. Wireless Personal Communications, 80(3), 1095–1112.
Khssibi, S., Idoudi, H., Van Den Bossche, A., Val, T., & Saidane, L. A. (2013). Presentation and analy-
sis of a new technology for low-power wireless sensor network. International Journal of Digital
Information and Wireless Communications, 3(1), 75–86.
Kohvakka, M., Suhonen, J., Hämäläinen, T. D., & Hännikäinen, M. (2010). Energy-efficient reserva-
tion-based medium access control protocol for wireless sensor networks. EURASIP Journal on
Wireless Communications and Networking, 2010(1), 1–22.
Kohvakka, M., Suhonen, J., Kuorilehto, M., Kaseva, V., Hännikäinen, M., & Hämäläinen, T. D. (2009).
Energy-efficient neighbor discovery protocol for mobile wireless sensor networks. Ad Hoc
Networks, 7(1), 24–41.
Kong, P., & Ku, H. (2016). Efficiency optimising scheme for wireless power transfer system with two
transmitters. Electronics Letters, 52(4), 310–312.
Kuhn, T., Jaitner, T., & Gotzhein, R. (2008). Online-monitoring of multiple track cyclists during train-
ing and competition (p81) The Engineering of Sport, 7, 405–412.
Langhammer, N., & Kays, R. (2012, August). Enhanced frequency hopping for reliable interconnection of
low power smart home devices. Paper presented at the 8th International Wireless Communications
and Mobile Computing Conference (IWCMC), Limassol, Cyprus.
Lanzisera, S., Mehta, A. M., & Pister, K. S. (2009, June). Reducing average power in wireless sensor net-
works through data rate adaptation. Paper presented at the IEEE International Conference on
Communications, Dresden, Germany.
Li, H., Li, J., Wang, K., Chen, W., & Yang, X. (2015). A maximum efficiency point tracking con-
trol scheme for wireless power transfer systems using magnetic resonant coupling. IEEE
Transactions on Power Electronics, 30(7), 3998–4008.
Li, M., & Zhuang, M. (2012, August). An overview of Physical layers on wireless body area network. Paper
presented at the International Conference on Anti-Counterfeiting, Security and Identification
(ASID), Taipei, Taiwan.
Liu, J., C. Chen, and Y. Ma (2012). Modeling and performance analysis of device discovery in blue-
tooth low energy networks, 1538-1543. Paper presented at the 6th IEEE Global Communications
Conference (GLOBECOM) Anaheim, CA.
Liu, J., Li, M., Yuan, B., & Liu, W. (2015). A novel energy efficient mac protocol for wireless body area
network. China Communications, 12(2), 11–20.
Liu, J., Sun, G., Zhao, D., Yao, X., & Zhang, Y. (2012). MCU-Controlling based Bluetooth data transfer-
ring system. Procedia Engineering, 29, 2109–2115. doi:10.1016/j.proeng.2012.01.271.
Maharjan, B. K., Witkowski, U., & Zandian, R. (2014, September). Tree network based on Bluetooth 4.0
for wireless sensor network applications. Paper presented at the 6th IEEE European Embedded
Design in Education and Research Conference (EDERC), Milano, Italy.
Mansoor, N., Islam, A. M., Zareei, M., Baharun, S., Wakabayashi, T., & Komaki, S. (2015). Cognitive
radio ad-hoc network architectures: A survey. Wireless Personal Communications, 81(3),
1117–1142.
Marin-Perianu, R., Marin-Perianu, M., Havinga, P., Taylor, S., Begg, R., Palaniswami, M., & Rouffet,
D. (2013). A performance analysis of a wireless body-area network monitoring system for pro-
fessional cycling. Personal and Ubiquitous Computing, 17(1), 197–209.
Mayne, K. (2016a). Smarter cycling series: What’s the future for the bike industry in the world of
driverless cars? Retrieved from https:ecf.com/news-and-events/news/smarter-cycling-series-
what’s-future-bike-industry-world-driverless-cars.
Mayne, K. (2016b). Cyclists and public bike sharing—The best kept secret in Smart City data collec-
tion? European Cyclists’ Federation. Retrieved from https://ecf.com/news-and-events/news/
cyclists-and-public-bike-sharing.
Mesin, L., Aram, S., & Pasero, E. (2014). A neural data-driven algorithm for smart sampling in
wireless sensor networks. EURASIP Journal on Wireless Communications and Networking,
2014(1), 1–8.
Mitola, J., Guerci, J., Reed, J., Yao, Y.-D., Chen, Y., Clancy, T. C., … McGwier, R. (2014). Accelerating 5G
QoE via public-private spectrum sharing. IEEE Communications Magazine, 52(5), 77–85.
Mraz, L., Cervenka, V., Komosny, D., & Simek, M. (2013). Comprehensive performance analysis of
zigBee technology based on real measurements. Wireless Personal Communications, 71(4), 2783–
2803. doi:10.1007/s11277-012-0971-1.
Naeem, M., Illanko, K., Karmokar, A., Anpalagan, A., & Jaseemuddin, M. (2013). Energy-efficient
cognitive radio sensor networks: Parametric and convex transformations. Sensors, 13(8),
11032–11050.
Nagarajan, R., & Dhanasekaran, R. (2013, April). Implementation of wireless data transmission in moni-
toring and control. Paper presented at the International Conference on Communications and
Signal Processing (ICCSP), Melmaruvathur, India.
Nechibvute, A., Chawanda, A., & Luhanga, P. (2012). Piezoelectric energy harvesting devices: An
alternative energy source for wireless sensors. Smart Materials Research, 2012, 13.
Nintanavongsa, P., Naderi, M. Y., & Chowdhury, K. R. (2013, April). Medium access control protocol
design for sensors powered by wireless energy transfer. Paper presented at the Proceedings of IEEE
INFOCOM, Turin, Italy.
Olieman, M., Marin-Perianu, R., & Marin-Perianu, M. (2012). Measurement of dynamic comfort in
cycling using wireless acceleration sensors. Procedia Engineering, 34, 568–573.
Padgette, J., Scarfone, K., & Chen, L. (2012). Guide to Bluetooth security. NIST Special Publication,
800, 121.
Qin, T., Meng, Y., Li, L., Wan, H., & Zhang, D. (2014). An energy-saving algorithm for wireless
sensor networks based on network coding and compressed sensing. China Communications,
11(13), 171–178.
Ramakrishnan, S., & Krishna, B. T. (2014, February). Closed loop fuzzy logic based transmission power
control for energy efficiency in wireless sensor networks. Paper presented at the International
Conference on Computer Communication and Systems, Chennai, India.
Rasouli, H., Kavian, Y. S., & Rashvand, H. F. (2014). ADCA: Adaptive Duty Cycle Algorithm for
energy efficient IEEE 802.15.4 beacon-enabled wireless sensor networks. IEEE Sensors Journal,
14(11), 3893–3902. doi:10.1109/JSEN.2014.2353574.
Rault, T., Bouabdallah, A., & Challal, Y. (2014). Energy efficiency in wireless sensor networks: A top-
down survey. Computer Networks, 67, 104–122.
Ravindranath, M. (2014). Cisco CEO at CES 2014: Internet of things is a $19 trillion opportunity. The
Washington Post. (accessed on January 8, 2014). Retrieved from http://www.washingtonpost.
com/business/on-it/cisco-ceo-at-ces-2014-internet-of-things-isa-19-trillion-opportunity/2014/
01/08/8d456fba-789b-11e3-8963-b4b654bcc9b2_story.html
Rawat, P., Singh, K. D., Chaouchi, H., & Bonnin, J. M. (2014). Wireless sensor networks: A survey on
recent developments and potential synergies. The Journal of Supercomputing, 68(1), 1–48.
Roman, R., J. Zhou, and J. Lopez, On the features and challenges of security and privacy in distributed
internet of things. Computer Networks, 2013. 57(10): p. 2266-2279.
Rosas, F., & Oberli, C. (2012, May). Modulation optimization for achieving energy efficient communications
over fading channels. Paper presented at the IEEE 75th Vehicular Technology Conference (VTC
Spring), Yokohama, Japan.
Rout, R. R., & Ghosh, S. K. (2013). Enhancement of lifetime using duty cycle and network coding in
wireless sensor networks. IEEE Transactions on Wireless Communications, 12(2), 656–667.
Ruiz-Garcia, L., Lunadei, L., Barreiro, P., & Robla, I. (2009). A review of wireless sensor technologies
and applications in agriculture and food industry: State of the art and current trends. Sensors,
9(6), 4728–4750.
Sendra, S., Lloret, J., García, M., & Toledo, J. F. (2011). Power saving and energy optimization tech-
niques for wireless sensor networks. Journal of Communications, 6, 439–459.
Setiawan, M. A., Shahnia, F., Rajakaruna, S., & Ghosh, A. (2015). ZigBee-based communication system
for data transfer within future microgrids. IEEE Transactions on Smart Grid, 6(5), 2343–2355.
Shin, H.-Y., Un, F.-L., & Huang, K.-W. (2013, July). A sensor-based tracking system for cyclist group. Paper
presented at the Seventh International Conference on Complex, Intelligent, and Software
Intensive Systems (CISIS), Taichung, Taiwan.
Skylock. The smartest bike lock. Retrieved from www.skylock.cc.
Smart Pedal. The connected bike enabling solution. Retrieved from connectedcycle.com/pedal.html.
Smith, P. (2011). US Patent CS-213199-AN. Comparisons between low power wireless technologies.
A&V Elettronica: Italty.
Soini, M., Nummela, J., Oksa, P., Ukkonen, L., & Sydänheimo, L. (2008). Wireless body area network
for hip rehabilitation system. Ubiquitous Computing and Communication Journal, 3(5), 42–48.
Solos. Performance eyewear. Retrieved from www.solos-wearables.com.
Srisooksai, T., Keamarungsi, K., Lamsrichan, P., & Araki, K. (2012). Practical data compression in
wireless sensor networks: A survey. Journal of Network and Computer Applications, 35(1), 37–59.
SRM. Retrieved from http://www.srm.de.
Sudevalayam, S., & Kulkarni, P. (2011). Energy harvesting sensor nodes: Survey and implications.
IEEE Communications Surveys & Tutorials, 13(3), 443–461.
Tabish, R., Ben Mnaouer, A., Touati, F., & Ghaleb, A. M. (2013, November). A comparative analysis of ble
and 6lowpan for u-healthcare applications. Paper presented at the 7th IEEE GCC Conference and
Exhibition (GCC), Doha, Qatar.
Talele, A. K., Patil, S. G., & Chopade, N. B. (2015, January). A survey on data routing and aggregation tech-
niques for wireless sensor networks. Paper presented at the International Conference on Pervasive
Computing (ICPC), Pune, India.
Thielen, D., Lorenz, T., Hannibal, M., Köster, F., & Plättner, J. (2012, September). A feasibility study
on a cooperative safety application for cyclists crossing intersections. Paper presented at the 15th
International IEEE Conference on Intelligent Transportation Systems, Anchorage, AK.
USDOT. (2010). The national bicycling and walking study: 15-year status report., FHWA-PD-94-023.
Washington, DC: Federal Highway Administration.
VeloCittà. Better use of bicycle share systems. Retrieved from http://velo-citta.eu/.
Walker, W., Aroul, A. P., & Bhatia, D. (2009, September). Mobile health monitoring systems. Paper pre-
sented at the 2009 Annual International Conference of the IEEE Engineering in Medicine and
Biology Society, Minneapolis, MN.
Wang, J. (2013). Zigbee light link and its applications. IEEE Wireless Communications, 20(4), 6–7.
Wegman, F., Zhang, F., & Dijkstra, A. (2012). How to make more cycling good for road safety?
Accident Analysis & Prevention, 44(1), 19–29.
WHO. (2015). Global status report on road safety, NLM classification: WA 275. Geneva, Switzerland:
World Health Organization.
Wong, A. W., Dawkins, M., Devita, G., Kasparidis, N., Katsiamis, A., King, O., … Burdett, A. J. (2013). A
1 V 5 mA multimode IEEE 802.15.6/Bluetooth low-energy WBAN transceiver for biotelemetry
applications. IEEE Journal of Solid-State Circuits, 48(1), 186–198. doi:10.1109/JSSC.2012.2221215.
Xie, L., Shi, Y., Hou, Y. T., & Lou, A. (2013). Wireless power transfer and applications to sensor
networks. IEEE Wireless Communications, 20(4), 140–145.
Yan, R., Sun, H., & Qian, Y. (2013). Energy-aware sensor node design with its application in wireless
sensor networks. IEEE Transactions on Instrumentation and Measurement, 62(5), 1183–1191.
Yang, C.-M., Wu, C.-C., Chou, C.-M., & Yang, C.-W. (2009, November). Textile-based monitoring system
for biker. Paper presented at the 9th International Conference on Information Technology and
Applications in Biomedicine, Larnaca, Cyprus.
Yang, Y., Yeo, J., & Priya, S. (2012). Harvesting energy from the counterbalancing (weaving)
movement in bicycle riding. Sensors, 12(8), 10248–10258.
Yongtai, H., Lihui, L., & Yanqiu, L. (2012). Design of solar photovoltaic micro-power supply for
application of wireless sensor nodes in complex illumination environments. IET Wireless Sensor
Systems, 2(1), 16–21.
Yoshida, S., Noji, T., Fukuda, G., Kobayashi, Y., & Kawasaki, S. (2013). Experimental demonstra-
tion of coexistence of microwave wireless communication and power transfer technologies for
battery-free sensor network systems. International Journal of Antennas and Propagation, 2013, 10.
Yu, K.-M., Zhou, J., Yu, C.-Y., Liu, J.-Y., Lee, C.-C., Chang, H.-W., & Hsieh, H.-N. (2009, December). An
event-based wireless navigation and healthcare system for group recreational cycling. Paper presented
at the 5th International Conference on Mobile Ad-hoc and Sensor Networks, Fujian, China.
Zebbane, B., Chenait, M., & Badache, N. (2015). A group-based energy-saving algorithm for
sleep/wake scheduling and topology control in wireless sensor networks. Wireless Personal
Communications, 84(2), 959–983.
Zhan, B., & Yu, X. (2013). Wireless node design in smart home system. In Y. Yang & M. Ma (Eds.),
Proceedings of the 2nd International Conference on Green Communications and Networks (GCN 2012)
(Vol. 3, pp. 545–552). Berlin, Germany: Springer.
Zhang, P., Sun, L., Zhang, P., Hou, R., Tian, G., & Liu, X. (2013, November). Wireless network design and
implementation in smart home. Paper presented at the 6th International Conference on Intelligent
Networks and Intelligent Systems (ICINIS), Shenyang, China.
13
Self-Adaptive Cyber-City System
Iping Supriana, Kridanto Surendro, Aradea Dipaloka, and Edvin Ramadhan
CONTENTS
13.1 Introduction: Cyber-City Systems and Seven Pillars of Life ..................................... 295
13.2 The Basis of System Development ................................................................................. 296
13.2.1 Representation of System.................................................................................. 297
13.2.2 Construction of Model ...................................................................................... 298
13.3 Self-Adaptive Model ........................................................................................................300
13.3.1 Rule Representation........................................................................................... 301
13.3.2 Structure of Knowledge Base ...........................................................................304
13.3.3 Reconfiguration Strategy ..................................................................................306
13.3.4 Component-Based Software ............................................................................. 310
13.4 Case Study of Cyber-City System .................................................................................. 311
13.4.1 Modeling of System ........................................................................................... 311
13.4.2 System Configuration ........................................................................................ 314
13.5 Discussion and Conclusion ............................................................................................ 316
13.5.1 Discussion ........................................................................................................... 316
13.5.2 Conclusion .......................................................................................................... 318
References..................................................................................................................................... 319
13.1 Introduction: Cyber-City Systems and Seven Pillars of Life

The growth of urban population in the world nowadays encourages the growth of the
level of needs which is increasingly out of control. This is based on the fact that the level
of knowledge in the population continues to increase, so (ITU-T, 2014) it led to the social,
economic, and environmental issues becoming connected to each other. This fact shows
that the characteristics of the system to control a city has the diversity of interrelated
elements and will lead to a variety of challenges and problems in the provision of public
services (PSn).
Therefore, the city management needs to have an intelligent solution and creates
a sustainable environment to (ITU-T, 2014) manage various infrastructure resources,
environmental resource, monitoring the activities in the city, and more needs. All of this
management will relate to the system requirements of the city-management system,
such as transportation management, energy management, water management, waste
management, municipal administration management, health services management,
and other PSn management.
295
Internet of Things (IoT) is a concept that can answer the challenges. Nowadays, there
are several definitions of it, but basically, they have the same goal to extend the benefits
of Internet connectivity that can connect a variety of real-world objects as physical and
virtual representation continuously. One of the implementation of IoT for the needs of
city-management system is how to utilizing the functions of information and communica-
tion technology is developed on a concept called the cyber city or smart city or intelligent
city or some other term, that basically aimed to meet the requirement of city manage-
ment development through the systematic formulation. The proposed system is packed
with various facilities, such as the efficiency and effectiveness of implementation, speed
of access, accuracy of services, and others. But this various advantages will not be able to
make a significant contribution if only temporary. This is related to the detail and com-
pleteness of the concept to meet the characteristic of the growth in a city.
In this discussion, we will describe a model about how to establish a system that can
accommodate a variety of changes in the application environment, how to manage the
change, and how to make the system has the capability to adapt and fit the changes. So that
this model can be an inspiration in developing a scenario of systems that needed to man-
age a city. The model that has been designed inspired by the seven pillars of life (taklif)
(Nabulsi, 2010), which is composed of the universe, law, reason, nature, lust, freedom of
choice, and time. These seven pillars are represented through an agent-oriented approach
BDI (belief, desire, intention). So the points of the inspiration are, we need to consider
which part is associated with the domain model and which part is associated with control
model in building the management software system, in this case, the city-management
system. It is expected to create a computational model that can be used as guidance in
developing a system of cyber-city systems and other IoT software systems in general, with
the self-adaptive ability which can handle the issue of change and growth of systems.
13.2 The Basis of System Development

Self-adaptive systems (SAS) is presented as an alternative solution to the problems
associated with the complexity of the IoT system, including the demands of autonomy,
automation, adaptability, flexibility, scalability, reliability, speed, and others (Supriana and
Aradea, 2015). SAS is a system that can automatically take appropriate action based on
the knowledge that the system has about what is happening in the system itself, guided
by the goal, and assisted by the users who are given the access (Ganek and Corbi, 2015).
SAS can modify the system behavior in response to the changes in the system itself or the
changes in its environment (Cheng et al., 2014). This definition shows that the system has
the knowledge, which can be used to achieve its objectives through several means, and is
able to make the adaptation in behavior based on events in the environment.
The characteristics of each element in the environment of an IoT-management system
require a mapping and alignment with the system that will be the ultimate driving
machine in the achievement of a city manager. So (Aradea et al., 2014), all of the activity that
occurs in the environment of this IoT system requires a mechanism that can represent the
behavior of the system in real-world change. SAS concept was developed as a solution to
this problem. The constructed cyber-city system must have the capability to reflect the
requirements of each element of a city, to the adaptability of changes in the system and its
environment.
Self-Adaptive Cyber-City System 297
13.2.1 Representation of System

Basically, the universe was created for man to be given the mandate as the leader on earth,
who is able to organize and manage all the resources of the earth. This practice is called
the taklif (Nabulsi, 2010). The process of taklif comprises seven pillars (Nabulsi, 2010),
namely (ITU-T, 2014) (1) universe, (2) reasonable, (3) nature, (4) religion or laws, (5) lust,
(6) the freedom to choose, and (7) time. The explanation of all seven pillars of this taklif as
can be seen in Table 13.1.
TABLE 13.1
Seven Pillars of the Taklif System
Pillars Explanation
The Universe All content of the earth, including the behavior of the components of the universe, can be
grouped into two kinds of roles in the taklif system, that is, as the knowledge and as
utilization. Human reasoning works on the principles of harmony with the universe and
the principles of law. There are three principles of human reasoning, namely the principle
of cause and effect (something happens for a reason), the principle of goal (something exists
must be the goal), and the principle of anticontradictory (two things can apply when they
are opposites).
Reasonable The human mind will process two facts, namely the fact sensuous (reality) and logical
facts (explanation). The logical fact has a higher position than the sensuous fact. Humans
have the potential to think broadly but limited by the information provided by the
senses. There is a lot of information about the past and the future that cannot be reached
by the senses. If we relate this thinking to the concept of religion, we can state that
religion is the only thing that can be delimiters in the laws or reason.
Nature Naturally, humans were created with moral instincts. This moral instinct that defines the
rules of human life. But there are some human attitudes which are contrary to this rule, this
attitude appears because of character flaws and physical to the human. This weakness
makes the natural rule to be violated, distorted, and even erased. This weakness is a
challenge for humans, this rule can be restored to its natural state through learning or
management that is able to organize and disenchant the people to respect the rule of nature.
Religion or Laws Religion is a concept of how people should behave toward the creator, fellow human
beings, and its natural surroundings. The truth of human thought and the natural order
must be calibrated with the rules of religion; truth in religion is generally considered an
absolute, whereas reason and nature can be distorted by lust. To determine the truth as
the basis of the rule, the truth must meet four requirements, which is carried by religion
or laws, approved by human reasoning, in accordance with nature, and confirmed by
objective reality.
Lust Lust is the desire to achieve the goal, with their desire for a need for pleasure and love of
something inherent in man, the lust will make him it will strive to achieve it. But lust is
neutral. Lust is controlled by the choice of reason and nature, whether just as satisfying a
need or as a means of achievement of objectives as stated in the rules of religion or laws.
Freedom of Choice Freedom of choice is central to the taklif system. In this case, the religious rules known as
the belief which acts as the basis of the freedom to choose. In this world, there are many
choices and each choice has its own consequences. Religious rules and nature that will
be a reference for these choices.
Time All interactions of the six pillars of the taklif system which have been described previously
run in the dimension of space and time. Its space is the earth and the universe, whereas
the time is a series of events recorded since humans are born until his death.
Source: Nabulsi, R. M. 7 Pilar kehidupan: Alam semesta, akal, fitrah, syariat, syahwat, kebebasan memilih, dan waktu,
Gema Insani, Jakarta, 2010; Agustin, R. D., Supriana, I., Model Komputasi Pada Manusia dengan
Pendekatan Agent, Kolaborasi BDI Model dan Tujuh Pilar Kehidupan sebagai Inspirasi untuk
Mengembangkan Enacted Serious Game, Konferensi Nasional Sistem Informasi (KNSI), ITB, 2012.
13.2.2 Construction of Model

Based on the model of the representation of system in Section 13.2.1, the construction model
of developed software in the IoT concept, referring to the seven pillars of the taklif system,
is mapped to the BDI Model (Bratman, 1987; Mora, 1999; Russel, 2003). A more detailed
description can be found in the paper (Agustin and Supriana, 2012). The BDI Model is a
theory of reasoning from a practical standpoint (practical reasoning) and were adopted in
an agent-oriented architecture.
There are two main processes in practical reasoning, namely deliberate and means-end
reasoning. Deliberate is the process of deciding specific circumstances to be achieved, called
the intention. The means-end reasoning is the process to draw up a plan to achieve that
intention. The deliberate process begins with understanding the stimulus from the environ-
ment using the belief, that the perception of the agent (which is assumed to be true) on the
environment. The outcome of this process is the captured problem classification by the agent
at the detected time. Based on these results, the system will search for the best reaction to solve
the given problem. In the desires of the BDI model, these solutions are emerging as an alter-
native option which generated sporadically, so the various alternative solutions appear; but
because it is generated sporadically, there may be a solution that is unrealistic or inconsistent.
Before adding a goal to the state of intention, all alternative choices in the state for this
goal needs to be verified and validated, to be realistic with the condition of the agent,
and consistent with the objectives set (Mora, 1999). Furthermore, in the state of inten-
tion, the system has obtained some options, to be realized as an action. In the end, the
mental state will generate an action into some subactions to be executed by the effector.
The mapping of these BDI models of the seven pillars of the taklif system is shown in
Figure 13.1.
1. Pillars of components in the universe are the facts of each entity and its environ-
ment. The behavior of the components in the universe in the form of events can
be captured by the information-model context. In this case, there are two kinds of
roles, namely the introduction of knowledge and utilization, which means how to
recognize and utilize the facts.
2. The next component is the representation of the pillars of sense. Pillars of sense
mean to know the nature and to work based on the principles that correspond to
the goal, nature, and religion or law (rules of how to behave/delimiters sense).
This component evaluates the condition by processing the facts of reality and
logical facts based on occurred events. In this evaluation phase, the facts will be
processed, diagnosed, and classified based on the problem, so we can determine
how to choose understanding.
3. The processing of facts would trigger some response that will be taken by the action
component in the process of adaptation. There will be some choices in response
to the prevailing condition in accordance with the rule of lust (functional require-
ments and nonfunctional requirements), and to generate the response options, it
will be processed by the system itself without being given the freedom to choose.
4. Components of sense will choose several response options in accordance with
the model of rules in religion or law (as described). In this process, selection and
filtering are performed between some alternatives of priority response, so that the
results are just a few alternatives with the highest priority.
6. Freedom of choice
Belief Desire
Conscious-
Pleasure
Sense ness Unconscious-
inference
Sequence of experience inference knowledge ness
engine
or special experience engine/ about knowledge
(Given)
independent universe (Lust)
instinct
environment
5. Lust
Dependent and be
1. The universe forced to sense
Environment Body and rational Independent and Goal

(fact-1... fact-n) inference engine full consciousness
Learning by Conscious- Rational
rational and ness inference
Unconsciousness knowledge engine/
knowledge happiness
Reward and inference engine about independent
(Nature) religion /reasonable
punishment (Reasonable and
experience nature)
Happiness 3. Religion 2. Reasonable
inference engine
Plan Intention
Conscious-
Planner Happiness
ness Conscious-
engine/ inference
Action affect to knowledge ness
independent engine/
environment (How to knowledge
(Reasonable independent
achieve about fitrah
and effector) (Nature)
Arrow of time intention)
7. Time 4. Nature
FIGURE 13.1
The mapping of the seven pillars in the taklif system into the BDI model.
5. Representation of the qolbu components will filter out some of the alternate
options again from remaining responses, in reference to the understanding of
nature’s rule (instinctive understanding of the religion or the law of nature) until
the obtained result is the best response that will be implemented.
6. Based on the selected response, some planning for the adjustment process can be
arranged, and the effector of the body part system to execute the selected response
can be determined in a certain order. The result of the execution of the effector is
called as action.
7. The actions taken will get a reward or punishment based on the accuracy of
the selected response with religious rules and nature should be. This stage will
be a learning process for the system and will become a trigger for updating
knowledge so that the system can determine the best response action in the
future.
The characteristics of the seven pillars of the taklif system in conducting adapta-
tion response are shown in Figure 13.1. The pillars are represented by 4 major rules in
3 BDI states, which is desires (R1), goal (R2 and R3), and the intention (R4); the states will
determine the understanding of belief. The development strategy for each of these rules
are as follows:
1. Rule-1 (generic structures): Represent the real world in the form of abstract classes
to model the goal.
2. Rule 2 (conceptual rule): Identifying the generic structure that meets the basic
specifications of fact; this rule will also guide the system in determining an
alternative initial option based on the understanding of belief to environmental
stimuli.
3. Rule-3 (configuration rule): Define the configuration of the components, verify and
validate an alternative option of conceptual results of operation rule.
4. Rule 4 (specific rule): A special rule that is used if needed or if operating results
do not meet certain criteria to obtain one or more options that are most relevant
actions.
13.3 Self-Adaptive Model

This section will discuss the proposed model to build software systems with self-adaptive
capabilities in the IoT concept. The framework was developed as shown in Figure 13.2,
consisting of two main parts, as follows:
1. The domain model is represented as goal model. The domain model is a compo-
nent that provides the IoT concept’s basic functions and application logic. This
model was developed based on goal-oriented approach, which can exploit the
human-oriented abstractions such as agents and some other concepts, so it can
represent real-world conditions.
2. The control model is represented as inference engine that controls or manages
the target system through the adaptation logic. These models apply some
patterns of action of the agent through the transition rules. The control model
considers context-aware scenarios to represent a control strategy to meet the
requirements of self-adaptation.
The domain model is a requirement of the IoT-goal system. Domain models are mapped
to the BDI models for defining goals, planning, and other elements. The results of the
mapping are represented by agent definition file and the plan specification. The concept
of this agent represents the context of the IoT system as a fact in belief base and environ-
ment class for the purposes of monitoring functions (Supriana and Aradea, 2016). This
approach quite provides the variability at run-time, but we extend it through the mapping
into the BDI models, which is based on seven-pillar taklif with rule-based, which are con-
structed in a more general and more flexible, so it can provide a better way of analyzing
the variability and can help in detailing the behavior of the system to meet the goals and
adaptation action.
The control model is the setting of the behavior of the IoT system, tasked to monitor
the environment and adjust the system if necessary, for example, reconfigure when
there is a change goal, optimize themselves when the operation changes, and able
Self-∗
- Agent
Mapping and definition file
updated model - Plan
Execute Action
specification
Goal model Inference engine
Context
NRules
Mapping and Editor

Knowledge
updated model - Generic structure
- Conceptual rule
Environment - Configuration rule
classes - Specific rule
FIGURE 13.2
Framework for self-adaptive systems.
to handle certain types of errors. The pattern applied is an extension of the architec-
ture event-condition-action (ECA) which in one or more event refers to the state of the
current system. The models will change the application or IoT-system environment.
Meanwhile, one or more condition refers to the time when a particular event occurs
and the action rule is activated. So one or more action can be activated under certain
conditions through the operation of rules in determining the behavior of the adaptive
system.
13.3.1 Rule Representation

Any changes occurred in the IoT system is seen as a model of automata and the relation-
ship of the possible state. Those changes can occur from an initial state until it reaches the
final state. So the description of automata can be expressed as the model of rules which
describes the relationship between the two states.
The representation of rules of these designed automata models, automatically, can detect
the changes in the states and determine the necessary actions for adaptations. Scenarios
are developed through the formation of the rule to identify similarities in the properties of
each list of knowledge. In general, the rule-based system consists of 11 classes (Wu, 2004)
as follows: rule, left-hand side (LHS), expression, fact, slot, composite expression, unary
expression, binary expression, right-hand side (RHS), action, and function, as shown in
Figure 13.3.
(nEn, vENn) nE1

nE3
vEN2
nE2 vEN1
(nEn, vENn)
nE4
nE2.1 vEN1.1
vEN1.2
(nEn, vENn)
1 -hasRHS
∗ Rule ∗
-ruleName
-hasLHS 1
LHS -description RHS
1
1 -operand-A 1 -consistsOf
Expression
-operand-B
0..∗
-operand
CompositeExpression Action
Fact 1
-factName 1
BinaryExpression
1 -holds
1..∗
Function
Slot UnaryExpression
-functionName
-slotName
-parameter
-slotValue
FIGURE 13.3
Knowledge tree and rule-based systems.
This rule-based system has a production rule in the form if condition > then action >.
Action > on the RHS is concrete. This production rule is a practical consequence of a par-
ticular condition, whereas the other rules in the rule base are known as derivation rule.
Conclusion > on the RHS is more abstract and becomes the logical consequence of certain
conditions. However, a production rule can apply derivation rule using certain actions such
as assert that expressed knowledge.
An action > defined on the RHS is determined by the expression on the LHS, which is
conformity between rules and facts. Therefore, to create a self-adaptive ability, the rule-
based system should be expanded as discussed in Section 13.3.2. The main objective is to
equate the LHS in accordance with the formulation of the property slot, so there is a set of
inner properties that become the source of facts, such as variables, constants, tuples, and
so on. The basis of the mechanism is comparing the two or more collections of properties
with a certain structure based on the criteria required.
In Figure 13.3 (bottom), we can see that the LHS is an expression that can be a fact,
a single expression (pattern) which is characterized by a name and a collection of
slots, or may be a composite expression with conditional elements (and, or, not). The
conditional element is also used to connect a single expression (facts) or a composite
expression. Class of fact has a containment reference in the class slot with the attribute
“slotName” and “slotValue.” The changes in this slot attribute will be detected based
on any of the elements contained in the facts. For example, in the structure of the fol-
lowing basic rule:
if ( fact-1) ( fact-2 )( fact-N ) then ( action-1)( action-N ) (13.1)
Information models of facts have variations on slotName and slotValue based on the defined
information context. The information models are defined to be a relation between entities in
the system and the environment. Information models are also represented as an intercon-
nected class and the facts in working memory. Thus, a system entity would have a slotName
such as “nameEntity” (nE1, nE2, nE3 … nEn) and has a view component for the environment
as slot “viewEnvironment” (vEN) and can be composed of other facts, for example (vEN1,
vEN2, vEN3,… vENn,), and each slot is possible to have more specific (vEN1.1, vEN1.2, …
vEN1.n), and the specification is also very possible to have other new slot. Representation of
fact is a list of knowledge owned by each entity, and we define the list of this knowledge in a
graph form or tree structure of knowledge, as we can see in Figure 13.3.
The tree structure of knowledge will add, delete, update, and customize every vertex
automatically in response to the occurred state changes in accordance with the facts. The
structure of property description of the knowledge tree can be managed by mapping the
patterns of the seven pillars of taklif into the BDI model. So the representation of rules will
be a reference for developed adaptation mechanisms to make a rule expansion of the ECA
model (Daniele, 2006; Pires et al., 2008). This model can handle a wider scope of the solu-
tion space, and flexible, by using the following basic structure:
WHEN < event > ; one or more transition state
IF <condition>; conditions that must be met to trigger action

(13.2)
THEN <action>; one or more actions when the event takes place
VALID TIME <time period>; period of adaptation actions

Operator selector
Knowledge
Tuple :
L : LHS R : RHS a. Context/fact (∑) : f1, f2 ... fn
NRules
b. Behavior/action (Q) : a1, a2 ... an
(q0 ... ... ... F) c. Transition (δ) : t1, t2 ... tn
L R L R L R d. The initial state (q0) and final state (F)
t1 t2 tn Rule based structure :

WHEN <event> ;
Contexts Behavior
IF <condition> ;
f1 a1
THEN <action> ;
f2 a2 VALID-TIME <time period> ;
. .
. . Note :
fn an LHS : Left-Hand Side
RHS : Right-Hand Side
Sensor Actuator
Environment
(fact-1 ... fact-n)
FIGURE 13.4
Model of rule representation.
Based on these descriptions, to define the needs of models, we set a tuple of the system,
which are as follows: (a) Facts context information (∑: f1 … fn) as the set of inputs, (b) Action
of system behavior (Q: a1 … an) as a set of finite status, (c) Transition function or opera-
tor (δ: t1 … tn) as a function of the change, (d) Preliminary data (q0) as initial status, and
(e) information targets status as the set of final status.
Figure 13.4 describes the relationship between variables and functions of the developed
model. Information context is a set of inputs to be monitored as an event of the facts from
any context information. The condition will be evaluated with reference to a specific event
that occurs within the set of states, and it will trigger the adaptation action through the
transition function of the prevailing condition; so the status of the target can be achieved
by the action of the expected behavior.
Descriptions of the model are as follows:
1. ∑: Context information in the form of a set of facts (f1 … fn) is captured from the
event environment and can determine the change in status.
2. Q: Behavioral actions (a1 … an) form the set of behaviors status of the system,
starting from the initial state (q0) to the status of which can be targeted (F).
3. δ: Transition function (t1 … tn) as time of action adaptation based on the evalua-
tion of the condition, to change the initial status (q0) to the status of the target (F).
13.3.2 Structure of Knowledge Base

With reference to the representation of prepared rules, any change in the IoT system state
can be expressed as series of a node changes (Supriana et al., 1989); each situation will be
modeled into a node, and changes of state will be described as the arc which connecting
two nodes. So the used structure to handle the dependencies between the elements of
knowledge is a directed graph.
The motivation for this development of the knowledge base structure is the reasoning
that the approach of rule-based systems in most applications only focused on the rules of
behavior of the software that is designed specifically, while the problems that may arise
relate to the variability that can be served. This problem triggers the idea to formulate a
model of the system structure to be more common and flexible, can be applicable to a wide
scope of variability in the system and its environment, and expanded flexibly. The devel-
opment of this model is targeted to meet the needs of the system to capture the variability
in context and behavior of the system itself and its environment. This model can also set
the criteria of assurance for the system management and its adaptation mechanisms.
In the design of the developed structures, each node in the graph has some slot cat-
egories, and each category has a number of attributes that can be increased, as shown in
Figure 13.5.
The basic mechanism for knowledge management is prepared to meet the scope of a life
cycle system: birth (create), use (use), updated (update), and deleted (delete). Developed-
management module consists of initiation slot, expansion slots, adjustment slot, and
inspection slot. The searching mechanism in a knowledge space is performed by using the
access to the slot module, control of the trajectory, and the search delay. This mechanism
can handle input consisting of context information, change and the output as behavioral
adaptation actions of the sensors and actuator module. Overview model development can
be seen in Figure 13.6.
Category Slot Slot Slot

?
Node
1
? ?
Node
2
Node x
Node y
Node
n
FIGURE 13.5
Structure of knowledge base.
Knowledge base structure

Operator selector Category Slot Slot Slot
Knowledge ?
Node
L : LHS R : RHS 1
NRules
? ?
(q0 ... ... ... F)
L R L R L R
t1 t2 tn
Node
Contexts Behavior
2
f1 a1
f2 a2
. . Node x
. . Node y
fn an
Node
Sensor Actuator n
Environment
(fact-1 ... fact-n)
FIGURE 13.6
Representation and structure of knowledge base.
13.3.3 Reconfiguration Strategy

The main objective of this strategy is to define the domain model and determine the most
relevant adaptation response options. The notation used to construct this algorithm con-
sists of the following:
1. A model of goal (G) consists of connected nodes by its property attributes (P); each
node also consists of a number of states (Sn), which may have a contribution to a
soft goal.
2. Some states (Sn) may consist of the initial state (q0) and the status of the target (F);
this state is influenced by several processing of facts (∑: fn) on the LHS, which will
determine the behavior of the action (Q: an) and on the RHS, through the transi-
tion function (δ: tn)
3. The process of recognition performed by observing ∑: fn as a trigger for Sn in each
G, until S can be set as a reference of plan preparation to realize a number of Q: an,
including its mapping to the component (C).
To realize the construction of the system and the need for changes, we define four basic
operations (create, read, update, and delete). This operation is mapped to the operation of
the components (Kramer & Magee, 2007) that can be seen in Chart 13.1.
In addition, we define the rule that represents each element in the components of goal
systems. This rule is compiled based on the rules of goal decomposition (Nakagawa et al.,
2008); if the goal decomposition is AND-decomposition, then the parent goal will require
multiple attributes of the relation (port) for each child goal with a one-to-one degree of
relationship. But if the goal decomposition is OR-decomposition, then the parent goal
will provide the conditional attributes of the relation (port) to each goal, with the one-to-
many degree of relation. In this activity, we also need to set the properties for each goal
(functional) and a soft goal (nonfunctional). Chart 13.2 illustrates the algorithm to define a
primitive component based on the goal model.
Chart 13.1 Configuration Algorithm

Components Configuration
G ← (goal, soft goal)
C ← (components)
for all N in goalModel do
G ← C: configuration components for operation
for each (∑, Q) ≠ ø do
(create, read, update, delete) ← DiagnoseNode (G)
create ← create component instance C from type G
read ← connect or disconnect C1 to C2
update ← set mode of C
delete ← delete component instance C
end for
goalModel m ← reconfiguration(create, read, update, delete)
enactModel(m)
end for
Chart 13.2 Diagnosis Algorithm

Goal Diagnosis
for all G in m do
G ← addProperty(P)
if G decomposition = AND-decomposition then
G parent ← add multiplePort // sum of G childrenPort
for all G in G children do
port ← add providedPort
C configuration ← DiagnoseNode
else if G.decomposition = OR-decomposition then
G parent ← add conditionalPort
for all G in G children do
port ← add conditional providedPort
C configuration ← DiagnoseNode
end for
end if
end for
end if
end for
The development-control strategy to observe and regulate every component of the system
that has been defined, by using the design pattern (Abuseta and Swesi, 2015) that was
inspired by the model of MAPE-K (IBM, 2005), and modified in accordance with the pre-
vious system requirements. Chart 13.3 shows the algorithm to monitor and analyze the
needs of the adaptation plan.
Chart 13.3 Observation Algorithm

States Observation
for all G in m do
m ← (∑ fn) // at run-time
G ← getValue(P) // time triggered or event triggered
for each value(S) in P do
S ← combining internal and external value(P)
if S in S.target(F) ≠ P.threshold then
systemState ← new S.system(S.instance) and
systemStateLog ← save(S.system) and
send information(S.system) to analyzerManager
end if
for each S.system in analyzerManager do
analyzer ← update(logs) actual S.system
search(S.system) in symptomRepository
if symptom ≠ ø then
create(adaptationRequest) and
update(adaptationRequest) for plan specification
else
addSymptom to symptomRepository and
create(adaptationRequest) and
send information(adaptationRequest) for plan specification
end if
end for
end for
end for
There are a number of properties on the model of goals that must be read and measured
(observation) in concurrency. In the Java programming language, multithreading tech-
niques are relevant to meet this need. This activity represents the state system at run-time
by using time-triggered or event-triggered; this process is performed in response to any
request or event. The state of the system at run-time is represented by a combination of the
internal and external property value; the desired state is directed by a goal and a soft goal.
Violations of the state detected by the threshold level of each goal property and a new run-
time system state are stored in the system state logs for analysis.
Violation of the goals and requirements of the system is analyzed based on the symp-
tom repository. The symptom repository is a collection of symptoms that has been set for
the system to avoid and heal itself. This component is part of the knowledge base. The
symptom repository is equipped with facilities to add new symptoms that occur during
runtime analysis through operating add symptom. The symptom class consisting of an
associative array is used to store any symptoms, which represent each event of symptoms
and its value represents the condition relating to the event. Some programming techniques
for this class and interfaces are (Abuseta and Swesi, 2015) map (Java), dictionary (Python),
and associative arrays (PHP). If the analysis detects the presence of symptoms, then the
plan component will receive a demand signal for adaptation and reconfigure the system
based on the engine policy. Chart 13.4 shows the algorithm reconfiguration plan.
The policy engine provides the high-level goals that control the operation and functions
of related systems. The general form is ECA rules; these rules are used to determine the
action when the event occurred and meet certain conditions. The policy engine is rep-
resented as a knowledge base, which provides an interface for system administrators to
determine and change the system policy. In our version, the engine is being expanded
with the model of rule editor, and also additions or changes in the specification can be per-
formed by editing the knowledge base directly, or reset back into the system.
Chart 13.4 Reconfiguration Algorithm

Reconfiguration Plan
for each adaptationRequest(S.system) do
init ← set work (∑, Q)
while δ {tn (fn, an) | n ≠ ø} do
δ ← find that the LHS of the operator match with work say it found
if found is only one then
RHS ← set work
else if work is equivalent with target then
stop succeed
end if
if found is more than one then
found ← set work one of the found
backs ← put rest of found
end if
if found is empty then
else if backs is empty then
stop failed
else
backs ← set work one of backs
end if
end if
for all δ(tn) is found do
a ← construct correctiveAction(addAction)
changePlan ← newChangePlan(an)
send changePlan to one or more executors
for each a in executor do
actuator ← update(an) // one or more actuators
S.system ← reconfiguration m with actuator
// set new value for C (DiagnoseNode)
systemStateLog ← saveState(S.system)
end for
end for
end for
Each request of adaptation will be represented as a state in the system, which detected
based on the occurred events. A set of δ can be expressed as δ {tn (fn, an) | n ≠ ø}, with
“fn” is the fact of context (the property of goal) and “an” is the behavior of the action that
expected to a specific contextual n. The quality of the inference engine depends on the
selection of state for the adaptive action on the RHS class. This quality is also determined
by the expression in the LHS class. The expression is a match between the rule and the
facts. For example, comparing two or more collections of properties with certain structures,
based on the required criteria. The used key strategies are the forward strategy, reusing
(reuse) the existing fundamental component, and matching the required specifications.
The changing plan contains the set of action for the execution component to perform
the adaptation action. This action should be carried out in some specific order, for exam-
ple, sequentially or concurrently or both. The execution component will use a number of
actuators to set new values of the property in the target system and its environment. The
corrective action of the execution component on a number of the actuator will bring the
system back to the desired state or an acceptable state, and then this state will be saved in
the log of the system state by the actuator.
13.3.4 Component-Based Software

In this section, a description of the reconfiguration strategy that has been discussed in
Section 13.3.3 will be implemented in the component-based software. The component is
a specific function unit which interacts with using the interface through the delivery of a
service. The component-based system is a collection of components which are connected
to each other, so it has the desired functions. In general, the interface between the compo-
nents is a standard form, so the components can be connected and have a good interoper-
ability. Based on the variety of this interface, the components can be divided into several
types, namely (a) the input component, (b) the process components, (c) output component,
and (d) the delivery components.
The input component is a component that only has an interface that generates
information. The process component is defined as a component which has a pair or
more connections between the input interface and an output interface. In general,
information for the output interface in the components of the process have a func-
tional relationship with the information on the input interface. Some of the functions
can be owned by any other process components, including structural, preparation,
grouping, and counting function.
Output component is a component that only has the interface that receives the output
information from the processing component. And the delivery component is a supporting
component that mediates the delivery between the input component and output compo-
nent. By using this delivery component, the system can become more compatible to be
accessed in the variety of platforms; it is mostly needed in a system with the concept of IoT.
Each component can provide independent services, connected or collaborated services,
and free active services. This independent service allows a component to perform its
function itself, actively, without requiring the presence of other components. The com-
ponents with the connected services or collaborated services are the component with
the interaction of two or more components connected. In particular, this process can
take place sequentially or in parallel, and the component of the free active services is a
component that is active and running but has not performed its function until there are
triggers that appear from any other active component.
One of the approaches that can be adopted to formulate the architecture of the software
components, for the needs of a cyber-city system in the concept of IoT, is to utilize
the architectural description languages, which was developed in the Darwin model
(Magee et al., 1996; Magee et al., 1996; Hirsch et al., 2006). This model is a declarative
component-based architectural description languages that supports a hierarchical model

and graphical modeling, easy to use in determining the software components, including
interconnection and structure by using formal modeling notation. The specification of
this software component architecture will be incorporated into the requirements of the
self-adaptive cyber-city system in Section 13.4, as a motivation of the case to provide a
more concrete.
13.4 Case Study of Cyber-City System

This section discusses the case about the development of the cyber-city system in the IoT
concept; the goal of this case is to make the cyber-city system with the ability to capture
requirements of each element of a city and has a good adaptability to changes and have the
capability to handle the growth of the system in its environment.
13.4.1 Modeling of System

The development of the cyber-city system can be initiated through the domain mod-
eling in the IoT concept; the system will identify each entity until the processes or
resources can be defined, and the relation between each other can be explained. Each
of these processes is connected to each interface that serves as executor of the system.
This activity can be done with the goal-based modeling approach. Furthermore, each
defined element in the system is mapped into the software components, until the opera-
tional mechanism of the system can be determined, whether the components are set to
dynamically or statically.
The stages are performed to identify and define the IoT context of adaptation mechanisms;
so in every identified context, we can build a knowledge tree, as shown in Figure 13.7. As
an example, according to the context of the user interface identification mechanism, the
structure of knowledge can anticipate the growth of user requirements, both of internal
or external stakeholders. In another example, with the basic mechanism of service-context
identification, the knowledge structure can anticipate the changes and growth of require-
ments on the activity and the process of converting data into system services. In addition,
it is possible to add an identification mechanism of other contexts if necessary, based on the
changes of facts or the growth of the system.
The illustration in Figure 13.7 shows that the structure of actor tree and data tree will be
automatically added, deleted, updated, and adapted to every vertex, in response to chang-
ing circumstances and growth occurs. As an illustration, every actor tree in the cyber-city
system will be linked to the requirements and changes of actors in government, private,
personal, and so on, as a user of the system. Data tree will be associated with the process of
change and growth activity of each actor in the service system, whether it is in the process
of trade, education, transportation, health, tourism, entertainment, and so on. Based on the
formulation of this knowledge-tree structure, we can define the requirements of PSn in the
form of the service catalog as shown in Table 13.2.
The service catalog describes details of the service for each actor. Duration is the service
availability of time, for example, 2 × 24 hours, 8:00 a.m. to 4:00 p.m., and so on. Options are
a class of selected service based on the level of interest or influence to the goal, for example,
level 1: A response (1-2), level 2: B response (2-3), and level 3: C response (3-4). Moreover, the
Interface Cyber city
I-1 P-1 P-2
Entity-1
P-4 P-5
I-2 P-3
Entity-2
P-6
P-7 P-8
I-n
Entity-n
P-n
I = Interface
P = Process
Mapping Mapping
Actor (A) Tree Data (D) Tree
(An, Dn) A1 D1
(Dn) D2
A4 D4
A2
A3 D3
(An, Dn)
A9 D5 D7
A6 D8
A5 A7 D6
A8
(Dn)
D9 (Dn)
(An, Dn)
FIGURE 13.7
The mapping of elements in a cyber-city system into the knowledge tree.
TABLE 13.2
Service Catalog
Description User
Service Catalog Duration Option Response U1 U2 U3 Un
Service-1.1 24 × 7 2 2 √ – √ √
Service-1.2 24 × 7 2 1 √ – √ –
Service-1.n 24 × 7 1 1 – √ – √
Service-2.1 24 × 7 1 2 √ √ √ –
Service-2.2 08–16 3 4 – √ √ √
Service-2.n 08–16 1 1 √ √ – –
Service-n.m 24 × 7 3 3 – – √ √
response is the incident recovery activity of the service class, which is classified based
on the level of seriousness of the incident, for example, level 1: 10 min, level 2: 30 min,
level 3: 1 hour, and level 4: 2 hours, depending on the impacts of incidents.
The changes that can lead to incidents in the service catalog can be categorized as the
facts of (fn) changes, for example:
1. f1: The political policy of the city can be associated with the changes in infrastruc-
ture and the structure of the city departments.
2. f2: The growth of activity, regarding changes to business processes and growth of
PSn.
3. f3: Service providers, changes legalization, are caused by the demand for a new
form, features, and algorithms, in the application of external services.
The category of service changes can be considered to set the service catalog for strategies,
management, and recovery. Request for a fast response to make changes in the system can
be made by performing a matching process between the occurred facts against the facts
in the structure of the knowledge tree. The service catalog consists of business-process
service which contains the entire PSn and technical service (Tsn). Tsn relates to all techni-
cal specifications of the technology. Figure 13.8 shows the adaptation mechanisms from
capturing events by the agent from the fact (f1, f2, …, fn) in the system environment. The
agents also capture the context of information from the service catalog (PSn, Tsn) until the
action implementation.
An event is the information of any context which is connected with the specification of
the new facts inside environmental and service catalog (∑ = f1, f2 … fn; PSn, TSn). The
created condition will evaluate the situation in accordance with the occurred specific
events, including the captured characteristics of the service and the changes in every
context (Q = fn ←→ PSn, fn ←→ TSn). So the evaluation of the service level agreement
(SLA) is performed to select the most appropriate action behavior.
Event Condition Action

Environment Service catalog
SLA Adaptation
Context-1 Context-1
Context-n Context-2 fn ← → PSn; E-tn ; R-tn ;
Ux: f1 (x) Ux: PSn (x)
Ux: fn (x) Ux: TSn (x) fn ← → TSn; →
Change fact-n
Change fact-n
Service fact-n
Service fact-n
act-1 ... act-n
FIGURE 13.8
The dynamic behavior of the system.
Based on the prevailing condition, the action of adaptation is performed (δ = E–tn; R–tn).
The reconfiguration process of evolution (E–t1, E–t2, … E–tn) at the time (t) made upon the
consideration of the SLA attainment targets evaluation, although the action of reconfigura-
tion (R–t1, R–t2, … R–tn) or the handling changes that have not been stipulated in the SLA.
Basically, this PSn adaptation action deals with the authorization of the service portfolio,
for example, when the service should be updated, replaced, maintained, refactored, dis-
missed, and rationalized. All of these services represented adaptation actions within the
control of changes to the structure of the tree of knowledge.
13.4.2 System Configuration

The mechanisms are developed to manage the service catalog, as shown in Figure 13.9.
The modeling process uses the TAOM4E tools (TAOM4E n. d.). A goal of the cyber-city
system services is managed using three subgoal activities such as the user interface, ser-
vice application, and service providers. Each goal can have subgoals and also plans that
contribute to the soft goal.
The modeling services through the goals element, as shown in Figure 13.9, are mapped
into the software components as shown in Figure 13.10. Each goal and plan will form
a primitive component so that there are four component groups—3 groups serve to
detect the context and 1 group as the reconfiguration components. The three groups are
“detect context-1” for the user interface, “detect context-2” for the service application,
and “detect context-3” for the service provider. All three groups of these components
require a group of components that has the function to perform the reconfiguration; the
Cyber city
service system
User interface
Service delivery
Service application
Interface
User
category
Analyze service
Detect Time Option
context-1 Features and
priority Service provider
Detect
role Detect
Detect context-2 Reconfiguration
device Detect
context-3
Request Detect
Mobile Tlp for change events
PC
++ Internal External
+ ++
++ + Relevance − + ++
− +
++
Comfortable Service catalog Response time + Security
+
HardGoal SoftGoal Plan Resource
AND-decomposition OR-decomposition Means-ends Contribution (++, +, −, −−)
FIGURE 13.9
The service catalog management for a cyber-city system.
User interface Service application Service provider
Detect device Request for change Internal
Detect role Detect events External
Detect context-1 Detect context-2 Detect context-2
Other contexts
Reconfiguration Detect context-n
Component-n Component-1 Component-2
Analyze service = Provided service port = Required service port
FIGURE 13.10
Software components configuration.
reconfiguration components will determine the adaptation strategies for each component
that needs it, so the presence of this component is generic. That means if at run-time sys-
tems require the addition of new components, then automatically, these components can
be added as shown in the dashed line in Figure 13.10. It is an example of adding a new
component to detect new context. Likewise, if a component is not needed or need to be
changed or need to be reused, it can be executed at run-time.
The request for change component is primitive for the component that requires recon-
figuration. The model is initialized by the variable the component dynamic, which is
determined by conditions at run-time. Achievement of the goal-service application con-
sists of features, time, options, and priority. This achievement is the target of the com-
ponent. This target is activated based on the soft-goal criteria and constraints. The soft
goal is a consideration of the contribution, against the nonfunctional system, whereas
constraints are the domain assumptions, which are set based on the strategy reconfig-
uration as described in Section 13.3. The components of “Detect Context-2” is a com-
posite component that describes the primitive component interconnect and needs with
the component reconfiguration through the type of OR relations in accordance with the
modeling goal.
Based on the configuration of developed components, each component will establish
a knowledge according to the needs of their respective functions. Figure 13.11 will illus-
trate the knowledge tree from the results of software components operations. For example,
the user interface and its goal plan “detect context-1” will form a tree of actors or users,
whereas the goal service application with the plan “detect context-2” will form a tree of
data or services, so does the other goals and plans.
Actor/User Data/Service
User interface Service application
Detect Detect
context-1 context-2
Mapping Mapping
Actor (A) Tree Data (D) Tree
(An, Dn) A1 D1
(Dn) D2
A4 D4
A2
A3 D3
(An, Dn)
A9 D5 D7
A6 D8
A5 A7 D6
A8
(Dn)
D9 (Dn)
(An, Dn)
FIGURE 13.11
The mapping of components into the knowledge tree.
13.5 Discussion and Conclusion

13.5.1 Discussion
SAS is a research area that has a broad spectrum because it deals with a variety of factors
and various conditions that must be recognized. Discussion performed by limiting the
scope of the basic goal-oriented in requirements engineering approach.
In this article, we introduced an approach that is based on the goal-oriented in require-
ments engineering approach by inserting the additional elements through the primitive
construction of the system as discussed in Sections 13.3 and 13.4. Figure 13.12 shows the
overall architecture models, which can be used as a guide for developers in constructing a
SAS in general, and self-adaptive cyber-city system in the IoT concept in particular.
Description of the system-development model in Figure 13.12 is as follows:
1. The development activities begin with the goal modeling to represent the IoT
domain model. The available tools at this stage can be adjusted with an adopted
goal-model approach, for example:
a. TAOM4E (TAOM4E n. d.) to model Tropos (Bresciani et al., 2004)
b. Objectiver (http://www.objectiver.com/) to model KAOS (Dardenne et al., 1993)
Goal model Inference engine
Specification
NRules
Model information Request Result
Goal (F, NF) priority

Control Formal model
and constraints Specification
component specification
description
Parser Components
Architectural
Architectural
Configurator configuration
description
(Component
languages
dependencies)
FIGURE 13.12
Development model of a system.
c. Open OME (http://www.cs.toronto.edu/km/openome/)

d. STS-ml (http://istar.rwth-aachen.de/tiki-index.php?page=i*+Tools)
e. OmniGraffle (http://www.omnigroup.com/omnigraffle), and others
Or we can arrange a description of this goal by developing its metamodels,
for example, using EMF (http://www.eclipse.org/modeling/emf/updates/ or
http://www.eclipse.org/modeling/emf/downloads/).
2. By the time of preparing the description of these goals, we need to ensure that
any functional (F) requirement should be represented as a goal and nonfunctional
(NF) requirements as soft-goals that can reflect the quality criteria of the system.
How to determine the parameters of the priorities and constraints are discussed
in Section 13.3.
3. After mapping the description of the goals that have been made in the architecture
description language, we configure it into a software component. This is accom-
plished by using a software to generate the required components.
4. The preparation of the software component architecture can refer to Section 13.3.4.
The supporting code generator tools such as java emitter templates (JET) Template
(http://www.eclipse.org/modeling/m2t/downloads/) can be used to transform
the eclipse modeling framework (EMF) meta-model. Or some other tools men-
tioned in point a, which has been equipped with the code generation facility.
5. The component control will coordinate all the components to give the inference
function. This component generates formal model specifications to perform the
selected reasoning tasks.
6. The components of inference engine can be developed by drafting rules, as discussed

in Section 13.3.3. This rule-based system can also be modeled using the EMF
meta-model tools. Some of the tools that can be used as an engine rule are CLIPS
(http://www.ghg.net/clips/CLIPS.html), JESS (http://www.jessrules.com/jess/
download.shtml or http://herzberg.ca.sandia.gov/jess/), jDREW (http://www.
jdrew.org/jDREWebsite/jDREW.html), Mandarax (http://mandarax.sourceforge.
net/), or using the Datalog Rules and DLV inference engine (Leone et al., 2006),
and others; they can be adapted to the description of the requirements of the goal.
The selection of tools that will be used in this IoT system development must be definitely
determined by studying the alignment between the description of goals and the needs of
the inference engine by considering any capabilities of tools to cover all requirements from
the developed concept.
13.5.2 Conclusion
Characteristics of the cyber-city system have much diversity of elements and are very
dynamic; it is caused by the very quick growth of the environment. The proposed
solution in this article is formulated through two approaches. First, importance to
understand and capture the variability in context (IoT concept) and behavior of the
system. This is realized through the domain modeling context with the logic-based
approach. Second, the availability of the knowledge structure and the quality of the
inference engine that can handle the scope of more broad and flexible solution space.
This is realized through the modeling of the inference context with the rule-based
approach.
The main highlight of discussion in this article is the provision of the SAS to meet the
scope of a system life cycle in the IoT concept; this relates to the control of automation
which is the part of the main concept of IoT. The given idea is an opportunity that can
be followed up in the control of automation of the future of the IoT. So the concept of IoT
system is expected to become more aware and adaptive; all orders can be performed auto-
matically, organized, intelligent, acting independently in context and situation, or event
environment.
The adaptation strategies as a mechanism for reconfiguration are proposed to overcome
this problem; a series of modifications were developed to represent the problem space.
In addition, this strategy also prepares the system to be able to conduct a search of the
solution space as an appropriate alternative solution automatically in the implementation
of the control strategy. The developed concept was inspired by the seven pillars of taklif sys-
tems which are the representations of life in the universe. These pillars are formulated into
the rules of systems with self-adaptive capabilities. We believe this proposed concept will
contribute to the provision of software systems in the context of a dynamic environment,
and especially in the construction of the cyber-city systems in the IoT concept. We expect
this concept to accommodate any issues of diversity and dynamism inside the system and
its elements.
References
Abuseta, Y. and K. Swesi. Design patterns for self adaptive systems engineering, International Journal
of Software Engineering & Applications (IJSEA), 6(4), pp. 11–28, 2015.
Agustin, R. D. and I. Supriana. Model Komputasi Pada Manusia dengan Pendekatan Agent,
Kolaborasi BDI Model dan Tujuh Pilar Kehidupan sebagai Inspirasi untuk Mengembangkan
Enacted Serious Game, Konferensi Nasional Sistem Informasi (KNSI), ITB, 2012.
Aradea, I. Supriana, and K. Surendro. An overview of multi agent system approach in knowledge
management model, International Conference on Information Technology Systems and Innovation
(ICITSI), IEEE, School of Electrical Engineering and Informatics, ITB, 2014.
Bratman, E. M. Intentions, Plans, and Practical Reason, CSLI Publications, Stanford, CA, 1987.
Bresciani, P., A. Perini, P. Giorgini, F. Giunchiglia, and J. Mylopoulos. TROPOS: An agent-oriented
software development methodology, Journal of Autonomous Agents and Multi-Agent Systems,
8(3): 203–236, 2004.
Cheng, B. H. C., K. I. Eder, M. Gogolla, L. Grunske, M. Litoiu, H. A. Muller, et al. Using models at
runtime to address assurance for self-adaptive, In [email protected] Foundations, Applications,
and Roadmaps, Lecture Notes in Computer Science (LNCS), Vol. 8378, pp. 101–136, Springer, 2014.
Daniele, L. M. Towards a rule-based approach for context-aware applications, Master Thesis,
University of Cagliari, Italy, 2006.
Dardenne, A., A. van Lamsweerde, and S. Fickas. Goal directed requirements acquisition, In Selected
Papers of the Sixth International Workshop on Software Specification and Design, pp. 3–50, Elsevier
Science Publishers B.V., Amsterdam, 1993.
Ganek, A. G. and T. A. Corbi. The dawning of the autonomic computing era, IBM Systems Journal,
42(1): 5–18, 2003.
Hirsch, D., J. Kramer, J. Magee, and S. Uchitel. Modes for software architectures. In V. Gruhn and
F. Oquendo (Eds.), EWSA 2006, Lecture Notes in Computer Science (LNCS), Vol. 4344, pp. 113–126,
Springer-Verlag, Berlin, Germany, 2006.
IBM, An Architectural Blueprint for Autonomic Computing, IBM, Hawthorne, NY, 2005.
ITU-T, Overview and role of ICT in smart sustainable cities Telecommunication Standardization
Sector, 2014.
Kramer, J. and J. Magee. Self-managed systems: An architectural challenge, Future of Software
Engineering, FOSE'07, pp. 259–268, The ACM Special Interest Group on Software Engineering
(SIGSOFT), IEEE Computer Society, Washington, DC, May 2007.
Leone, N., G. Pfeifer, W. Faber, T. Eiter, G. Gottlob, S. Perri, and F. Scarcello. The DLV system for
knowledge representation and reasoning. ACM Transactions on Computational Logic (TOCL),
7(3): 499–562, 2006.
Magee, J., N. Dulay, S. Eisenbach, and J. Kramer. Specifying distributed software architectures.
In Fifth European Software Engineering Conference (ESEC95), Barcelona, September 1995.
Magee, J., J. Kramer, and D. Giannakopoulou. Analysing the behaviour of distributed software
architectures: A case study. In 5th IEEE Workshop on Future Trends of Distributed Computing
Systems, pp. 240–245, 1996.
Mora, M. C. BDI models and systems: Reducing the Gap. In J. P. Muller et al. (Eds.), ATAL’98, LNAI
1555, pp. 11–27, 1999. Springer-Verlag, Berlin, Germany, 1999.
Nabulsi, Dr, R. M. 7 Pilar kehidupan: Alam semesta, akal, fitrah, syariat, syahwat, kebebasan memilih, dan
waktu, Gema Insani, Jakarta, 2010.
Nakagawa, H., A. Ohsuga, and S. Honiden. Constructing self-adaptive systems using a KAOS
model, In Proceedings of the SASOW, pp. 132–137, IEEE, 2008.
Pires, L. F., N. Maatjes, M. van Sinderen, and P. D. Costa. Model-driven approach to the imple-
mentation of context-aware applications using rule engines, In M. Mühlhäuser, A. Ferscha,
and E. Aitenbichler (Eds.), Constructing Ambient Intelligence. AmI Workshops. Communications in
Computer and Information Science, vol. 11, pp. 104–112, Springer, Berlin, Germany, 2008.
Russell, S. and P. Norvig. Artificial intelligence, a Modern Approach, 2nd Prentice Hall, Upper Saddle
River, NJ, 2003.
Supriana, I. and D. Aradea. Model self-adaptive sebagai landasan sistem untuk menunjang pen-
umbuhan komunitas, Keynote Paper Seminar Nasional Teknologi Informasi dan Komunikasi
(SENTIKA), Vol. 6, Yogyakarta, Maret 18–19, 2016.
Supriana, I. and D. Aradea. Automatically relation modeling on spatial relationship as self-
adaptation ability, Internationl Conference on Advanced Informatics: Concept, Theory and Application
(ICAICTA), Vol. 2, IEEE, Bangkok, Thailand, 2015.
Supriana, I., S. Wahyudin, and A. Mulyanto. Pengembangan motor inferensi untuk aplikasi
sistem pakar dalam model diagnosa, Technical Report, School of Electrical Engineering and
Informatics, Bandung Institute of Technology, Kota Bandung, Indonesia, 1989.
Tool for Agent-Oriented visual Modelling for Eclipse (TAOM4E) and its plugin t2x (Tropos4AS to
Jadex), developed by the Software Engineering group at Fondazione Bruno Kessler (FBK),
Trento, available, including the extensions, at http://selab.fbk.eu/taom.
Wu, C. G. Modeling rule-based systems with EMF, Eclipse Corner Article, Copyright (c) 2004 Chaur
G. Wu. All rights reserved, 2004.
14
The Role of Big Data, Cloud Computing,
and Mobile Technologies in the
Development of IoT Ecosystems
Janusz Wielki
CONTENTS
14.1 Introduction ...................................................................................................................... 321
14.2 The Concept of IoT and the Most Important Factors Stimulating Its
Development ..................................................................................................................... 322
14.3 The Most Important Opportunities and Possibilities Related to the Use of
IoT by Enterprises............................................................................................................. 323
14.4 The IoT Ecosystem and Its Most Important Components .......................................... 328
14.5 The Analysis of the Role and Relationships between Advanced Big Data
Business Analytics, Cloud-Based Solutions, and Mobile Technologies in the
Context of Creating the Architecture and Operations of the IoT.............................. 329
14.6 Conclusion .........................................................................................................................334
References.....................................................................................................................................334
14.1 Introduction
The dynamic development of information technology moves it into the next phase of its
evolution at a rapid pace. This is the phase in which the first computers are able to receive
data from virtually any type of physical object (Bisson et al. 2013). The situation results
from the fact that components such as sensors, processors, and software are becoming
much more a part of the product, and at the same time, have the capability to connect to
the accompanying outer infrastructure layer due to their connectivity. Moreover, connect-
ing literally everything to the Internet becomes economically justified (World Economic
Forum 2015). As a result, there were considerably more capabilities to create cyber-physical
systems, that is, systems integrating the physical and virtual worlds (Industrie 4.0 Working
Group 2013; Schwab 2016). The concept of the Internet of Things (IoT) has emerged in
this context (Barbier et al. 2013; Gartner 2014; Ericsson 2015). Simultaneously, it is not pos-
sible to create the technological infrastructure of the IoT system without the use of three
other leading technologies, which in itself carry an enormous transformation potential.
This applies to cloud computing, advanced big data analysis tools, and mobile solutions
(Burkitt 2014; Aharon et al. 2015; Heppelmann and Porter 2014, 2015; ITU 2015). In this
context, the main motivation connected with this chapter is to broaden knowledge about
321
the role of three technologies (big data, cloud computing, and mobile technologies) in a
process of creation solutions based on the IoT concept. Its basic goal is an attempt to iden-
tify the role and place which these three complementing technologies play in building IoT
ecosystems and their functioning, and what kind of mutual relations among them exists.
14.2 The Concept of IoT and the Most Important

Factors Stimulating Its Development
The analysis of the literature on the subject shows that there is no single universal defini-
tion of the IoT, and the authors emphasize issues differently. And so (Tapscott and Tapscott
2016), they use the term IoT in the context of installing intelligence into existing infrastruc-
ture, whereas Keen refers to the intelligent devices on the network (Keen 2015). Heppelmann
and Porter propose a more extensive definition. According to them, the phrase Internet of
Things was created to reflect a growing number of smart, connected products and highlight
the new opportunities they can represent (Heppelmann and Porter 2014). In this context,
Ericsson clarifies that a connected device is “a physical object that has an IP stack, enabling
two-way communication over a network interface” (Ericsson 2015). On the other hand, a
considerable group of definitions applies the concept of IoT to the use of sensors, actuators,
and communication technologies embedded in physical objects and the opportunities that
they bring about. And so, according to Bauer et al. (2014), “the Internet of Things refers
to the networking of physical objects through the use of embedded sensors, actuators,
and other devices that can collect or transmit information about the objects.” On the other
hand, according to Bisson et al. (2013), “the Internet of Things refers to the use of sensors,
actuators, and data communications technology built into physical objects—from road-
ways to pacemakers—that enable those objects to be tracked, coordinated, or controlled
across a data network or the Internet.” While Aharon et al. (2015) define the IoT as “sensors
and actuators connected by networks to computing systems. These systems can monitor
or manage the health and actions of connected objects and machines. Connected sensors
can also monitor the natural world, people, and animals.” Similarly, Dobbs et al. (2015)
define IoT. According to them, the concept of the IoT is related to the “embedded sensors
and actuators in machines and other physical objects that are being adopted for data
collection, remote monitoring, decision making and process optimization in everything
from manufacturing to infrastructure to health care.” On the other hand, the IoT according
to the European Commission is a term referring to “devices of all sorts (…) equipped with
sensors and actuators, connected to the Internet, allowing them to monitor their status
or the environment, to receive orders or even to take autonomous action based on avail-
able information” (European Commission 2015). However, the definition given by The
White House (2014) emphasizes the IoT as the capability of devices to communicate with
other devices. According to this definition, the IoT is “a term used to describe the abil-
ity of devices to communicate with each other using embedded sensors that are linked
through wired and wireless networks.” In turn, the consulting company PwC defines IoT
as an ecosystem created by the equipment, tools, and facilities. They define the IoT as the
“ecosystem of Internet-connected devices, operational tools and facilities” (PWC 2016). For
Schwab, the IoT is a relationship between things and humans that can be realized through
the use of various technological solutions. As he writes, IoT “in its simplest form, it can be
described a relationship between things (products, services, places, etc.) and people that
The Role of Big Data, Cloud Computing, and Mobile Technologies 323
is made possible by connected technologies and various platforms” (Schwab 2016). Rifkin
sees IoT in a similar spirit. As he writes, “the Internet of Things will connect everything
with everyone in an integrated global network” (Rifkin 2014). In turn, Cisco proposes a
completely different definition of the IoT. For them, “IoT is simply the point in time when
more things or objects were connected to the Internet than people” (Evans 2011). There are
a number of factors stimulating the development of the IoT. Undoubtedly, the continuous
technological progress is a key factor. It caused the critical building blocks of comput-
ing, needed to create IoT ecosystems and growing exponentially for years reached such a
level of technical maturity, and correspondingly low price that the universal implementa-
tion of solutions based on this concept became possible (Brynjolfsson and McAfee 2014).
This applies to such aspects that are a key to its implementation as capabilities and prices
of sensors,* computing power, disk space, network connections, both wired and wireless
networks, or progress in the area of creating low-cost power systems. Actions taken to
eliminate the limitations of connectedness between the components of the IoT ecosystems
have become an important element stimulating the development of the IoT. An example is
the introduction of IPv6 solving the problem of exhausting the available IP addresses, as a
result, removing one of the major barriers to the development of IoT (Barbier et al. 2013).
14.3 The Most Important Opportunities and Possibilities

Related to the Use of IoT by Enterprises
When it comes to the senior management view on the key benefits of using the concept
of the IoT, one can indicate top five categories according to a study conducted by KPMG
in 2015. The key ones are the improvement of productivity (20%) and the acceleration of
innovation cycle in organizations (20%). Further important aspects are the possibilities of
greater diversity of products or services offered by the company (16%), cost reduction (13%),
and an increase in the level of profitability (10%) (Figure 14.1).
At the same time, the senior management indicated the areas in the same study in which
they see the greatest potential for monetization for their companies resulting from their
use of solutions from the realm of the IoT. Those related to consumers and consumer
markets (22%) are by far dominant. The next such areas are as follows: technology (13%),
aerospace, space and defense (10%), education (10%), and automotive/transportation (9%)
(KPMG 2015). However, it is worth noting that, as McKinsey’s forecasts indicate, the target
value generated by IoT solutions for B2B will be twice the size of those created for the con-
sumer market (Aharon et al. 2015) (Figure 14.2).
When it comes to the possibilities related to the use of smart-connected products,
which allow the creation of value in IoT solutions, result from four basic functionalities
(Figure 14.3).
The first ones are sensors built into smart products that enable to monitor the following:
• Its health, operation, and use

• External environment
* For example, sales of sensors being a key component of each IoT ecosystem have been increasing at a rate of 70%
per annum since 2010 (Bisson et al., 2013).
Improved efficiencies/higher productivity 20.0%
Faster innovation cycles 20.0%
Product/service differentiation 16.0%
Cost reductions 13.0%
Increased profitability 10.0%
Enhanced customer adoption and loyalty 6.0%
Increased market share 6.0%
0% 5% 10% 15% 20%
FIGURE 14.1
The key benefits of using the concept of the Internet of things.
Consumer markets/retail 22.0%
Technology 13.0%
Aerospace and defence 10.0%
Education 10.0%
Automotive/transportation 9.0%
Health care 6.0%
Financial services 6.0%
Telecommunications 6.0%
Energy 5.0%
Manufacturing 5.0%
0% 5% 10% 15% 20%
FIGURE 14.2
IoT monetization opportunities.
Level 4
Autonomy
Level 3
Optimization
Level 2
Control
Level 1
Monitoring
FIGURE 14.3
Four types of opportunities offered by smart, connected products.
In turn, software contained both in the product itself and the infrastructure layer offers
even more possibilities; namely, it enables to remotely control the product and its func-
tions and personalization of its operation on a scale that previously was not possible to
achieve. Monitoring capabilities and the resulting wide-data stream combined with the
control opportunities offered by smart products allow the organizations to optimize their
performance in an extremely wide range. This applies to aspects such as the significant
improvement in the operation of the product itself or its predictive diagnostics and repair
(Heppelmann and Porter 2014). The three functionalities discussed above allow the smart
products to achieve an unprecedented degree of autonomy. Moreover, different levels of
autonomy can be achieved, which are as follows (Heppelmann and Porter 2014):
• Independent operation
• Independent coordination of operation with other products or systems
• Autonomous improvement of the product and its personalization
• Autonomous self-diagnosis and support service
These four functionalities offered by smart-combined products provide business organi-

zations with two basic types of opportunities. They are related to the following (Aharon
et al. 2015):
• Reconstruction of business processes

• Creation of new business models
When it comes to the reconstruction of business processes, there are a number of related
opportunities that differ depending on the settings to which they apply. According to
McKinsey, one can identify nine such areas with the greatest economic impact (Figure 14.4).
These are, for example, changes in the design of domestic appliances in the home
setting; namely the process is based on the analysis of usage-based design. In the retail
setting, the key application areas of the IoT solutions are as follows: checkout automation
and goods layout optimization or individualization of promotional activities in stores.
Factories 3,700
Cities 1,660
Human 1,590
Retail environments 1,160
Worksites 0,930
Outside 0,850
Vehicles 0,740
Home 0,350
Officies 0,150
0,000 0,500 1,000 1,500 2,000 2,500 3,000 3,500 4,000
FIGURE 14.4
Potential economic impact of IoT in 2025 (in trillions USD).
On the other hand, in the office setting, the IoT applications include organizational rede-
sign and the monitoring of employees or the use of augmented reality for training pur-
poses. Regarding the transformation of business processes in the factories, the key issues
include optimization of operations and the related improvement in productivity, optimi-
zation of the use of equipment and supplies, predictive maintenance, the maintenance of
equipment, and the occupational safety and health. Another setting that enables a deep
reconstruction of business processes is related to nonstandard production worksites, such
as oil and gas or construction sites. In this case, the most important capabilities are similar
to the previous case, with additional IoT-enabled R&D activities. When it comes to the
setting associated with the various types of vehicles, the key areas of IoT-based solutions
include repair and condition-based maintenance, equipment design based on an analysis
of their usage or presales analytics. Vehicles are also related to the capabilities of recon-
struction of logistics processes. This applies to things such as real-time routing, the use of
connected navigations, or the use of transport monitoring systems (Aharon et al. 2015).
Regarding the opportunities associated with the implementation of new business models,
we can identify ten basic types in this context. These include the following:
1. Business models based on anything-as-a-service concept
2. Business models based on the use of new forms of outsourcing
3. Business models based solely on the data and their use
4. Business models based on additional services related to the physical product
offered to customers
5. Business models based on smart products that are sources of added value for the
customer
6. Business models based on behavioral profiling

7. Hybrid business models
8. Business models based on offering IoT platforms
9. Business models based on offering comprehensive IoT infrastructure solutions
10. Business models based on offering extended services
The model being the product-as-a-service is the mainstream in the first group. Its develop-
ment is related to much more widely observed processes of migration from the customer
buying the product to the one in which the manufacturer retains ownership of it, and the
customer uses it, paying for its real use. The development of functionalities related to intel-
ligent, connected products provides big opportunities in this area. Rolls Royce, which offers
their engines to airlines in the power-by-the-hour model, is one of the companies pioneering
in this field. In this model, the airlines pay for the real engine use time, instead of incurring
a one-off cost of its purchase and additional costs of maintenance and repair. Xerox also
employs such a model by monitoring the actual use of their photocopiers via the installed
sensors (Heppelmann and Porter 2015). The development of smart systems also allows
for the implementation of business models based on offering new forms of outsourcing.
Another such example is the Pacific Control company operating in Dubai. It offers the
remote monitoring of buildings, airports, and hotels based on the IoT (The Economist 2010).
For the third category, the development of smart-connected devices enables one to collect
vast amounts of different types of data that can be used to create business models based
on their usage. Skyhook Wireless company, which offers specific information acquired
based on geolocation data they collect, is an example of this approach. They can include
information such as which local bars will be the most popular on a specific day and time,
how many people will go near the billboard at a given date and a specific time, or what is
the density of people in a specific urban area on a given day and time. The company uses
anonymous geolocation data collected from mobile users of its services in every major
American city during the past 24 months to carry out this analysis (The Economist 2010;
Mims 2010).
The development of the IoT also enables one to implement business models based
on providing customers with additional services related to the physical product they
purchased and use. Caterpillar company is one example of this approach. Specialized
teams advise customers on how to optimize the deployment of equipment, when a
smaller number of machines suffice and how to achieve better fuel efficiency through
the stock of machines based on analysis of data collected from each machine used on
the construction site (Heppelmann and Porter 2015). Heidelberger Druckmaschinen, a
manufacturer of printing presses, offers a similar type of service based on over 1000 sen-
sors installed in them (The Economist 2010). Another group is business models based
on providing customers with smart products that are sources of additional benefits to
them. Play Pure Drive is an example of such a solution. In this case, Babolat company
has transformed a traditional product into a smart one that provides players with the
opportunity to improve their technique by the use of a dedicated application, a tennis
racket equipped with appropriate sensors, and a system enabling the connection to the
smartphone. Clothing manufacturer Ralph Lauren made a similar move by offering smart
PoloTech Shirt. It collects all parameters including pulse, the intensity of the movement,
calories burned, and many others with built-in sensors in real-time during exercise and
transmits them to a smartphone or smartwatch (Heppelmann and Porter 2014; Ralph
Lauren 2016). Another group of business models are those based on behavioral profiling.
The system for establishing insurance rates based on monitoring of the driving behav-
ior through the suitable telemetry device mounted in the vehicle is an example of this
type of solution. The American insurance company Progressive offers this solution under
the name of Snapshot (Burkitt 2014; Progressive 2016). Coverbox uses a similar system
on the British market (The Economist 2010). Hybrid business models are a compromise
between the models of product-as-a-service and traditional purchase of products by cus-
tomers. They connect sales with, for example, different types of service contracts based
on the monitoring of the device operations. Another group is business models based on
offering IoT platforms to the users. Apple HomeKit is an example of such a solution.
It controls various home devices from different manufacturers through the smartphone
application. HealthKit platform, which enables the integration of devices for monitor-
ing people’s health and activity, is another example of a solution by the same company
(Burkitt 2014). The next group of business models is those based on providing compre-
hensive IoT infrastructure solutions. ThingWorx platform is one of well-known examples
of this approach. It provides comprehensive services for the creation of IoT solutions
(ThingWorx 2016). The last group is business models based on the provision of extended
services. This forward-looking category includes solutions based on the use of data and
information collected by the providers of various IoT services and providing their own
services based on them. Operations of insurance companies working on solutions which
include creating their own portfolio based on cooperation with companies offering various
types of IoT systems designed to monitor health and physical activity are an example of
this approach (Burkitt 2014).
14.4 The IoT Ecosystem and Its Most Important Components

The appropriate technological infrastructure is essential to achieve the functionalities
described earlier and the related new business models. Figure 14.5 shows the key elements.
Product hardware consists of sensors, processors, and elements providing commu-
nication (antenna ports, etc.) embedded in the product and supporting its traditional,
mechanical, and electrical components. In turn, the product software is the operating sys-
tem, on-board applications, user interface, and product control components embedded
in the product. At the communication level, suitable communication protocols such as
MQTT, CoAp, AMQP (Dwyer 2015; Perera 2015) providing communication between the
product and product cloud are the necessary element of the system. The latter element of
the IoT system is composed of four layers. The first one (smart product application) are
applications that run on remote servers, which manage the monitoring, control, optimiza-
tion, and autonomous operations of product features. Another one (product data database)
is a system that allows the aggregation, normalization, and data management both in real-
time and on historical data acquired from the product. Another layer of product cloud
(rules/analytics engine) is related to the rules, business logic, and big data analytical capa-
bilities providing algorithms used during the product operation and revealing the related
new product insights. The last layer (application platform) is an application development
and execution environment, which allows quick creation of smart networked business
applications that use data access, visualization, and run-time tools. On the contrary, tools
for managing system access and authentification of users and product security, connectiv-
ity, and individual layers of the product cloud are an extremely important component
Product cloud
Application platform
Rules/analytics engine
Product data database Integration

with
Smart product applications
existing
business
systems
Security
and
identity Connectivity External
Network communication sources
of
information
Product
Product hardware
Product software
FIGURE 14.5
IoT system technological infrastructure.
of the IoT system. The entire infrastructure also has to have access to external sources of
information (weather, traffic, energy prices, social media, etc.) necessary from the point of
view of the capabilities offered by the product. As the technological infrastructure of IoT
is part of the overall information technology (IT) infrastructure in an organization, the
tools that integrate data from smart-connected products into key IT systems of a company,
such as enterprise resource planning (ERP), customer relationship management (CRM), or
product lifecycle management (PLM) are necessary (Heppelmann and Porter 2014).
14.5 The Analysis of the Role and Relationships between Advanced

Big Data Business Analytics, Cloud-Based Solutions,
and Mobile Technologies in the Context of Creating
the Architecture and Operations of the IoT
The appropriate infrastructure layer is necessary to allow the operations of the IoT
ecosystem. The solutions offered as a part of it must provide the following (Burkitt 2014;
Heppelmann and Porter 2014; Deichmann et al. 2015):
• Continuous connection of ecosystem elements to the Internet at all levels and

through various transmission channels (persistent connections, Wi-Fi, cellular, or
Bluetooth)
• Adequate computing power and disk space necessary to collect and process vast
amounts of data supplied from numerous endpoints of the system
• Software running on external servers, management monitoring, control,

optimization, and autonomous operations of functionality offered by the product
• Connection to social networks
• Services related to data management and analysis
• Opportunities to develop the necessary software
It is not possible to create such an infrastructure of the IoT system without the use of three
other leading technologies, which in itself carry an enormous transformation potential
(Figure 14.6).
It is necessary to ensure reliable operation of the entire IoT ecosystem in the
communication layer at the basic level to think about creating solutions, IoT at all. Hence,
ubiquitous connectivity is seen as one of the key enablers that stimulate the development of
the IoT (Aharon et al. 2015). Generally speaking, one can distinguish two market segments
related to IoT connections, that is, massive IoT connections and critical IoT connections.
The first one is characterized by high connection volumes, low cost, requirements on low-
energy consumption, and small data traffic volumes. The characteristics of the other one
are ultrareliability and availability and low latency (Ericsson 2016). IoT systems use dif-
ferent kinds of technical solutions, such as fixed connections, Wi-Fi, cellular, or Bluetooth
connections to ensure adequate connectivity. According to the report by Ericsson, the
compounded annual growth rate of the number of IoT connections is to be 23% between
2015 and 2021. At the same time, the highest level of growth is expected in cellular IoT
and is projected at 1.5 billion of various types of connected devices (Ericsson 2016). These
forecasts clearly show that mobile technologies grow into a key solution when it comes
to providing connectivity in IoT systems. At the same time, the development of 5G net-
work, characterized by greater capacity and lower energy requirements, make them very
important due to these qualities from the point of view of stimulating the growth of IoT
(Ericsson 2016). At the same time, the role of cloud technologies is crucial due to the num-
ber of devices operating in the IoT ecosystem and the amount of obtained, collected, and
processed data. It would be difficult to imagine the creation of IoT without these solu-
tions, both for technical (e.g., scalability) and economical reasons. The progress in this
Bi
g
ing
d
te ata
t
pu
ch t
no oo
om
lo ls a
dc
gi n
es d
ou
Internet
Cl
of
Things
Mobile technologies
FIGURE 14.6
Technologies supporting development of IoT ecosystems.
area taking place in recent years has become a driver of IoT market development, and
they are related to such aspects as follows (Dwyer 2015; RightScale 2016; Wielki 2015b):
• Increasing throughput and reliability of Internet connections

• The development of web services and their increasing availability
• The processes of commoditization of information technology
• Decreasing consumer concerns related to the security of cloud solutions
There are several key factors that determine the usefulness of cloud technologies in the
creation of IoT solutions. Undoubtedly, the use of a cloud infrastructure provides a whole
range of opportunities to reduce costs. They apply to both physical and software-related
elements. As for the first one, it concerns such aspects as follows (Wielki 2015a):
• Reduction or elimination of waste related to the low level of hardware utilization

• Reduction of costs connected with hardware maintenance
• Lowering costs related to energy consumption
• Possibilities for the permanent analysis of costs and the selection of the optimal
service level
In the case of software, it is related to the following (Wielki 2015a):
• Purchase and installation of software, its maintenance, and upgrade

• Purchase of wrongly selected software
• Low level of software usage
• Developing and testing of applications
In the case of critical IoT systems, it is also a question of cloud infrastructure’s scalabil-
ity. It makes it possible to add the necessary computational capacity, disk space, band-
width, and new services or applications in a way that is not only cost effective, but also
much faster (Bisson et al. 2013; Clutch 2016). All three major cloud-service models are
useful in the development of the IoT infrastructure, that is, software-as-a-service (SaaS),
infrastructure-as-a-service (IaaS), and platform-as-a-service (PaaS). The first two are
necessary for the operation of three layers of product cloud, that is (Figure 14.4), smart-
product application, product-data database, and rulet/analytics engine. The third one is
very important from the point of view of the application platform layer. At the same time,
while creating the IoT infrastructure in the cloud model, organizations can take advantage
of each of the three basic types of cloud, that is, private, public, and hybrid. Research
shows that private and hybrid clouds record the highest growth among businesses
(RightScale 2016). However, the connection of devices and providing the infrastructure
for their proper operation is just the first step toward building a competitive advantage
based on the IoT systems. The real value is when organizations are able to transform vast
amounts of collected data, often in real time, into real business intelligence that can be
taken advantage of, for example, in the creation of new business models. Here comes the
crucial role of advanced-analytical systems, that is, big data systems, also known as big
data and advanced analytics (Magnin 2016). Such solutions are, according to IDC, “a new
generation of technologies and architectures, designed to economically extract value from
very large volumes of a wide variety of data by enabling high-velocity capture, discovery,
TABLE 14.1
The Differences between the Big Data Analytical Systems and Traditional Ones
Big Data Analytical Systems Traditional Analytics
Data type Unstructured data Data formatted in rows and columns

Volume of data 100 TBs to PBs Tens of TBs or less
Flow of data Continual data flow Static pool of data
Utilized analysis methods Machine learning Hypothesis-based
Primary purpose of system usage Products based on data Internal decision support and services
and/or analysis” (Gantz and Reinsel 2012). Table 14.1 shows the comparison of big data
analytical systems and the traditional ones.
The amount of data from IoT systems is growing and will grow exponentially (Brown
et al. 2011). According to IDC, the global amount of data will increase tenfold from 4.4 ZB
to 44 ZB in 2014–2020. In 2020, ten percent of all the world’s data will come from the IoT sys-
tems. At the same time, we should emphasize that what is important is not the amount of
data, but the amount of payload. At the moment, not all of the data collected are used, and
those that are used in many cases are not fully utilized (e.g., in the case of massive amounts
of data coming from devices operating within IoT systems, their use in the prediction and
optimization is critical) (Aharon et al. 2015). However, this situation will change according
to IDC. According to these forecast, the number of useful data shall increase in 2013–2020
from 22% to 35% (Adshead 2014). In these circumstances related to the development of IoT
and rapid growth of data available to companies from different devices, the growing inter-
est in tools that provide them with the capability to conduct advanced business analytics
should not be unusual. According to the forecast by ABI Research, the market for this type
of tools that enable to integrate, analyze, and create reports based on data from IoT systems
is expected to reach the value of $30 billion in 2021 (Stackpole 2016). At the same time, the
implementation, maintenance, and development of advanced analytical solutions, such
as business intelligence (BI), require the use of an appropriate but costly infrastructure
or having the staff with appropriate qualifications (Olszak 2014). When they are based on
cloud computing technologies, and with the implementation of the analytics-as-a-service
(AaaS) model, there is the possibility of reducing those costs, use the latest solutions avail-
able on the market, and at the same time, eliminating potential staffing issues (Thompson
and Van der Walt 2010; Intel IT Center 2014). The situation is similar with regard to big data
projects. In the case of projects of this type, known as cloud-based big data analytics (Intel
IT Center 2014), organizations can build their own analysis ecosystem based on big data
technology and the abovementioned three basic cloud service models (Figure 14.7).
Solutions available in the IaaS model enable the organizations to build infrastructural
foundations for the whole big data ecosystem and implement the advanced analytical
services available in the SaaS model. Examples of services available in the first one are
solutions that are as follows:
• Amazon Web Services

• Windows Azure
• Citrix CloudPlatform
Cloud-based Big Data

ecosystem
SaaS
IaaS
PaaS
Cloud computing
FIGURE 14.7
Big data ecosystem based on cloud technologies.
• Vmare Cloud Suite

• Rackspace
When it comes to big data analytics services available in the SaaS model, the examples of
these solutions are as follows:
• Amazon Elastic MapReduce

• Google BigQuery
• Rackspace Hadoop
• Windows Azure HDInsight
• FICO Big Data Analyzer
At the same time, the organization can combine the use of different analytical solutions for
their various business purposes, such as customer sentiment analysis, risk management,
and asset management. Services available in the PaaS model, which can be used by orga-
nizations as a platform to develop their advanced analytical tools, are also a complementary
part for the big data ecosystem (Intel IT Center 2014). Examples of the services offered in
the model PaaS include solutions such as
• Google App Engine

• Windows Azure
• Force.com
• Red Hat OpenShift
• VMare Cloud Foundary
At the same time, there are attempts to develop standards for creation of advanced
big data business analytics based on cloud solutions (ITU 2015).
14.6 Conclusion
Business organizations are on the verge of profound changes in their operations.
They result from the dynamic developments in IT, which enable to increasingly use
the Internet infrastructure to receive data from virtually any type of physical object.
This gives the organizations a whole range of unprecedented opportunities in terms
of value-creation processes based on the implementation of new business models and
reconstruction of those used thus far. At the same time, it is necessary to build a reliable,
adequately efficient, scalable, and cost-effective infrastructure layer to allow the use
of the emerging opportunities related to the IoT and build the competitive advantage
based on it. With regard to these issues, the role of the three leading technologies, which
dynamically develop in the last few years and significantly change today’s economic
reality, that is, mobile technologies, cloud computing, and advanced analytical tools for
big data, is very important. The implementation of new business models outlined in this
chapter or deep transformation of processes in various social and economic areas would
not be possible without them, and without them reaching the appropriate level of matu-
rity. It should also be noted that the abovementioned technologies bring a whole range
of various challenges (Wielki 2015c, 2015d), which can overlap and intensify in certain
circumstances. This applies to, for example, such important aspects as data security or
privacy issues. In this case, each organization must evaluate and weigh the potential
benefits and risks, building the IoT ecosystem which is optimal from their point of view
and which integrates solutions which are cheaper, but with a lesser degree of control
(i.e., operating in the public cloud), with more expensive solutions having the higher
degree of control (i.e., operating in the private cloud). Undoubtedly, the creation of
adequately reliable IoT system is not a simple task, but nevertheless the continuous
progress in three technologies discussed in this chapter significantly facilitates this task.
References
Adshead, A. 2014. Data set to grow 10-fold by 2020 as Internet of things takes off. http://www.
computerweekly.com/news/2240217788/Data-set-to-grow-10-fold-by-2020-as-internet-of-
things-takes-off (accessed April 10, 2014).
Aharon, D., Bisson, P., Bughin, J. et al. 2015. The Internet of Things: Mapping the value beyond
the hype. http://www.mckinsey.com/~/media/McKinsey/dotcom/Insights/Business%20
Technology/Unlocking%20the%20potential%20of%20the%20Internet%20of%20Things/
Unlocking_the_potential_of_the_Internet_of_Things_Full_report.ashx (accessed July 1, 2015).
Barbier, J., Bradley, J., Handler, D. 2013. Embracing the Internet of everything to capture your share
of $14.4 Trillion. https://www.cisco.com/web/about/ac79/docs/innov/IoE_Economy.pdf
(accessed September 6, 2014).
Bauer, H., Patel, M., Veira, J. 2014. The Internet of things: Sizing up the opportunity. http://www.
mckinsey.com/industries/high-tech/our-insights/the-internet-of-things-sizing-up-the-
opportunity (accessed December 23, 2014).
Bisson, P., Bughin, J., Chui, M. et al. 2013. Disruptive technologies. http://www.mckinsey.
c o m / ~ / m e d i a / M c K i n s e y / d o t c o m / I n s i g h t s % 2 0 a n d % 2 0 p u b s / M G I / R e s e a rc h /
Technology%20and%20Innovation/Disruptive%20technologies/MGI_Disruptive_
technologies_Full_report_May2013.ashx (accessed June 1, 2013).
Brown, B., Bughin, J., Byers, A., Chui, M., Dobbs, R., Manyika, J. 2011. Big data: The next frontier
for innovation, competition, and productivity, http://www.mckinsey.com/mgi/publications/
big_data/pdfs/MGI_big_data_full_report.pdf (accessed May 30, 2011).
Brynjolfsson, E. and McAfee, A. 2014. The Second Machine Age. New York: W. W. Norton &
Company.
Burkitt, F. 2014. A strategist’s guide to the Internet of things. strategy+business, Issue 77 (November).
http://www.strategy-business.com/article/00294?tid=27782251&pg=all (accessed November
19, 2014).
Clutch. 2016. Enterprise cloud computing survey. https://clutch.co/cloud#survey (accessed February
11, 2016).
Davenport, T. 2014. Big data@work. Boston, MA: Harvard Business School Press.
Deichmann, J., Roggendorf, M., Wee, D. 2015. Preparing IT systems and organizations for the
Internet of Things. http://www.mckinsey.com/industries/high-tech/our-insights/preparing-
it-systems-and-organizations-for-the-internet-of-things (accessed November 30, 2015).
Dobbs, R., Manyika, J., Woetzel, J. 2015. No Ordinary Disruption. New York: PublicAffairs.
Dwyer, I. 2015. Internet of things. http://cdn2.hubspot.net/hubfs/553779/PDFs/Ironio-IoT-
Whitepaper-Jun2015_4.pdf?t=1455213785378 (accessed July 9, 2015).
Ericsson. 2015. Ericsson mobility report. http://www.ericsson.com/res/docs/2015/mobility-report/
ericsson-mobility-report-nov-2015.pdf (accessed November 30, 2015).
Ericsson. 2016. Ericsson mobility report. http://www.ericsson.com/res/docs/2016/ericsson-
mobility-report-2016.pdf (accessed June 29, 2016).
European Commission. 2015. Monitoring the digital economy & society 2016–2021. http://ec.europa.
eu/newsroom/dae/document.cfm?doc_id=13706 (accessed December 30, 2015).
Evans, D. 2011. The Internet of things. http://www.cisco.com/web/about/ac79/docs/innov/IoT_
IBSG_0411FINAL.pdf (accessed April 30, 2011).
Gantz, J. and Reinsel, D. 2012. The digital universe in 2020: Big data bigger digital shadows, and
biggest growth in the far east. http://www.emc.com/collateral/analyst-reports/idc-the-
digital-universe-in-2020.pdf (accessed December 29, 2012).
Gartner. 2014. Gartner says 4.9 Billion connected “Things” will be in use in 2015. http://www.
gartner.com/newsroom/id/2905717 (accessed November 20, 2014).
Heppelmann, J. and Porter, M. 2014. How smart, connected products are transforming competition,
Harvard Business Review. November, pp. 64–88.
Heppelmann, J. and Porter, M. 2015. How smart, connected products are transforming companies,
Harvard Business Review. October, pp. 96–114.
Industrie 4.0 Working Group. 2013. Recommendations for implementing the strategic initiative
Industrie. 4.0. http://www.acatech.de/fileadmin/user_upload/Baumstruktur_ nach_Website/
Acatech/root/de/Material_fuer_Sonderseiten/Industrie_4.0/Final_report__Industrie_4.0_
accessible.pdf (accessed July 30, 2014).
Intel IT Center. 2014. Big data in the cloud. http://www.intel.com/content/dam/www/public/
us/en/documents/product-briefs/Big Data-cloud-technologies-brief.pdf (accessed April 29,
2014).
ITU. 2015. Big data—cloud computing based requirements and capabilities. https://www.itu.
int/rec/dologin_pub.asp?lang=e&id=T-REC-Y.3600-201511-I!!PDF-E&type=items (accessed
December 23, 2015).
Keen, A. 2015. The Internet is Not the Answer. London: Atlantic Monthly Press.
KPMG. 2015. The changing landscape of disruptive technologies. https://techinnovation.kpmg.
chaordix.com/static/docs/TechInnovation2015-Part2.pdf (accessed December 26, 2015).
Magnin, C. 2016. How big data will revolutionize the global food chain. http://www.mckinsey.
com/business-functions/digital-mckinsey/our-insights/how-Big Data-will-revolutionize-
the-global-food-chain (accessed August 29, 2016).
Mims, C. 2010. Follow the Smart Phones. https://www.technologyreview.com/s/418451/follow-the-
smart-phones/ (accessed April 23, 2011).
Olszak, C. 2014. Business intelligence in cloud. Polish Journal of Management Studies, 10(2): 115–125.
Perera, S. 2015. IoT analytics. http://resources.idgenterprise.com/original/AST-0163543_wso2_

whitepaper_iot-analytics-using-Big Data-to-architect-iot-solutions.pdf (accessed December 29,
2015).
Progressive. 2016. The fair way to pay for car insurance. https://www.progressive.com/auto/snapshot/
(accessed February 9, 2016).
PWC. 2016. Industry 4.0: Building the digital enterprise. https://www.pwc.com/gx/en/industries/
industries-4.0/landing-page/industry-4.0-building-your-digital-enterprise-april-2016.pdf
(accessed May 1, 2016).
Ralph Lauren. 2016. The polo tech shirt. http://www.ralphlauren.com/product/index.jsp?
productId=69917696 (accessed April 30, 2016).
Rifkin, J. 2014. The Zero Marginal Cost Society. New York: Palgrave Macmillan.
RightScale. 2016. State of the cloud report. http://assets.rightscale.com/uploads/pdfs/RightScale-
2016-State-of-the-Cloud-Report.pdf (accessed July 30, 2016).
Schwab, K. 2016. The Fourth Industrial Revolution. Cologny/Geneva: World Economic Forum.
Stackpole, B. 2016. Making IoT magic with analytics. http://www.computerworld.com/article
/3078754/internet-of-things/making-iot-magic.html#tk.drr_mlt (accessed June 3, 2016).
Tapscott, A. and D. Tapscott. 2016. Blockchain Revolution. New York: Penguine Random House.
The Economist. 2010. Augmented business. http://www.economist.com/node/17388392/ (accessed
November 29, 2010).
The White House. 2014. Big data: Seizing opportunities, preserving values. https://www.whitehouse.
gov/sites/default/files/docs/big_data_privacy_report_may_1_2014.pdf (accessed May 26, 2014).
ThingWorx. 2016. One platform—Limitless possibilities. http://www.thingworx.com/ (accessed
April 16, 2016).
Thompson, W. and J. Van der Walt. 2010. Business intelligence in the cloud. SA Journal of Information
Management, 12(1). http://www.sajim.co.za/index.php/SAJIM/rt/printerFriendly/445/444&.
Wielki, J. 2015a. An analysis of the opportunities and challenges connected with utiliza-
tion of the cloud computing model and the most important aspects of the migration
strategy. In Annals of Computer Science and Information Systems, Vol. 5. Proceedings of the
2015 Federated Conference on Computer Science and Information Systems. Ed. M. Ganzha, L.
Maciaszek, M. Paprzycki, pp. 1569–1574. New York City/Warszawa: Institute of Electrical
and Electronics Engineers/Polskie Towarzystwo Informatyczne.
Wielki, J. 2015b. Analiza możliwości wykorzystania modelu cloud computing w kontekście reduk-
cji kosztów związanych z funkcjonowaniem infrastruktury IT współczesnych organizacji.
Problemy Zarządzania, 13(2) (52): 204–216.
Wielki, J. 2015c. The opportunities, impediments and challenges connected with the utilization of
the cloud computing model by business organizations. In Information Management in Practice.
Ed. B. Kubiak, and J. Maślankowski, pp. 11–26. Gdańsk: Faculty of Management University of
Gdańsk.
Wielki, J. 2015d. The social and ethical challenges connected with the big data phenomenon. Polish
Journal of Management Studies, 11(2): 192–202.
World Economic Forum. 2015. Deep shift: Technology tipping points and societal impact. http://
www3.weforum.org/docs/WEF_GAC15_Technological_Tipping_Points_report_2015.pdf
(accessed September 30, 2015).
15
Data Processing in IoT Using
an Enterprise Service Bus
Siddharth S. Prasad, Robin Singh Bhadoria, and Mohit Mittal
CONTENTS
15.1 Introduction ...................................................................................................................... 337
15.2 Enterprise Service Bus .................................................................................................... 338
15.3 Integration Core Principles............................................................................................. 339
15.4 Service Modeling and Realization ................................................................................ 339
15.4.1 How Modeling Improves Service-Oriented Architecture ........................... 339
15.4.2 Service Realization ............................................................................................340
15.5 Enterprise Integration Patterns for Data Handling in IoT ......................................... 341
15.5.1 What Is the Need for Integration? ................................................................... 341
15.5.2 What Makes Integration So Hard? .................................................................. 341
15.5.3 Asynchronous Data Architecture ................................................................... 341
15.5.4 How Can Patterns Help?...................................................................................342
15.6 Challenges for IoT Data...................................................................................................343
15.7 Conclusion ........................................................................................................................344
References.....................................................................................................................................345
15.1 Introduction
The Internet of Things (IoT) is a collection of several physical devices or embedded
gadgets that are interconnected through well-known network such as internet. Such
devices need to be followed the same border of guidelines while accessing and facilitat-
ing their client. There have been several challenges that need to be addressed in related
to interoperability between different gadgets, central coordination for handling data,
and utilization of resources. For example, developers are working with applications that
create massive volumes of new and rapidly changing data types. These data types could
be in the form of structured, semistructured, unstructured, and polymorphic (Miorandi
et al., 2012; Swan, 2012). The processing of all these different types of data is a challeng-
ing task. If an application requires the processing of just anyone of that, it creates lot
of ambiguity in recognition of such data type and formats. ESB is software framework
that provides multiple service integration over common and shared platform. ESB also
support multiple data format and type that helps any service to recognize its format.
337
Sensor
networks Home and
Sensor Home and
Bluetooth health health devices
networks
devices
IP IP6LoWPAN Actuation
networks
IoT’s
Actuation architecture
networks
Propretory NFC
Entertainment Entertainment
networks ISO NFC
networks
14443
Interconnectivity by IoT’s architecture
FIGURE 15.1
From different technologies to Internet of Things.
It also enabled the system with known set of pattern, namely, Enterprise Integration
Pattern (EIP). When there are many types of data that need to be processed, current solu-
tions could be implemented using ESB.
In the past few years, there has been a massive increase in the interest of develop-
ing such as NFC, Sensor network, and many more as shown in Figure 15.1. With the
increase in the popularity of IoT in today’s world, efficient processing of the data sent
from different device can be merged with the IoT platform. The major task that needs
to be dealt with is to provide integrating platform which could help in handling data
and its format (Dastjerdi & Buyya, 2016). As in many applications, even the slightest of
latency could be problematic. Making this transmission of data more efficient would
have a great impact on such applications. Such impaction could efficiently be done by
using ESB.
15.2 Enterprise Service Bus

An Enterprise Service Bus (ESB) fundamentally is architecture. It is a set of principles and
rules for the integration of many applications together over an infrastructure that is bus-
like. There are various ESB products. They all enable users to build architecture of this type
but differ in the way they do it and capabilities that are offered by them. ESB architecture’s
core concept is that you integrate various applications by inserting a communication bus
between them and then enable all applications to talk to the bus (Bhadoria, 2015; Bhadoria
et al., 2017). This reduces coupling between systems so that they can communicate without
dependency on or knowledge of each other on the bus. The ESB concept was created as a
result of the need to do away with point-to-point integration, which becomes very hard
to manage and brittle over time. Point-to-point integration leads to custom integration
code being distributed among applications with no central way to track or troubleshoot.
Spaghetti code, is the popular term used to describe this, and it does not scale as it creates
close dependencies between applications (Aggarwal et al., 2013).
Data Processing in IoT Using an Enterprise Service Bus 339
An ESB’s primary duties:
• Control and monitor routing of the exchange of messages between services

• Resolve contention between the components of communicating service
• Control versioning and deployment of services
• Marshal usage of services that are redundant
• Cater for services such as message and event queuing and sequencing, security or
exception handling, event handling, data transformation and mapping, enforcing
good quality of communication service and protocol conversion
15.3 Integration Core Principles

Mapping ESB architecture to five core principles of integration:
• Service integration: Composing many existing fine-grained components into an

individual higher order composite service to achieve appropriate granularity of
services and promote manageability and reuse of the basic components.
• Protocol transformation: Data transformation between standard data formats and par-
ticular data formats required by each connector of the ESB. Standard (or Canoncial)
data formats greatly simplify the requirements for transformation associated with
implementation of a large ESB in which there are lots of providers and consumers,
each with their own data formats and definitions. For example, conversion of for-
mats between CSV, Cobol to either JSON or SOAP/XML, and vice-versa.
• Incorporate transportation: Transport protocol negotiation between multiple formats
(such as JMS, JDBC, and HTTP).
• Mediation: It interprets different interfaces for the following reasons: (a) incor-
porate different service versions to provide compatibility with existing services,
(b) support implementation for multiple channels with underlying service
component. It may involve multiple interfaces simultaneously to handle same
component with standards formats such as SOAP/XML.
• Nonfunctional consistency: It includes the consistency associated with security and
monitoring policies that are applied in handling service. Moreover, scalability and
availability could be achieved through multiple service instances that are plugged-in
to ESB. It provides with improved throughput (in terms of scalability) and eradicate
single-points-of-failure (SPOFs), which is also covers the primary objective availability.
15.4 Service Modeling and Realization

15.4.1 How Modeling Improves Service-Oriented Architecture
Service-Oriented Architecture’s power is in its ability to enable agility of business through
reuse and integration of business process (Amsden, 2007). This is achieved by SOA in two
ways: (a) By encouragement of solutions organized around services that are reusable and
encapsulate functional capabilities separated from their implementations and (b) by pro-
viding methods to manage coupling between functional capabilities.
Modeling can be used to lean the gap between today’s business demands and service
deployed based solutions. SOA models the level of abstraction in handling data in service so
that one can focus on business goals to achieve high reusability and availability. Such
development approaches can be used in implementing SOA features such as scalability,
reusability, and availability with platforms of Java (J2EE) or IBM CICS. This helps in meet-
ing the agile goals of business functional and nonfunctional objectives.
The term service-oriented architecture (SOA) has several overtones. SOA is used com-
monly by practitioners to both define a style of architecture and describe a common
infrastructure that enables IT systems that are built using that style of architecture to
operate. These are useful perspectives that are technology focused, but they are not
enough by themselves.
To reach its potential, an SOA-based IT infrastructure (or SOA) needs to be relevant to
business, thus business-driven and implemented to support the business. SOA solutions
need to be designed in ways that are connected to the requirements of the business that
they fulfill. This is difficult to accomplish if the requirements of the business are captured
in several XML documents that describe collection of web services.
What do we need? A way to formalize the requirements of business and increase the
level of abstraction such that SOA can resemble business services more closely and how
those services might meet the goals and objectives of the business. The deployed solution
is thus tied to its intended business value. Simultaneously, we need a method to isolate
business concerns from the SOA platforms that support them.
These goals can be achieved with the help of Modeling and model-driven development
(or MDD). Models allow us to abstract the implementation details away and focus on the
issues that drive architectural choices. To some degree, the approach we will depict applies
one of the fundamental principles of SOA to the development of SOA solutions: loose cou-
pling, and separation of concerns. Here, we neatly separate the responsibilities and tasks
of business analysts from those of IT staff.
Generally, business analysts will be centered on business operational and organiza-
tional requirements necessary to meet business objectives and goals that accomplish
some business vision. Frequently, they are not related to (nor skilled enough to deal
with) IT concerns, such as cohesion and coupling, reuse, security, distribution, data
integrity, persistence, failure recovery, concurrency, and others. Further, tools for busi-
ness process modeling often do not have the necessary capabilities to address these
concerns, and, if they did, they most probably would not be effective tools for business
analysts.
15.4.2 Service Realization

The first step in the design of service implementation is to provide the services. That is, to
figure out what service capabilities will be provided by which service providers. This is an
integral part of designing an SOA, as the choice of providers establishes the relationships
between service providers and consumers. Therefore, this establishes both the coupling
between the parts of a system and its capabilities.
All the operations could be put into one service and have a straightforward solution.
But all clients would depend on that single service, which would lead to very high degree
of coupling. Any change in the provider would lead to a possible change in all consum-
ers. This was a typical issue with module libraries in the old days of programming in C.
A separate service could also be created for each functional capability, but this would
result in a system that is very complex that would not reflect good cohesion and encapsu-
lation. The communication between different clients would be difficult in implementing
different interfaces for different data format and type. The reason for such mismatch
in interoperability is the support for different protocols by different clients in interac-
tion. The prominent solution for such interoperability could be ESB. It not only organizes
data and services together but also mediates the system with global security of Denial of
Service (DoS) attack.
In the end, choosing the participants of the service is something that takes a bit of skill
and can be affected by lots of compromises. Distribution can play a key role in handling
data from different interfaces that could be managed by means of different connectors.
Designing SOA-based solution should be independent of service clients and provider’s
location, but that is not very practical solution in general. Services that are deployed to
global solution must not affect the overall performance, security, and availability of SOA
system. A system that ignores all these features in its solution may lead to unacceptable
implementations of service that decreases the service proliferation.
15.5 Enterprise Integration Patterns for Data Handling in IoT

15.5.1 What Is the Need for Integration?
Business applications today rarely live in isolation. Instant access is what is expected by
users to all business functions that can be offered by an enterprise, regardless of which
system the functionality may reside in. This requires disparate applications to be linked
with integrated solution that is based on middleware. This gap is lean with the aid of ESB
that provides the plumbing such as routing, data transport, and data transformation.
15.5.2 What Makes Integration So Hard?

It is a complex task to handle multiple service integrations for different data patterns
together to build a solution for specific application. There are so many other conflicts that
are related to different drivers that support data functionality. Middleware architecture
plays an important role in handling services more efficiently and provides inevitable
additions to overall monitoring for system architecture (Bonomi et al., 2012). Commonly,
different service vendors provide best practices and methodologies for handling service
through their well-defined interface, but these interfaces are inefficient in recognizing dif-
ferent data format and its type. This gives a rise to conflict for software counterfeit in con-
trolling data and its associated format.
15.5.3 Asynchronous Data Architecture

The best strategy for service integration has been supported by architectures that handle
data asynchronously. It allows the loose coupling of services that overcomes the limitations
of remote execution. That is why asynchronous messaging that exchange data from mul-
tiple services are enabled by Enterprise Integration Patterns (EIPs). Unfortunately, asyn-
chronous messaging has its pitfalls. A lot of assumptions are valid when developing single,
synchronous applications but do not support dynamic environment. Service providers

neutrality make the design of system independent of these pitfalls so that it enhance the
system performance and robustness for data architectures (Atzori et al., 2010).
15.5.4 How Can Patterns Help?

Patterns are a proven approach to capturing knowledge of experts, especially in those
fields, where there are no simple universal (one size fits all) answers, such as object-oriented
design, application architecture, or message-oriented integration. A specific problem is
tackled by each pattern by the discussion of design considerations and presentation of an
elegant solution that balances the forces. Many a time, the first approach that strikes us is
not the solution, but an approach that, through actual use, has evolved over a period of
time. Consequently, each pattern incorporates into it gets the experience that past archi-
tects have gained by building solutions frequently and learning from their shortcomings.
Thus, no pattern is invented. Rather, they are observed and discovered from actual practice
(Kulkarni et al., 2016).
Integration of services on common platform of IoT can be handled with the help of
potential end-points. These end-points provide interface to interact with other services
from different domains. This also facilitate to exchange data presented in different format
and type. ESB play a vital in handling and organizes services (Sarma & Girão, 2009). This
scenario is well depicted in Figure 15.2.
Software
Services
User Terminal
Laptop
acts as
acts as
router Router Enterprise
router
service bus
Cellphone User
acts as router
User Vehicle’s
connectivity
Sensor network
FIGURE 15.2
End-point communications in IoT.
15.6 Challenges for IoT Data

Devices having network interface can be categorized into three depending on resources it
possess:
• Devices with multiple processors of high speed with large memory (gigabytes)
and multiple networking options, such as PC, laptops, tablets, and smartphones.
• Devices with few megabytes of memory used in different industrial or home
automation.
• Devices with very small processing power, memory and usually having a single
network interface. These devices might operate for a longer period on long-lasting
batteries. This class includes sensor nodes (such as 8-bit microcontrollers with
32-kB RAM and 512-kB flash storage).
The first two classes of devices mentioned above have many frameworks and protocols that
are confirmed to be efficient enough already, whereas devices specified above in last case
make it tough to figure an interoperable system with aid of same protocols. Henceforth,
there is limited applicability of such techniques that are used to integrate systems. Such
devices should be synchronized with real-time requirements (Cooper & James, 2009).
Cloud infrastructure is frequently used with IoT, and it provides solution based on
devices that are producing data. Figure 15.3 represents the service flow for IoT that
gives an idea of actual stream of dataflows. Initially, some data are created from IoT
device software development kit (SDK) that is responsible for set of client libraries
Rule engine
Services
SDK
IoT Authentication Device gateway

device SDK and
authorization
Device
shadow Applications
Registry IoT API
FIGURE 15.3
Service flow control in IoT.
and exchange of information. When this information is validated through predefined

technique authentication and authorization, this step is secured with encryption. The
device gateway channelizes this secure message to various services that could receive/
send this secure message to further service with the help of rule engines. This rule-
based engine initially transforms the message based on certain logic that follows some
business information exchange. Applications that receive this secure message through
device shadows would keep the device state during entire service transaction and main-
tains this connection. For each information stream, registry is maintained, which assigns
unique identity to each device.
The main power of the IoT paradigm lies in its ability to provide real-time data from
numerous different distributed sources to other smart entities, machines, and people for
many different types of services. One major issue is that the underlying data from the vari-
ous resources are immensely heterogeneous. This can be very noisy and is usually very
distributed and of large scale (Zorzi et al., 2010).
It is therefore hard for other entities to use the data productively, without a clear
definition of what is accessible for processing. To enable productive use of this very
distributed and heterogeneous data, frameworks are required to describe the data in
a way that is intuitive enough, so that it becomes more easy to use, that is, the issue
of semantic interoperability gets addressed. This leads to challenges both in terms of
providing scalable, high quality, and real-time analytics and also in terms of describing
intuitively to users information about what sort of services and data are available in a
variety of scenarios (Ali et al., 2016).
Thus, techniques to analyze, query, manage, and clean the data in the distributed way are
required. The cleaning is performed usually at the time of data collection and is often built
into the middleware that interfaces with the sensor devices. So, the research on the cleanli-
ness of data is often studied in regards to things-oriented vision. The issues of implement-
ing standardized descriptions and data access for smart services are studied in the context
of standardized web protocols and interfaces, and querying/description frameworks as
offered by semantic web technology. The idea is basically to reuse the already existing
web infrastructure intuitively, so the distributed nature and heterogeneity of the various
sources of data can be integrated seamlessly with the different services (Bandyopadhyay &
Sen, 2011).
These problems are examined usually within the web of things’ context and the semantic
web visions. Therefore, the end-to-end data management of IoT technology requires the
collaboration and unification between the various aspects of the development of these tech-
nologies in order to provide an infrastructure that is seamless and effective (Negash et al.,
2015).
15.7 Conclusion
The data and objects in IoT are heterogeneous, as opposed to the World Wide Web of
documents wherein objects are described themselves in terms of a natural lexicon.
Also, the objects and data in the IoT paradigm may not be naturally available in a
way that is descriptive enough to be searchable, unless, of course, an effort is made to
create descriptions that are standardized for these objects in terms of their properties.
The cleaning of data in IoT may be required for different reasons that are concluded
as follows:
• When data are obtained from conventional sensor devices, it may be incomplete,
noisy or may require probabilistic uncertain modeling.
• RFID data are duplicate, redundant, noisy, and even incomplete with regards to
sensor readers.
• The method of privacy-preservation may require a reduction of data quality inten-
tionally, in which case methods are required for processing of privacy-sensitive
data.
References
Aggarwal, C. C., Ashish, N., & Sheth, A. (2013). The internet of things: A survey from the data-
centric perspective. In C. C. Aggarwal (Ed.), Managing and Mining Sensor Data (pp. 383–428).
NY: Springer.
Ali, M. I., Ono, N., Kaysar, M., Shamszaman, Z. U., Pham, T. L., Gao, F., … Mileo, A. (2016). Real-time
data analytics and event detection for IoT-enabled communication systems. Web Semantics:
Science, Services and Agents on the World Wide Web, 42, 19–37.
Amsden, J. (2007). Modeling SOA, Part 1. Service Identification. IBM Developer Works. Retrieve from:
https://www.ibm.com/developerworks/rational/library/07/1002_amsden/index.html?
S_TACT=105AGX15&S_CMP=LP.
Atzori, L., Iera, A., & Morabito, G. (2010). The internet of things: A survey. Computer Networks, 54(15),
2787–2805.
Bandyopadhyay, D., & Sen, J. (2011). Internet of things: Applications and challenges in technology
and standardization. Wireless Personal Communications, 58(1), 49–69.
Bhadoria, R. S. (2015). Performance analysis for enterprise service bus in SOA system. International
Journal of IT Business Strategy Management, 1(2015), 9–16.
Bhadoria, R. S., Chaudhari, N. S., & Tomar, G. S. (2016). The performance metric for enterprise
service bus (ESB) in SOA system: Theoretical underpinnings and empirical illustrations for
information processing. Information Systems, 65(2017), 158–171.
Bonomi, F., Milito, R., Zhu, J., & Addepalli, S. (2012, August). Fog computing and its role in the inter-
net of things. In Proceedings of the first edition of the MCC workshop on Mobile cloud computing
(pp. 13–16). Helsinki, Finland: ACM.
Cooper, J., & James, A. (2009). Challenges for database management in the internet of things. IETE
Technical Review, 26(5), 320–329.
Dastjerdi, A. V., & Buyya, R. (2016). Fog computing: Helping the internet of things realize its potential.
Computer, 49(8), 112–116.
Kulkarni, P. H., Kute, P. D., & More, V. N. (2016). IoT based data processing for automated industrial
meter reader using Raspberry Pi. In Internet of Things and Applications (IOTA), International
Conference on (pp. 107–111). IEEE.
Miorandi, D., Sicari, S., De Pellegrini, F., & Chlamtac, I. (2012). Internet of things: Vision, applications
and research challenges. Ad Hoc Networks, 10(7), 1497–1516.
Negash, B., Rahmani, A. M., Westerlund, T., Liljeberg, P., & Tenhunen, H. (2015). LISA: Lightweight
internet of things service bus architecture. Procedia Computer Science, 52, 436–443.
Sarma, A. C., & Girão, J. (2009). Identities in the future internet of things. Wireless Personal
Communications, 49(3), 353–363.
Swan, M. (2012). Sensor mania! the internet of things, wearable computing, objective metrics, and
the quantified self 2.0. Journal of Sensor and Actuator Networks, 1(3), 217–253.
Zorzi, M., Gluhak, A., Lange, S., & Bassi, A. (2010). From today’s intranet of things to a future
internet of things: a wireless-and mobility-related view. IEEE Wireless Communications,
17(6), 44–51.
Section IV
Future Research, Scope, and Case

Studies of the Internet of Things
and Ubiquitous Computing
16
Infection Tracing in i-Hospital
Mimonah Al Qathrady, Ahmed Helmy, and Khalid Lmuzaini
CONTENTS
16.1 Introduction and Background ........................................................................................ 349
16.2 i-Hospital Nodes Tracking and Processing High-Level Framework ........................ 352
16.2.1 Encounter Sensing and Collection Side ........................................................... 352
16.2.1.1 Sensing Devices ................................................................................... 353
16.2.1.2 Network Architectures.......................................................................354
16.2.1.3 Filters and Aggregation......................................................................354
16.2.1.4 Sending the Tracking and Sensing Information to the Server ......355
16.2.2 Server Side ............................................................................................................ 355
16.2.3 Application Side .................................................................................................. 355
16.3 Infection Tracing: The Infection-Tracing Framework ................................................. 356
16.3.1 Encounters Side ................................................................................................... 356
16.3.1.1 Encounter Collection and Processing .............................................. 356
16.3.1.2 Encounter Range ................................................................................. 358
16.3.1.3 Direct/Indirect Encounter ................................................................. 358
16.3.1.4 Encounter with Human and Objects ............................................... 358
16.3.2 Tracing Side .......................................................................................................... 359
16.3.2.1 Tracing Parameters ............................................................................. 359
16.3.2.2 Disease Spread Modeling .................................................................. 359
16.3.2.3 Infection Tracing ................................................................................. 360
16.3.3 Evaluation for i-Hospital Infection Tracing System ....................................... 365
16.3.3.1 Evaluation Metrics .............................................................................. 366
16.3.3.2 Simulation and Evaluation ................................................................ 366
16.4 Conclusion ......................................................................................................................... 368
References..................................................................................................................................... 369
16.1 Introduction and Background

Infection transmission risks are present in all hospital settings as well as other com-
munities. According to the CDC (2007), humans—patients, healthcare personnel, or
visitors—constitute the primary source of infectious agents, along with other objects
such as contaminated equipment. Direct contact and indirect contact with individuals or
objects are the primary methods of transmission of infectious agents. Also, Middle East
Respiratory Syndrome has an outbreak in South Korea was traced to a patient who has
been admitted to a hospital (Tina, 2016), where the virus is transmitted to 82 other people,
mostly visitors and fellow patients. Moreover, according to a survey that was based on a
349
large sample of U.S. acute care hospitals, about 1 in 25 hospital patients have at least one
health-care-associated infection (Magill et al., 2014). These, as well as cases of H1N1 Swine
Flu and Ebola, in which more than 1,300 patients were admitted in a Singapore Hospital
because of H1N1 (Subramony et al., 2010). Also, there are more than 4K reported deaths in
West Africa in the first nine months of the Ebola epidemic (Team, 2014). These cases pro-
vide concrete examples that motivate our work of infection tracing.
When there is a case of epidemic disease, it is vital to identify the original sources to
control and prevent further spread of infection by tracing the sources and then tracing
forward from the sources and identifying the population with higher risk including the
nodes that contacted the sources and might be infected as well.
As many smart hospitals and communities are equipped with mobile and sensing
devices to track the objects’ movement and their encounters, we propose to enhance these
technologies and act immediately whenever a case of infection is detected. Our target is an
intelligent hospital that deploys Internet of Things and able to track the encounter between
individuals and objects.
Using a mobile application to collect encounter information has been used in the past
different studies for different purposes. In iTrust (Kumar & Helmy, 2012), the encoun-
ter information is filtered and ranked for social reasons to identify trusted encounters. It
ranks the encountered nodes based on the duration and frequency of encounters.
Other researchers have worked on studying the infection transmission and dynamics
in different populations. They captured the encounters between peoples using sensing
devices. A wireless sensor network technology was used to obtain close proximity data
between individuals during a typical day at an American high school; then, the data are
used to construct the social network relevant for infectious disease transmission (Salathé
et al., 2010). Also, face-to-face interactions between the attendees at a conference were cap-
tured using radio-frequency identification (RFID), the spread of epidemics along these
interactions was simulated (Stehlé et al., 2011).
Mobile phone devices and Bluetooth have been used in other works as well to collect
the encounters and study or map the epidemic disease. Mobile phones are used to collect
human contact data and record information such as locality, user symptoms for flu or
cold, and human interactions. Then, the data were used to develop mathematical models
for the spread of infectious diseases (Yoneki & Crowcroft, 2014). The sensed interaction
by mobile phone and Bluetooth data is proved to be suitable for modeling the spread of
disease and increase the predictive power of epidemic models (Farrahi et al., 2015). Also,
GPS in mobile phones was used for tracking the epidemic on the map; after the patient
is diagnosed with Dengue fever, the patient’s location is used to draw the epidemic map
(Reddy et al., 2015)
The previous researches have focused on modeling the disease transmission, which is
only one component of the i-hospital infection-tracing framework.
More intelligent hospitals in the future will deploy Internet of Things which consists
of heterogeneous devices. As a result, the encounter collection framework in this chap-
ter is designed to be used with different heterogeneous devices as it will be the case
with the Internet of Things. The sensing devices will gather the encounter data. Thus,
the applications that use the framework are intended to be technology independent.
They could potentially work with data collected from various technologies, for exam-
ple, RFID or mobile phones.
Another important part of our infection-tracing framework is tracing back to infection
sources. The concept of tracing back has been usually used as a defense mechanism in com-
puter networks to identify the source of attacks (Peng et al., 2007). Kim and Helmy (2010)
Infection Tracing in i-Hospital 351
introduced a protocol framework to trace back the attackers in mobile multihop networks;
the problem is complicated by mobility.
Tracing back the attackers in a computer network is less complicated than tracing
the disease infection for several reasons: the main reasons are related to the nature of the
agents’ spread methods and ranges. There may be a different range (and method) for the
disease spread, sensing, and communication. Such differences must be taken into con-
sideration during the design, simulation, and analysis. Also, to make the attack happen
in computer networks, there must be some kind of communication between nodes, so
information about the communicating nodes can be saved. On the other hand, when the
disease spreads from one person to another, it is impossible to record the exact moment
of the infected nodes during the transmission process.
Trace back and forward for a source of problem and subjects that have been affected
by a problem were used in the foodborne outbreak to identify the source of a product
that was implicated in the foodborne outbreak (Weiser et al., 2013). Tracing back the sup-
ply chain is different than tracing the disease. This is because the supply chain database
has certain data, whereas the certain data when the infection transmitted between two
individuals are not available.
Several researchers have tried to solve the problem of identifying a source of differ-
ent kind of infection like rumors when it spread in the society. For example, Shah and
Zaman (2011) obtained an estimator for the rumor source based on the infected nodes
and the underlying network structure. Others considered the problem of identifying
an infection source based on an observed set of infected nodes in a network (Luo &
Tay, 2013). The works of finding the source out of the infected population usually have
knowledge about the group of infected nodes and attempt to figure out which one is
the source.
The idea of infection tracing in this chapter has more challenging scope where the knowl-
edge about infected nodes is not available, and the system attempts to find the sources and
population at risk after knowing only one case.
Contact tracing is an important mean of controlling infectious diseases. Armbruster
and Brandeau (2007a) developed a simulation model for contact tracing and used it to
explore the effectiveness of different contact-tracing policies in a budget-constrained set-
ting. A simulation model of contact tracing is used to evaluate the cost and effectiveness
of different levels of contact tracing (Armbruster & Brandeau, 2007b).
This chapter will explain the infection tracing framework that exploits the Internet
of Things in the intelligent hospital that collects encounter information. Also, it will
explain how this encounter data can be used to trace automatically the source of infec-
tion and population at risks as soon as one case of infection is detected.
An i-hospital system that traces back from infected node to identify the original sources
of infection will control and prevent further spread of infection and then guide the trace
to identify infected population even before the patients report their cases.
The first part of the chapter is about an i-hospital high-level framework that gathers
the encounter data from different heterogeneous devices of the Internet of Things (IoT),
processes and utilizes them for various applications. It focuses on the encounter collection
and processing.
The second part is about the infection-tracing framework. It shows how the encounter
data that collected from IoT can be utilized to tackle the tracing problem in the intelligent
hospital. Then, the disease infection source trace back problem is defined. Selection meth-
ods that work on traced to identify the more likely infected nodes are described. Then,
simulation and evaluation results that use real wireless data are presented.
16.2 i-Hospital Nodes Tracking and Processing High-Level Framework

The i-hospital nodes tracking and processing high-level framework consists of three main
components: encounter sensing and data collection side, server side, and application
side. First, the encounter data are collected, aggregated, processed, and sent to the server,
where it is processed further. The data will serve a wide variety of applications. One of
the applications is infection tracing. The framework components interact with each other,
by sending data, control messages, or parameters between each other. This framework
assumes the availability of the infrastructure; other architectures will be described in the
next section where the hospitals lack or have limited infrastructures.
16.2.1 Encounter Sensing and Collection Side

This side is represented by the block A in Figure 16.1. The encounter sensing and data
collection side consists of three main components: the sensing devices in the IoT that
capture their encounters, the network architectures, and the filtration and processing of
encounter data before sending them to the server.
(A) Encounter sensing and (C) Application side

collection side (B) Server side
Filters:
Noise reduction Mining
aggregation Searching Domain specific
in network processing Predicting guidelines and metrics:
missing
Information Encounter duration
Back tracking Encounter distance
Encounter frequency
Network Encounter context:
architectural Network architectures Location
design
Sensing devices
Risk group and

assessment
FIGURE 16.1
i-Hospital high-level framework for collecting encounters and processing them. Block (A) shows the encounter
collection side. Block (B) shows the server side that is used for storing and running the applications. Block (C)
represents the application sides, where various applications with different parameters can use the collected
encounter data.
16.2.1.1 Sensing Devices

Smart hospitals project used an RFID as the main device to track objects and people
(Fuhrer & Guinard, 2006). However, our i-hospital framework works with differ-
ent heterogeneous sensing devices in IoT to collect their encounter information. For
example, RFID tags, mobile and Bluetooth devices, access points, IoT-based wristbands,
and IoT-enabled beds and equipment. These devices have different ranges when sensing
encounters, and the i-hospital framework takes the sensing ranges into consideration.
RFID could be passive or active tags attached to human or objects. Its range depends
on its type. Tags could record its encounter with other tags as in active tags, or the read-
ers track tags. RFID-passive tag has a microcircuit and an antenna. It receives its power
from the energizing electromagnetic field of an RFID reader (or interrogator). The energy
coupled from the electromagnetic field undergoes voltage multiplication and rectification
to power the passive tag’s microelectronics. The tags should be placed within the range
of an RFID reader. Passive RFID tags are different in how they receive the power from
readers or how to communicate data to readers. Active teas used batteries and can serve
longer ranges than passive tags. Cisco (2014) gives more explanation for the differences
between active and passive tags can be found in the Cisco (2014).
The read range of passive tags (tags without batteries) depends on many factors: the power
of the reader, interference from metal objects, or other RF devices and the frequency of oper-
ation. In general, low-frequency tags are read from a foot or less. High-frequency tags are
read from about three feet and ultra-high frequency (UHF) tags are read from 10 to 20 feet.
Active tags use batteries to boost read ranges to 300 feet or more (RFID journal, 2016).
Regular mobile device with Bluetooth capability can use Bluetooth to record encounters.
Also, some devices running Bluetooth can be attached to objects to record encounters.
Bluetooth record encounters within 10 to 100 m.
Another example of sensing devices is access points, in which they record the MAC
address of associated devices. Two devices are more likely to be encountered if they associ-
ated with the same access point at overlapped intervals of time.
Each sensing device has its capabilities and limitation. For example, access points have a
wider range, but it registers encounters that might not be captured by RFID. Thus, incorpo-
rate different sensing devices and combining their encounter records will potentially help
increase the accuracy of encounter tracking data. For example, data from different sources
could identify more encounters than what could be reported by only one sensing device.
They also could identify that the probability of two subjects did not encounter each other
even if they have been reported as being encountered by one sensing device.
In fact, human or objects in the intelligent hospital will be more likely be equipped with
heterogeneous devices that cooperatively perform the sensing and tracking functions.
Other tracking devices may be placed in static places like RFID readers. Other sensing
devices are expected to be carried by mobile users like mobile devices. Mobile devices
may not only capture its encounter with each other by using Bluetooth, for example, but
also they enhance the functionality of RFID technology. For example, RFID readers may
miss reading tags when they are out of its range, but some mobile device may have the
ability to read tags and could supplement the tag information to the server.
Other devices in the environment are special nodes and have more capabilities than the
rest such as filters. They process the data before sending them to the server.
When talking about the heterogeneity of sensing technology, we have two aspects: the
heterogeneity of the device’s capabilities and the heterogeneity of nodes that are equipped
with the sensing devices.
Not only RFID passive and active have different capabilities, but also the mobile phones
themselves have different capabilities. For example, sensing between mobile phone is not
bidirectional where one device has more power and can read the other device while the
other one cannot read it. This device heterogeneity should be addressed. Their effect on
the collected data must be measured and predicted.
Nodes are not only heterogeneous because of the sensing devices that sense their
encounters, but they are also different based on their roles. For example, some nodes are
mobile and move from one place to another, whereas other nodes are static. Nodes similar
to equipment are different based on their frequency of use, will they be used only for one
time like a bed sheet or could be used multiple times like a wheelchair or an expensive
X-ray device. If the nodes are human, their role can be defined as if they are patient, visitor,
or hospital staff. The node’s mobility can be based on their role too, by taking into consid-
eration the area they are allowed to move to or restricted from.
16.2.1.2 Network Architectures

The high-level framework assumes the availability of an infrastructure that receives the
encounter data. When the devices send their data to the server, there are different possible
architectures of collecting the data. One approach is that the devices send the data directly
to the server, and then the data are processed and stored in the server. The drawback of
this method is contention on the server side. Another approach is that the devices send
their data to other dedicated devices in the networks, where these dedicated devices pro-
cess the data, and then send a cleaner version of the data to the server. Having dedicated
devices in the network to clean the data before sending it to the server will reduce the
overhead on the server, as only limited devices connected to the server, the dedicated ones,
and not all the devices in the area.
This concept of collecting data is different than network architecture concept when there
is a lack of infrastructure or servers, which will be explained in the next section.
16.2.1.3 Filters and Aggregation

The purpose of the filters is to reduce the overhead on the server by removing noise,
aggregate data, and remove redundancy. They process the data before sending them
to the server. Filters reduce the pressure on the server by reducing the amount of data
flood to it, especially when they remove the redundant data. They are also reducing the
overhead on the server processing as they are cleaning the raw data, filtering the noisy
and redundant raw data, and cross-correlation of reported information from different
devices before sending them. There are many levels of aggregation performed by the fil-
ters; the level depends on the granularity of the system. The basic level is an aggregate of
the result reported by a device or multiple devices. For example, if the mobile device
keeps sensing the same equipment, and the equipment does not change its position, then
instead of sending the same information to the server, the report could be aggregated
into one report indicating that the time period and that equipment is still in its posi-
tion. They may also detect missing information and run some algorithm to fill the gap.
Detecting missing encounter information and inferring them is also run on the server
side as it needs much information from multiple sources similar to layout planning and
rich encounter data.
Filter nodes could be special ones with storage and processing capabilities, in which
the mobile devices send the collected information to them similar to dedicated devices
described earlier in the network architecture; The other option is installing some filter
functionality within all or selected groups of mobile devices.
16.2.1.4 Sending the Tracking and Sensing Information to the Server

Sending encounter information can be done either by pulling or pushing the data. When
nodes push their data, they do it constantly or periodically without waiting to be asked.
On the other side, if the data will be taken by pulling, the nodes wait until it being queried
about their data or asked to send them. In this problem, push and pull methods may be
used at the same time, with different spectra. Nodes could push information after a spe-
cific amount of time and after processing and aggregating them.
Also, it may depend on kind of the tracked objects; for instance, information about
expensive objects is constantly pushed. Another way is to recognize the frequency of
query about different kinds of objects; if the object is frequently queried, then information
about it is constantly pushed; other information is pushed when the node, for example, is
leaving the facility; otherwise, its encounter data are sent when they are requested.
16.2.2 Server Side

The server plays an important role in this framework, in which it receives the mobile and
encounter data, processes them further, and runs the application—mining the massive
amount of data such as classifying nodes or clustering them into groups.
After receiving the encounter data, the server will put the encounter data into further
processing, such as detecting the gap and filling the missing information. Detecting the
presence of missing encounter information is very crucial in an environment similar to a
hospital.
When the server detects missing data, it is expected to infer the missing data.
In some cases, when there is something abnormal in the data it receives, it will notify
the administration about the case. These notifications might send to a human being who
is in charge of interacting with the system. Then, the human will answer the server if
the situation is normal or not. For example, the situation is normal if the crowd is not
available in the facility for that period of time for some reason, and eventually there is no
sensing information. In this case, the system may not go further to fill the gap and adjust
the model.
The presence of missing information can be detected by either direct inferring or more
complex model that relies on analysis of historical information.
16.2.3 Application Side

Various applications will utilize the tracking and sensing information that has already
been collected. The applications supply the server with their domain-specific parameters.
Using an appropriate algorithm with the application parameters, the servers will return
back a suitable solution or recommendations for each problem.
An example of problems that will highly benefit from knowledge about encounters
between objects is improving the hospital layout. In a previous study (Brown et al., 2014),
the impact of building space on social interactions was measured by using encounter data
that were collected by wearable sensing devices.
Another significant problem, which is the core of this chapter, is infection tracing.
Examples of application parameters are duration, frequency, distance, or encounter
definition. Duration is the amount of time two nodes encounter each other, whereas fre-
quency is the number of times two nodes have encountered each other regardless of the
encounter duration. Distance can be defined as how many meters two nodes were far
apart. The distance is usually a probabilistic distance, and it depends on the sensing device
that records the encounters of two nodes. The encounter is used to be defined as two nodes
have been in specific distance from each other at the same time. However, an application
similar to infection tracing may have various definitions for encounters. It may define at
the nodes that have been in the same place within T time, as in indirect encounter, which
will be described in the next section.
16.3 Infection Tracing: The Infection-Tracing Framework

Infection tracing is an important application that utilizes IoT in the intelligent hospital and
the collected encounter data. The application goal is to trace infection by identifying the
sources of infection and population that might be at risk of being infected.
The framework that is used for tracing is shown in Figure 16.2. It consists of three
main parts: encounter, tracing, and system evaluation side. Encounter side focuses on the
encounter definition and parameters while dealing with infection. Tracing side describes
infection tracing algorithms. Finally, the evaluation side defines the metrics that will be
used to measure the infection tracking system.
16.3.1 Encounters Side

This side concerns about the encounter collection and processing for tracing infection.
Block A in Figure 16.2 shows its parts. The high-level framework described in Section 16.3
has already explained the general parts of the encounter data collection. The encounter
side in this section will point to encounter specifications that should be taken into consid-
eration when we are dealing with infection tracing.
Different architectures of encounter collection will be explained briefly. Then, we will
point to other encounter issues such as encounter ranges, direct and indirect encounter,
and the subject of encounter: human or objects.
16.3.1.1 Encounter Collection and Processing

Collecting the encounter information can be performed using one of several network
architectures. Some architecture uses a centralized server (or group of servers) to which
nodes push their sensed data. The server then processes the data using sanitation,
aggregation, prediction, and trace-back algorithms. This architecture was explained in
Section 16.3 and represented by block B in Figure 16.1. Semicentralized architecture
is another useful architecture when there is limited infrastructure; it uses small throw
boxes in parts of the area (e.g., floors of a hospital building). When a node encounters
with a throwbox, it pushes its collected data. The throw boxes communicate with each
(A) Encounter side
Sensing devices
Encounter processing
architectures: Encounter range:
Centralized Infection range

Semi-distributed Sensing range
Fully-distributed Communication
range
Inferring the missing Type of encounter:

Subject of encounter:
encounters:
Direct
Human object
Inferring algorithms Indirect
Area layout
(B) Tracing side
Parameters: Infections period, latent period,

immunity parameters, distance, ...
Disease spread modeling:

Tracing algorithms:
SEIR
SIS Tracing backward for sources
SIR Tracing forward from sources
.. Tracing forward from infected
Filtering and selection method:

Mostly named
Link infection probability All selected nodes
Random selection
Final suspected node
(C) Evaluation side

System evaluation metrics:
Infection percentage, coverage, accuracy, specificity, sensitivity
FIGURE 16.2
The infection-tracing framework contains: Block (A) shows the Encounter sensing parts for the collection and pro-
cessing of encounter data. Block (B) shows the tracing parts, where all suspected nodes are traced and then
group of nodes will be selected by using any of the filtering methods. Block (C) shows the evaluation metrics
for tracing algorithms.
other during a distributed trace-back process. Another possible architecture is a fully dis-
tributed one, in which nodes save their encounters locally and only share it on-demand
as requested by other mobile nodes. The mobile nodes can also have different roles or
designations based on the heterogeneity of their characteristics (e.g., staff/doctors versus
patient/visitor). Clearly, there is a trade-off between the cost, energy, delay, and accuracy
of information between the architectures.
16.3.1.2 Encounter Range

There are three different ranges:
1. Sensing range: The range in which two devices can sense each other.
2. Communication range: The range in which two devices are able to communicate
directly with each other.
3. Infection range: The distance between an infectious person and others that allows
for the infection to be transmitted.
For each range, the reported encounters are different. Hence, when the encounters are
ordered on the basis of frequency and duration of encounters, the ranking will differ
between ranges. For example, using Bluetooth encounters to identify individuals with
maximum patient-contact time may not return the accurate list of people that have
encountered the patient the most. This is because the range of agent infection is much
smaller than Bluetooth. Thus, such approach may not obtain the correct ranking of our
group of interest. This issue needs to be taken into consideration when utilizing encoun-
ters in infection trace-back.
16.3.1.3 Direct/Indirect Encounter

Direct encounters are when nodes come in encounter range of each other. These encounters
can be recorded directly using mobile devices, unlike indirect encounters with a patient.
An example of an indirect encounter is when an individual enters a room after a patient
has left, though the disease agent is still present. Indirect encounters cannot be recorded
directly as the two nodes are no longer within a sensing range of each other but can be
inferred using encounter records, distance, time, and the layout of the hospital.
16.3.1.4 Encounter with Human and Objects

A point—usually neglected but important in our case—is the human–object encounter.
Usually, humans are the main carrier of the devices that recognize the contacts and save
them, but they are not the only source of infection. Encountering contaminated equip-
ment can be as dangerous as well. Therefore, there is a need for recording encounters
with sharing equipment such as wheelchairs, operation tables, X-ray devices, among
others.
These devices usually are tracked in smart hospitals. Therefore, the encounter between
them and human can be easily found and should be taken into consideration when there
is an infection case.
6 7 6 5 6
7 5 7 5 6
8 4 7 5
8 4 8 4 8 4
9 3 9 3 9 3 9 3
10 2 10 2 10 2 10 2
11 1 11 1 11 1 11 1
12 20 12 20 12 20 12 20
13 19 13 19 13 19 13 19
14 18 14 18 14 18 14 18
15 17 15 16 17 15 17 15 17
16 16 16
(a) (b) (c) (d)
FIGURE 16.3
Illustration for the contact-tracing concept. In (a) the nodes after the disease spread, nodes from 1 to 10 are
infected, and 11 to 20 are not, (b) the beginning of the tracing where we only know one infected node (node 1),
(c) shows the result of tracing all the suspected nodes, and (d) random selection from the traced nodes.
16.3.2 Tracing Side

The contact tracing process is invoked as soon as an infected case is confirmed (in the
hospital). Figure 16.3 illustrates the concept of tracing. The infected nodes after the infec-
tion spread is shown in Figure 16.3a in which colored nodes are the infected ones. In
Figure 16.3b, the confirmed infected case is represented by node 2. Tracing starts imme-
diately after one case has confirmed. In Figure 16.3c, the tracing has been completed, and
all the suspected nodes will be included in the coverage. However, the covered nodes
will include some noninfected nodes. Therefore, a selection method might be necessary to
identify the population that is more likely to be infected. Figure 16.3d shows the result of
a random selection method in which the coverage is reduced; however, there are multiple
missing infected cases.
The tracing process consists of three steps: finding sources by tracing back, tracing
forward from sources, and tracing forward from the infected population.
16.3.2.1 Tracing Parameters

Several main parameters of the infection tracing process are informed by knowledge of
the disease itself. Examples of parameters include the immunities of the population to
specific diseases, and the number of confirmed infected cases to trace-back from, latent
period, infection period. As a result, there will be various forms of the infection tracing
techniques for different circumstances and diseases. Main parameters that shape the
tracing that explained in this chapter are as follows:
• Infectious period (pi): The period length in which an infected node is infecting others.
• Latent period (pl): The period between getting the infection and start infecting others.
16.3.2.2 Disease Spread Modeling

The disease spread modeling is an important component in tracing framework. Tracing
infection will mimic in reverse to trace the infection spread. There are several models
for disease spread in a population. The models depend on the disease itself, for example,
if the patients will be immune after recovery or still be susceptible after recovery, and
they might be infected again.
The following words are usually used to define the models:
Susceptible (S): The state in which the node might get infected.
Infected (I): The state in which the individuals have the disease and may infect others
whom encounter with.
Recovered (R): The state after the node has recovered from the infection and will not
infect others while in this state.
Exposed (E): The state in which the node has already infected. However, during this
state, the node will not infect others.
Examples of models that have been used to simulate the disease spread in a population
are as follows:
Susceptible-Infected (SI) model: Susceptible nodes may get infected, and infected nodes
do not recover during the simulation.
Susceptible-Infected-Susceptible (SIS) model: In this model, the infected nodes may
recover from the disease and get infected again.
Susceptible-Infected-Recovered (SIR) model: Infected nodes may recover, and recovered
nodes will not be infected again.
Susceptible-Infected-Recovered (SEIR) model: When the nodes got an infection, they
become exposed in a number of days for the latent period then become infectious
for a period of time. After the infectious period, the nodes are recovered and will
not be infected again.
More explanation of various mathematical modeling for the disease spread can be found
in the mathematical theory of infectious diseases and its applications book (Bailey, 1975).
Also, there are other factors that should be taken into consideration when modeling
the disease spread, especially if the information is available. Examples are immunity
factors, like if the individual was immunized or not. Another factor when simulates the
disease is the ages of persons in which elderly and less immune people may be at higher
risk of infection than others.
16.3.2.3 Infection Tracing

The process of tracing begins from an identified case of infection. The tracing backward
and forward concepts are illustrated in Figures 16.4 and 16.5. The figures show the encoun-
ter of nodes during specified days. Nodes are identified by numbers, and the edge between
two nodes represents an encounter between the two nodes during that day. For example,
on day 1, there is an encounter between node 20 and node 19, and node 19 encountered
with nodes 3 and 12 on the same day.
Node 2 in the figures is the confirmed cases of infection. Therefore, the tracing process
will start from this node. We will assume for now that the trace window is only one-day
backward or forward. The trace-back process begins on day 3 by tracing node 2 encounters,
which are nodes 8 and 12. Then, the encounters of 8 and 12 on day 2 will be traced, which
are nodes {6,4,19}. The tracing back will continue from 6, 4, and 19, and tracing back its
Day 1 Day 2 Day 3
6 5 7 6 6
7 5 7 5
8 4 8 4 8 4
9 3 9 3 9 3
10 2 10 2 10 2
11 1 11 1 11 1
12 20 12 20 12 20
13 19 13 19 13 19
14 18 14 18 14 18
15 16 17 15 17 15 16 17
16
FIGURE 16.4
Back tracing example for three days, starts from day 3.
Day 1 Day 2 Day 3
6 6 7 6
7 5 7 5 5
8 4 8 4 8 4
9 3 9 3 9 3
10 2 10 2 10 2
11 1 11 1 11 1
12 20 12 20 12 20
13 19 13 19 13 19
14 18 14 18 14 18
15 17 15 17 15 17
16 16 16
FIGURE 16.5
Forward tracing example for three days starts from day 1.
encounters on day 1. At the end of tracing back process, six nodes will be selected, which
are {3,4,5,6,7,8,12,19}. Table 16.1 shows the summary of tracing back each day:
The trace-forward will start from traced nodes that were selected on the trace back pro-
cess. It starts from day 1 and trace the encounters of {3,5,7,12} as they are the nodes traced
on day 1 on the tracing back. The dark gray arrows in Figure 16.5 show the trace-forward
process. Table 16.2 summarizes the trace forward process:
TABLE 16.1
Summary of Tracing Back; the Nodes it Starts
from and the Traced Nodes on Each Day
Day Trace from Traced Nodes
3 2—the confirmed case {8,12}

2 {8,12} {6,4,19}
1 {6,4,19} {7,5,3,12}
Total: {3,4,5,6,7,8,12,19}
TABLE 16.2
Summary of Tracing Forward; the Nodes it Starts from and the
Traced Nodes on Each Day
Day Trace from Traced Nodes
1 {7,5,3,12} {6,16,4,19,18}
2 {6,16,4,19,18} {3,5,6,7,12,16,17}
3 {3,5,6,7,12,16,17} {1,2,4,8,11,15,17,20}
Total-> {1,2,3,4,5,6,7,8,11,12,15,16,17,18,19,20}
Identifying the Infection sources, which are the nodes that infect other nodes, is achieved
by tracing backward. The population at risk is identified by tracing forward from sources.
The population at risk might or might not infect others. The difference between finding
the sources or population at risk is the process of tracing and the tracing window, as we
will explain in the next section. Parameters of disease like (pi) and (pl) are used to identify
the tracing backward or forward windows.
16.3.2.3.1 Tracing Windows

There are three tracing process types: Tracing backward for a source, trace forward from a
source, or tracing forward from an identified infected. Different windows of tracing are
estimated on the basis of tracing process type. Figure 16.6 shows the tracing windows.
1. Trace-Back process for a node: It starts from an observed infected node. It aims to
find the source of infection for this node. The process produces a group of nodes
that contains the source. The windows of tracing depend on our knowledge of
infection day. If the infection day is known, then the encounters will be traced
1st scenario for 2nd scenario for

Tracing window day(x) day(x)
pl pi
(a)
1st scenario 2nd scenario for

day(y) day(y)
Tracing window
pl pi
(b)
Day(z) Tracing window
pl pi
(c)
FIGURE 16.6
Time lines that shows the scenarios and the tracing windows. (a) shows the trace back for sources scenarios
and tracing windows. (b) shows the first- trace forward process scenarios and tracing windows. (c) shows the
second-trace forward process tracing window where the infection day is assumed to be known.
for that day only. However, if it is not, we need to estimate a tracing window.
For example, on the day(x), we only know that this node has been infected, but
the day of infection is unknown, and we do not have other knowledge of where
the nodes are on its infectious period. Consequently, we estimated the knowledge
of tracing windows on two scenarios, in which every possible day of infection
will be included, and the source of infection will be on the traced list.
a. First scenario: The infectious node is on the first day of its infectious period.
Consequently, the node was infected on day(x) − pl.
b. Second scenario: The infected node is on the last day of its infectious period.
Hence, the nodes’ infection day was on the day(x) − (pl + pi) + 1.
Figure 16.6a shows the day(x) on both cases.
All other scenarios are in between two assumptions. As a result, all of the nodes
encounter during this window will be traced:
{day( x) − (pi + pl ) + 1, day( x) − pl }
The nodes that resulted from this window of tracing will be considered as sus-
pected sources of infection for an input node.
2. Trace forward from a source: The node that we trace from here is a suspected source
of infection. We aim to identify the nodes that might be infected by encounter-
ing this node. Also, the window of tracing depends on our knowledge of this
nodes infectious period. The worst case is when the only known is the node that
was infectious on the day(y) and when the node started the infectious period or
ended it are unknown. Consequently, we have two possible scenarios.
a. First scenario: The node is on the first day of its infectious period; hence, it will
continue infecting its contact until the day(y) + pi − 1.
b. Second scenario: The node is on the last day of its infectious period, so it has
infected others since the day(y) − pi + 1.
Figure 16.6b shows the day(y) on both cases.
All other scenarios are in between two scenarios. As a result, all of the nodes
encounter during this window will be traced:
{day( y) − pi + 1, day( y) + pi − 1}
The resulted nodes will be considered as population with the risk of being infected
by encountering this node.
3. Trace forward from an infected node: It applies to a node in which the day of infec-
tion for this node is known. The process aims to find a set of nodes that might
be infected by this node. Day(z) is assumed to be the day when the node got
infected, and the node will start infecting its encounters after its latent period
until it finishes its infectious period. Therefore, the contacts in the following win-
dow will be traced.
{day( z) + pl , day( z) + pl + pi − 1}
Figure 16.6c shows the estimated scenarios and the tracing window.
16.3.2.3.2 Tracing Approaches

The tracing approaches used the tracing windows that have been described earlier. We
will describe one approach that consists of two parts:
• Tracing all suspected nodes S.

• Selecting groups of nodes from the S.
16.3.2.3.2.1 Tracing All Suspected Nodes (S) The result S will include all the infected
nodes, but it will also include some noninfected ones, the false positive cases, as shown
in Figure 16.3c in which nodes {1,5,8,11,17,20} are included in the trace, even they are not
infected.
Tracing S has three phases: tracing sources, tracing the infected nodes from sources,
then final tracing of infected nodes.
a. Tracing sources phase: The result of this process is a group of nodes that are sus-
pected to be the sources of infection on specific days. The process starts with the
confirmed infected case “I” that is known to be infected on day N. Day 1 is the first
day where the infection started in the network.
b. Trace-forward from the sources phase: This process works on the result of the previous
phase. Let us call the result of the previous phase Sources. Sources[i] contains the
list of nodes that are suspected to be sources of infection on the day(i). As a result,
Sources[i] were infectious on the day(i), and they are transmitting the infection to
their encounters during their suspected period of infection.
c. Trace-forward from the infected phase: The previous phase will list nodes that are sus-
pected to be infected on defined days. If we assume that they are infected, they are
in return will spread the infection. In this phase, we trace from nodes that their
days of infection are assumed to be known, whereas the previous phase trace from
nodes that the days they got their infection are unknown.
The above tracing phases will produce all suspected nodes S. S will include all the infected
nodes along with other noninfected nodes. For the purpose of controlling the disease
spread, nodes in S are monitored. However, it might be infeasible to contact and monitor all
of them due to their large numbers and the limitation of the resources such as time, money,
and staff. Consequently, selection methods have been advised to select a group of S. The
selection method aims to select the infected nodes in S and ignore the noninfected ones.
16.3.2.3.2.2 Selecting Subset Nodes of S The selection methods attempt to classify nodes in S
as infected or not, then select the ones that are classified as infected. This section will intro-
duce three selection methods: random selection, mostly named* (Mn*), and encounter
infection probability (EncP). Figure 16.3d is an example of selecting nodes randomly from S.
a. Random selection: This method is used mainly for reference purposes. The subgroup
of S is selected randomly where each one has the same probability pr to be selected.
b. The mostly named* (Mn*): Mostly named as the contact tracing method has been
proposed by Armbruster and Brandeau (2007a). The index case is asked to name
its contacts. Then, the named contact is assigned a score based on the number of
index cases who named it. Each time, an index case has to be identified to continue
the contact tracing and update the scores. The named contacts are ranked from
highest to lowest, and the top K contacts are selected. This policy is modified in
i-hospital infection tracing to work on mobile and wireless data, and we called it
mostly named* (Mn)*. In Mn*, the patient will not be asked about their contacts.
After an infected case is detected, the contact tracing is invoked; the tracing
continues automatically through wireless encounter data, and attempt to find
all the populations at risk. During tracing, a counter is assigned to each node.
The node counter will be incremented by one each time the node is traced. After
that, the nodes are ranked on the basis of their counters from highest to lowest,
and the top “K” will be selected as the most suspected ones.
c. Encounter infection probability (EncP): The probability of infection for a node during
its encounter with an infected node is p for each second, and (1 − p) is the probabil-
ity for the node to not be infected in this second. As a result, if the node encounter
with an infected node for a duration (d), the probability of infection (pe) will be
pe = 1 − (1 − p)d
Consequently, the probability of infection for a node depends on its encounter duration
with other infected ones. During the tracing process, the probability is computed and
updated each time the node is traced. The probability of infection is computed for multiple
encounters with different durations (pn) as
 k

pn = 1 − 


∏(1 − p)
i =1
d( i )



where:
k is the number of encounters
d(i) is the duration of encounter i
p is the probability of infection per second
Finally, the nodes will be ranked from highest to lowest based on the probability of infection
and the top K will be selected.
16.3.2.3.3 Assigning Score to Nodes Based on Their Encounter History

Assigning a score to the node may help one to cluster nodes into defined groups. It also
will help one to understand how much the epidemic is spread given it starts from a node
from a specific cluster. Also, it could be used efficiently in the process of tracing. The score
summarizes the node’s mobility, and its encounters during its presence in the i-hospital.
The score could be based on:
• The number of the nodes’ distinct encounters

• The nodes’ encounter durations
• Frequency of encounters regardless of its duration
• Combination of any of the above
• The number of distinct places the nodes visited during its presence
16.3.3 Evaluation for i-Hospital Infection Tracing System

This part will define several metrics that are used to evaluate the tracing metric. Then, a sim-
ulation that used real extensive data is shown. The result of the simulation will be discussed.
16.3.3.1 Evaluation Metrics

Various metrics will be used to evaluate the tracing system. Examples are coverage, speci-
ficity, sensitivity, true positive, true negative, and accuracy.
Coverage corresponds to the overhead, and it is defined as the number of nodes that
have been traced during the trace-back process to the total number of nodes.
The traced nodes are considered as the infected ones, and the nodes that are not traced
are classified as noninfected. Therefore, it defines the true positives (TP) as the number of
infected nodes that are classified as infected during a tracing phase. Similarly, the nodes
that are not traced (i.e., classified as noninfected in the trace-back phase) and not infected
are called the true negatives (TN).
Sensitivity (SV) is computed on the basis of TP and the number of all infected nodes
population. Sensitivity reflects the ability of the system to identify all infected cases. In
our system, when the tracing all suspected node ends, it will have sensitivity equal to 1.
However, it will have high coverage that will affect the specificity.
Specificity (SP) is computed based on TN and the number of nodes that are not infected.
It reflects the tracing system ability to filter out noninfected nodes.
Accuracy is the ratio of (TN + TP) to all nodes.
16.3.3.2 Simulation and Evaluation

Real data were used for i-hospital infection tracing evaluation. On campus, Wi-Fi traces
are used for simulation purposes. The data were collected by our group from our univer-
sity campus for years. For this study, we use one week from April 2 to April 8, 2012. Six
buildings were chosen where they have different mobility characteristics and different
density of users. The traces reflect when mobile devices associate with an access point.
Two nodes are considered to be encountered if they associate with the same access point
in an overlapping time frame. Although it does not represent the continuous mobility
trajectory, the traces are believed to reflect many instances in which two nodes are in
proximity of each other and are likely to encounter. Information about the buildings’
users and record numbers are shown in Figure 16.7. The information is processed to
reflect the contact information. We assume the data encounter include the encounters
within infection range. The simulation consists of two phases: disease propagation phase
and infection tracing phase.
16.3.3.2.1 Disease Propagation Simulation

The infection is assumed to spread through direct contact from human-to-human. The
infection range Ri is the same as the encounter range Re; that is, Ri = Re. To simulate the
disease propagation, SEIR disease propagation model is used. In SEIR model, nodes flow
between four states: Susceptible (S), Exposed (E), Infectious (I), and Recovered (R). The
node is susceptible until it contacted an infectious node, then it becomes exposed with
the rate β. It remains exposed a number of days for the latent period. After the latent
period, the node becomes infectious and starts infecting its encounter for a number of
days for the infectious period. Then, the node will be recovered. When the node is recov-
ered, it will not be infecting others or getting infectious again. The disease simulation
uses the parameters (β = 0.005/s, pl = one day, pi = two days). A random node is chosen to
be in an infectious state. Then, it infects the contacts; the disease will propagate between
nodes using the above parameters. At each building the percentage of infected nodes (pf)
180,000 16,000
160,000 14,000
Number of records
140,000 12,000
Number of user
120,000
10,000
100,000
8,000
80,000
6,000
60,000
40,000 4,000
20,000 2,000
0 0
CSE EYN INF MCSB TUR HUB CSE EYN INF MCSB TUR HUB
(a) Buildings (b) Buildings
Building names
70%
Percentage of infection
Computer science (CSE)
60%
Entomology-Nematology (EYN) 50%
40%
Infirmary (INF)
30%
Microbiology and cell research (MCSB) 20%
10%
Turlington hall (TUR)-Mainly classes
0%
The hub (HUB)-dining and CSE EYN INF MCSB TUR HUB
computer facilities
(c) Buildings
FIGURE 16.7
Buildings information: User and record numbers; and the infection percentage per building. In (a) records
numbers, (b) building’s users, and (c) infection percentage.
is computed. The percentage of infected nodes is the ratio of the number of infected nodes
and the total number of nodes that have been in the building. It varies for each building
as it is shown in Figure 16.7.
16.3.3.2.2 Tracing Simulation

The simulation starts with a random node that was infected on the last day. We run the
tracing simulation for 10 times, each time the tracing starts from a new random node.
First, the group of suspected nodes S is traced, and then the selection methods that were
described in Section 16.3 are applied to S to produce the final selected nodes. The follow-
ing metrics are computed: coverage, true positive, true negative, and accuracy. Figure 16.8
shows the result of the simulation.
i-Hospital infection tracing system will find all infected cases, even if it starts with only
one confirmed case of infection. The problem is the overhead in terms of number of the
traced nodes. The result of tracing will include a large number of nodes: all of the infected
cases and some-noninfected cases as well. Minimize the coverage (the number of selected
nodes) and hence reducing the overhead is an important goal of this system. As a result,
the selection methods that were explained before are applied to the traced nodes to iden-
tify the nodes that are most likely to be the infected nodes. The results of the simulation
are shown in Figure 16.8. The figures show the coverage versus accuracy for the studied
algorithms in six buildings. Mn* or EncP selection methods achieve higher accuracy than
using random selection. The accuracy of Mn* and EncP reaches up to 80% in some cases,
especially when the coverage is closer to the infection percentage. The accuracy in some
buildings increases with coverage, but it decreases in others, as it is seen in Figure 16.8.
mn* EncP Random

CSE INF EYN
0.9 1 0.9
0.8 0.9 0.8
0.7 0.8 0.7
0.6 0.7 0.6
Accuracy
Accuracy
Accuracy
0.6
0.5 0.5
0.5
0.4 0.4
0.4
0.3 0.3
0.3
0.2 0.2 0.2
0.1 0.1 0.1
0 0 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Coverage Coverage Coverage
mn* EncP Random
MCSB HUB TUR
0.8 0.9 0.9
0.7 0.8 0.8
0.6 0.7 0.7

0.6 0.6
Accuracy
0.5
Accuracy
Accuracy
0.5 0.5
0.4
0.4 0.4
0.3
0.3 0.3
0.2
0.2 0.2
0.1 0.1 0.1
0 0 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Coverage Coverage Coverage
FIGURE 16.8
Result of tracing in six different buildings.
The reason for these various trends of accuracy is related to the node that was chosen to
be the source of infection and the encounter patterns of population in each building that
will affect the percentage of infections. As the percentage of the infected population gets
larger, the accuracy will increase with coverage and vice-versa.
To question is if the choice of starting node (the confirmed case of infection from which the
tracing starts) has an effect on the accuracy versus coverage trend or not, we have chosen
multiple different random nodes to start tracing from and we compared the results. The
figures show that the result has similar trend of accuracy versus coverage with very little
variation. On other words, the graph increasing and decreasing trends with coverage seem
to remain quite consistent with different choices of nodes where the tracing starts from.
16.4 Conclusion
The first part of this chapter presented i-hospital framework that utilizes different het-
erogeneous sensing devices in IoT for collecting and processing encounter information.
Then, i-hospital infection tracing framework is explained in the next section. In this
work, encounter-based information in infrastructure-based server architecture was used
for tracing of sources of infection and at-risk population. Our systematic framework of
building blocks offers a suite of trace-back algorithms that mimic (in-reverse) the infec-
tion spread model, to provide probabilistic forward and backward search techniques. We
also provide one of the most extensive studies on evaluation of disease spread using
WLAN traces of over 34k users. IoT equipment such as RFID readers and tags, IoT-based
wristbands, and IoT-enabled beds encounter data can apply the same tracing algorithms
seamlessly. Findings show the potential promise of this method to reach high accuracy
while investigating a relatively small population.
References
Armbruster, B., & Brandeau, M. L. (2007a). Who do you know? A simulation study of infectious disease
control through contact tracing. In proceedings of the 2007 International Conference on Health
Sciences Simulation (ICHSS), part of the 2007 Western MultiConference on Modeling &
Simulation (WMC) (pp. 79–85). San Diego, CA.
Armbruster, B., & Brandeau, M. L. (2007b). Contact tracing to control infectious disease: When
enough is enough. Health Care Management Science, 10(4), 341–355.
Bailey, N. T. J. (1975). The mathematical theory of infectious diseases and its applications. Charles Griffin &
Company Ltd, 5a Crendon Street, High Wycombe, Bucks HP13 6LE.
Brown, C., Efstratiou, C., Leontiadis, I., Quercia, D., Mascolo, C., Scott, J., & Key, P. (2014, September).
The architecture of innovation: Tracking face-to-face interactions with ubicomp technologies. In proceed-
ings of the 2014 ACM international joint conference on pervasive and ubiquitous computing
(pp. 811–822). Seattle, WA: ACM.
CDC (Center for Disease Control and Prevention). (2007). Healthcare infection control practices advisory
committee (HICPAC). Retrieved from http://www.cdc.gov/hicpac/index.html.
Cisco (2014). Wi-Fi Location-based services 4.1 design guide. Retrieved from http://www.cisco.com/c/
en/us/td/docs/solutions/Enterprise/Mobility/WiFiLBS-DG.html.
Farrahi, K., Emonet, R., & Cebrian, M. (2015, February). Predicting a community's flu dynamics with
mobile phone data. In proceedings of the 18th ACM conference on computer supported coopera-
tive work & social computing (pp. 1214–1221). Vancouver, BC: ACM.
Fuhrer, P., & Guinard, D. (2006). Building a smart hospital using RFID technologies: Use cases and imple-
mentation. Fribourg, Switzerland: Department of Informatics-University of Fribourg.
Kim, Y., & Helmy, A. (2010). CATCH: A protocol framework for cross-layer attacker traceback in
mobile multi-hop networks. Ad Hoc Networks, 8(2), 193–213.
Kumar, U., & Helmy, A. (2012, November). Discovering trustworthy social spaces. In proceedings of the
third international workshop on sensing applications on mobile phones (p. 7). Toronto, ON: ACM.
Luo, W., & Tay, W. P. (2013, May). Finding an infection source under the SIS model. In 2013 IEEE international
conference on acoustics, speech and signal processing (pp. 2930–2934). Vancouver, BC: IEEE.
Magill, S. S., Edwards, J. R., Bamberg, W., Beldavs, Z. G., Dumyati, G., Kainer, M. A., ... & Ray, S. M.
(2014). Multistate point-prevalence survey of health care–Associated infections. New England
Journal of Medicine, 370(13), 1198–1208.
Peng, T., Leckie, C., & Ramamohanarao, K. (2007). Survey of network-based defense mechanisms
countering the DoS and DDoS problems. ACM Computing Surveys (CSUR), 39(1), 3.
Reddy, E., Kumar, S., Rollings, N., & Chandra, R. (2015). Mobile application for dengue fever monitor-
ing and tracking via GPS: Case study for Fiji. arXiv preprint arXiv:1503.00814.
RFID journal (2016), Frequently asked questions. Retrieved from http://www.rfidjournal.com/site/
faqs#Anchor-Is-53555.
Salathé, M., Kazandjieva, M., Lee, J. W., Levis, P., Feldman, M. W., & Jones, J. H. (2010). A high-resolution
human contact network for infectious disease transmission. Proceedings of the National Academy
of Sciences, 107(51), 22020–22025.
Shah, D., & Zaman, T. (2011). Rumors in a network: Who’s the culprit? IEEE Transactions on Information
Theory, 57(8), 5163–5181.
Stehlé, J., Voirin, N., Barrat, A., Cattuto, C., Colizza, V., Isella, L., … Vanhems, P. (2011). Simulation
of an SEIR infectious disease model on the dynamic contact network of conference attendees.
BMC Medicine, 9(1), 1.
Subramony, H., Lai, F. Y., Ang, L. W., Cutter, J. L., Lim, P. L., & James, L. (2010). An epidemiological
study of 1348 cases of pandemic H1N1 influenza admitted to Singapore Hospitals from July to
September 2009. Annals Academy of Medicine Singapore, 39(4), 283.
Team, W. E. R. (2014). Ebola virus disease in West Africa—The first 9 months of the epidemic and
forward projections. The New England Journal of Medicine, 2014(371), 1481–1495.
Tina, H. S. (2016). How one patient spread MERS to 82 people. Retrieved from https://www.sciencenews.
org/blog/science-ticker/how-one-patient-spread-mers-82-people.
Weiser, A. A., Gross, S., Schielke, A., Wigger, J. F., Ernert, A., Adolphs, J., … Appel, B. (2013). Trace-
back and trace-forward tools developed ad hoc and used during the STEC O104: H4 outbreak
2011 in Germany and generic concepts for future outbreak situations. Foodborne Pathogens and
Disease, 10(3), 263–269.
Yoneki, E., & Crowcroft, J. (2014). EpiMap: Towards quantifying contact networks for understanding
epidemiology in developing countries. Ad Hoc Networks, 13, 83–93.
17
Emergency Department, Sustainability, and
eHealth: A Proposal to Merge These Elements
Improving the Sanitary System
Andrea Malizia, Laura Morciano, Jacopo Maria Legramante, Pasqualino Gaudio,

Sandro Mancinelli, Francesco Gilardi, Carlo Bellecci, and Leonardo Palombi
CONTENTS
17.1 Introduction ...................................................................................................................... 371
17.2 Design of Interventions to Optimize the ED ............................................................... 380
17.2.1 Hi Triage .............................................................................................................. 380
17.2.2 H24 Medical Assistance .................................................................................... 382
17.2.3 eHealth................................................................................................................. 382
17.3 Feasibility Study and Creation of Indicators................................................................384
17.4 IoT and Health Care ......................................................................................................... 385
17.5 Conclusion......................................................................................................................... 385
References..................................................................................................................................... 386
17.1 Introduction
With a population of nearly 60 million, Italy is the sixth most populous country in Europe
and the 23rd in the world, representing 0.81% of the global population. The territory covers
301,316 km2, with a population density of 193 inhabitants per km2. A range of indica-
tors shows that the health of the population has improved over the last decades. Average
life expectancy reached 79.4 years for men and 84.5 years for women in 2011, the second
highest in Europe (compared with 77.4 years for men and 83.1 years for women for the
European Union as a whole). Italy has one of the lowest total fertility rates in the world: in
2011, it was 1.4 births per woman, far below the replacement level of 2.1. The population
growth rate is, therefore, very low (0.3% in 2012), one of the lowest in the European Union,
and immigration is the source of most of this growth. Italy is divided into 20 regions
(Figure 17.1), each region has a certain number of provinces (that to date are 93), and the
provinces are divided into municipalities (comuni) that are around 8100 in all the country.
Regions differ in terms of demography, economic development, health care infrastructure,
and health expenditure. There are marked regional differences for both men and women
in most health indicators, reflecting the economic and social imbalance between the north
and south of the country.
371
Italy
Repubblica
Italiana
Boundaries of regions
Capitals of regions
FIGURE 17.1
Regions of Italy.
The World Health Organization has ranked the Italian health care system second best in
the world with only French system ranked higher (WHO 2000). Italy has a National Health
Service established in 1978 with the declared goal of providing uniform and comprehensive
care, financed by general taxation.
The system is regionally based and organized at three levels: national, regional, and
local. At national level, the Ministry of Health (supported by several specialized agen-
cies) sets the fundamental principles and goals of the health system, determines the core
benefit package of health services guaranteed across the country, and allocates national
funds to the regions. The regions are responsible for organizing and delivering health
care: They have virtually exclusive responsibility for the organization and administra-
tion of publicly financed health care (Ferré et al., 2014). At local level, health services are
delivered through a network of population-based health local unit deliver public health,
community health services, and primary care directly, and secondary and specialist care
directly or through public hospitals or accredited private providers. The main source
of financing is national and regional taxes, supplemented by copayments for pharma-
ceuticals and outpatient care. In 2012, total health expenditure accounted for 9.2% of
Emergency Department, Sustainability, and eHealth 373
FIGURE 17.2
Lazio demographic situation.
gross domestic product (GDP) (slightly below the European Union average of 9.6%).
Public sources made up 78.2% of total health care spending.
Lazio is a region placed in the center of Italy; it is characterized by a number of inhabit-
ants that is lower than 6 million, with an average age of 43.6 years and a quasi-equally
division between the percentage of men (48.2%) and women (51.8%) (Figure 17.2). The
young population of the region (0–24 years) is the one with the lowest number of people,
whereas around the 75% of the Lazio’s population is in the range of 25 to over 75, as is
shown in Figure 17.3 (year 2014). The total number of foreign persons is around the 10%
(Figure 17.2), and there are several foreign countries represented in the Lazio population
(Figure 17.4, referred to 2010). For this reason, for each initiative/implementation pro-
posed, it is essential to consider cultural differences. Lazio region has 43 medical structures
with emergency department (ED). A website developed by the health service of the Lazio
Region monitors, continuously, the number of access per single structures.
In this chapter, the authors analyzed the problem of the EDs. As previously mentioned,
self-funding is an important option for the regions, and the reduction of costs for the ED is
a proper solution for increasing the economic capability of the regional health systems. To
get this result, the authors proposed the following series of implementations based on the
main concept of the sustainability:
• Interventions with a social impact

• Interventions with an environmental impact
• Interventions with an economical impact
To achieve this, each intervention should considering cultural differences (Figure 17.5).
Sustainable development is development that meets the needs of the present without compromis-
ing the ability of future generations to meet their own needs (Brundtland definition, 1987).
0 5 10 15 20
2.6%
0–2 age
2.8%
3–5 age
5.6%
6–11 age
5.4%
12–17 age
6.7%
18–24 age
11.6%
25–34 age
15.8%
35–44 age
45–54 age 16.3%
55–64 age 12.4%
65–74 age 10.4%
75 e più 10.3%
Inhabitants Males Females
FIGURE 17.3
Lazio inhabitants age distribution (2014).
Nationality (Year 2010)
% on % on
Nationality (n.)
foreigners population
Romania 196,410 36.19 3.43
Filippine 32,126 5.92 0.56
Polonia 24,392 4.49 0.43
Albania 23,337 4.30 0.41
Ucraina 18,922 3.49 0.33
Bangladesh 16,161 2.98 0.28
Perù 14,895 2.74 0.26
Cina Rep. Popolare 14,890 2.74 0.26
India 14,586 2.69 0.25
Moldova 12,413 2.29 0.22
Marocco 11,606 2.14 0.20
Ecuador 9,676 1.78 0.17
Egitto 9,209 1.70 0.16
Sri Lanka 8,150 1.50 0.14
Bulgaria 7,722 1.42 0.13
Rankings
> is on 5° place among 20 regions per % foreigners on total inhabitants
> is on 11° place among 20 regions per % underage foreigners
> Foreigners growth rate[1]: 31.6% (2° place among 20 regions)
FIGURE 17.4
Lazio foreign inhabitant’s distribution.
The three spheres of sustainability
Social-Environmental Environmental-Economic
Environmental justice Environmental
Natural resource use Energy efficiency
Natural resources stewardship Subsidies/incentives for
Locally and globally Environmental management
Pollution prevention use of natural resources
(air, water, land, waste)
Social Sustainability
Standard of living Economic
Education Profit
Community Cost savings
Equal opportunity
Emergency Department, Sustainability, and eHealth
Economic growth
Research and
development
Economic-Social
Business ethics
Fair trade
Worker’s rights
+ The fourth sphere:
Cultural diversity
FIGURE 17.5
Sustainability scheme.
375
The case study considered in this work will be the ED of the “Policlinico Tor Vergata”
(PTV), the University of Rome Tor Vergata’s hospital. The PTV is in the south area of Rome,
in the Local Health Unit Roma2, one of the biggest of the region and that with the lowest
socioeconomic level.
The authors analyzed a 1-year access to the ED, which comprises an area of about
1.5 million people and almost 45,000 annual accesses.
A skilled-nurse staff based on Italian Triage guidelines assesses the first patient
evaluation:
• White tag: Nonurgent condition

• Green tag: Less urgent condition/low priority
• Yellow tag: Urgent, potentially life-threatening emergency conditions
• Red tag: Critical, immediately life-threatening emergency condition
A total number of 46,820 patients’ visit occurred during the study period. The ED was
accessed by 38,016 users, with a mean age of 49.6 ± 21.6 years. Patients with an age ≥65 years
make more than 25% of those accesses (about 11,000 accesses/year). The average distribu-
tion of the Triage code in the overall ED population was the following:
1. Green tag 59%

2. Yellow tag 26%
3. White tag 10%
4. Red tag 4%
5. Not executed 1%
The hospitalization percentage of the patients entering in the ED is 17%, the percentage of
patients that are sent home (home discharge) is 56% and the percentage that leave the ED
without a visit (left without being seen [LWBS]) is 15%. Among those who are LWBS in the
ED, the percentage rises three-fold among patients defined as frequent users. Frequent users
are those patients with four or more accesses in ED per year. Despite they are poorly repre-
sented in the examined ED, comprising 6% of the overall accesses, they greatly contribute
to the ED overcrowding. Recent data suggest a strong correlation between older age and
frequent use of the ED, due to the multiple chronicle conditions of these patients and a lack
of alternative care choices.
One of the main indicators of the ED overcrowding is the percentage of LWBS, which
suggests too long waiting time. This percentage in the ED of the PTV is the higher of the
Lazio region (15% vs 6%).
Older patients have a strong influence on the numbers mentioned above. Elderly patients
require significantly more emergency care resources than their younger counterpart. This
means both higher number of visits and the use of more complex resources, due to the
higher number of existing comorbidities. The greater complexity of this kind of patients
is also clear from rate at which they arrive in the ED with grey or black tag (OR 3.1: LC
2.9–3.2), as shown in Figure 17.6.
Another important aspect is the number of hospitalization occurred in this kind of
patients. Due to their complexity, the elderly were more often hospitalized than younger
adults. In our case study, the risk of hospitalization increased 5-fold in patients aged ≥ 65
(OR 5.2; LC 4.9–5.3). This is one of the highest rates of hospitalization ever reported in the
literature (Table 17.1 and Figure 17.7).
8000
6913
7000
6000
5260
5000
4000
3000
2000
1146
1000 727
0
Population <65 years Population >65 years
Grey tag Black tag
FIGURE 17.6
Yearly number of black and grey code tags at PTV divided per age.
TABLE 17.1
Comparison of Odds Over >65 Years versus Odds <65 Years between PTV and Other Foreign
Hospital Structures
Odds >65 Years versus
Reference Year Type of Study % of Over >65 Years Odds <65 Years
Aminzadeh 2002 Systematic revision 32–47 2,5–4,5

Yim VWT 2009 Retrospective analysis 45 3
Sona A 2012 Prospect 39, 8 –
Lowthian JA 2012 Retrospective analysis 39 3,9
Albert M 2013 Data collection CDC 36, 5 –
Latham LP 2014 Retrospective transversal 21 –
Keyes DC 2014 Case-control 51 2
Legramante 2016 Retrospective analysis 44 5,2
As highlighting in the previous data, older individuals and frequent users had a signifi-
cantly higher number of ED accesses, contributing to the overcrowding of EDs. Although
numbers of interventions have been introduced to limit this problem, PTV is the second
structure per number of access in short-time period (in Figure 17.8 is reported, for example,
a bar graph taken in 2 min of observation of the Institutional website).
Each access to the ED has been estimated costing 360 € per patient, whereas the hospital-
ization is calculated in 700 € per day (with an average length of stay of 5 days per patient).
The accesses cost in the PTV ED settle around 20–25 on millions € per year. Considering
the workdays lost due to each access, these numbers seem to further increase.
The authors estimate, for PTV, a catchment area of patients divided in the following:
• 25% from Rome
• 20% from Frosinone
• 20% from the other provinces of Lazio
25000
20996
20000
15000
10000
5303
5000 4563
3479
0
Population <65 years Population >65 years
N. patients N. hospitalizations
FIGURE 17.7
Yearly number of hospitalizations per patients at PTV per age.
March 18, 2016

Time range: 2.57 p.m.−2.59 p.m.
Total number of access
140
120
100
80
60
40
20
0
San Giuseppe
Ospedali Riuniti
San Filippo Neri
Belcolle
Ospedale Civile
Ospedale Civile
Santa Scolastica
F. Spaziani
San Camillo De Lellis
Andosilla
Pol. Univ. Tor Vergata
Sant'Andrea
Sandro pertini
Policlinico Casilino
Madre G. Vannini
San Giovanni Calibita-FBF
Sant'Eugenio
C.T.O. Andrea Alesini

Santo Spirito
Pol. Univ. Umbertol
Pol. Univ. A. Gemelli
Bambino Gesu'
San Giovanni Addolorata

San Camillo-Forlanini
Santissima Trinita'
San Benedetto
Dono Svizzero
San Giovanni di Dio
A. Fiorini
Santa Maria Goretti
P. Colombo
San Sebastiano Martire
Padre Pio
San Giovanni Evangelista
A. Angelucci
Coniugi Bernardini
L. Parodi Delfino
San Paolo
Oftalmico
G. Battista Grassi
Cristo Re
Santissimo Gonfalone
Aurelia Hospital
FIGURE 17.8
Number of access in the Lazio ED & First Aid in 2 min, March 18, 2016.
The total estimation is around 1.5 million of potential patients. Considering that
the estimated number of accesses to the first aid of PTV is around 540,000, the esti-
mated percentage of accesses is around the 33% of the potential catchment area of the
patients. It is plausible considering a percentage of potential accesses in a range from
18% to 33%.
FIGURE 17.9
Contamination produced by one travel of 38 min of car (33 of train).
The impact of these huge numbers of accesses has consequences even on the environ-
ment, as people usually use public or private transportation to reach the PTV. The authors
have interviewed (from April 2016 to July 2016) the patients to investigate how they arrive
at PTV, and how far they live from the hospital. The results are that the 85% of the inter-
viewed patients reach the PTV using a private transportation, and the average time of
travel is around 35–42 min.
The authors used these data as boundary conditions to run a calculation with the free-
license software ecopassenger, considering a travel of 38 min (from the city of Colleferro
near to Rome). The results obtained are shown in Figure 17.9.
It is clear that multiplying this value for the number of the yearly accesses, the contami-
nation values produced to reach the PTV (considering the estimation previously done) are
very high (Table 17.2).
TABLE 17.2
Estimated Contaminations due to the Travels of the Patients to Reach the PTV ED
Data Values
Carbon dioxide (kg/year) 1.320300–2.430000

Energy resource consumption (l/year) 586800–1.08000
Particulate matter (kg/year) 130–238
Nitrogen oxides (kg/year) 5428–9990
Nonmethane hydrocarbons (kg/year) 440–810
Data from the state-of-the-art suggest that reducing the accesses in the ED is
the main factor to reduce costs and improve quality of the services through the following:
• Reduction of both length of stay and waiting time

• Reduction of diagnostic and therapeutic errors
• Improvement of psychological states of both patients and operators
• Reduction of air pollution
17.2 Design of Interventions to Optimize the ED

The interventions to optimize the PTV ED need to have the following triple impact:
• Social impact: Accesses reduction, improves the appropriateness of access points,

improves the capability to do right diagnosis, and assigns right therapies, waiting
times reduction.
• Environmental impact: Reduction of pollutants into the atmosphere, reduction of
fossil fuels.
• Economic impact: Reduction of the costs, both for the hospital and the patients.
To do that the main interventions proposed are based on the following:
• Hi triage
• H24 medical assistance
• eHealth
17.2.1 Hi Triage
To determine the quality of patient care, there are four main actions of the emergency
process that have to work correctly: triage, testing and evaluation, handoffs, and admitting
(Eisenberg et al., 2005).
Triage derived from the French word trier, and it is defined as the sorting of patients
according to urgency and need for care (Conrand Stoppler, 2014). To date, triage is used
to assess priority for health care in many different conditions, which are as follows:
• In mass causalities, it is used to identify priorities and decide who need to be

transported in medical structure
• In the emergency rooms to decide who treat urgently
• To optimize the use of space in crowded medical structures
At the hospital, a skilled-nurse staff assesses triage, and it could be influenced by overcrowd-
ing. O’Connor, Gatien, Weir, and Calder (2014) hypothesized that patients would be more
often undertriaged if the ED was crowded. This has striking impacts on the health of patients
and the quality of care. In fact, undertriaged ones could lead the patients to leave the hos-
pital without being visited increasing the length of stay. It can provokes adverse outcomes
and adverse events such as an increase in mortality or delayed antibiotic administration in
patients with infectious diseases (Apker, Mallak, & Gibson, 2007; Edwards, & Sines, 2008;
Cheung et al., 2010; Maughan, Lei, & Cydulka, 2011).
In a simulated emergency situation, quantitative information requiring frequent updates and

information generated early were most likely to be inaccurate (Bogenstatter et al., 2009).
Metaanalyses (Bost, Crilly, Wallis, Patterson, & Chaboyer, 2010; Calleja, Aitken, & Cooke,
2011) reveal that structured handovers that involved both written and oral communication
improved information exchange. (Lori, Eric, & Forde, 2016)
From this analysis, emerges the main role of the nurse, which cannot be substituted. He
provides to
• Evaluate the conditions of the patient according to the main vital functions, symp-
toms, and vital parameters.
• Assign to the patient the personal data, the reason of the access in the emergency
room, the name of the family doctor, and the pharmacologic history.
• Assign the priority code to the patients.
• Check periodically the conditions of the patient to evaluate variation that can
determine a different priority.
This work can be improved by a more efficient organization and the development of new
technologies as follows:
• The introduction of innovative diagnostic tools with automatized measures of the

following:
• Body temperature, during the check in, through a proper collocation of ther-
mocameras and customized software
• Heart rate
• Blood pressure
• Hematologic and biochemic fast-check to improve the evaluation of vital
functions
• Saturometry
• Stress index
• Adoption of hardware and software to monitor patients with green tag codes to
constantly check the body parameters and the psychological conditions of the
patients.
• Hardware and software (similar to a pc station) allow the patients in nonurgent
conditions to partially compile their medical records, realized in different lan-
guages: These tools could improve communication between patients and medical
staff during the assessment of health status.
• Touch screen for nurse and operator that can process a first/rapid medical register
with a direct access to the previous data of the patients and the information get
during his passage through the diagnostic paths.
In the United States of America, a lot of hospitals have started an important use of technolo-
gies in ED (Geisler et al., 2010). The patient that is registered through an electronic medical
record is easily controllable, but it has to pay attention to the incapability to register
important details that can be excluded (or hidden) by the software. Another critical issue
could be represented by poor flexibility by which the electronic systems log information
(Atak et al., 2005; Bogenstatter et al., 2009; Eisenberg et al., 2006; Schubert et al., 2013).
Hi triage ones hypothesized by the authors combined the essential role of the skilled
nurses in the ED and the improved use of new technologic tools to simplify, speed up, and
optimize the triage process, while reducing errors caused by overcrowding.
17.2.2 H24 Medical Assistance

The last report from the Italian Society of Emergency Medicine depicted that older adults
need hospitalization as the main cause of ED overcrowding. As mentioned above, older
adults represent about 25% of all ED users, and they are hospitalized in almost 50% of the
cases. One critical point of this elderly patients flow is the availability of continuing care after
hospital discharge. Indeed, many of them require, once discharged, other forms of assis-
tance that would not allow them to return to their homes. To avoid this issue, the authors
hypothesized two relevant measures. First, an early taking charge of frailty patients, with a
discharge service directly managed by the hospital to reduce patients’ visits to the ED and
further reducing the number of hospitalizations and the average length of stay. Second,
introducing the concept of health smart home, a home equipped with several medical
devices, that can be remotely controlled, to monitor the patients 24 h. (Bailenson et al., 2008;
Banerjee et al., 2015; Kientz et al., 2008; Lehmann et al., 2004; Rialle et al., 2002; Riva, 2003).
There are many solutions studied; a (main) resume is reported in Table 17.1 from Mano
et al. (2016).
17.2.3 eHealth
eHealth refers to the use of tools based on information and communication technologies to
support and promote the prevention, diagnosis, treatment, and monitoring of disease and
the management of health and lifestyle (Ministry of Health). To improve ED performances,
the authors suggest the introduction of technologies tools to the PTV hospital, which are
as follows:
• Electronic health record: An electronic health record (EHR), or electronic medical

record, refers to the systematized collection of patient and population electronically
stored health information in a digital format. These records can be shared
across different health care settings. Records are shared through network-con-
nected, enterprise-wide information systems, or other information networks and
exchanges. EHRs may include a range of data, including demographics, medical
history, medication and allergies, immunization status, lab results, diagnostic
images, vital signs, detection of anthropometric data, and billing information.
In Italy, the use of EHR is regulated by the Ministry of Health (Italian Ministry
of Health, 2011). The use of EHRs has been widely demonstrated, and this system is
applied in a lot of medical structure worldwide. (Asan, 2017; Blumenthal, 2010; Bates &
Bitton, 2010; Embi et al., 2013; Collins et al., 2011; Carayon et al., 2014; Carayon, 2012;
Häyrinen et al., 2008; Weir et al., 2011).
• Telemedicine: Telemedicine is the use of telecommunication and information tech-
nology to provide remote medical care. It helps us to eliminate distance barriers
and can improve access to medical services that would often not be consistently
available in distant rural communities. It is also used to save lives in critical care
and emergency situations. Although there were distant precursors to telemedicine,
it is essentially a product of the twentieth century telecommunication and informa-
tion technologies. These technologies permit communications between patient and
medical staff with both convenience and fidelity, as well as the transmission of
medical, imaging, and health informatics data from one site to another. Early forms
of telemedicine achieved with telephone and radio have been supplemented with
video telephony, advanced diagnostic methods supported by distributed client/
server applications, and additionally with telemedical devices to support in-home
care. (Audebert et al., 2005; Bairagi & Sapkal, 2013; Castiglione et al., 2015; Hsu,
2017; Legris et al., 2016; Tao et al., 2016; Zhu et al., 2015).
Actually, in Italy, Minister of Health has prepared telemedicine guidelines. The
authors want to propose this particular branch of ehealth for PTV because it can
be used for the following services:
• Providing a consultation with a patient or a specialist assisting the primary
care physician in rendering a diagnosis. This may involve the use of live inter-
active video or the use of store and forward transmission of diagnostic images,
vital signs, and/or video clips along with patient data for later review.
• Remote patient monitoring, including home telehealth, uses devices to remotely
collect and send data to a home-health agency or a remote diagnostic testing facil-
ity for interpretation. Such applications might include a specific vital sign, such
as blood glucose or heart ECG or a variety of indicators for homebound patients.
The following services can be used to supplement the use of visiting nurses.
• Consumer medical and health information includes the use of the Internet and
wireless devices for consumers to obtain specialized health information and
on-line discussion groups to provide peer-to-peer support.
• Medical education provides continuing medical education credits for health
professionals and special medical education seminars for targeted groups in
remote locations.
• Consumer Health Informatics: Consumer Health Informatics (CHI) is a subbranch
of health informatics that helps us to bridge the gap between patients and health
resources. The American Medical Informatics Association defines it as “the field
devoted to informatics from multiple consumer or patient views.” The Consumer
Health Informatics Working Group of the International Medical Informatics
Association defines it as “the use of modern computers and telecommunications
to support consumers in obtaining information, analysing unique health care
needs and helping them make decisions about their own health.” CHI includes
patient-focused informatics, health literacy, and consumer education. The focus of
this field is to allow consumers to manage their own health, using Internet-based
strategies and resources with consumer-friendly language. Currently, CHI stands
at a crossroads between various health care-related fields such as nursing, public
health, health promotion, and health education.
• Virtual Health Care teams: Virtual Health care teams (VHT) are groups of health
experts cooperating and digitally sharing information about patients. The creation
of VHTs would be fundamental in big medical structures such as PTV to create
good communications between colleagues and between operator of PTV and the
family doctors.
The acquired data coming from the eHealth application on PTV ED can allow the develop-
ment of software, based on self-learning codes, to be able to give a preliminary evaluation
of clinical risk, needs of hospitalization, and social assistance to support the following:
1. The regional health record of Lazio (project in progress in the Department of

Biomedicine and Prevention)
2. The health system
3. The research and teaching activities of University of Rome Tor Vergata
17.3 Feasibility Study and Creation of Indicators

Following two phases are essential to get the proposals effective:
1. A feasibility study: Necessary to support the evaluation that is necessary to adopt

the choices and increase the operability spectra. The first phase is a technical veri-
fication of the real capacity to realize all the proposals from an economic-manage-
rial and organizational point of view. Each proposal has to be deeply analyzed to
evaluate the real possibility to be realized to calculate the time for its realization
and schedule the so called phases of investment recovery.
2. The creation of sustainability indicators: As instrument to support all the sustain-
ability procedure and policies adopted to realize the proposals. The authors have
identified a list of indicators divided per themes and types (Table 17.3).
TABLE 17.3
List of Indicators to Reach a Sustainable ED for PTV
Theme Subtheme Indicator Type of Indicator
Social indicators PTV ED People influx reduction Descriptive
Quality improve of operator’s work Performance
Access improvements Descriptive
Patients manage Understanding of patients Descriptive
problems
Correct sorting wards Performance
Reduction of waiting times Performance
Environmental Energy Reduction of energy consumption Composite
indicators inside PTV
Reduction of energy consumption Composite
of the people
Pollutants and Reduction of pollution emission in Composite
contamination atmosphere
Reduction of consumption of fossil Composite
combustibles
Economic indicators PTV costs Reduction of cost for operators Performance
Reduction of accesses Descriptive
Reduction of FU Composite
Reduction of managerial costs Composite
People costs Reduction of work days lost Performance
Reduction of patient costs Descriptive
17.4 IoT and Health Care

The Internet of Things (IoT) concept entails the use of devices to collect data and connect
those data in private or public cloud, enabling them to automatically trigger certain events
and be available to properly manage emergency situations.
Internet-connected devices have been introduced to patients in various forms. Whether
data comes from fetal monitors, electrocardiograms, temperature monitors, or blood-
glucose levels, tracking health information is vital for some patients. Many of these
measures require follow-up interaction with a health care professional. This creates an
opening for smarter devices to deliver more valuable data, lessening the need for direct
patient–physician interaction. By embedding IoT-enabled devices in medical equipment,
health care professionals will be able to monitor patients more effectively—and use the
data gleaned from the devices to figure out who needs the most hands-on attention. In
other words, by making the most of this network of devices, health care professionals
could use data to create a system of proactive management—as they say, prevention is
better than the cure. A number of technologies can reduce overall costs for the prevention
or management of chronic illnesses. These include devices that constantly monitor health
indicators, devices that autoadminister therapies, or devices that track real-time health
data when a patient self-administers a therapy. Many patients have started to use mobile
applications (apps) to manage various health issues because they have an increased access
to high-speed Internet and they have tablets or smartphones to do that. These devices and
mobile apps are now increasingly used and integrated with telemedicine and telehealth
via the medical IoT. A new category of personalized preventative health coaches (Digital Health
Advisors) will emerge. These workers will possess the skills and the ability to interpret
and understand health and well-being data. They will help their clients to avoid chronic
and diet-related illness, improve cognitive function, achieve improved mental health, and
achieve improved lifestyles overall. As the global population ages, such roles will become
increasingly important (Dimitrov, 2016).
17.5 Conclusion
The analysis of the available data has been necessary to properly evaluate a number of
interventions that can improve the performance of the whole ED of PTV in a sustain-
able direction. These interventions are useful to optimize the functionalities of PTV ED
not only in condition of normal operability but also in extraordinary condition such as
earthquakes, tsunami, and chemical, biological, radiological, and nuclear (CBRNe) events
(Di Giovanni et al., 2014; Malizia, 2016).
The attended results are as follows:
• Risks reduction.
• Attention to social and cultural diversities through software and hardware for the
patients available in different languages but able to immediately translate each
communication between operators and patients.
• Usage of the indicators in Table 17.3 to evaluate the ED of PTV to propose it as a
pilot model in Italy.
• University courses creation to form specialized personnel with a multidisciplinary

preparation to also launch multidisciplinary and innovative research projects.
• Reduction, around 15%, of patients accesses that means (according to the number
previously mentioned)
• Reduction of access cost (and of the connected hospitalizations) around 2.2–3
million €/year that can be reinvested to recover the expenses.
• Pollution reduction that, considering the number in Figure 17.9, could be of
3,500 kg/year of carbon dioxide, 3.5 kg/year of particulate, 150 kg/year of
nitrogen oxides, and 12 kg/year of nonmethane hydrocarbons.
• Reduction of fuel consumption around 16,000 liters/year that, together with
the working days, has a direct impact on the patient’s costs.
• Environmental study to evaluate the impact on humans, environment, ani-
mals, materials, and cultural resources.
The conclusion of this paper is that the abovementioned proposals and the results attended
provoke an improvement in term of the following:
• Appropriateness of access in ED
• Reduction of inflows in ED
• Education of waiting times in ED
• Appropriate selection of patients’ problems
• Best operator performances
• Best and most appropriate care for the patient
• Availability of supplementary educational tools with the development of remote
or on-site presentations of cases, their parameterization, the results of clinical and
surgical interventions, and so on
• Improvement of the economic environment through the reduction of personnel
costs, reduction of overall costs (visits, hospitalizations, and management struc-
ture), and reduction of the costs incurred by the patient
• Reducing the impact on the environment thanks to the reduction of patients’
travel to reach the ED has a direct impact on reducing fuel consumption, reducing
pollutants emissions into the atmosphere and reducing the energy consumption
in general
The future development for/of this work is the punctual analysis of each proposal from
a sustainability point of view through a collection of experimental evidences worldwide
collected.
References
Albert M. (2013). National Hospital Ambulatory Medical Care Survey: Emergency Department
Summary Tables. Available from: https://www.cdc.gov/nchs/data/ahcd/nhamcs_emergency/
2013_ed_web_tables.pdf
Aminzadeh, F., & Dalziel, WB., (2002). Older adults in the emergency department: a systematic
review of patterns of use, adverse outcomes, and effectiveness of interventions. Annals of
Emergency Medicine, 39(3), 238–247.
Apker, J., Mallak, L. A., & Gibson, S. C. (2007). Communicating in the “gray zone”: Perceptions about
emergency physician hospitalist handoffs and patient safety. Academic Emergency Medicine, 14,
884–894.
Asan, O. (2017). Providers’ perceived facilitators and barriers to EHR screen sharing in outpatient
settings. Applied Ergonomics, 58, 301–307.
Atak, L., Rankin, J. A., & Then, K. L. (2005). Effectiveness of a 6-week online course in the Canadian
triage and acuity scale for emergency nurses. Journal of Emergency Nursing, 31, 436–441.
Audebert, H. J., Kukla, C., Clarmann von Claranau, S., Kühn, J., Vatankhah, B., Schenkel, J., …
TEMPiS Group. (2005). Telemedicine for safe and extended use of thrombolysis in stroke: The
Telemedic Pilot Project for Integrative Stroke Care (TEMPiS) in Bavaria. Stroke, 36(2), 287–291.
Bailenson, J. N., Pontikakis, E. D., Mauss, I. B., Grossd, J. J., Jabone, M. E., Hutchersond, C. A. C., …
John, O. (2008). Real-time classification of evoked emotions using facial feature tracking and
physiological responses. International Journal of Human-Computer Studies, 66(5), 303–317.
Bairagi, V. K., & Sapkal, A. M. (2013). ROI-based DICOM image compression for telemedicine.
Sadhana. Academy Proceedings in Engineering Sciences, 38(1), 123–131.
Banerjee, T., Keller, J. M., Popescu, M., & Skubic, M. (2015). Recognizing complex instrumental
activities of daily living using scene information and fuzzy logic. Computer Vision Image
Understand, 140, 68–82.
Bates, D. W., & Bitton, A. (2010). The future of health information technology in the patient-centered
medical home. Health Affairs, 29(4), 614–621.
Blumenthal, D. (2010). Launching HITECH. New England Journal of Medicine, 362(5), 382–385.
Bogenstatter, Y., Tschan, F., Semmer, N. K., Spychiger, M., Breuer, M., & Marsch, S. (2009). How
accurate is information transmitted to medical professionals joining a medical emergency?
A simulator study. Human Factors, 51, 115–125.
Bost, N., Crilly, J., Wallis, M., Patterson, E., & Chaboyer, W. (2010). Clinical handover of patients
arriving by ambulance to the emergency department—A literature review. International
Emergency Nursing, 18, 210–220.
Calleja, P., Aitken, L. M., & Cooke, M. L. (2011). Information transfer for multi-trauma patients
on discharge from the emergency department: Mixed-method narrative review. Journal of
Advanced Nursing, 67, 4–18.
Carayon, P. (2012). Sociotechnical systems approach to healthcare quality and patient safety. Work,
41, 3850–3854.
Carayon, P., Li, Y., Kelly, M. M., DuBenske, L. L., Xie, A., McCabe, B., Orne, J., … Cox, E. D. (2014).
Stimulated recall methodology for assessing work system barriers and facilitators in family-
centered rounds in a pediatric hospital. Applied Ergonomics, 45(6), 1540–1546.
Castiglione, A., Pizzolante, R., De Santis, A., Carpentieri, B., Castiglione, A., & Palmieri, F. (2015).
Cloud-based adaptive compression and secure management services for 3D healthcare data.
Future Generation Computer Systems, 43–44, 120–134.
Cheung, D. S., Kelly, J. J., Beach, C., Berkeley, R. P., Bitterman, R. A., Broida, R. I., & White, M. L. (2010).
Improving handoffs in the emergency department. Annals of Emergency Medicine, 55, 171–180.
Collins, S. A., Bakken, S., Vawdrey, D. K., Coiera, E., & Currie, L. (2011). Model development for EHR
interdisciplinary information exchange of ICU common goals. International Journal of Medical
Informatics, 80(8), 141–149.
Conrand Stoppler, M. (2014). Medical triage: Code tags and triage terminology. MedicineNet.com.
Available from: http://www.medicinenet.com/script/main/art.asp?articlekey=79529
Di Giovanni, D., Luttazzi, E., Marchi, F., Latini, G., Carestia, M., Malizia, A., … Gaudio, P. (2014).
Two realistic scenarios of intentional release of radionuclides (Cs-137, Sr-90)—The use of
the HotSpot code to forecast contamination extent. WSEAS Transactions on Environment and
Development, 10, 106–122.
Dimitrov, D. (2016). Medical internet of things and big data in healthcare. Healthcare Information
Research, 22(3), 156–163.
Edwards, B., & Sines, D. (2008). Passing the audition—The appraisal of client credibility and
assessment by nurses at triage. Journal of Clinical Nursing, 17, 2444–2451.
Eisenberg, E. M., Baglia, J., & Pynes, J. (2006). Transforming emergency medicine through narrative:
Qualitative action research at a community hospital. Health Communication, 19, 197–208.
Eisenberg, E. M., Murphy, A. G., Sutcliffe, K., Wears, R., Schenkel, S., Perry, S., & Vanderhoef, M.
(2005). Communication in emergency medicine: Implications for patient safety. Communication
Monographs, 72, 390–413.
Embi, P. J., Weir, C., Efthimiadis, E. N., Thielke, S. M., Hedeen, A. N., & Hammond, K. W. (2013).
Computerized provider documentation: Findings and implications of a multisite study of
clinicians and administrators. Journal of the American Medical Informatics Association, 20(4),
718–726.
Ferré, F., de Belvis A. G, Valerio, L., Longhi, S., Lazzari, A., Fattore, G., Ricciardi, W., & Maresso, A.
(2014). Italy: Health system review. Health Systems in Transition, 16(4), 1–168.
Geisler, B. P., Schuur, J. D., & Pallin, D. J. (2010). Estimates of electronic medical records in U.S.
Emergency departments. PLoS One, 5(2), e9274.
Häyrinen, K., Saranto, K., & Nykänen, P. (2008). Definition, structure, content, use and impacts of
electronic health records: A review of the research literature. International Journal of Medical
Informatics, 77(5), 291–304.
Hsu, W. Y. (2017). Clustering-based compression connected to cloud databases in telemedicine and
long-term care applications. Telematics and Informatics, 34(1), 299–310.
Italian Ministry of Health. (2011). The electronic Health Record National Guidelines. Official web-
site of Italian Minister of Health. Available from: http://www.salute.gov.it/portale/documentazi-
one/p6_2_2_1.jsp?lingua=italiano&id=1654
Keyes, D. C., Singal, B., Kropf, C. W., & Fisk, A. (2014). Impact of a new senior emergency department
on emergency department recidivism, rate of hospital admission, and hospital length of stay.
Annals of emergency medicine, 63(5), 517–524.
Kientz, J. A., Patel, S. N., Jones, B., Price, E., Mynatt, E. D., & Abowd, G. D. (2008). The Georgia tech aware
home. CHI’08 extended abstracts on human factors in computing systems, ACM, Florence, Italy,
pp. 3675–3680.
Latham, L. P., & Acroyd-Stolarz, S. (2014). Emergency department utilization by older adults: a
descriptive study. Canadian Geriatrics Journal, 17(4), 118–125.
Legramante, J. M., Morciano, L., Lucaroni, F., Gilardi, F., Caredda, E., Pesaresi, A., Coscia, et al.
(2016). Frequent use of emergency departments by the elderly population when continuing
care is not well established. Plos One, https://doi.org/10.1371/journal.pone.0165939.
Legris, N., Hervieu-Bègue, M., Daubail, B., Daumas, A., Delpont, B., Osseby, G.V., ... Béjot, Y. (2016).
Telemedicine for the acute management of stroke in Burgundy, France: An evaluation of effec-
tiveness and safety. European Journal of Neurology, 23(9), 1433–1440.
Lehmann, O., Bauer, M., Becker, C., & Nicklas, D. (2004). From home to world-support-ing context-aware
applications through world models. In proceedings of the second IEEE annual conference on per-
vasive computing and communications, 2004. PerCom 2004, IEEE, Orlando, FL, pp. 297–306.
Lori, A. R., Eric, M. E. & Forde, C. (2016). The role of patients’ stories in emergency medicine triage.
Health Communication, 31(9), 1155–1164.
Lowthian, J. A., Curtis, A. J., Jolley, D. J., Stoelwinder, J. U., McNeil, J. J., & Cameron, P. A. (2012).
Demand at the emergency department front door: 10-year trends in presentation. Medical
Journal of Australia, 196(2), 128–132.
Malizia, A. (2016). Disaster management in case of CBRNe events: An innovative methodology to
improve the safety knowledge of advisors and first responders. Defense and Security Analysis,
32(1), 79–90.
Mano, L. Y., Faiçal, B. S., Nakamura, L. H. V., Gomes, P. H., Libralon, G. L., Meneguete, R. I., …
Ueyama, J. (2016). Exploiting IoT technologies for enhancing health smart homes through
patient identification and emotion recognition. Computer Communications, 89–90, 178–190.
Maughan, B. C., Lei, L., & Cydulka, R. K. (2011). ED handoffs: Observed practices and communication
errors. The American Journal of Emergency Medicine, 29, 502–511.
O’Connor, E., Gatien, M., Weir, C., & Lisa Calder, L. (2014). Evaluating the effect of emergency
department crowding on triage destination. International Journal of Emergency Medicine, 7, 16.
Rialle, V., Duchene, F., Noury, N., Bajolle, L., & Demongeot, J. (2002). Health “Smart” home:
information technology for patients at home. Telemedicine Journal and e-Health, 8(4), 395–409.
Riva, G. (2003). Ambient intelligence in health care. Cyber Psychology Behaviour, 6(3), 295–300.
Romero, E., Araujo, A. M., Moya, J. M., Goyeneche, J. M., Vallejo, J. C., Malag, J., … Fraga, D. (2009).
Image processing based services for ambient assistant scenarios. Distributed Computing,
Artificial Intelligence, Bioinformatics, Soft Computing, and Ambient Assisted Living. Lecture Notes in
Computer Science, 5518, 800–807.
Schubert, C. C., Denmark, T. K., Crandall, B., Gnome, A., & Pappas, J. (2013). Characterizing novice-
expert differences in macrocognition: An exploratory study of cognitive work in the emergency
department. Annals of Emergency Medicine, 61, 96–109.
Sona, A., Maggiani, G., Astengo, M., Comba, M., Chiusano, V., Isaia, G., Merlo, C., Pricop, L.,
Quagliotti, E., Moiraghi, C., Fonte, G., Bo, M. (2012). Determinants of recourse to hospital treat-
ment in the elderly. European Journal of Public Health, 22(1), 76–80.
Tao, L., Paiement, A., Damen, D., Mirmehdi, M., Hannuna, S., Camplani, M., Burghardt, T., &
Craddock, I. (2016). A comparative study of pose representation and dynamics modelling for
online motion quality assessment. Computer Vision and Image Understanding, 148, 136–152.
Weir, C. R., Hammond, K. W., Embi, P. J., Efthimiadis, E. N., Thielke, S. M., & Hedeen, A. N. (2011).
An exploration of the impact of computerized patient documentation on clinical collaboration.
International Journal of Medical Informatics, 80(8), 62–71.
Yim, V. W. T., Graham, C. A., & Rainer, T. H. (2009). A comparison of emergency department uti-
lization by elderly and younger adult patients presenting to three hospitals in Hong Kong.
International Journal of Emergency Medicine, 2(1), 19–24.
Zhu, N., Diethe, T., Camplani, M., Tao, L., Burrows, A., Twomey, N., … Craddock, I. (2015). Bridging
e-Health and the internet of things: The SPHERE project. View at Publisher. IEEE Intelligent
Systems, 30(4), 39–46.
18
The Contiki Operating System: A Tool for Design
and Development of IoT—Case Study Analysis
B. Venkatalakshmi, A. Pravin Renold, and S. Vijayakumar
CONTENTS
18.1 Introduction ...................................................................................................................... 391
18.2 Architecture of Contiki OS ............................................................................................. 393
18.3 Communication Stack in Contiki OS ............................................................................ 394
18.4 Case Studies ...................................................................................................................... 395
18.4.1 A Case Study on the Cooja Simulator with Its Program .............................. 395
18.4.2 A Case Study on the Routing Protocol for Low Power Lossy Networks
with Network Coding........................................................................................ 396
18.4.3 A Case Study on the Routing Protocol for Low Power Lossy Networks
for Body Area Network ..................................................................................... 397
18.4.4 A Case Study on the Constrained Application Protocol for Smart
Parking ...................................................................................................................398
18.4.5 A Case Study on the Privacy Homomorphism-Based Secured Data
Aggregation......................................................................................................... 399
18.5 Conclusion .......................................................................................................................... 403
References..................................................................................................................................... 403
18.1 Introduction
The Contiki operating system (OS) is an open-source OS specially designed for Internet
of Things (IoT). It supports Internet connectivity to monitor and control the low cost and
resource-constrained embedded devices. The Contiki OS is based on an event-driven
kernel that supports multithreading. It supports a micro-transmission control protocol/
internet protocol (TCP/IP) stack for IoT and Rime communication stack for wireless sen-
sor networks. The Contiki OS is designed using the C language. The advantage of uIP
(micro IP) is to provide TCP/IP protocol suite even for resource constrained 8-bit micro-
controllers. As the embedded device able to run TCP/IP makes it suitable to connect to
Internet. The uIP has minimal features needed for a full TCP/IP stack. The main control
function of uIP does the following activities repeatedly:
1. Check for the arrival of packet from the network

2. Check for the periodic timeout
391
The memory is effectively managed in uIP. The explicit dynamic memory allocation
is not supported. When a packet reaches a node, the device driver places it in the
global buffer (long enough to hold one packet of maximum size). To avoid overwrit-
ing, the application will process the data immediately or save it in the secondary
buffer for later processing. To further optimize the memory usage, during collision,
the application will be able to generate sent data rather than retransmitting the data
from buffer as in TCP/IP. The application program interface specifies the mechanism
of application program that interacts with the uIP TCP/IP stack.
Contiki OS has an inbuilt simulator called Cooja from version 2.0 of Contiki. It is a
Java-based simulator. The simulator aims to minimize the step between code develop-
ment and executing it on different platforms. Cooja is a cross-level simulator. It sup-
ports simulation and emulation; Cooja executes the program on the host CPU or in
an instruction-level TI MSP430 emulator. Cooja performs simulations at three different
levels such as application level, OS level, and machine-code instruction level (Osterlind
et al., 2006).
1. Application level: It deals with the design and implementation of routing protocols,
radio medium, radio devices, and duty cycles of sensor nodes.
2. Operating system level: It includes the entire Contiki OS, the process, and its associ-
ated libraries.
3. Machine-code instruction level: It includes the emulation of Tmote sky board’s exe-
cution using the assembly code pertaining to the architecture of the board.
The two main components of the Cooja simulator are interface and plug-in. The inter-
face denotes the property of a node such as radio, position of nodes, and hardware
peripherals. The plug-in takes care of the interactions with the simulation and the nodes
in the simulator.
The various features of Contiki OS are as follows:
1. Support for recently approved standards by IETF (Internet engineering task force)
in the domain of 6LoWPAN such as RPL (routing protocol for low-power lossy
networks), CoAP (constrained application protocol).
2. Support for emulation by means of the simulator Cooja.
3. Usage of lightweight, stackless thread called protothread. Protothreads avoid con-
ditional blocking and reduce explicit state machines in event-driven programs.
A protothread is stackless, that is, all protothreads in a system run on the same
stack.
4. Mechanism to operate nodes in low power state and to support tool for monitoring
energy consumption by the nodes.
5. Support for a light weight flash file system called as coffee. Coffee file system is
used as storage device in resource-constrained system. It uses application program
interface to provide the functionality such as storage, access, and handle data on
storage.
6. Usage of shell-based debugging environment. The shell commands allow text-
based interaction with sensor nodes based on Unix-like terminal.
The Contiki Operating System 393
18.2 Architecture of Contiki OS

As mentioned in above section, Contiki OS is based on C language. So the input file sup-
ports the C language which contains the information like operating mode of the node
such as time driven or event driven. The application level configuration such as proces-
sor details, sensor details, sensing duration, network, and medium access control (MAC)
type are specified in input file. The core components of Contiki OS shown in Figure 18.1
includes kernel, program loader, supporting libraries such as shell, flash file system, energy
monitoring module, and the communication stack. The core of Contiki OS is implemented
in C language. The input file is mapped with the core files component such as commu-
nication stack and other associated files. The compiled program could be loaded to the
Contiki OS supported hardwares such as MSP430, Atmel, Tmote sky boards, and so on.
The communication stack refers to a group of protocols that implements network pro-
tocol suite. The communication stack of Contiki OS supports sensor networks commu-
nication using Rime and uIP with 6LoWPAN network communication framework as
shown in Figure 18.2.
The device drivers act as an interface between the application and the hardware.
Moreover, they are responsible for reading the packet and forwarding them to the higher
layers. The communication stack processes the packets and forward them to the appli-
cation for which the packet is intended for. In the case of any response to be done, the
application program responds via the communication stack.
Input Application Core Log file/

(filename.c) components binary file
FIGURE 18.1
Overview of Contiki OS.
Application_1 Application_2
Core components
Communication
stack
Rime uIP Others
Device drivers
Hardware
FIGURE 18.2
Communication stack interaction with other modules.
18.3 Communication Stack in Contiki OS

a. RIME: The protocols in Rime communication stack shown in Figure 18.3 support
networking paradigm in wireless sensor networks. The wireless sensor networks
are considered as the basic building blocks of IoT. The Rime communication
stack provides a set of lightweight communication primitives. The Rime com-
munication stack supports single-hop, multihop, unicast, mesh, tree routing, and
broadcast-based communication primitives. The packets forwarded by a node to
neighboring nodes are maintained in packet queues, whereas each packet queue
has a lifetime. The packet is removed from the queue once the lifetime expires.
The packet queue is defined with the name of the queue and the maximum size of
the queue. The nodes are addressed on the basis of an unsigned integer data type
with a width of 8 bits (uint8_t). The sink node is always provided with node id as
“1” by default. The routing protocols determine the path based on the occurrence
of event or based on the expiration of time toward the sink node. Once the path is
determined, the packet is transmitted in unicast fashion on the chosen path.
b. Micro-IP (uIP): uIP is the world’s smallest full TCP/IP stack. Intended for tiny
microcontroller systems in which code size and RAM are severely constrained, uIP
only requires 4–5 kB of code space and a few hundred bytes of RAM. uIP has been
ported to a wide range of systems and has found its way into many commercial
products.
uIP has the network frameworks as 6LoWPAN. 6LoWPAN open standard was defined by
IETF to address the need of wide address space required by IP-driven connected devices
that can only be provided by IPv6. 6LoWPAN is an acronym of IPv6 over Low-Power
Wireless Personal Area Networks. IETF developed this standard to make IPv6 to be used
in low power and lossy networks that are based on IEEE 802.15.4. 6LoWPAN contains an
adaptation layer that allows IPv6 packets can be carried out through IEEE 802.15.4 link
layer frames.
Application protocols like

HTTP RTP Application layer MQTT, HTTP, COAP,
and so on
TCP UDP ICMP Transport layer UDP ICMP
IPv4/IPv6 Network layer IPv6 RPL
LoWPAN adaptation
Ethernet MAC Data link layer
IEEE 802.15.4 MAC
Ethernet PHY Physical layer IEEE 802.15.4 PHY
IP protocol stack 6LoWPAN protocol stack
FIGURE 18.3
6LoWPAN protocol stack.
The 6LoWPAN protocol stack shown in Figure 18.3 is equivalent to the traditional IP
protocol stack with some differences; 6LoWPAN supports IPv6 only, so a LoWPAN adap-
tation layer is introduced to optimize the IPv6 over IEEE 802.15.4. Next, the commonly
used transport layer protocol is user datagram protocol (UDP); TCP is not used because of
performance reasons.
Following are the main features of 6LoWPAN stack:
• Physical layer (IEEE 802.15.4 PHY) is used to convert data bits into signals that are
transmitted and received through radio frequency (RF) communication.
• The data link layer provides reliable connection between two nodes. The data link
layer also contains LoWPAN adaptation layer for adapting the IPv6 packets over
IEEE 802.15.4.
• Network layer initiates the assigning of IP address to individual nodes and routing
the packets either through mesh under (link layer) or route over (network layer)
routing protocols. The packets are forwarded hop-by-hop from source to destina-
tion. Packets are fragmented and sent to next hop whereas these fragments are
collected in the receiving hop and sent to either upper layer or to the next hop
according to routing table information.
• Transport layer is responsible for the applications to have their own communication
sessions. Usually, two types of transport layer protocols are used—TCP and UDP.
But due to some performance reasons of TCP-like power consumption, packet over-
head, UDP is preferred over TCP as a transport layer protocol. Other protocols such
as transport layer security and the Internet control message protocol are also used.
• Application layer uses sockets for communication between applications. Each
6LoWPAN application uses sockets for sending and receiving the packets that are
associated with protocols such as UDP and ports (source/destination). Application
layer is also used for data formatting. Widely used application protocol is HTML.
But due to its text-based approach, it has a large overhead that is not suitable for
6LoWPAN systems. For this, an alternative application protocol such as CoAP
and Message Queue Telemetry Transport (MQTT) is used. The sicslowpan mod-
ule supports the use of IPV6 on IEEE802.15.4 in Contiki OS. The functionalities
such as addressing, fragmentation, and header compression are implemented.
18.4 Case Studies

For a beginner, to experience coding in the Cooja simulator, a small sample program exer-
cise has been provided as Case Study 18.4.1.
18.4.1 A Case Study on the Cooja Simulator with Its Program

#include "contiki.h" // Core of contiki operating system
#include "stdio.h"
PROCESS(first,"HELLO_WORLD_PROCESS"); // Process control block
AUTOSTART_PROCESSES(&first); // Autostart of process during system boot
PROCESS_THREAD(first, ev, data) // Protothread with name of the process,
event identifier and pointer to data
{
PROCESS_BEGIN(); // Beginning of protothread
printf("First Program \n")
PROCESS_END(); // End of process protothread
}
The Contiki-built system is composed of a number of makefiles. These are as follows:
• Makefile: The project’s makefile, located in the project directory.

• Makefile.include: The system-wide Contiki makefile, located in the root of the Contiki
source tree.
• Makefile.$(TARGET) (where $(TARGET) is the name of the platform that is currently
being built): Rules for the specific platform, located in the platform’s subdirectory
in the platform/directory.
• Makefile.$(CPU) (where $(CPU) is the name of the CPU or microcontroller architecture
used on the platform for which Contiki is built): Rules for the CPU architecture, located
in the CPU architecture’s subdirectory in the cpu/directory.
• Makefile.$(APP) (where $(APP) is the name of an application in the apps/directory): Rules
for applications in the apps/directories. Each application has its own makefile.
The Makefile of the above code is as follows:
CONTIKI =../..
all: hello
include $(CONTIKI)/Makefile.include
The first line specifies the location of Contiki followed by the name of the application and
system wide makefile location that contains definitions of the core of Contiki system. The
latest version of Contiki OS can be downloaded at www.contiki-os.org. This website pro-
vides the Contiki OS and its simulator Cooja along with VMware. The website provides
step-by-step tutorial for the execution of program.
The rest of the case studies of the chapter demonstrate an approach for exploring the
communication stacks of Contiki OS. For example, Case Study 18.4.2 helps in understand-
ing that RPL protocol in uIP stack that can be suitably programed for overall network
performance enhancement using network coding. Case Study 18.4.3 explores feasibility
of RPL protocol for body area networks. CoAP, an application layer protocol of uIP stack,
has been customized for an application through Case Study 18.4.4. A secure data aggre-
gation based on privacy homomorphism (PH) technique has been demonstrated in Rime
stack of Contiki through Case Study 18.4.1.
18.4.2 A Case Study on the Routing Protocol for Low Power

Lossy Networks with Network Coding
IoT demands next-generation access networking with cost effective, high reliability, and
secure access networking trends for industrial applications (Rojas, 2015). This can be
achieved through a novel technique called network coding that was briefed into work
done by Weiner et al. (2014). It insisted the need for PHY/MAC layer enhancement to
address low-latency and high-reliability requirements of wireless nodes. Swamy et al.
(2016) ensured that network-coding technique could be a solution for this. The case study
Sink
Relay nodes
Information flow
Source nodes
NC node
Monitoring
environment
FIGURE 18.4
Network model for network coding.
given below also demonstrates that the network-coding technique can be used at IoT nodes
with reduced retransmissions and increased throughput at the sink node.
Network coding is a technique which can be used to improve a network’s throughput,
efficiency, and scalability, as well as resilience to attacks and eavesdropping. Instead of
simply relaying the packets of information, the nodes of a network can combine the packets
together for transmission. This can be used to increase the maximum possible information
flow in a network. Moreover, it may help for resilient transmission of packets. Network
coding is a method of optimizing the flow of digital data in a network by transmitting
digital evidence about messages. The case study considered a network scenario, as shown
in Figure 18.4.
Based on the position and the status of incoming contents, the nodes are configured
for modes like simple forward or encode and forward. Simple Ex-Or based encoding has
been used. In the Cooja simulator, the RPL at the network layer has been modified for
achieving this status variation of nodes. A function called forwarding function in IPV6
has been updated in synchronization with timers to diagnose the contents intended for
same destination. If the condition is true and if the node is an eligible encoding node, then
the packets are exported and then forwarded. The procedure of this case study includes
three modules of programming, namely source node, forwarding node, and receiving
node coding. Using this procedure, we can obtain higher throughput at receivers, packet
backup under node failures, effective distributed storage, and so on. The result statistics
can be obtained by integrating the Cooja simulator with wireshark analyzer. More details
on this case study on network coding are available in the work done by Venkatalakshmi
and Vasanth Kumar (2015).
18.4.3 A Case Study on the Routing Protocol for Low Power

Lossy Networks for Body Area Network
The wireless body area network is used for continuous monitoring of vital signs of human
body. The network consists of implanted sensor nodes and a sink for data collection as
shown in Figure 18.5. The sensor nodes are responsible for sensing and transmitting the
data to the sink. In implanted sensor nodes of wireless body area network, the heat gener-
ated by the node’s circuitry causes damage to the tissues. As communication consumes
more power and heat, designing an effective routing protocol for wireless body area net-
works by incorporating temperature in the routing metrics becomes essential. The thermal
Sink
Sensors
FIGURE 18.5
Architecture of body area network (BAN).
aware protocols maintain the node in active state if the node’s temperature is below the
safety level.
Thermal aware RPL leaves the hotspot node while forwarding packets. The neighbor
nodes are determined on the basis of the metrics such as distance and temperature. The
nodes with minimal distance and having temperature value below the threshold are cho-
sen for forwarding the packets. The objective function of RPL is modified to include the
new metrics temperature along with distance to build the destination-oriented directed
acyclic graph. If the temperature of the node is greater than the threshold, an alternate
path is chosen to forward the packet. The node with high temperature is allowed to go to
sleep state for certain time duration. The performance of thermal aware RPL gives better
packet delivery ratio (Renold, 2014).
18.4.4 A Case Study on the Constrained Application Protocol for Smart Parking
6LoWPAN usage for parking lot monitoring reduces the time taken by users to determine
the free parking lot. The advantage of this approach is that the drivers can decide their
duration time of parking their vehicles via Internet. The parking lot is allocated on the basis
of requirement. The system comprises of parking lot, sky motes, and mobile devices with
Internet connectivity. The free parking slots are intimated to the users via web address as
per request (Aarthi & Renold, 2014). The data transmission from Internet to that of user
uses a protocol termed CoAP. The motes used for the parking application can be classified
into three categories such as one mote acts as server, one as border router, and clients may
be in any number as shown in Figure 18.6.
• Server: A Restful server shows how to use the REST layer to develop server-side
applications.
• Border router: Border router keeps radio turned on. Enabling of it helps in connec-
tion between that of client as well as server to that of CoAP web address. Border
router has the same stack and fits into mote memory.
• Client: A CoAP client that polls the/actuators/toggle resource for certain-time
duration.
C
C
BR
S
C C
BR—Border router
C S—Server
C
C C—Client
FIGURE 18.6
Categories of motes.
Parking lot is considered to be a service in random order queue, and Markov chain is
used to explain memory-less property. The following equation shows the characteristic
of nondependence of the past:
pi j (t) : = P(Xτ+t = j Xτ = i ) = P(Xt = j X0 = i )
where P(Xτ+t = j Xτ = i ) represents the future occurrence

P(=
Xt j= X0 i ) represents the present occurrence. If the user wants to know the free
parking lot, he enters the vehicle number and the duration of time to park the vehicle.
CoAP uniform resource identifier (URI) is made for this work was coap: //vec:6000. The
communication between server and clients are of multihop fashion. The data from the
motes are sent to that of Copper Web Browser that can be viewed by the different users
(Kovatsch, 2011). Based on the duration of parking, the slots are allocated for the vehicle.
The methodology achieves less time in determination of free parking lot.
18.4.5 A Case Study on the Privacy Homomorphism-Based Secured Data Aggregation

Data-aggregation schemes are vulnerable to various security attacks. In such cases,
attackers maliciously change the aggregated data collected from the sensor nodes and
send them to cluster heads (CHs). Data-aggregation schemes combined with PH give
better security than traditional methods. In PH-based encryption, the incoming data
from sensor nodes are encrypted and fed to the CHs that in turn directly aggregate the
ciphered texts without decryption (Steffi Diana and Vijayakumar, 2014).
In data-aggregation scheme, the sensor nodes collect information from the end nodes.
Whenever the base station requires data, at data aggregator “A” that collects the information
from the sensor nodes, computes the addition, subtraction, and exclusive-or of the incom-
ing data and sends the aggregated data to the base station “B” over a multihop path, and
node “E” is leaf node in this path as shown in Figure 18.7. It provides better bandwidth and
energy consumption in the network. In the base station, these aggregated data are decrypted.
Without proper encryption, the adversary nodes can inject malicious data into the network.
For this reason, we are using data aggregation combined with PH as a standard mechanism
for encryption. PH enables us to do direct computation on encrypted data. Here, a specific
type of computation is performed on the cipher text and obtained an encrypted result in the
form of cipher text that is a result of the input plaintext. It can do addition and multiplica-
tion of the encrypted data and obtain the decrypted result without knowing the actual value
A A
E E
E E
E
E
FIGURE 18.7
Sensor nodes controlled by base station.
of the data. Moreover, the sensor nodes collect differential data rather than raw data from
the sensors for energy efficiency and minimize the burden of CHs in processing the data.
The case study follows a network model described as follows. The protocol executes in
two sessions—reference data transfer session and subsequent data transfer session. In
reference data session, the reference data are transferred from sensor nodes to base sta-
tion in a secure and energy-efficient way and perform verification of it in the base station.
Base station can keep the reference data as a correspondence to the node id for recovery of
original sensed data in differential data transmission session. The data-transfer session
transfers the differential data from sensor nodes to base station in a secure and energy-
efficient way. Hence, the base station can recover the individual sensed data and can
check the integrity of all sensing data. To demonstrate the proposed protocol, a cluster
1 may be considered. Here, wth sensor node (SNw) is chosen as CH of cluster 1, and the
remaining sensor nodes (SN1 … SNw–1) are considered as cluster members. Both CH and
cluster members perform sensing and the CH’s performs data aggregation. The entire
scenario is well depicted in Figure 18.8.
id1 − PUSN1
idw − PUSNw
H, PRBS
BS
id1 .... idw Ĉ, σ

CH
3
H, PUBSPRSNw CH1 id
CH2 (SNw) w−
1 C
w−
1
σ
1,σ
1
w−
id 3
C
2 2
1
id2 C , σ
1
id
C 3,
SNw−1
σ3
SN1
H, PUBS,PRSNw
H, PUBS,PRSNw
SN2 SN3
H, PUBS,PRSNw H, PUBS,PRSNw
FIGURE 18.8
Cluster formation in sensor networks.
In the setup phase, the base station uses this phase to generate and install all necessary
secrets for each sensor node and for them too.
PU SNi = PR SNi × g2
where:
PR- is the private key
PU- is the public key
g2 is the generator
In the initial phase, the sensor node executes this phase for transferring cipher text
appended to the node id with digital signature as a pair to CH.
cipher text of sensed data = (Ci ) = (ri , si ) = (ki × G , Mi + kr × γ)
digital signature σi = PR SNi × hi
In aggregate phase, the CHi uses this phase to aggregate the cipher texts, signatures com-
ing from its cluster members, and to generate aggregated cipher text and aggregated sig-
nature without performing any decryptions and encryption at the CHi.
∑ r ,∑
w w
Aggregated Cipher text (C 1 ) = (R 1 , S 1 ) = i Si
i =1 i =1
∑
w
Aggregated signature (σ̆) = σ i = σ1 + σ w
i =1
In the recovery phase, the base station goes through this phase to recover each individual
sensing data from the aggregated cipher text reached at the base station.
Decryption: Base station generates M′ by decrypting the ciphertext with Private key
of base station (PRBS).
Cˆ 1 = (Rˆ 1 , Sˆ1 ) and M′ = − PR BS × Rˆ 1 + Sˆ1
Remapping: Remapping using rmap() function to remap elliptic curve points into
plain text.
Remapped data m′ = rmap( M + ) = M + G
Decoding: Base station recovers the sensing data.
di = m′[l. (i − 1), (l, i ) − 1]
In the verify phase, the base station checks the integrity and authenticity of all sensing
data in this phase. So that base station can validate the correctness and the source of the
sensing data. The entire scenario is well depicted in Figure 18.9.
The IDi of each recovered data at the base station is checked with the IDs stored in base
station. If it matches, base station confirms that the data are from a valid node (ID).
Integrity of each recovered data is checked by base station.
∏
w
en (σ , g2 ) = en (hi , PU SNi )
i =1
hi = H(di )
Idi, Ci, Si Id1 ... Idw, Ci Si

Node BS
CHi
Digital signature Recovery of
generation Aggregation
differential data
Sensing data Cipher text
Differential data Recovery of
calculation original data
Digital
signatures Authenticity
Encryption of
differential data checking
Integrity checking
FIGURE 18.9
Nodes secured communication.
This network scenario can be simulated in the Cooja simulator of Contiki OS as shown below.
1. The sensor node collects the differential temp data from sensors and encrypts it
difdata = llabs(data-refvalue);
printf("Sensor data value = %llu \n",difdata);
difdata = difdata << betavalue;
difdata = binary(difdata);
difdata = difdata * G;
return difdata;
2. And then it generates digital signature and sends ciphertext and digital signature
with $ appended to the aggregator.
unsigned long long r, encryptdata;

difdata = encode();
r = k * G;
encryptdata = difdata + (k * PUBS);
return encryptdata;
3. The aggregator collects the differential data from sensor nodes and sends the
aggregated info to base station.
if(((UIP_IP_BUF->srcipaddr.u8[sizeof(UIP_IP_BUF->srcipaddr.u8)
– 1])==nodeid[i])&&(flag[i]==0))
{
s1=s1+aggdata;
s2=s2+aggsig;
flag[i]=1;
aflag=0;
}
uip_udp_packet_sendto(client_conn1, buf, strlen(buf),

&server_ipaddr1,UIP_HTONS(UDP_BASE_PORT));
4. The original data is recovered in the base station.
data[i]=mapdata%1000;
mapdata=mapdata/1000;
num=binary(data[i]);
printf("\noriginal data recovered at Base station = %llu",num);
5. Finally, it is decrypted in the base station.
data1=(long long*)malloc(sizeof(long long));

Mapdata1=(-PRBS*r1)+s1;
Mapdata1=Mapdata1/G;
data1 = remapping(Mapdata1);
return data1;
Thus, the above technique recovers sensing data from the aggregated result and verifies
data integrity and authenticity. It improves the network lifetime by achieving energy and
bandwidth efficiency while preserving security requirements such as confidentiality, data
authentication, and integrity. It reduces the energy consumption by transferring differ-
ential data rather than raw data from nodes so that the CH also wants to process small
amounts of bits.
18.5 Conclusion
Contiki OS is more user friendly and users can configure for different core operations
of a wireless networks. All the layers can be comfortably configured for the application-
specific requirements. Its compatibility with Skymotes, Zolertia (Z1) motes, Open motes,
and so on provides real-time application development using the Contiki OS and Cooja
simulator as an effective tool for IoT.
References
Aarthi, R., and Renold, A. P. (2014). COAP based acute parking lot monitoring system using sensor
networks. ICTACT Journal on Communication Technology, 5(2): 923–928.
Kovatsch, M. (2011). Demo abstract: Human-coap interaction with copper. In Distributed Computing
in Sensor Systems and Workshops (DCOSS), 2011 International Conference on (pp. 1–2). IEEE.
Osterlind, F., Dunkels, A., Eriksson, J., Finne, N. and Voigt, T. (2006). Cross-level sensor network
simulation with Cooja. In Proceedings of 31st IEEE Conference on Local Computer Networks
(pp. 641–648). IEEE.
Renold, P. (2014). Routing protocol for low power lossy networks. In Advanced Communication Control
and Computing Technologies (ICACCCT), 2014 International Conference on (pp. 1457–1461). IEEE.
Rojas, B. (2015). Toward next-generation access networking technologies in industrial/enterprise internet
of things. Retrieved from: http://download.peplink.com/resources/IDC-iot-whitepaper.pdf.
Steffi Diana, B. and Vijayakumar S. (2014). A concealed data aggregation using privacy homo-
morphism in wireless sensor networks. International Journal of Innovative Research in Science,
Engineering and Technology (IJIRSET), 3(3): 2404–2410.
Swamy, V. N., Rigge, P., Ranade, G., Sahai, A., and Nikolic, B. (2016). Network coding for high-
reliability low-latency wireless control. In Wireless Communications and Networking Conference
(WCNC), 2016 IEEE (pp. 1–7). IEEE.
The Contiki OS. Retrieved from http://dunkels.com/adam/software.html (accessed January 11,
2017).
The Contiki OS. Retrieved from http://www.contiki-os.org/ (accessed January 11, 2017).
Venkatalakshmi B. and Vasanth Kumar C. (2015). Design and development of network coding
algorithm for IoT devices. Australian Journal of Basic and Applied Sciences, 9(16): 367–372.
Weiner, M., Jorgovanovic, M., Sahai, A., and Nikolie, B. (2014). Design of a low-latency, high-
reliability wireless communication system for control applications. In Communications (ICC),
2014 IEEE International Conference on (pp. 3829–3835). IEEE.
19
Novel Trends and Advances in Health Care
Using IoT: A Revolution
Awanish Kumar and Archana Vimal
CONTENTS
19.1 Introduction ...................................................................................................................... 406
19.2 Health Care Implications ................................................................................................ 407
19.2.1 Wireless Patient Monitoring ............................................................................. 407
19.2.2 Mobile System Access ....................................................................................... 407
19.2.3 Medical Devices and Sensors...........................................................................408
19.2.4 Telemedications ..................................................................................................408
19.3 Enable Technologies Making IoT a Reality .................................................................. 410
19.3.1 Low Power Operations...................................................................................... 410
19.3.2 Integrated Precision-Analog Capabilities ...................................................... 411
19.3.3 Graphical User Interface ................................................................................... 411
19.3.4 Connectivity Standards .................................................................................... 411
19.3.5 Health Care Solutions Using Smartphones—Medical IoT Apps ................ 411
19.4 Challenges to Overcome ................................................................................................. 412
19.4.1 Scalability ............................................................................................................ 412
19.4.2 Interoperability .................................................................................................. 412
19.4.3 Safety of Patients ................................................................................................ 413
19.4.4 Security and Personal Privacy ......................................................................... 413
19.4.5 Lack of Government Support ........................................................................... 413
19.5 Accomplishments through IoT ...................................................................................... 414
19.5.1 Glucose Level Sensing ....................................................................................... 414
19.5.2 Electrocardiogram Monitoring ........................................................................ 414
19.5.3 Rehabilitation System ........................................................................................ 414
19.5.4 Medication Management .................................................................................. 415
19.5.5 Wheelchair Management.................................................................................. 415
19.5.6 Monitor an Aging Family Member ................................................................. 415
19.5.7 Other IoT Technologies ..................................................................................... 416
19.6 Conclusion and Future Perspective .............................................................................. 416
References..................................................................................................................................... 417
405
19.1 Introduction
The health care sector is a major considerate area worldwide. The government of all over
the world invests a huge amount of GDP in the health care sector for public health and
welfare. According to World Bank Report, 2014, the total percent of GDP spent on health
care throughout the world was 9.9%. However, high-income countries are paying more
attention toward the public health compared with low-income countries. In Figure 19.1,
a comparative GDP expenditure graph is shown (The World Bank, 2014).
In India, there are only 40 doctors per 100,000 people (Charania et al., 2016). However,
the health care sector is growing at the rate of 15% annually and is expected to cover
market value United States Dollar (USD) 280 billion by the year 2020 (Nishith Desai
Associates, 2016). It is changing by adopting emergent technology-based devices and
services. In such situation, the intervention of Internet of Things (IoT) in health care
would be useful to improve people’s life. It is expected to earn USD 14tn revenue glob-
ally (Srinivasa, 2016). The main aim of IoT in the health care sector is to enhance the
quality of people’s life and that too in an economical way. At the same time, it should
provide equity and ease of access. The revolution in the mode of information and com-
munication technologies (ICTs) has a great impact in addressing global health prob-
lems. Such modes include computers, the Internet, and smartphones that transform
the way of communication between individuals, pursue and exchange of information,
and enrich their lives (WHO, 2009). With the intervention of advancing technology, it
gets easier to connect physical objects to the Internet, opening a new world of oppor-
tunities. IoT-based health care systems make use of a set of interconnected devices that
produces a network system. The function of this system is health care assessment such
as monitoring patients and automatically detecting situations of medical emergency.
The general architecture of IoT begins with different sensing devices that collect patient
data by using different devices and sensors. Then there is a middleware that provides
interoperability and security. The next level is monitoring devices that are connected
14
12 12.3
10
9.9
8
% GDP
6
6.2
5.8 5.7 5.8
4 4.5 4.7
0
World High Low and Low Lower Middle Upper India
income middle income middle income middle
income income income
FIGURE 19.1
Percent of GDP expenditure on the health care sector for the year 2014.
Novel Trends and Advances in Health Care Using IoT 407
via wireless-networking technology. It functions in a similar fashion as of a network

manager that is responsible for storing, aggregating, consolidating, and security alarm-
ing (Fan et al., 2014). IoT interconnects all the objects and makes them smart, which
could be termed the next technological revolution (Fan et al., 2014). These technologies
empower public-health care system through various advanced applications but alto-
gether it brings some challenges. The major concerns are ethical, legal, and social issues
such as patient’s privacy and safety (Omogbadegun & Ayo, 2007).
19.2 Health Care Implications

19.2.1 Wireless Patient Monitoring
The introduction of wireless technology and the mobile system plays a major role in
the transformation of the health care industry. With the introduction of wireless net-
working, the limitation imposed by a traditional wired solution like Ethernet and USB
are overcome. The present era is the third generation of wireless technology starting
from pagers to basic mobile phones and now to smartphones. The health care profes-
sionals substitute personal digital assistant with smartphones for their routine work
within the clinical environment. They are taking benefits of smartphones with cellular
integration and built-in Wi-Fi, and Bluetooth wireless functionality. It helps them to
maintain patient’s records and to provide other health care services in a better and
efficient way.
It was appraised by Ericsson Mobility Report and Allied Business Intelligence (ABI)
Research that there will be 30 billion gadgets connected through a wireless network by
the year 2019 (Charania et al., 2016). As the development in wireless technology advances,
it upsurges the quality of services provided to the patients. It serves with patients with
more individual care, enhances the efficiency of caretaker, and decreases the chances
of medical negligence. In spite of the well establishment of wireless technology in vari-
ous other domains, few problems arise in the health care sector. The reason is that the
requirement of the medical domain is completely different. It demands high quality and
assurance in terms of operation and implementation (Ngoc, 2008). Freescale provides
microcontrollers that support wireless connectivity for devices. The system relies on the
use of wireless standard connections. Personal area networks used for personal gadgets
employs Bluetooth and Bluetooth low energy. For local area network used in health care
center, Wi-Fi and Bluetooth-based connections are used (Niewolny, 2013).
19.2.2 Mobile System Access

In recent years, a drift from PC to mobile devices is observed. The mobile devices may
be smartphones, notebook, or tablet computers. Such devices are equipped with sensors
and actuators. This makes the gadget smart by empowering them to compute, act, sense,
and interchange information with the associated network. Apart from this, mobile devices
are provided with radio-frequency identification, quick response codes, global positioning
system, and/or radio-frequency identification scanners or quick response codes readers.
All these technology enhancers act as a connector between the physical world and cyber-
space (Coetzee & Eksteen, 2011).
Battery
Photo button
IP camera
Headphone
Smart glasses
Smart watches Pregnancy tracking device
Smart
wearable
devices
Smart headbands
Medical
garments
Fitness bands Wireless pain reliever
FIGURE 19.2
Some smart health/medicine-related wearable devices and gadgets available in the market.
19.2.3 Medical Devices and Sensors

IoT-based medical devices and sensors have drastically changed the health care monitoring
and treatment in a positive way. It supports a better living standard by giving person-
alized and optimized services. It is also helpful in providing timely and cost-effective
services, and many times prove as a life saver in critical medical conditions. The recent
development of technologies such as sensor, the Internet, cloud, mobility, and big data
technologies add on the benefits of the existing medical devices. In addition, it supports
better connectivity to various health programs across the world (Khanna & Misra, 2013).
The developing countries such as India emerges a huge market for wearable devices that
includes smartwatches, fitness bands, health and fitness trackers, smart glasses and others.
Some of the smart wearable devices are shown in Figure 19.2.
International Data Corporation revealed that about 400,000 units of wearable devices
were sold during the first quarter of 2016 in India (Wearable Tech News, 2016). The smart
wearable market share captured by the fitness band was 87.7%, whereas the devices run-
ning on third-party applications hold 12.3%. An increase of 67.2% delivering wearable
devices is observed from the year 2015. The International Data Corporation also suggests
that smartwatches will singly occupy 70% market share of wearable devices by the year
2019 (Singh, 2016).
19.2.4 Telemedications
The term telemedicine was introduced in the 1970s that mean healing at a distance (Strehle &
Shabde, 2006). The World Health Organization has accepted the following wide depiction
of telemedicine and explained it as “The delivery of health care services, where distance
is a critical factor, by all health care professionals using information and communication
technologies for the exchange of valid information for diagnosis, treatment and preven-
tion of disease and injuries, research and evaluation, and for the continuing education of
health care providers, all in the interests of advancing the health of individuals and their
communities” (World Health Organization, 1998). It can be explained as an approach of
delivering distant health care with the help of telecommunication and information
technology. It is a real-time, two-way communication between patient and clinician at
a distant site. The communication takes place by means of advanced telecommunication
devices such as smartphones or real-time video conferencing. Health care practitioners
evaluate, diagnose, and treat patients by sharing medical data with peers and specialist
from worldwide. This increases the reach of the clinician to remote areas. This proves as
a boon for the communities that are deprived of health care benefits due to the geographi-
cal barrier (Med, 2016). Other than this, it is also a mean to modernize the health care
itself by the involvement of the user in the management. Moreover, it offers a new insight
to develop a healthy standard of living (Puskin, Johnston, & Speedie, 2006). To provide
the benefits of eHealth care or telemedicine facilities in rural areas, many health care cen-
ters have implemented the public–private partnership mode. This sector is growing very
rapidly in India with a compound annual growth rate (CAGR) of 20% and is expected to
grow USD 18.7 million by 2017 (eHealth, 2015). Indian Government has also come up with
the scheme such as National eHealth Authority to encourage the concept of telemedicine
in India (Nishith Desai Associates, 2016). The various advantages that telemedicine offers
are summarized in Figure 19.3.
There are broadly three categories of telemedicine (Smith, 2015) that are as follows:
• Store and forward

• Remote monitoring
• Real-time interactive
The various fields of medication in which telemedicine transformed the health care sector
are summarized in Figure 19.4.
Time
reduction
Peoples’
Connectivity participation/
to rural area self-help
Telemedicine
for universal
Exchange of
health care Cost
information/
sharing of saving
experience
Disaster
Improved relief/
quality emergency
care help
FIGURE 19.3
The various advantages offered by telemedicine in improving the health of people.
Teleaudiology
Teledentistry Telepsychiatry
Telemedicine
Teledermatology Telepathology
Teleradiology
FIGURE 19.4
The various applications of telemedicine in the different medical fields.
One of the major applications is in telecardiology. It revolutionizes the cardiac care sys-
tem and has the potential to save time money and lives. It provides the means for the
remote specialist interpretation of electrocardiographic recordings through telephone
transmission. It brings the efficient way of diagnosis and management of acute and chronic
cardiac disease along with identifying patient at risk. It is beneficial for both physicians
and patients as it serves benefits in terms of ease of access, the speed of diagnosis, the
efficiency of management, and the freeing up of resources. It reduces the number of sec-
ondary referrals, misdiagnosed cases, and reducing door-to-balloon time. It empowers
clinicians to improve clinical assessment of the patient and enables them to safely manage
more of their own patients (Backman et al., 2010).
19.3 Enable Technologies Making IoT a Reality

The execution of IoT for practical use is possible through the use of several enabling
technologies. For the well-organized functioning of smart sensors and microcontroller,
components with certain capabilities are needed. That device should have low power oper-
ation, integrated precision-analog capabilities, and graphical user interfaces (GUIs). Other
than this the device should be small with extended battery life to increase its usability and
should offer high accuracy at the relatively cheaper rate (Ullah et al., 2016).
19.3.1 Low Power Operations

For making IoT a reality, there are some features that must be satisfied. Low power
demand is one of them. This is particularly useful when devices have no or limited access
to power sources. The devices should only be active when there is a need for its operation.
Its communication with large network devices demands huge power; therefore, it must be
occasional. Moreover, human and system intervention should be less frequent. The practi-
cal implementation of this is possible when it is supported by control over idle and sleep
mode, signaling, uplink power adaptation to link and others. Collaboration with other
devices is also an alternative to decrease power consumption (Wu et al., 2011). The example
is IP (Internet protocol) stack, a light protocol that links a large number of communicating
devices all over the world. Its operation required only small and battery-operated embed-
ded devices (Atzori et al., 2010).
19.3.2 Integrated Precision-Analog Capabilities

For making IoT implementation practically possible, it is a mandatory to have high accu-
racy technology achievable at low cost. Integrated precision-analog capabilities are one of
the ways by which sensors can achieve this. There are technologies enabling this by the
use of microcontrollers. Such microcontrollers have analog components such as low-power
op-amps and analog-to-digital converters with high-resolution capacity (Niewolny, 2013).
This can be clarified with the assistance of a case of temperature control. Temperature
control (a specific temperature) is required for a large number of applications whether it
is storing milk at the cold, a geyser maintaining optimum temperature, or glucose meter
having a temperature within specified limit for correct result prediction. The temperature
control in all such devices is possible with the help of microcontroller-based temperature
sensors (Smith, 2013).
19.3.3 Graphical User Interface

Along with above mentioned technologies, user-friendly and easily accessible GUIs are
compulsory for people to take benefit of smart technologies. A good GUI is the one that
enables display devices to deliver more information with vivid detail and makes it easy
to access that information (Niewolny, 2013). Presently, smartphone app provides a GUI
for accessing and controlling the devices from home through server real IP. For that users
need any Wi-Fi- or 3G/4G-enabled smartphone that supports Java (Piyare, 2013).
19.3.4 Connectivity Standards

The term IoT can be interpreted in a simple way as a world-wide network of intercon-
nected objects that exchange information based on standard communication protocols. It
is predicted that three-time growth in the Internet connected IoT devices will be observed
by the year 2020. It is expected to increase from 10 to 34 billion (BI Intelligence, 2016).
To deal with increasing burden, the ideal connectivity should be in such a way that it is
accessible anytime, anyplace, by anyone, and for anything. The intelligence of the network
should rely on ensuring trust, privacy, and security in addition to the lower requirement
of resources. It should provide better computation capability at low-energy consumption
(Atzori et al., 2010).
19.3.5 Health Care Solutions Using Smartphones—Medical IoT Apps

The health care solution provided by smartphones is termed mobile health or m-Health.
It is described as the use of mobile and wireless technologies in the practice of med-
icine and the monitoring of public health. The most recent example is a smartphone
with health sensors as accessories like wrist gear (Sermakani, 2014). To serve Indian
people, Indian Medical Associations (IMA) come forward with a digital partnership
scheme with Lybrate (private mobile health care provider). Around 0.25 million doctors
are associated with it, get trained to serve the people better through advanced technol-
ogy based services. It will be a bridging step toward connecting rural and urban heath
care system (WI Bureau, 2015). The Government of India launched a mobile app called
National Health Portal India targeting rural residents. The Gram Panchayats (a local self-
government organization) are connected to Government through broadband and assist
semiliterate or illiterate rural in accessing health-related information (National Health
Portal of India, 2015). NGOs such as Gramvaani brought mobile and interactive voice
response-based technology to facilitate health benefits to 700 billion rural population of
India (Gramvaani, 2013).
19.4 Challenges to Overcome

19.4.1 Scalability
In spite of the fast growth of IoT in the health care sector, there are only 6% people are
aware or taking benefit out of it (Madhuri & Sowjanya, 2016). There are few challenges
need to be addressed for its universal application. One of such obstacles is scalabil-
ity. In simple terms, scalability is the ability of a system to maintain or increase its
efficiency with increased workload. The increased workload can be in the form of a
large volume of data storage, its processing and analysis. With the enhancement of
technology-based facility in medical and health care sector, a large volume of data
are collected. The raw data generated demands big data analytics and cloud storage
for meaningful interpretation of data (Sermakani, 2014). To meet the dynamically grow-
ing requirements/demands of people, society, and health organizations, IoT networks
need to be scalable and adaptable. (Islam et al., 2015). Scalability in IoT can be attained
by bearing in mind few important recommendations. Whenever deploying an IoT sys-
tem, long-term support for devices based on technology should be incorporated. The poten-
tial longevity of the network technology should be considered and must be compared
with expected lifespans of devices. The system must be created in such a way that its
expansion is possible in an easy way, thinking of future prospects. The next consider-
ate parameter is device durability. Ensure quality of device to cut down the opera-
tional cost for the device along with the network for long-term operation (John Horn,
2016).
19.4.2 Interoperability
Health care information and management systems society (HIMSS) explained the term
interoperability in the context of health care as the potentiality of different information
technology systems and software applications to communicate, exchange data, and use
the information that has been exchanged. The data exchanged are shared between clini-
cians, lab, hospital, pharmacy, and patient. The aim is to improve the effective delivery
of health care for individuals and communities (HIMMS, 2013). As wireless technology
is emerging and changing rapidly, it causes interoperability in IoT more difficult. This
raises intercommunication problems leading to integration issues. The devices used
in IoT ranges from low power to high power, and capabilities are a major problem
that arises during integration. Apart from the heterogeneity of devices and technology,
few more interoperability associated problems are security and privacy-related issues
(Elkhodr et al., 2016). The solution of this can be achieved in the form of cloud comput-
ing that aids in IoT data collection and data processing, supports rapid setup, and inte-
gration of new things at low costs (Hodkari & Aghrebi, 2016).
19.4.3 Safety of Patients

One of the major concerns of IoT is patient safety. Some legal issues arise from time to time
with respect to patient’s safety. There are apps and wearables available in the market that
is not subjected to food and drug administration (FDA) regulation. There are reliability
issues like whether an app or device provides the quality service that the developer claims.
Such mobile apps and medical devices pose a risk to a patient’s safety (Terry, 2016). There
must be strict Government regulation and policy to have a check on the launching of such
devices and apps to secure patient’s safety.
19.4.4 Security and Personal Privacy

Privacy is another issue that should be dealt with while expanding IoT. Knowingly or
unknowingly, everyone has to counter face IoT at some point. Sometimes, huge data
generated by the means of IoT application remain invisible and unnoticeable causing
loss of control over it. Some other problems also arise like shifting of control toward
devices and algorithm due to automatic decisions causes security issues (Gong, 2016).
There are works going in the direction of providing security and personal privacy to
the user. One of such examples is the patient-centered access to secure systems online
project. It was intended to apply cutting edge security to the correspondence of clinical
data throughout the Internet (Masys et al., 2002). The use of encryption algorithm and
hashing techniques will also ensure data privacy and integrity, respectively (Devi &
Muthuselvi, 2016).
19.4.5 Lack of Government Support

IoT in the health care sector is still going through developing phase. Industry hesitates
to invest in the lack of assurance of profitable returns. The government should come up
with financial support for social welfare and prosperity. Taking inspiration and assurance
from Government projects, more private firms will participate in IoT-based schemes for
development of people and society (Kim & Kim, 2016). Many countries like the United
Kingdom, the United States, China, and Japan came forward with an investment plan in
IoT sector. UK government has funded a £5 million project, whereas China invests USD
800 million in the IoT industry. Such type of motivation is needed for growth and develop-
ment in IoT industry (Xu et al., 2014).
Another demand from Government is to set regulatory bodies such as FDA for people’s
safety and security. Such authorities must actively participate in setting up the standards
committee, bringing up the regulations, and to make sure strict persuasion of laws by
people and industry (Sermakani, 2014).
19.5 Accomplishments through IoT

19.5.1 Glucose Level Sensing
Diabetes is a chronic disorder affecting a large number of the population worldwide.
To address this global problem, it is needed to have non-invasive glucose-level sensing
technique. Its effective management could be possible with the help of a combination of
IoT and m-health (mobile health). It is called m-IoT (m-healthy things). Researchers have
developed a new technique that connects IPv6-based communication technologies such
as 6LoWPAN with emerging 4G networks. It comprises a temperature sensor to quantify
the body temperature of the patient. Along with this, it also has a noninvasive optophysi-
ological sensor attached for glucose-level monitoring in the body of the patient. By using
an IPV6 access point that is connected to an IP-based health care center makes clinicians
access patient’s data. This helps in real-time monitoring and management of glucose level
in patients (Istepanian et al., 2011).
19.5.2 Electrocardiogram Monitoring

According to a WHO report, cardiovascular diseases are the top most cause of death
throughout the world. With the expansion of Internet facility, secure sharing of patient’s
information and clinical data are possible even for remote patient’s monitoring applica-
tions. A web-based ECG system helps in real monitoring of cardiac patient from home.
This is an improved and cheaper way of cardiac-disease management from home and can
be applied to other diseases such as diabetes and asthma (Magrabi et al., 1999). Continuous
ECG monitoring is required to tackle and manage an emergency situation. In such a case,
wireless body sensor networks emerge as an economical, continuous, and remote health
monitoring for next generation of ambulatory personal telecardiology or e-cardiology sys-
tems. With the help of such mini and wearable sensors, doctors are able to record and
monitor real-time cardiogram of chronic patient. This wireless technology is providing
personalized care and is a boon for such patient (Kanoun et al., 2011). Another such tool is
Bluetooth low-energy technology that helps in wireless monitoring of ECG. This method
lessens power utilization as well as gives long monitoring strategy with the assistance of
cell phones (Yu et al., 2012).
19.5.3 Rehabilitation System

IoT-based rehabilitation system comes with the solution of better supervision of patients
and old-age people. IoT is a proficient way to interconnect all the accessible resources
and provide the immediate action required. The benefit it offers over conventional hospi-
tal rehabilitation is ease-effective treatment, adequate interaction, and quick reconfigura-
tion. It is a better way to fulfill patient’s specific requirements by maximum utilization of
medical resources (Fan et al., 2014). One of such examples includes smart rehabilitation
system for poststroke patients. Around 80% of poststroke patients face difficulty in body
movement. It is difficult for them to perform day-to-day activity due to muscle weak-
ness, paralysis, stiffness, or changes in sensation. The intervention of technology helped
patients with impaired motor activity to regain their health. The introduction of smart
devices helps patient to do exercise on their own, tracking of their performance became
easy, and reduces their dependency on occupational therapists (Harris & Sthapit, 2016).
A progress toward smart rehabilitation system is exemplified through brain–computer

interface controlled functional electrical stimulation that has been proposed for spinal
cord injury and stroke patients. It comes with a durable and easily accessible system
that activates the motor pathways by means of motor imagery and sensory pathways
(Al-Taleb & Vuckovic, 2016). Another example includes the use of ankle–foot rehabilita-
tion system using the robotic system to assist elderly people suffering from joint spastic-
ity (Chen et al., 2016).
19.5.4 Medication Management

Smart medication system can empower elderly people with independent and quality
life. Elderly patients suffering from one or more chronic diseases need timely medi-
cation, specific therapies, periodic/routine monitoring of health parameters, specific
diet plan, personal care, and attention. Information and communication technology
and robotic technologies provide a new care process (Fiorini et al., 2016). An automatic
drug dispensers or talking pillboxes constantly remind about timely intake of medicine
and help to avoid any ambiguity (Powe, 1999). A step-ahead solution is cloud robotics
that enables lucrative relationships among patients, family members, caregivers, phar-
macists, and physicians. It served with a combined service of cloud computing along
with a residential robot, an android robot application, and an online interface (Fiorini
et al., 2016). Another good example of IoT-based medication system is in the case of
Parkinson’s disease and multiple sclerosis management in which wearable sensors
are used for quality improvement of patient’s life (Uem et al., 2016). Apart from this,
the personalized care could be provided by interconnecting existing medical device
products such as inhalers and insulin pens with sensors and connectivity technologies
(Dimitrov, 2016).
19.5.5 Wheelchair Management

Another application of IoT is supporting patients with declining motor and sensory
capabilities by using wheelchair-management techniques. The traditional assistive
devices such as canes, crutches, walkers, and wheelchairs embedded with modern tech-
nology ease patient’s life. Such devices with the help of sensor gain knowledge about
the patient’s physiological and physical state and functions accordingly (Ko et al., 2010).
Use of such systems not only decreases patient’s dependency on others but also lessens
family member’s worries about his/her safety. This can be clarified with the assistance
of an illustration in which scientists designed wheelchair with a locator and ultrasonic
radar. The locator made it easy to detect the patient’s activity and inform the respon-
sible authority, such as if patients fall from a wheelchair in the toilet, the emergency
signal reaches the nurse to take immediate action to help the patient, whereas the radar
system uses ultrasonic pulses to monitor patient’s activity in their beds (Alemdar &
Ersoy, 2010). Tracking of wheelchairs at airports has also become easier by the use of
IoT (Srinivasa, 2016).
19.5.6 Monitor an Aging Family Member

The data collected by a poll of 12,000 experts from health care sector unveil that aging pop-
ulation would be the greatest challenge of this domain by the year 2025. The respondents
are majorly from Asia Pacific (20%), Europe, and Middle East Africa (30%) (Staff, 2015).
Aged family members required continuous monitoring to tackle any emergency medical
situation. Their activity can be traced by the deployment of emergent technology such
as ultrasound-based technology. It is very helpful in a critical situation such as detecting
person fall. There is an emergency-call facility available based on battery-operated system.
Apart from this, there are waterproof sensor systems that can be worn like a wristwatch.
The sensor is programed to locate user at every 15–20 s and send relevant data to the
homecare gateway by means of inbuilt-wireless-wide area network connection. Authority
is warned if any critical incidence (such as cardiac arrest) occurs to provide medical assis-
tance without delay (Rghioui et al., 2014).
19.5.7 Other IoT Technologies

Other than above mentioned assistances, IoT technologies impart other benefits in the
health care sector. The advantage it offers includes collecting data and its automatic
processing, identification, and verification of individuals, tracing patients’ activities,
wearable gadgets, and others (Atzori et al., 2010). One of such example is Zio Patch. It is a
cardiac rate observer/recorder for transient ischemic attack (TIA) or stroke patient. During
its implication, a waterproof patch is attached to the patient chest for 14 days. The data
gathered during this period and the patch are supposed to be mailed to a central data
processing center. It is then analyzed by the cardiologists to take required remedial steps
(Tung et al., 2014).
19.6 Conclusion and Future Perspective

Although many accomplishments are achieved in the health care sector using IoT, still
there are many shortcomings that need to be remedied for the well-being of people and
society. There are many supporting technologies and devices that make IoT, a reality.
However, few challenges are there as a barrier which need to be overcome in future. One
of such challenges is electrical power requirement by IoT-based devices, especially in
remote areas or inaccessible areas. Providing benefits of IoT-based services that too at an
affordable price is a major challenge. Solar powering in the form of photovoltaics (PV) to
power devices seems to be a potent futuristic solution for this problem. Other sustain-
able energy sources could also be used as energy-scavenging technique such as kinetic,
thermal, or wind energy. These sources could be a striking alternative of photovoltaics
(Haight & Haensch, 2016). Other than this, the future of IoT also lies on Internet of nano-
things. The development of nanomachines with communication capabilities and their
interconnection with micro- and macrodevices will enable the Internet of nano-things.
This internet of nano-things will be effective in terms of existing technology by faster
mode of communication, less power consumption, and economically beneficial (Akyildiz &
Jornet, 2010).
Introducing wearable gadgets linked to mobile apps meet current medical demand and
will be advantageous for public health. It is estimated that by the year 2020, smart wear-
able devices will achieve business returns of USD 22.9 billion (Jegan, 2016). However, some
problems persist and are imperceptible. For example, a large number of Indian population
(around 63 million people) is suffering from hearing impairment (Mohan Kumar, 2015).
To improve their life quality, mobile phone linked with the wearable device can be used
as an audiometry device. Another example of wearable technology is smart lens that is in
the phase of development by the joint collaboration effort of Google and Novartis. It could
be useful for 1.7 billon people suffering from presbyopia (vision defect). It will also detect
glucose level in a diabetic patient through tear (Senior, 2014). Another milestone in the
medical and health care science that is yet to be achieved is the introduction of surgical
robots. To make this true, renowned name from medical and IT field (Johnson & Johnson
and Google) has come ahead with a tie up Verb Surgical. They are developing high-end
technology smart surgical robots of relatively smaller size (20% less) and that too at a rela-
tively low cost (Pierson, 2015).
References
Akyildiz, I. F., & Jornet, J. M. (2010). The Internet of nano-things. IEEE Wireless Communications, 17(6),
58–63.
Alemdar, H., & Ersoy, C. (2010). Wireless sensor networks for healthcare: A survey. Computer
Networks, 54(15), 2688–2710. doi:10.1016/j.comnet.2010.05.003
Al-Taleb, M. K., & Vuckovic, A. (2016, May). Home-based rehabilitation system using portable brain-
computer interface and functional electrical stimulation. In International brain-computer interfer-
ence meeting (p. 100). doi:10.3217/978-3-85125-467-9-100
Atzori, L., Iera, A., & Morabito, G. (2010). The Internet of things?: A survey. Computer Networks, 54(15),
2787–2805. doi:10.1016/j.comnet.2010.05.010
Backman, W., Bendel, D., & Rakhit, R. (2010). The telecardiology revolution: Improving the manage-
ment of cardiac disease in primary care. Journal of the Royal Society of Medicine, 103(11), 442–446.
doi:10.1258/jrsm.2010.100301
BI Intelligence. (2016). There are two major challenges to implementing IoT solutions in healthcare. Retrieved
from http://www.businessinsider.com/there-are-two-major-challenges-to-implementing-iot-
solutions-in-healthcare-2016–7?IR=T
Charania, C., Nair, G., Rajadhyaksha, S., & Shinde, A. (2016). Healthcare using Internet of things.
International Journal of Technical Research and Applications, 41(March), 42–45.
Chen, G., Zhou, Z., Vanderborght, B., Wang, N., & Wang, Q. (2016). Proxy-based sliding mode control
of a robotic ankle-foot system for post-stroke rehabilitation. Advanced Robotics, 30(15), 992–1003.
doi:10.1080/01691864.2016.1176601
Coetzee, L., & Eksteen, J. (2011, May). The Internet of things-Promise for the future? An Introduction. IST
Africa, Gaborone, Botswana (pp. 1–9).
Devi, K. N., & Muthuselvi, R. (2016). Secret sharing of IoT healthcare data using cryptographic algo-
rithm. International Journal of Engineering Research, 5(4), 790–991.
Dimitrov, D. V. (2016). Medical Internet of things and big data in healthcare. Healthcare Informatics
Research, 22(3), 156–163.
eHealth. (2015). Telemedicine: Metamorphosing Indian healthcare. Retrieved from http://ehealth.eletson-
line.com/2015/11/telemedicine-metamorphosing-indian-healthcare/
Elkhodr, M., Shahrestani, S., & Cheung, H. (2016). The Internet of things: New interoperability, man-
agement and security challenges. International Journal of Network Security and Its Application,
8(2), 85–102.
Fan, Y. J., Yin, Y. H., Xu, L. D., Zeng, Y., & Wu, F. (2014). IoT-based smart rehabilitation system. IEEE
Transactions on Industrial Informatics, 10(2), 1568–1577. doi:10.1109/TII.2014.2302583
Fiorini, L., Esposito, R., Bonaccorsi, M., Petrazzuolo, C., Saponara, F., Giannantonio, R., … Cavallo, F.
(2016). Enabling personalised medical support for chronic disease management through a
hybrid robot-cloud approach. Autonomous Robot, 40(7), 1–14. doi:10.3837/tiis.0000.00.000
Gong, W. (2016, July). The Internet of things (IoT): What is the potential of the Internet of things (IoT) as
a marketing tool? 7th IBA Bachelor Thesis Conference, Enschede, The Netherlands (pp. 1–13).
Gramvaani. (2013). Rural health care: Towards a healthy rural India. Retrieved from http://www.
gramvaani.org/?p=1629
Haight, B. R., & Haensch, W. (2016). Solar-powering the Internet of things. Science, 353(6295), 124–125.
Harris, N. R., & Sthapit, D. (2016). Towards a personalised rehabilitation system for post stroke treatment.
Sensors application symposium (pp. 1–5).
HIMMS. (2013). Definition of interoperability. Retrieved from http://www.himss.org/library/
interoperability-standards/what-is
Hodkari, H. K., & Aghrebi, S. G. M. (2016). Necessity of the integration Internet of things and cloud
services with quality of service assurance approach. Bulletin de La Société Royale Des Sciences de
Liège, 85, 434–445.
Islam, S. M. R., Kwak, D., & Kabir, H. (2015). The Internet of things for health care: A comprehensive
survey. Access, IEEE, 3, 678–708.
Istepanian, R. S. H., Hu, S., Philip, N. Y., & Sungoor, A. (2011). The potential of Internet of m-health Things
m-IoT for non-invasive glucose level sensing. Proceedings of the annual international conference
of the IEEE Engineering in Medicine and Biology Society, EMBS (pp. 5264–5266). doi:10.1109/
IEMBS.2011.6091302
Jegan. (2016). Transforming healthcare through IoT-TVS NEXT. Retrieved from http://tvsnext.io/blog/
transforming-healthcare-through-iot/
John Horn. (2016). 3 keys to ensuring scalability in the IoT. Retrieved from http://www.itproportal.
com/2016/02/02/3-keys-to-ensuring-scalability-in-the-iot/
Kanoun, K., Mamaghanian, H., Khaled, N., & Atienza, D. (2011). A real-time compressed sensing-based
personal electrocardiogram monitoring system. Design, automation & test in Europe conference
and exhibition, Grenoble, France (pp. 1–6).
Khanna, A., & Misra, P. (2013). The Internet of Things for Medical Devices-Prospects, Challenges and
the Way Forward (A white paper). Retrieved from http://234w.tc.tracom.net/resources/white_
papers/Pages/Internet-of-Things-Medical-Devices.aspx
Kim, S., & Kim, S. (2016). A multi-criteria approach toward discovering killer IoT application in
Korea. Technological Forecasting and Social Change, 102, 143–155. doi:10.1016/j.techfore.2015.05.007
Ko, B. J., Lu, C., Srivastava, M. B., Stankovic, J. A., Ieee, F., Terzis, A., & Welsh, M. (2010). Wireless
sensor networks for healthcare. Proceedings of the IEEE, 98(11), 1947–1960.
Madhuri, T., & Sowjanya, P. (2016). A study on public health surveillance using Internet of things
and its applications. International Journal of Innovative Research in Computer and Communication
Engineering, 4(4), 5143–5147. doi:10.15680/IJIRCCE.2016
Magrabi, F., Lovell, N. H., & Celler, B. G. (1999). A web-based approach for electrocardiogram
monitoring in the home. International Journal of Medical Informatics, 54(2), 145–153. doi:10.1016/
S1386–5056(98)00177–4
Masys, D., Baker, D., Butros, A., & Cowles, K. E. (2002). Giving patients access to their medical
records via the Internet: The PCASSO experience. Journal of the American Medical Informatics
Association, 9(2), 181–191. doi:10.1197/jamia.M1005
Med, G. (2016). What is telemedicine? Retrieved from https://www.globalmed.com/additional-resources/
medical-technology
Mohan Kumar. (2015). Wearable medical devices and end-to-end healthcare. Retrieved from https://
www.mdtmag.com/blog/2015/06/wearable-medical-devices-and-end-end-healthcare
National Health Portal of India. (2015). m-Health. Retrieved from http://www.nhp.gov.in/
miscellaneous/m-health
Ngoc, T. V. (2008). Medical applications of wireless networks abstract?: Table of contents: Retrieved from
http://www.cse.wustl.edu/~jain/cse574–08/index.html
Niewolny, D. (2013). How the Internet of things is revolutionizing healthcare. Freescale Semiconductor, Inc.
(Vol. October). Retrieved from http://cache.freescale.com/files/corporate/doc/white_paper/
IOTREVHEALCARWP.pdf
Nishith Desai Associates. (2016). Investment in healthcare sector in India. Retrived from http://
www.nishithdesai.com/fileadmin/user_upload/pdfs/Research%20Papers/Investment_in_
Healthcare_Sector_in_India.pdf
Omogbadegun, Z. O., & Ayo, C. K. (2007). Impact of mobile and wireless technology on healthcare. In
3GSM & Mobile Computing: An Emerging Growth Engine for National Development (pp. 164–171).
Covenant University, Ota, Nigeria.
Pierson, R. (2015, December 10). J&J, Alphabet aim for smarter, smaller, cheaper surgical robot. Retrieved
from http://www.reuters.com/article/us-alphabet-johnson-johnson-robots-idUSKBN0TT1SB
20151210#xG8lkORQElOQ6M7U.97
Piyare, R. (2013). Internet of things: Ubiquitous home control and monitoring system using
android based smart phone. International Journal of Internet of Things, 2(1), 5–11. doi:10.5923/j.
ijit.20130201.02
Powe, P. S. (1999). Date of Patent: US Patent No. 5,954,225. Washington, DC: U.S. Patent and Trademark
Office. Retrieved from https://www.google.es/patents/US5954225.
Puskin, D., Johnston, B., & Speedie, S. (2006). Telemedicine, Telehealth, and Health Information Technology.
Washington, DC. Retrieved from http://www.who.int/goe/policies/countries/usa_support_
tele.pdf?ua=1
Rghioui, A., L’aarje, A., Elouaai, F., & Bouhorma, M. (2014). The Internet of things for healthcare
monitoring: Security review and proposed solution. 2014 Third IEEE International Colloquium in
Information Science and Technology (CIST), 5(5), 384–389.
Senior, M. (2014). Novartis signs up for Google smart lens. Nature Biotechnology, 32(9), 856.
Sermakani, V. (2014, April). Transforming healthcare through Internet of things. Project management prac-
titioners' conference (pp. 1–26). Banglore, India. Retrieved from http://pmibangalorechapter.
in/pmpc/2014/tech_papers/healthcare.pdf
Singh, A. (2016). India is becoming an important market for wearable devices. Retrieved from http://
www.sundayguardianlive.com/opinion/7513-india-becoming-important-market-wearable-
devices
Smith, D. (2013). An introduction to MSP430TM microcontroller-based temperature-sensing solutions. Texas
Instruments Inc. Retrieved from http://goo.gl/UKaURK
Smith, Y. (2015). Types of telemedicine. Retrieved from http://www.news-medical.net/health/Types-
of-Telemedicine.aspx
Srinivasa K. (2016). View from India: Educate today for tomorrow’s internet of medical things.
Engineering and technology. Retrieved from https://eandt.theiet.org/content/articles/2016/11/
view-from-india-educate-today-for-tomorrow-s-internet-of-medical-things/
Staff, F. (2015). IoT, Big data key to overcoming healthcare bottlenecks by 2025: Survey-Firstpost.
Retrieved from http://www.firstpost.com/business/iot-big-data-key-overcoming-healthcare-
bottlenecks-2025-survey-2373440.html
Strehle, E., & Shabde, N. (2006). One hundred years of telemedicine: Does this new technology
have a place in paediatrics? Archives of Disease in Childhood, 91(12), 956–959. doi:10.1136/
adc.2006.099671
Terry, N. P. (2016, April). Will the Internet of things disrupt healthcare? Indiana University Robert H.
McKinney School of Law Research Paper No. 2016–21. doi:10.2139/ssrn.2760447
The World Bank. (2014). Health expenditure, total (% of GDP). Retrieved from http://data.worldbank.
org/indicator/SH.XPD.TOTL.ZS?locations=IN
Tung, C. E., Su, D., Turakhia, M. P., & Lansberg, M. G. (2014). Diagnostic yield of extended cardiac
patch monitoring in patients with stroke or TIA. Frontiers in Neurology, 5, 266. doi:10.3389/
fneur.2014.00266
Uem, J. M., Maier, K. S., Hucker, S., & Scheck, O. (2016). Twelve-week sensor assessment in parkin-
son's disease: Impact on quality of life. Movement Disorder, 1–2. doi:10.1002/mds.26676
Ullah, K., Shah, M. A., & Zhang, S. (2016). Effective ways to use Internet of things in the field of medical
and smart health care. International conference on intelligent system engineering, Islamabad,
Pakistan (pp. 1–8).
Wearable Tech News. (2016). India wearable market hits 400,000 units in Q116 with fitness bands the
key. Retrieved from http://www.wearabletechnology-news.com/news/2016/jun/09/
india-wearable-market-hits-400000-units-q116-fitness-bands-key/
WHO. (2009). Telemedicine-Opportunities and developments in Member States (Vol. 2).
WI Bureau. (2015). Mobile healthcare platform lybrate ties-up with IMA; aiming to turn doctors technically
trained. Retrieved from http://www.franchiseindia.com/wellness/Mobile-healthcare-
platform-Lybrate-ties-up-with-IMA;-aiming-to-turn-doctors-technically-trained.5717
World Health Organization. (1998). A health telematics policy. Report of the WHO group consultation on
health telematics. Retrieved from http://scholar.google.com/scholar?hl=en&btnG=Search&q=in
title:A+health+telematics+policy#5
Wu, G., Talwar, S., Johnsson, K., Himayat, N., & Johnson, K. D. (2011). M2M: From mobile to embed-
ded Internet. IEEE Communication Magazine, 49(4), 36–43.
Xu, L. Da., He, W., & Li, S. (2014). Internet of things in industries: A survey. IEEE Transactions on
Industrial Informatics, 10(4), 2233–2243.
Yu, B., Xu, L., & Li, Y. (2012). Bluetooth Low Energy (BLE) based mobile electrocardiogram monitoring sys-
tem. IEEE International Conference on Information and Automation (pp. 763–767). doi:10.1109/
ICInfA.2012.6246921
20
Mining Ubiquitous Data Streams for IoT
Nitesh Funde, Meera Dhabu, Ayesha Khan, and Rahul Jichkar
CONTENTS
20.1 Introduction ...................................................................................................................... 421
20.2 IoT and Big Data Fundamentals ....................................................................................422
20.2.1 Comparisons between Traditional Data and Dynamic Data in IoT ...........423
20.2.2 Concept Drift ......................................................................................................423
20.2.3 Ubiquitous Data Stream Mining ..................................................................... 424
20.3 Ubiquitous Data Stream Processing and Mining Algorithms .................................. 424
20.3.1 Preprocessing .....................................................................................................425
20.3.2 Ubiquitous Data-Stream-Mining Algorithms ............................................... 427
20.3.2.1 Clustering .......................................................................................... 428
20.3.2.2 Classification ..................................................................................... 431
20.3.2.3 Frequent Pattern Mining .................................................................433
20.4 Ubiquitous Data-Stream-Mining Applications for IoT .............................................. 435
20.4.1 Smart Vehicles .................................................................................................... 435
20.4.2 Remote Health Monitoring............................................................................... 436
20.4.3 Public Safety ....................................................................................................... 437
20.5 Challenges in the Ubiquitous Data Stream Mining toward the Next
Generation of IoT.............................................................................................................. 437
20.6 Conclusion ........................................................................................................................ 438
References..................................................................................................................................... 439
20.1 Introduction
People send and receive information anytime, anywhere all across the globe with the help
of Internet. It is not only just people that are using the Internet; nowadays, physical objects
also connect with each other in the internetwork with the help of Internet. Internet of
Things (IoT) basically connects all the physical objects in the world to the Internet (Gubbi
et al., 2013). Machine-to-machine communication is useful in the manufacturing indus-
try to monitor machinery operations, detect fault, and raise alerts. It will technologically
revolutionize the next generation having a similar impact brought by the revolution of the
computer and the Internet.
With the advancement in the networking industry, more number of people access to
the network that generates a huge amount of data, making it difficult to extract useful
421
knowledge. The data generated by IoT are massive; it contains potentially infinite high-
speed data streams from various heterogeneous sources. Therefore, the traditional data-
mining models are not sufficient to mine this new type of data. Big data analytics is
one of dynamic areas in IoT, which satisfies the characteristics of this new type of data.
The main motivation of this chapter is the real-time analytics of the data generated by
the sensors in the various IoT-based systems. The real-time data are nothing but the
data streams that are sequential, online, and multidimensional in nature. In the real-
world application of data streams, underlying distribution of data changes over the time.
These changes will make a greater impact on the prediction accuracy because of the
default static predictive model. The state is called concept drift. Thus, the predictive
model will have to be updated as the data distribution changes over the time. This
area of data mining has acquired the attention of the researchers to develop new data-
stream-mining techniques since the last decade. Various excellent techniques have been
proposed to address these issues of analyzing fast data streams (Nguyen et al., 2015;
Pinage et al., 2016).
With ever-increasing computational power and wireless-networking capability of small
ubiquitous devices such as mobile, sensor devices, the area of ubiquitous computing
provides the best technological infrastructure for the IoT. It provides an excellent oppor-
tunity to perform intelligent data analysis in the ubiquitous computing environments.
Ubiquitous data mining is the process of performing data analysis on embedded, mobile,
and ubiquitous devices to support applications such as mobile-activity recognition, smart
homes, and health care (Gama & May, 2011). Thus, ubiquitous data-stream-mining (UDM)
algorithms must be computationally efficient, resource-aware, context-aware, and energy-
efficient (Gaber et al., 2014). This chapter aims to present the detailed survey of recent
research developments of data stream mining in the ubiquitous computing environment
and discuss the challenges of ubiquitous data stream mining in IoT.
This chapter is organized as follows. It begins with an overview of big data fundamen-
tals in IoT. Then, a comprehensive survey of ubiquitous data-stream-mining algorithms is
presented. Finally, the ubiquitous applications and challenges in IoT are discussed, fol-
lowed by the conclusion.
20.2 IoT and Big Data Fundamentals

The next era will be of IoT computing differently from the traditional desktop comput-
ing. In IoT, many of physical things that surround us will be in the network of each other.
Business and individuals will be able to take control of their environment with the help
of IoT. Modern day home, vehicles, and office environment can be controlled remotely
with just a smartphone. This new challenge of high degree of connectivity is addressed
by radio-frequency identification (RFID) and sensor networks technologies. Thus, it is
possible to collect huge amount of data about these devices, specifically the information
provided by the sensors embedded in the environment. IoT refers to the world of Internet-
connected devices by which big data are gathered, stored, processed, and managed. Big
data are having main 3 V’s—volume, variety, and velocity. The data generated by devices
in IoT are massive and contain potentially infinite high-speed data streams from hetero-
geneous, distributed sensor sources. Thus, IoT needs the big data methodology for IoT
analytics.
Mining Ubiquitous Data Streams for IoT 423
TABLE 20.1
Difference between Traditional Data and Data in IoT
Characteristics Traditional Data Data in IoT
Data schema Static Dynamic

Available resources Flexible Limited
Computation results Accurate Approximate findings
Data modeling Persistent Modeled as transient data stream
Algorithms Offline processing in which Online processing in which time
time is not constraint is important as data may skip
Dimension One dimensional Multidimensional
20.2.1 Comparisons between Traditional Data and Dynamic Data in IoT

With the continuous advancement in the hardware and software technologies, nature
of the data is always changing in the today’s world. Infinite volume of dynamic data
are generated in various domains such as sensors, Twitter, Facebook, network moni-
toring and traffic engineering, financial market and web logs, web page click streams,
and so on. These data are nothing but the data streams that are different from the
conventional static database. The difference between conventional and new type of
data is given in Table 20.1. Thus, it is interesting how to process and manage the data
streams in a computationally efficient way. This becomes a hot topic of data mining in
IoT environments.
20.2.2 Concept Drift

In the data-mining community, the concept is defined to represent the overall data distri-
bution that is used to perform classification, clustering, and other problems at any time.
Generally, the underlying data distribution is expected to be stable, that is, concept does
evolve over time. However, in many real-world applications of data stream, the statisti-
cal properties of the target environments change over time. This problem is known as
concept drift.
Thus, concept drift occurs when underlying data distribution changes over time in an
unpredictable ways. For example, the behavior of customer in an e-commerce website may
change over time. This shopping behavior of customer changes seasonally. The concept
drift of the unknown probability function can be defined as follows.
Pi(x, ω) ≠ Pj(x, ω) (20.1)
where:
x is a data instance
ω represents the class, and the change occurs from time ti to time tj , where ti < t j .
This mean that optimal predictive model for Pi(x, ω) is no longer suitable (or optimal) for
Pj(x, ω) in a changing environment (Ang et al., 2013). Concept drift also compromises the
prediction accuracy in the above settings.
There are four types of concept drifts such as gradual, incremental, abrupt, and recur-
ring concepts. In gradual concept drift, concept is gradually replaced by the new concept.
Here, both the concepts coexist in a time period in which changes are detected by using
the methods based on time windows that can scan the data. In incremental concept drift,
the concept evolves slowly over time. Here, the new concept replaced the old concept com-
pletely, and both the concept cannot exist in the same time period. In abrupt concept drift,
old concept is suddenly replaced by the new concept. In recurring concept, old concept
disappears but may arrive in the future.
20.2.3 Ubiquitous Data Stream Mining

The effort taken by the researchers in the creation of human-to-human interaction in the
late 1980s resulted in the creation of ubiquitous computing in which the objective is to
bring the technology in the everyday life. Thus, it opens a new opportunity to make intel-
ligent data analysis in the ubiquitous computing environment.
UDM is the process of analyzing real-time sensory data coming from heterogeneous
sources onboard ubiquitous devices (Gaber et al., 2014). UDM is the part of knowledge
discovery in ubiquitous computing environment. It consists of different technologies such
as data mining, machine learning, sensor networks, and activity recognition.
The ubiquitous knowledge discovery is having following characteristics:
• Limited resources: Battery power, bandwidth, computational power, and memory.

• Real-time: The data and other constructs are real time in nature, and learning mod-
els should to be updated in accordance with the evolving environment.
• Locality: The ubiquitous devices know only their local spatiotemporal environment.
• Information processing power: The devices are capable of information processing.
• Mobility: These devices can change their location.
• Distributed: These devices are distributed in nature because of their exchange of
information with other devices.
It is necessary to use data-mining techniques to extract meaningful information from data

in ubiquitous devices with the above mentioned characteristics. The learning takes place
locally within the ubiquitous devices. However, the learning algorithm is quite different
than the traditional data-mining algorithms. The algorithm should be online, incremen-
tal, anytime, and anywhere rather than offline batch scenario in traditional data-mining
algorithm. In the UDM, learning takes place under unreliable, highly resource-constraint
environment such as battery power and bandwidth. Thus, the algorithm should be
lightweight in terms of resource utilization. The next section describes the existing ubiq-
uitous data-stream processing and mining algorithm onboard ubiquitous devices.
20.3 Ubiquitous Data Stream Processing and Mining Algorithms

The real world consisting of interconnected physical ubiquitous objects, exchanging the
information constantly, the volume and velocity of the data generated, and the procedure
involved in handling of those data become critical. There is a range of IoT applications
that requires mining of data streams in ubiquitous environments such as eHealth and
Internet of Things Knowledge

Application
Internet
Access gateway Preprocessing Data stream mining Visualization
Sensing entity
Ubiquitous devices
FIGURE 20.1
Conceptual architecture of ubiquitous data stream mining in the Internet of Things (IoT).
intelligent transportation systems (ITSs). The data generated by the ubiquitous devices
need systematic approach to extract knowledge for successful IoT applications.
Nowadays, efforts have been taken by researchers on big data analytics in the ubiquitous
devices and IoT. The conceptual architecture of UDM process in IoT is shown in Figure 20.1.
The ubiquitous devices such as sensor network and smartphones consist of various sen-
sors; inbuilt sensors (accelerometer in case of smartphones) collect the raw data that need
to be preprocessed. The data-stream-mining techniques consist of various classifications,
clustering, and frequent pattern mining (FPM) techniques; extract the useful information
that can be visualized; and appropriate decision is made from this. This knowledge can be
shared with other ubiquitous devices in the network.
The knowledge-discovery process in the ubiquitous computing environment is explained
in detail in the next upcoming subsections.
20.3.1 Preprocessing
Preprocessing always plays an important role in the process of data mining. Here, in IoT
of ubiquitous computing environment, raw data come from heterogeneous, sensor sources
that contain noisy, missing, incomplete values, and the data may be structured, semistruc-
tured, and unstructured. Preprocessing is one of the basic, important components of big
data mining system. The raw data should be well prepared before it is feeding to data-
stream-mining process.
The raw data can be related to the context data (temperature, GPS-data), data coming
from different applications or data of different behaviors of users. There are three basic
traditional preprocessing approaches that need to apply to the data stream in ubiquitous
environments. They are as follows: outlier removal, normalization, and handling missing
values.
Outliers are referred to as the data points that are far from the mean of the correspond-
ing normal variables. The behavior of outlier is different from the normal data point.
This type of data points are erroneous values and have to be removed from further
process of analysis. The basic way to remove outliers is to use hypothesis testing method
when the data are normally distributed. However, researchers found this method not to
be robust.
Normalization is an important step of preprocessing that scales down or scales up the

data range before it feeds to the further stage. It is given by computing the mean x and
the variance σ2 of the data points xi using following equations:
N N
∑ ∑(x − x )
1 1
x= xi and σ2 = i i
2
(20.2)
N i =1
N −1 i =1
where N is the number of data points. Then, the original values are transformed using the
following equation.
xik − xk
zik = (20.3)
σk
The variable zik is normally distributed with a mean of 0 and a variance of 1. Hence, all the
data points will have identical values of mean and variance. If the mean is subtracted from
the data points, then that process is called centering.
The scaling technique is a special case of normalization that reduces the data pints to a
predefined interval [0, 1] or [−1, 1]. Sigmoid scaling is a renowned scaling technique that
reduces all data points to interval [0, 1] using nonlinear transformation on the original data
points.
1
zik = (20.4)
1 + exp(− xik )
Moreover, some simple functions f (·) can be applied on the data values instead of convert-
ing it into the interval of 0 to 1. The common function is the logarithmic function that
transforms original data points to the logarithmic scale.
The third approach of preprocessing is the handling of missing values. The most popu-
lar technique is the missing data imputation which replaces the missing value with the
random or average value. For example, the average of data points can be used in the place
of missing value.
Figo et al. (2010) described various preprocessing techniques for the context recogni-
tion of smartphone user from the accelerometer data. The key focus of context-aware
services is the capability of mobile devices to collect, process, and manage the raw sensor
data for extracting useful information. From this, device must be able to determine the
features of sensor signal. Sensors in the mobile devices generate a large volume of raw-
data points contaminated with noise in environment that require to be filtered out. The
built-in accelerometer sensor gives the signals instead of a value (given by other sensor)
and requires complex preprocessing techniques to characterize the physical activity of
the user, that is, running and walking (context) within a certain time frame. The authors
have classified sensor signals processing techniques in three broad domains for feature
extraction, namely, the time domain, the frequency domain, and discrete representation
domains.
In the time domain, mathematical and statistical functions are used to extract informa-
tion from raw sensor data. It includes several mathematical and statistical functions such
as mean, median, standard deviation, min, max, range, root mean square metric, correla-
tion, cross correlation, and integration. They are mostly used to select the main features in
the raw sensor data.
In the frequency domain techniques, the repetition of the sensor signal is captured. It
often correlates with the periodic nature of a particular activity such as running or walk-
ing. Fourier transform is the commonly used technique of signal transformation that
allows the important features of time-based signal such as its average, coefficient sum,
and information entropy to represent it in the frequency domain.
In the discrete representation domains, the transformation of accelerometer signal and
other sensor signals into strings of discrete representation symbols. Although there is an
information loss after discretization process, the sensor signal is compressed well into
string. For example, consider a case of a 3-axis accelerometer signal, in which a series of n
input samples split into windows of w consecutive samples. The first step is to compute the
average value for each of window followed by the discretization process over an alphabet
of fixed size. The domain value function is used by the discretization process that defines
the values of intervals corresponding to a given symbol. The symbolic aggregate approxi-
mation is developed recently, which uses piecewise aggregate approximation for mapping
range of values to string symbols (Bondu et al., 2013). Edit distance techniques (such as
Euclidean distance) are used to match the string with known patterns of user activity.
Plötz et al. (2011) used principal component analysis and deep learning-based feature
learning method for activity recognition in ubiquitous computing.
After preprocessing, the stream data are now well prepared for analysis. In the next sec-
tion, the data-stream-mining techniques in the ubiquitous computing environments are
reviewed.
20.3.2 Ubiquitous Data-Stream-Mining Algorithms

As the UDM is the process of analyzing real-time sensory data coming from heteroge-
neous sources onboard resource-constraint ubiquitous devices, there is a requirement of
algorithms that are cost efficient in terms of computations and adaptive to their varying
and dynamic operational context. Here in UDM, the primary and important is not only the
limitation of resources but also these resources and factors such as data rates are subjected
to continuous variability. Thus, algorithm needs to be adaptive to the changing concept.
There are three approaches for adaptation in UDM algorithms, namely resource-aware
approach, situation-aware approach (or context-aware), and the combination of resource
and situation-aware approach, that is, hybrid approach.
In the ubiquitous computing environments, various devices are connected in the
network. Analyzing the real-time data in this environment is a challenging task because
of the limited resource constraints such as battery power, memory, and network band-
width in the ubiquitous devices. Thus, the adaptation in UDM implies that there is
continuous processing or mining of data streams onboard a ubiquitous device and
continuous variability in resource availability. For example, the adaptation of various
parameters in data stream processing algorithms of real time environmental monitoring
such as resource levels, data rates etc. The common approach is of temporal window
size of data streams. The size of the temporal window can be increased or decreased
on the basis of the current status. Thus, the larger window states that the process-
ing of the stream data occurs less frequently, whereas the smaller window performs
processing more frequently. Hence, higher frequency of processing of small temporal
window leads to higher consumption of resources, and accuracy of the stream process-
ing will be higher during the window. This happen vice-versa to the larger temporal
window. Researchers have proposed the granularity-based adaptation algorithms that
adapt according to availability of resource and data stream rate (Gaber et al., 2014).
Data stream mining

algorithms
Frequent pattern
Clustering Classification
mining (FPM)
Lightweight Lightweight
Lightweight
clustering frequent (LW
class (LWClass)
(LWC) FPM)
Distributed FPM in mobile-

RA-VFKM classification for commerce
PDM
Naive Bayes FPM in activity

RA-HCluster classifier recognition
Fuzzy
PcStream classification
algorithm
FIGURE 20.2
Ubiquitous data stream mining algorithms.
This approach is classified into three different forms of variation such as algorithm
input granularity (AIG), algorithm processing granularity, and algorithm output granu-
larity (AOG). AIG is the process in which data rates feeding into the mining algorithm
are adapted according to battery charge. AOG is the process in which the adaptivity is
achieved by adjusting the algorithm output rate (number of clusters). This process is
discussed in more detail in the next paragraph. Algorithm processing granularity does
the adaptation of processing of algorithm according to CPU usage.
As discussed, the data-stream-mining algorithm requires performing granularity-based
processes that adjust its parameters operations to deal with reduced or increased resource
availability and speed of data rates. This adjustment will have the impact on the accuracy
of mining algorithms. Therefore, resource availability and the accuracy of mining algo-
rithms have a strong corelationship.
All the UDM algorithms are reviewed, which are lightweight and cost-efficient in terms
of resource-utilization. The algorithms follows the granularity-based approaches specifi-
cally the AOG form and resource-aware adaptation, situation-aware approach, and hybrid
of two. The lightweight clustering (LWC), lightweight class, and lightweight FPM algo-
rithm follow the AOG approach particularly. The structure of UDM algorithms review is
as shown in Figure 20.2.
20.3.2.1 Clustering
Clustering is the process of grouping the high similar data points in one group, whereas
these data points are dissimilar to other data points that exist in the different group or
cluster. It is based on the principle of maximizing the intraclass similarity and minimizing
the interclass similarity. The clustering is the unsupervised learning method. Various clus-
tering algorithms are proposed till now. The algorithms in the ubiquitous environments
are reviewed as follows.
1. Lightweight Clustering (LWC): The LWC follows the AOG-based approach in

which algorithm rate, data rate, time needed to fill the available memory, and
accuracy are interrelated (Gaber et al., 2005). In AOG, algorithm threshold is
the parameter for controlling the creation of the new outputs based on the three
factors that vary over the time. They are available memory, time remaining to
fill the available memory, and the rate of data stream. The threshold value is
the maximum acceptable distance element of the data stream and the mean
of the group (centroid). The higher the threshold, the lower will be the output
size produced. The threshold of algorithm used the Euclidean or Manhattan
distance functions. This threshold of algorithm is also used in the lightweight
classification and FPM.
LWC is a one-pass LWC algorithm. The basic idea of LWC proposed by Gaber
et al. (2005) is to add the new data points to existing clusters incrementally in
accordance with adaptive threshold value that is the smallest distance from
the data point to the centroid of cluster. If the distance between the new data
element of the data stream and the centroid of existing cluster is greater than
the threshold values, then create the new cluster. After creation of cluster, if
the memory is full then perform the merge operation to accommodate the new
cluster in memory. If the measured distance is not greater than the threshold,
then increase the weight of existing centroid. The flowchart of LWC is shown in
Figure 20.3.
2. RA-VFKM (Resource-Aware Very Fast K-Means): Shah et al. (2005) proposed
RA-VFKM which is the extended version of Very Fast K-Means (VFKM) algorithm
for execution on resource-constrained ubiquitous devices. VFKM (Domingos &
Hulten, 2001) is a data stream-clustering algorithm developed as faster extended
version of K-means. VFKM uses only particular number of data item in each
step for faster execution instead of using all the available data items used by the
K-means. Thus, VFKM observes an error εi with the probability δi in the results
obtained after mining the particular number of data items in each step i as com-
pared with the results that would have been obtained by mining all the data items
in K-means. The goal of VFKM is to perform a clustering in minimum time with
atmost ε∗ and atleast 1−δ∗ probability with the given parameters.
In a RA-VFKM, resource-awareness and adaptation concepts are applied using
AOG approach. The two new error parameters are added such as ε∗∗ the maximum
allowable error limit and δ∗∗ maximum allowable probability. These two parameters
are controlled and managed in each processing cycle based on the availability of
resource (when the available memory decreases or in critical stage). The algo-
rithm will consume less computational resources, if these parameter values are
increased by some calculated factor to satisfy the goal of fast K-means (FKM) as
mentioned above. Overall, if greater accuracy is needed (decrease in values of error
parameter), then algorithm consumes computational resources quickly which can
result into an application failure. Therefore, RA-VFKM improved the longevity
and continuity in resource-utilization onboard a ubiquitous device compared with
nonresource-aware version of algorithm.
Initialization
Data point x = 1, c = 1, M = # of
memory blocks available
For each cluster, calculate the

distance between the input data
point x and cluster centroid
No For each cluster, calculate

If distance >
minimum distance of x to its
threshold
existing cluster centroids
Yes
Assign the data point x to the
cluster whose distance is
Create a new cluster c
minimum from it and increase
the centroid weight by that
minimum distance
If memory is No
Centroid = (centroid
not full
*short.distance + x)/
(short.distance +1)
Yes
The data point x becomes Merge the data

the centroid of cluster c points into one
cluster
Stopping criteria
FIGURE 20.3
Lightweight clustering (LWC) flowchart.
3. RAH-Cluster (Resource-Aware High Quality Clustering): It is a clustering algorithm

that adapts according to the available resources, so that the ubiquitous device can
continue with the process of clustering at acceptable accuracy under limited mem-
ory resources (Chao & Chao, 2011). The process is divided into the following two
parts: online maintenance and offline clustering. In the online maintenance, the
statistics of summary of data stream are computed and used by the process-offline
clustering. In the first part, the sliding window model is used for processing the
data stream, and summary statistics are computed in the form of microclusters
that updates incrementally. The hierarchical summary frame is used for stor-
ing the microcluster, and the level of it can be adjusted on the basis of resource
availability. The hierarchical summary frame can be adjusted to a higher level by
reducing the amount of data to be processed in a limited resources scenario to
decrease the resource consumption.
In the second part of offline clustering, the summary statistics are used for
the clustering process, and adaptation is carried out on the basis of the available
memory resource. Here, first the resource observing module decides whether
the memory is sufficient or not by computing its usage and remaining memory.
When the memory is critically low, then sliding window size and hierarchical
summary frame are adjusted using AIG approach, and threshold distance value
is decreased to reduce the amount of data to be processed. The distance thresh-
old can be increased to improve the accuracy of RAH cluster. Compared with
RA-VFKM, RAH cluster perform well in terms of accuracy and maintains low and
stable usage of memory resource.
4. PcStream: PcStream is a stream-clustering algorithm for finding temporal context
of an entity in an unsupervised manner (Mirsky et al., 2015). It mines the sensor
stream data for contexts. It dynamically detects the present context as well as pre-
dicts the coming context. For example, it used to detect the different behavior of
data point, that is, anomaly in sensor networks.
The name pcStream is used for the principal components of the data distri-
butions in the stream data that are used dynamically to detect and compare
contexts. The basic idea of the core pcStream algorithm is to capture the data
stream’s distribution and determine when it change to known or unknown (new)
distribution. There are various parameters of pcStream—threshold distance, the
minimum context drift size, available memory size of model, and the percent of
variance to retain in the context models. The general process of the algorithm is
started with calculation of statistical similarity between the new data point x and
the known context. If the result is greater than the defined threshold, then it will
be the potential point for outlier or new cluster. If such points are continuous and
are able to fill the minimum context drift size, then the new cluster (context) is
discovered. If it is not greater than the user defined threshold, then it will belong
to known context. Here, the memory resource is maintained stable by merging the
context models. The implementation of pcStream algorithm is ongoing on android
mobile platform.
20.3.2.2 Classification
The classification is the supervised learning problem, important for the decision making.
Give a data point, assigning it to one of predefined target class label or category called clas-
sification. The classification aims to predict the class labels for each of the data instances
accurately. There are many challenges of ubiquitous stream data learning system for the
problem of classification such as limited resource constraints, adaptation, handling recur-
ring concepts, use knowledge from other ubiquitous devices to improve learning local
concepts, and others. A large number of classification algorithms are available, but out of
them, only a few algorithms are suitable for classifiying data streams in ubiquitous envi-
ronments. The algorithms of classification are surveyed, which have been customized to
operate in UDM based on the principle of anytime and anywhere data stream mining.
1. Lightweight classification algorithm: The lightweight classification algorithm starts

with setting the number of data points according to available memory (Gaber et al.,
2005). The algorithm follows the AOG approach that has been discussed in the
above sections. Whenever a new data point comes, the algorithm searches for the
Initialization
Data point x = 1, c = 1, M = # of
memory blocks available
For each c, calculate the

distance between the centroid [j]
and input data point x
No For each c, find the shortest

If distance >
distance centroid [j] and
threshold
input x
Yes
Determine whether the class

Increment c by one label of input data item x match
to the existing class labels
If memory is No Increase weight of centroid [j] by

not full shortest distance as:
centroid = (centroid *short.
distance + x)/(short.distance + 1)
Yes
The data item itself Merge the data

becomes centroid points to accommodate
in memory
Stopping criteria
FIGURE 20.4
Lightweight classification (LWClass) flowchart.
instance nearer to it based on predefined distance threshold. The distance thresh-

old represents the acceptable similarity measure to consider the new data point
according to existing element’s attribute values. If the algorithm finds this data
element, then it checks the class label of stored data items and new data element.
If it is similar with the accepted distance, then the weight of stored data item
is increased by one; otherwise, the weight is decreased by one, and new data
item is ignored. If the weight becomes zero, then stored data elements will be
released from the memory. The flowchart of the lightweight classification algo-
rithm is shown in Figure 20.4.
2. Distributed classification for pocket data mining: Pocket Data Mining (PDM) is the
process of analyzing data streams for distributed environments (Stahl et al., 2012).
Advancements in wireless mobile computing devices such as laptops, tablets, and
smartphones have made it possible to run a wide-range of ubiquitous applications
in this environment. Stahl et al. (2012) proposed the adoption of data-stream-

classification techniques such as hoeffding trees and naïve Bayes for PDM. PDM
perform collaborative and distributed data mining using stream mining technolo-
gies. The PDM architecture consists of different agent miners, mobile resource
discoverers, and mobile agent decision makers. Here, agent miner is the mobile,
distributed in the environment, which implements the basic stream-mining algo-
rithms. The naïve Bayes classifier is originally developed for batch learning; but
its incremental nature made it applicable to data streams. Here, the naïve Bayes
classifier uses its implementation from massive online analysis (MOA) tool (Bifet
et al., 2010). Naïve Bayes classifier is driven from the Bayes Theorem that states that
if E is an event of interest, and event E occurs with P(E) probability, and P(E|X) is
the conditional probability that event E occurs under that X occurs then:
P(E|X ) = P(X|E)P(E)
P(X )
The naïve Bayes classifier is to assign a class label to a data instance to which
it belongs with the highest probability. Osmani et al. (2014) used adaptive naïve
Bayes classifier with kernel density estimation on smartphone to detect the prox-
imity of two or more individuals (devices) which infers the interpersonal distance
(Osmani et al., 2014).
3. Fuzzy classification algorithm: The unsupervised data-stream-mining techniques
need human intervention for further analysis and understanding of the clustering
results. It becomes challenging in UDM applications when real-time data analy-
sis or decision making is required. Horovitz et al. (2005) presented approach to
annotate the results obtained from ubiquitous data stream clustering to facilitate
interpretation to enable real-time decision making.
The approach is divided into three stages. First stage is the online LWC of the data stream.
In the second stage, principles of fuzzy logic in fuzzy labeling clustering algorithm is used
to label the clusters in real time based on the domain or expert knowledge. The third stage
is the fuzzy classification algorithm that uses the labeled clusters onboard the ubiquitous
device. In this stage, new incoming stream data will be assigned to labeled clusters using
fuzzy degree of membership. The following factors are considered when classifying an
incoming data stream to one of the available classes: distance of new data item from cen-
troid of class, the weight of the class (number of data elements in the class), and the dis-
tance of data item from other class centroid. The classification results include the degree of
probability that the data item belonging to the particular class.
This fuzzy approach of clustering and classification algorithm is used and verified in
the road-safety application to detect the drunk-driving behavior efficiently and accurately.
20.3.2.3 Frequent Pattern Mining

A frequent pattern is a pattern such as a set of items, sequences that occurs frequently
in a dataset. For example, a set of items, such as milk and bread that are appearing fre-
quently together in transaction database, is a frequent itemset. The other example is of a
subsequence such as buying first a smartphone, and then a memory card and a screen
guard, if it occurs frequently in an online shopping database history, then it is called a
Initialization
set number of top frequent
items = k and counter
for each k
Receive the new data

item x one by one
If item is new Yes Put this item in the

and one of the list and increase the
counters = 0 counter by one
No
It item is
Yes Increase the counter
present in one
of k counters by one
No
Increase
The item is new If time >
Yes Check the No the
and all threshold
time counter
counters are time by one
full
Yes
No Reset at least n of k
Stopping criteria counters = 0 and put
Stopping criteria
the item and increment
the counter
FIGURE 20.5
Lightweight frequent pattern mining (LWFPM) flowchart.
frequent sequential pattern. FPM is a key player in association mining, correlations, and
many other relationships among the data. It also helps in clustering, classification and
other data-mining tasks. Thus, it has become a vital task of data-mining research. The
frequent pattern-mining algorithms were surveyed in the mobile or ubiquitous environ-
ments of limited resources for the various applications in mobile commerce and human
activity recognition (Figure 20.5).
1. Lightweight frequent pattern-mining (LWFPM) algorithm: The lightweight frequent

pattern-mining algorithm (LWFPM) sets the number of frequent items that will
be estimated according memory available (Gaber et al., 2005). It follows the AOG
approach. This number of frequent items changes over the time to cope with high
stream data rate. On receiving the new data points one by one, algorithm tries
to find counter and increases the item for the existing registered items. The new
item will be ignored if all the counters are occupied. Counters will also have to
decrease by one till the LWF algorithm reaches some threshold time. The number
of least frequent items will be ignored also, and their counters are reset to zero. If
the new data item is similar to one of existing items in memory then counters will
be increased by one. The major parameters that may affect the accuracy of algo-
rithm are threshold time, number of calculated frequent items, and the number of
items ignored. The flowchart of the LWF algorithm is shown in Figure 20.5.
2. Frequent pattern mining (FPM) in personal sensing device: The research of frequent
pattern-mining algorithm for personal sensing device based data-mining systems
is still ongoing and is at the initial stage. There are only two studies that adapt
FPM algorithms in mobile commerce and human activity recognition.
Lu et al. (2012) proposed the Personal Mobile Commerce Pattern (PMCP-Mine) as the part
of Mobile Commerce Explorer framework to find the personal shopping pattern online of
mobile users. The approach of PMCP-Mine first mines the frequent transactions of a mobile
user from a local purchase dataset and removes the infrequent transactions; updates the
local transactions database. Finally, PMCP-Mine will predict the new pattern of transac-
tions based on the local updated patterns of transactions.
Wang et al. (2012) proposed hierarchical approach based data mining techniques of
emerging patterns (EP) for activity recognition using body sensor networks (BSN). It
works at two layers. In the first layer, data are processed at BSN node and transmitted
it to a mobile device for further analysis. At BSN node, lightweight algorithms are used
for real-time recognitions of gesture. EP is a set of frequent items in one class and a set
of infrequent items in other classes. The main assumption of this EP-based technique is
that data instances of EP items are most likely belong to the corresponding EP class. The
proposed algorithm of EP-based technique performs better than the single layer based and
HMM-based algorithms.
The use of learning methods with FPM-based algorithms is gaining popularity in adaptive
systems. For example, incremental learning methods are used at first stage to adapt continu-
ously with concept drift. Second stage includes the discovery of context correlations using
adaptive a priori algorithms (Kishore Ramakrishnan et al., 2014). It is important to note that
the combination of FPM algorithms with different learning methods is investigated rigor-
ously in resourceful environment, but their usefulness in resource-constrained environments
(RCE) is still an unexplored research area. Therefore, there is need of detailed study of perfor-
mance of FPM-based algorithms to find frequent patterns in RCEs.
20.4 Ubiquitous Data-Stream-Mining Applications for IoT

There are several UDM applications that are important for the IoT.
20.4.1 Smart Vehicles

For developing a smart traffic management system, it is necessary for our vehicles to be
smart, by incorporating multiple sensors connecting them to the Internet and making it a
part of system of systems. Such advanced vehicles will have the capability to sense, speak,
and hear. It will have the capacity to collect data about the distance between surrounding
vehicles, temperature of the engine. Thus, it will contribute to the ITS or smart traffic man-
agement system application of IoT.
Li et al. (2015) designed road vehicle monitoring system based on intelligent visual IoT.
The system is able to identify and extract the visual tags of vehicles on the urban roads.
Visual tags consist of license plate number, vehicle type, and color. The visual sensor
nodes are installed on the urban roads and mobile sensing vehicles for collecting basic
information of it. The design of wireless sensor network consists of different intelligent
visual sensor nodes that extract the visual tags of vehicles on the roads and transmit video
streaming. All the nodes are distributed on the urban roads, and vehicles together con-
struct a large scale intelligent visual IoT. It can effectively help the traffic law enforcement
officers to discover the blacklisted, stolen, and illegal vehicles. Each sensor node contains
a high-resolution camera and an embedded processor, and a wireless link between these
nodes and the central server is established. The central server can receive and analyze
the visual tags transmitted by all the nodes. The estimated route of the target vehicle can
be chronologically linked with the characteristics of its visual tag that can be associated
and mined from the central database. The data mining related tasks can also be done on
the sensing vehicles to collect localize important information. The concept of a Smart vehi-
cle can play a major role in road safety by performing real-time data analysis of sensory
data in moving vehicles for accident prevention. One may also think of drunk-driving
detection through onboard analysis of data streams in moving vehicle.
Horovitz et al. (2005) integrate fuzzy approach in the LWC and classification algorithms
for detecting the drunk-driving detection. They have simulated the overall scenario by ana-
lyzing the data of blood alcohol concentration and collecting the different characteristics of
the drivers such as reaction time to peripheral signals (sec), correct responses to peripheral
signals (number), lane position deviation (ft), speed deviation (mph), and times over speed
limit (number), collisions (number). Some rules are defined to detect the unusual events.
If there are higher number of correct responses to peripheral signals, then driver will be the less
drunk and if there are higher number of times over speed limit then the driver will be more.
20.4.2 Remote Health Monitoring

There are many uses of the systems and products that connect to the IoT, which are chang-
ing business in health care industry. Both patients and health care providers benefit from
IoT making a bigger presence in health care. Health care IoT comes handy by means of
mobile medical applications or wearable devices that can capture health data of patients.
Hospitals can use IoT to keep tabs on the location of medical devices, personnel, and
patients. Internet-connected devices have been introduced to patients in various forms.
Data might come from electrocardiograms, fetal monitors, blood glucose levels, or tem-
perature monitors; tracking this health information is important for patients. Follow-up
interaction with a health care specialist is required for many of such measures. This cre-
ates a need for smarter devices to deliver more valuable data, reducing the need for direct
patient-physician interaction.
Remote health monitoring is an application UDM system can be used to provide 24 × 7
monitoring of patients wearing biosensors. Typical remote health-monitoring applications
use the smartphone as a communications device to transfer the data collected from the
patients wearing the biosensors. Sensors collected the real-time data about the health condi-
tion of a patient. Nowadays, earphones of smartphones can read a patient’s blood pressure;
accelerometer could detect the patient’s physical activity and thus smartphone is used to
detect the consciousness of the patient. The sensory data can be mined to find behavioral
patterns of the patient. The medical staff could also be equipped with smartphones so, if
any critical conditions of patient will send alerts to the nurse. Thus, it is possible to use
UDM to analyze the data immediately onboard mobile device and send immediate alerts to
the nurse when required. The algorithms need to be scalable and cost-efficient in terms of
energy consumption and other computational resources. The LWC algorithm that is dis-
cussed above can be used for remote health-monitoring application.
Hassanalieragh et al. (2015) discussed the opportunities and challenges for IoT in real-
izing this vision of the future of health care monitoring. They also reviewed the differ-
ent approaches for integrating the remote health-monitoring technologies into the clinical
practice of medicine. Here, the LWC type algorithm is used for onboard mining of data
stream with fuzzy logic rules.
20.4.3 Public Safety

IoT technologies are transforming our living area into smart environment typical exam-
ples such as smart city, smart building. Variety of sensors is deployed in smart environ-
ments for capturing and analyzing the data in a smart way, adapting to the behavior of
people. Public safety is one of the challenges of smart city. As the sensors are deployed in
smart cities to monitor the pulse and report situations all around the cities in an 24 × 7
basis, the safety problems can be identified early and services can be localized quickly
that will improve the safety of the citizens. Many studies show that safety issues can be
addressed even with large cheap sensors, for example in crowd detection, fire breakout,
blocked street, or a burst pipe (Soliman et al., 2010).
There are many challenges in that have been taken into account for public safety such
as establishing dynamic communication channels, dealing with big data in real time, and
preserving privacy issues (Vermesan & Friess, 2015). The real time data analysis is the first
important concern in public safety because critical situation needs to be detected early
so that authorities will have enough time to take action to provide safety solutions. Here,
various operations of data mining need to be done onboard ubiquitous device in order to
transfer the results or data for further analysis.
20.5 Challenges in the Ubiquitous Data Stream Mining

toward the Next Generation of IoT
With the rapid development in IoT, big data technologies, the fundamental challenge is
to explore the large amount of stream data and extract the meaningful information for
further actions. In the ubiquitous data stream systems, following challenges need to be
addressed.
1. Large amount of stream data need to collect, mine, and analyze at real-time
onboard ubiquitous device thus need the mining data stream techniques in ubiq-
uitous environment.
2. Data come from heterogeneous distributed sensor sources, and communication is
done with different devices, the data format may be different or unstructured thus
need preprocessing techniques.
Applications Health care Public safety Smart vehicles
Mining algorithms Clustering Classification Freq. pattern mining
Data preprocessing Outlier removal Normalization Handling missing values
Constraints Resource-aware Context-aware Other constraints
Ubiquitous devices Sensor devices Smartphones Other Ubiq devices
FIGURE 20.6
Big data mining system for ubiquitous devices in IoT.
3. The knowledge is deeply hidden in the data and knowledge extraction is not
straight forward so need to analyze different properties of data and find the asso-
ciations in the data.
4. The computational resources in the ubiquitous environments such as battery
power, memory are limited. Thus big data techniques have to be performed well in
the RCE of the ubiquitous computing. The UDM methods should also be managed
well. The accuracy of method should be compromised according to the situations
depending on the specific types of applications. There is need of context-aware
systems in today’s IoT applications.
A big data mining system for ubiquitous devices in IoT is suggested as shown in Figure 20.6.
20.6 Conclusion
This chapter presents current progresses and challenges associated with UDM in running
application; considering time, memory, and resource constraints such as battery power
in the ubiquitous devices. Nowadays, ubiquitous computing environment is popular
because of its ever-increasing computational power and capacity of wireless networking
of the small devices such as sensor devices, smartphones. These ubiquitous devices play
an important role in the various IoT applications such as health care, smart homes, ITSs.
This chapter aims to present the detailed survey of recent research developments of data
stream mining in the ubiquitous computing environment and discuss the challenges of
UDM in IoT.
The different UDM methods such as classification, clustering and FPM in RCEs are
reviewed. Also the challenges of UDM are presented, which needs to be addressed in the
future research of IoT. Big data mining system for ubiquitous data mining in IoT is given.
There is a need for further study of this ubiquitous data mining for the next generation of IoT.
References
Ang, H. H., Gopalkrishnan, V., Zliobaite, I., Pechenizkiy, M., & Hoi, S. C. (2013). Predictive handling
of asynchronous concept drifts in distributed environments. IEEE Transactions on Knowledge
and Data Engineering, 25(10), 2343–2355.
Bifet, A., Holmes, G., Kirkby, R., & Pfahringer, B. (2010). Moa: Massive online analysis. Journal of
Machine Learning Research, 11(May), 1601–1604.
Bondu, A., Boullé, M., & Grossin, B. (2013, August). SAXO: An optimized data-driven symbolic representation
of time series. Neural networks (IJCNN), The 2013 international joint conference on (pp. 1–9). IEEE.
Chao, C. M., & Chao, G. L. (2011). Resource-aware high quality clustering in ubiquitous data streams.
Proceedings of the 13th International Conference on Enterprise Information Systems, Beijing,
China (Vol. 1, pp. 64–73).
Domingos, P., & Hulten, G. (2001, June). A general method for scaling up machine learning algo-
rithms and its application to clustering. In ICML (Vol. 1, pp. 106–113).
Figo, D., Diniz, P. C., Ferreira, D. R., & Cardoso, J. M. (2010). Preprocessing techniques for context
recognition from accelerometer data. Personal and Ubiquitous Computing, 14(7), 645–662.
Gaber, M. M., Gama, J., Krishnaswamy, S., Gomes, J. B., & Stahl, F. (2014). Data stream mining in ubiq-
uitous environments: state-of-the-art and current directions. Wiley Interdisciplinary Reviews:
Data Mining and Knowledge Discovery, 4(2), 116–138.
Gaber, M. M., Krishnaswamy, S., & Zaslavsky, A. (2005). On-board mining of data streams in sensor
networks. In Advanced methods for knowledge discovery from complex data (pp. 307–335).
London: Springer.
Gama, J., & May, M. (2011) Ubiquitous knowledge discovery. Intelligent Data Analysis, 15, 1.
Gubbi, J., Buyya, R., Marusic, S., & Palaniswami, M. (2013). Internet of Things (IoT): A vision, archi-
tectural elements, and future directions. Future Generation Computer Systems, 29(7), 1645–1660.
Hassanalieragh, M., Page, A., Soyata, T., Sharma, G., Aktas, M., Mateos, G., & Andreescu, S. (2015,
June). Health monitoring and management using Internet-of-things (IoT) sensing with cloud-
based processing: Opportunities and challenges. In Services Computing (SCC), 2015 IEEE
International Conference on (pp. 285–292). IEEE.
Horovitz, O., Krishnaswamy, S., & Gaber, M. (2005, October). A fuzzy approach for interpretation
and application of ubiquitous data stream clustering. In Proceedings of Second International
Workshop on Knowledge Discovery in Data Streams, in conjunction with the 16th European Conference
on Machine Learning and the 9th European Conference on the Principals and Practice of Knowledge
Discovery ECML/PKDD.
Kishore Ramakrishnan, A., Preuveneers, D., & Berbers, Y. (2014). Enabling self-learning in dynamic
and open IoT environments. Procedia Computer Science, 32, 207–214.
Li, Q., Cheng, H., Zhou, Y., & Huo, G. (2015). Road vehicle monitoring system based on intelligent
visual Internet of things. Journal of Sensors.
Lu, E. H. C., Lee, W. C., & Tseng, V. S. M. (2012). A framework for personal mobile commerce pattern
mining and prediction. IEEE transactions on Knowledge and Data engineering, 24(5), 769–782.
Mirsky, Y., Shapira, B., Rokach, L., & Elovici, Y. (2015, May). pcStream: A stream clustering algo-
rithm for dynamically detecting and managing temporal contexts. In Pacific-Asia Conference on
Knowledge Discovery and Data Mining (pp. 119–133). Springer International Publishing.
Nguyen, H. L., Woon, Y. K., & Ng, W. K. (2015). A survey on data stream clustering and classification.
Knowledge and information systems, 45(3), 535–569.
Osmani, V., Carreras, I., Matic, A., & Saar, P. (2014). An analysis of distance estimation to detect
proximity in social interactions. Journal of Ambient Intelligence and Humanized Computing, 5(3),
297–306.
Pinage, F. A., dos Santos, E. M., & da Gama, J. M. P. (2016). Classification systems in dynamic envi-
ronments: An overview. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery,
6(5), 156–166.
PlÖtz, T., Hammerla, N. Y., & Olivier, P. (2011, July). Feature learning for activity recognition in ubiq-
uitous computing. In IJCAI Proceedings-International Joint Conference on Artificial Intelligence
(Vol. 22, No. 1, p. 1729).
Shah, R., Krishnaswamy, S., & Gaber, M. (2005, October). Resource-aware very fast k-means for ubiq-
uitous data stream mining. In Proceedings of the Second International Workshop on Knowledge
Discovery in Data Streams, held in conjunction with ECML PKDD 2005.
Soliman, H., Sudan, K., & Mishra, A. (2010, November). A smart forest-fire early detection sensory
system: Another approach of utilizing wireless sensor and neural networks. In Sensors, 2010
IEEE (pp. 1900–1904). IEEE.
Stahl, F., Gaber, M. M., Aldridge, P., May, D., Liu, H., Bramer, M., & Philip, S. Y. (2012). Homogeneous
and heterogeneous distributed classification for pocket data mining. In Transactions on Large-
Scale Data-And Knowledge-Centered Systems V (pp. 183–205). Berlin, Germany: Springer.
Vermesan, O., & Friess, P. (Eds.). (2015). Building the hyperconnected society: Internet of things research
and innovation value chains, ecosystems and markets (Vol. 43). River Publishers.
Wang, L., Gu, T., Tao, X., & Lu, J. (2012). A hierarchical approach to real-time activity recognition in
body sensor networks. Pervasive and Mobile Computing, 8(1), 115–130.
21
IoT toward Efficient Analysis of Aging,
Cardiometabolic, and Neurodegenerative
Diseases—An eHealth Perspective
Leandro Cymberknop, Parag Chatterjee, Diego Dujovne,

Luis Romero, and Ricardo Armentano
CONTENTS
21.1 Introduction ...................................................................................................................... 441
21.2 IoT for Cardiovascular Diseases and Aging ................................................................442
21.2.1 Arterial Indicators: Blood Pressure and Stiffness .........................................442
21.2.2 Cardiovascular Disease: Monitoring Aging ..................................................444
21.3 IoT in Neuro-Cognitive Systems ....................................................................................445
21.3.1 Focal Issues toward Treating Parkinson’s Disease ........................................446
21.3.2 Use of Wearable Systems for Parkinson’s Disease.........................................446
21.4 Networking for the Medical IoT Devices......................................................................448
21.4.1 Use of IoT to Connect Heterogeneous Devices ..............................................448
21.4.1.1 Paradigms ........................................................................................... 449
21.5 Data Management and Intelligence .............................................................................. 450
21.5.1 Intelligent Wearables and Machine Learning ................................................ 451
21.5.2 In-Device-Automated Reporting and Analysis ............................................. 452
21.6 Conclusion......................................................................................................................... 453
References..................................................................................................................................... 453
21.1 Introduction
Internet of Things (IoT) has been a path-breaking technology backed up by many
handshaking research areas to establish a high-end connectivity and communication
between several mutually related devices to share information and interact, toward a better
user experience. In the field of health care the task of IoT is not only to proffer a truly effi-
cient and personalized health care to the users but also to redefine the health care system
by connecting all the stakeholders and the state-of-the-art technologies making the most of
the information shared across the closely communicating devices using the IoT platform.
Cardiovascular diseases (CVD), cerebrovascular accidents, and cancer have always
been prevalent largely in the elderly population. This includes an increasing incidence
of chronic conditions, such as osteoarthritis, chronic airways disease, and diabetes com-
bined with sedentary lifestyles (e.g., obesity). Thus, major health care challenges are posed,
441
primarily focusing on prevention, early detection, and minimally invasive management

of such diseases. As a result, new technologies are applied to assist in patient monitoring
and care. Accordingly, wearable systems offer users the ability to interact with other tools
and physical objects around them, being capable of continuously monitoring vital signs and
electrical signals generated by heart and brain, including posture and physical activities
(Andreu-Perez et al., 2015; Chan et al., 2012; Krohn et al., 2016). Here, IoT constitutes an
emerging paradigm in which everyday objects, devices, and sensors exchange data with
little to no human intervention, exploiting the advanced ways of connectivity and comput-
ing ability.
This chapter would be providing an approach of IoT in some sectors of health care such
as CVD and Parkinson’s disease (PD), where this technology can redefine the entire health
care experience.
21.2 IoT for Cardiovascular Diseases and Aging

Heart diseases stand as one of the leading causes of human fatalities, in which myocardial
infarction constitutes the primary killer worldwide. Risk of several diseases such as coro-
nary heart disease, ischemic stroke, and diabetes mellitus increases because of alterations of
blood pressure (BP) and cholesterol. In this perspective, when a chronic disease is diagnosed,
continuous and long-term monitoring is required (Kakria et al., 2015; Krohn et al., 2016).
Therefore, cardiovascular-focused mobile health technologies count helpful in terms of
noninvasive (or implantable) devices that could provide essential physiological information
such as continuous BP measurements, continuous multilead electrocardiogram (ECG)
monitoring (by means of patches or shirts), transthoracic impedance and cardiac output,
arterial oxygen saturation, among others (Chan et al., 2012; Steinhubl and Topol, 2015).
Some approaches can address to some diseases as shown in Figure 21.1.
21.2.1 Arterial Indicators: Blood Pressure and Stiffness

BP constitutes a powerful CVD risk factor, which is responsible in part for various cardio-
vascular events. Traditionally, two specific points of the BP curve, peak systolic BP and
end-diastolic BP, are used to define the cardiovascular (CV) risk factor (Nichols et al., 2011).
There is also evidence that nocturnal BP is one of the stronger predictors, even when con-
trolling for clinical BP (Steinhubl and Topol, 2015). The determinants of the BP waveform
are the cushioning capacity of arteries (influenced by arterial stiffness) and the timing and
intensity of wave reflections. It can be seen as the result of the combination of a forward-
pressure wave (coming from the heart) and a backward wave (returning from particular
sites characterized by specific reflection coefficients) that propagates (in the opposite direc-
tions) at a given speed, defined as Pulse Wave Velocity (Nichols et al., 2011). Here, wearables
are used for self-monitoring and preventing health conditions such as hypertension (sys-
tolic BP become persistently elevated) and stress. Wireless upper-arm (or even finger’s) BP
monitors can be used for extraction and transmission of BP information, toward the IoT
approach (Kakria et al., 2015). However, the periodic inflation of the cuff is still annoy-
ing and painful in long-term use (Noh et al., 2014). Furthermore, conventional sphygmo-
manometry underestimates the effect of arterial stiffening on central BP (CBP, which is
measured at the output of the left ventricle) and provides less information concerning
IoT for Efficient Analysis of Aging, Cardiometabolic, and Neurodegenerative Diseases 443
Cardiovascular
disease
(ECG, BP, SpO2 ,
respiration)
Neurological
and neuro- Respiratory
degenerative diseases
disorders
(Respiration
(Wearable and
rate)
implantable
devices)
IoT
monitoring
for chronic
conditions
Diabetes Cancer
mellitus
(Invasive
(Blood glucose blood flow
concentration) analysis)
Renal
diseases
(MEMS
technology
therapy)
FIGURE 21.1
Chronic conditions monitoring and IoT approach.
systemic arterial stiffness. Several studies have demonstrated that CBP is a better predictor
of carotid intima-media thickness, restenosis after coronary angioplasty, coronary artery
disease severity, and mortality in end-stage renal disease (Nichols et al., 2011). As this
methodology is still challenging, with the help of nonobtrusive wearable devices, moni-
toring beat-to-beat BP (particularly CBP) 24 h per day could become simple and routine
(Steinhubl and Topol, 2015). In this sense, pulse transit time (PTT, the time delay for the
pressure wave to travel between two arterial sites) can provide the basis cuff-less BP mea-
surement. Generally, aortic PTT is determined from BP waveforms for large artery stiff-
ness quantification and improved cardiovascular risk stratification. Physical models based
on the Moens-Kortweg and Bramwell-Hill equations are usually applied to estimate BP
from PTT, parameters of which are adjusted by means of traditional BP cuff measurements
(Laurent et al., 2006; Mukkamala et al., 2015).
A simplification of the process has been proposed (named pulse arrival time, PAT), in
which PTT is assessed by the time delay between the R-wave of an ECG waveform and
the distal arterial waveform foot (Walsh et al., 2014). However, PAT is equal to the sum
of PTT and the preejection period, being thus dependent on the ventricular electrome-
chanical delay and isovolumic contraction phase (which can vary with contractility and
afterload of the heart). As a result, PAT may not be an adequate surrogate for PTT as a
marker of BP, exceeding the food and drug administration (FDA) BP bias and precision
error (Zhang et al., 2011). Another approach, focused specifically on CBP assessment, is
based on ECG combined with ballistocardiogram recordings (a measurement of the body
response caused by the blood ejected during the cardiac systole) to provide an estimation
of PTT within the aortic domain. The use of ballistocardiogram recordings to monitor
heart rhythm is advantageous in terms of user compliance as very lightweight sensors
are utilized (e.g., accelerometers) (Fierro et al., 2016; Noh et al., 2014). It has to be noted that
next to the accuracy of readings, the user’s compliance with continuous usage is what
guarantees the success of a ubiquitous-health system (Noh et al., 2014). In general terms,
accurate estimation of PTT (in terms of wearable systems design) is the main determinant
in BP, CBP, and stiffness assessment. A perfect complement to this technology could be the
pocket mobile echocardiography, a small device capable to provide high-resolution two-
dimensional ultrasound imaging, color Doppler, and measurement capabilities, thus revo-
lutionizing the bedside and outpatient management of cardiac patients (Walsh et al., 2014).
21.2.2 Cardiovascular Disease: Monitoring Aging

A considerable share of elderly people has reduced mobility due to age-related diseases
leading to a greater need for care and assistance. High age related incidence of CVD arises
in part due to broader aging processes, including accumulating morbidities, diminish-
ing homeostasis, and prolonged injurious effects of CV risk factors (Forman et al., 2001).
Vascular stiffening predisposes to functional decrements and ultimately to ischemia,
heart failure, arrhythmia, and other CV disorders (Forman et al., 2001). An example of CV
disease remote monitoring for the elderly is proposed by Armentano and Kun (2014) in
which complimentary, gold-standard, noninvasive, and low-cost techniques are applied
to obtain parameters used to characterize the structural and functional vascular state.
Sphygmomanometry, applanation tonometry, and ultrasound techniques are used to mea-
sure the presence of atherosclerotic plaques, intima media thickness variations, instan-
taneous BP and vessel diameter waveforms, and local and global vessel distensibility
(Figure 21.2). It becomes clear that pocket mobile echocardiography devices should be
combined with continuous BP waveforms estimations and powerful signal and image-
processing systems to achieve this proposal.
Carotid intima Local and global Peripheral and

Carotid artery
media thickness stiffness central pressure
plaque presence
assessment estimation wave analysis
(ultrasound
(ultrasound (applanation (applanation
measurements)
measurements) tonometry) tonometry)
FIGURE 21.2
Integrated approach to characterize the structural and functional vascular state for the elderly.
21.3 IoT in Neuro-Cognitive Systems

After Alzheimer’s disease, PD is the second most common neurodegenerative disorders
of the central nervous system and affects motor skills and speech (Jankovic, 2008). The
decrease in quality of life associated with this disease affects adversely both the patients
and their families.
Obtaining an accurate diagnosis is critical, as it is now strongly based on clinical obser-
vation and information provided by the patient through his personal impressions and
records. Although the clinical symptoms of PD are quite typical, in many cases it is neces-
sary to ensure that these characteristics do not respond to disorders associated with other
neurodegenerative diseases.
The primary biochemical abnormality in PD is the deficiency of dopamine due to
degeneration of neurons in the substantia nigra (Figure 21.3) pars compact (DeLong and
Wichmann, 2007). The four characteristics of motor features associated with PD are brady-
kinesia, rest tremor, rigidity, postural, and gait impairment. It should be noted that these
manifestations do not always occur together in each patient.
Bradykinesia refers to slowness of movements with a progressive loss of amplitude or
speed during attempted rapid alternating movements of body segments (Jankovic, 2008;
Marsden, 1994). It is crucial to distinguish true bradykinesia from simple slowness. In fact,
failing to acknowledge this is a major source of misdiagnosis.
Rest tremor is a rhythmic oscillatory involuntary movement that arises when the affected
body part is relaxed and supported by a surface, thus removing the action of gravitational
forces (Bain, 2007). It vanishes with active movement. In PD, rest tremor frequency is usu-
ally in the low-to-mid range (3–6 Hz), whereas the amplitude is quite variable, from less
than 1–10 cm wide.
Rigidity refers to an increased muscle tone felt during examination by passive move-
ment of the affected segment (limbs or neck), involving both flexor and extensor muscle
groups (Jankovic, 2008). This resistance is felt throughout the full range of movement and
does not increase with higher mobilization speed.
With regards to postural and gait impairment, Parkinsonian patients tend to adopt
a stooped posture, owing to the loss of postural reflexes, a major contributor to falls
(Jankovic, 2008; Sethi, 2008). The pull test is performed to assess postural stability; the
examiner stands behind the supine patient who is previously warned of the pull applied
to her shoulders, then allowing her to step back to regain balance—some patients will fall
without any sort of postural response.
Substantia nigra
(Dopamine
producing cells)
FIGURE 21.3
Location of the substantia nigra in the brain.
21.3.1 Focal Issues toward Treating Parkinson’s Disease

In the current medical practice that is widely used, assessment of PD motor disabilities
is based on neurological examination during patient’s visits to the clinic and home dia-
ries that the patient or the caregiver keeps. However, the short-time examination may not
reveal important information to the neurologist, whereas data from the daily diaries are
highly subjective as they rely on the patient’s memory and perception of his own symp-
toms. Moreover, the pattern and severity of PD symptoms may vary considerably during
the day. So, measurements of motor functions made in the clinic may not reflect in an
accurate way the actual motor disabilities experienced by the patients in their daily life.
The first step in the evaluation of a patient with tremor to determine if it has a
Parkinsonian origin or not is to categorize the tremor based on its activation condi-
tion, topographic distribution, and frequency (Budzianowska and Honczarenko, 2008).
Establishing the underlying cause is very important because prognosis and specific
treatment plans vary considerably.
The most regularly used instrument for the assessment of PD is the revised version of
the unified PD rating scale (UPDRS) (International Parkinson and Movement Disorder
Society-UPDRS). The Movement Disorder Society-UPDRS stands for numerous clinical
scales and questionnaires that have been regularly used in clinical routine and in studies
to evaluate the presence, severity, and progression of PD symptoms. However, the disad-
vantages of these methods, such as high expenditure of time, investigator and location
dependency, limitations in repeating the assessments regularly, and the lack of quanti-
tative outcomes, motivate to search additional, complementary or even supplementary
quantitative, and objective assessment strategies to manage PD.
21.3.2 Use of Wearable Systems for Parkinson’s Disease

Quantitative assessment and management of PD using new technology-based tools, worn
or operated by patients preferentially in the domestic environment, have attracted consid-
erable attention during recent years.
Currently, assessment of motor abnormalities in PD is mainly clinical, based primarily
on the UPDRS (part III). Unfortunately, this scale presents intra- and interobserver dif-
ferences, and its use is limited to the patient’s visit to the hospital. To overcome these
limitations and difficulties, ambulatory monitoring of PD motor and nonmotor symptoms
method or procedure is required.
It is expected that the development of an effective technical solutions to measure specific
PD features for clinical management in either a controlled or an uncontrolled environment
would be possible with current technology.
Using various techniques, several groups have proposed objective methods to detect and
quantify tremor, bradykinesia, and postural instability (Rajaraman et al., 2000; Salarian
et al., 2007). Recently, there has been a growing interest in applications of body-fixed sen-
sors or very close to the body over clothes, that is, wearable sensors (WSs), for long-term
monitoring of these patients. These sensor units can perform not only motion and physi-
ological data sensing but also some low-level processing.
Although having in mind the possibility of some mean of integration or net of WSs,
there has been detailed several aspects of such sensors that need to be taken into consider-
ation (Pastorino et al., 2013). These aspects are resumed in Table 21.1.
Pasluosta et al. (2015) have proposed a new methodology to estimate the outcomes of
the Pull Test using inertial sensors attached to patient’s shoes. Other research groups have
TABLE 21.1
Aspects to be Taken into Consideration at the Time of Integrate a Network of Wearable Sensors
Aspect Desired Conditions to Meet
Size The size of a wearable sensor needs to be considerably small, which is even in the range of
micrometer, predominantly in the MEMS-systems. However, small size restricts enough
room for power sources and storage
Storage To compromise on size, small sized sensors have extremely limited storage capacity, which
makes it necessary to transmit data to larger repositories or to the cloud
Processing Due to size and power constraint, processing tried to be minimal in a sensor. However,
newer technologies intend to reduce the turnaround time and hence recent trends focus
on edge computing for processing the data close to its source (i.e., the sensor)
been used WSs placed at the patient’s truck and lower back to measure postural sway as a
mean for estimating postural instability (Foerster et al., 1999; Patel et al., 2008).
A group of researchers from Harvard University has developed a wearable, wireless-
sensor platform for motion analysis of patients (Lorincz et al., 2009). The system uses eight
wireless nodes equipped with sensors for monitoring movement and physiological con-
ditions. In contrast to previous systems intended for short-term use in a laboratory, this
system was designed to support long-term, longitudinal data collection on patients in hos-
pital and home settings. Although some of the processings are done at each individual
node, a base station is needed to collect data from sensors. This point is a limitation if you
want to cover a wide area of circulation of the patient, as it will be limited to the near vicin-
ity of the base station.
Pastorino et al. (2013) have proposed a WS network, called perform system, that uses
four triaxial wearable accelerometers (one in each patient’s limb) and one accelerometer
and gyroscope wear on a belt used to record body-movement accelerations and angular
rate. All these information are received and stored on an SD card. The data collected then
need to be downloaded to a PC to identify and quantify the patients’ symptoms.
Although several technologies and systems have been proposed for the monitoring of
some motor symptoms associated with PD, still there is a lack in having a system that
allows recording information in the patient remotely using WSs, including the ability to
perform this task in an outdoor environment, as well as providing objective parameters
near real time to take a certain action (e.g., adjust the dosage of medication delivered to the
patient; Figure 21.4). This leads to the development of a WS network that allows, with the
Smart device:
Drug (levodopa) - Low-level processing
injection pump, - Parameter extraction
controlled remotely - Communication
Medical doctor
analyzes results and
takes some related
Wearable sensors
actions
(wrist and waist): Cloud processing
Senses and sends data Big data analytics
to the smart device
FIGURE 21.4
A possible configuration of an ambulatory monitoring system, with a feedback loop for drug delivery adjustment.
minimum number of sensors, the registration of parameters to achieve a precise diagnosis

and following of patients with PD (Romero et al., 2016). A system with such characteris-
tics should include a mobile device (e.g., smartphone or tablet) that would be in charge of
making the data collection from the sensors and the parameter extraction and would also
upload this information to the cloud, in which additional processing should be done. At
the other end of the link, a medical doctor would analyze the results from cloud process-
ing (big data analytics) and could, for example, adjust the dosage of medication needed in
a remote way.
21.4 Networking for the Medical IoT Devices

Engineering experience says that one of the key elements in IoT (Da Xu et al., 2014) is the
ability to connect, create, and maintain networks autonomously, with minimal user inter-
vention. Connected things are meant to be alone, managing to fulfill their task and to last
as long as possible, many of them in harsh environments. Even though IoT networks can
be wired, wireless is the most common transmission media, because of the deployment
and relocation flexibility it provides and the cost reduction in cables. However, the wire-
less channel are not free from problems, such as interference and propagation issues, and
bandwidth restrictions from local regulations. Following the geographical distribution of
the nodes according to the specific measurement requirements from the application, the
generated topology has not only to deal with communication impairments but also with
range and limited energy challenges. IoT is a very different ecosystem than the original
Internet; moreover, given the exponential growth of the number of connected things, they
will soon outnumber the amount of people connected to the Internet, raising scalability
problems.
From an evolutionary perspective (Mainetti et al., 2011), the ability to create networks
from number-constrained devices has progressed during the last 25 years from research
proposals at the beginning to commercially available solutions nowadays. Brainstorming
in the research community resulted in hundreds of ideas spanning from physical to appli-
cation network layers; but only the selection of the simplest and efficient ones was done.
Those whose design was thought with the whole networking stack interaction in mind
and engineering criteria are the current winners. The future is an open opportunity to
create new ways to improve both communications and networks together in an effort to
enable universal interoperability among all constrained and mainstream devices. It is the
task of standardization boards (and the Internet Engineering Task Force in particular) to
ease integration and adapt each technology to a single protocol: IPv6 (Ishaq et al., 2013).
21.4.1 Use of IoT to Connect Heterogeneous Devices

Connecting all devices on the IoT is ideal that is pushing industry today to overcome
physical limitations, legacy requirements, scalability considerations, and standardization
efforts. IoT devices use different type of media to interconnect, betting either on wired
solutions on standard cables, or to wireless links, based on traditional wireless local area
network (WLAN) infrastructure, Cellular Data (Currently 3G, 4G, and future 5G), and
satellite links such as Iridium for remote access.
21.4.1.1 Paradigms
Three main paradigms in IoT for health applications in terms of connectivity are as fol-
lows: IoT devices for personal mobile use, such as portable cardiac monitors; devices for
home use, such as home activity sensors for the elderly; and integrated systems for medical
facilities, such as hospital-wide network deployments. (Baig et al., 2015; Chen et al., 2013).
For the first case, these devices are aimed to be off-the shelf items for massive applica-
tions, commonly used during daily activities, either with a body-area network (e.g., to
monitor movement using accelerometers) or for cardiac or BP ambulatory instrumenta-
tions. These devices are meant to transmit the information via low-bandwidth wireless
links, mostly based on cellular data where available. If worldwide connectivity is needed,
links based on satellite connections such as Iridium are also available. A special subclass
of this case covers temporary deployments to serve special events in which monitoring is
required, but a standard infrastructure cannot be installed in a short time. For example, in
catastrophe scenarios, such as earthquakes or flooding situations, a portable system must
be available and completely functional in a very short time, so as to allow remote access to
patient’s information and telemedicine links.
The second case implies a system with a local range for multiple IoT interconnected
devices at home and a wide-area network access using a third-party Internet service pro-
vider. Nowadays, the typical path of the information flow is to send the data to a cloud-
based service, which is generally a part of the offer with available analytics both for the
user and for the doctors. The main difference with the first case is the concept of having
multiple sensors in the same area (namely, home or office where the patient interacts) that
corresponds to an integrated view of the behavior of the user, which may be complemented
with a body-area network. What is expected in this kind of context is synchronization of
the generated data against a single-time source, relative high bandwidth availability to ful-
fill online video analysis, cardiac data transmission, and environmental sensing data. This
type of traffic is highly variable; there is a baseline traffic of constant and periodic data
transmitted and a variable component from a different kind of events. If a remote monitor-
ing system is established to react, the balance between local (home or office) processing,
called Fog computing, and Cloud-based access must be balanced to provide more reli-
ability in case the domestic Internet service fails. Failure should lead to the use of backup
redundancy to a lower bandwidth network, for example, cellular data, which may lead to
data loss and reaction delay. It is also relevant to notice that home-based networks do not
provide industrial reliability and real-time capabilities, so the design and operational poli-
cies must agree with this disadvantage.
The third case includes systems built for hospitals or clinics for remote (in-building)
monitoring purposes. These are both mobile devices that are worn by the patients using
body-area networks and devices installed on the infrastructure, using both wired and
wireless links. Although the devices can be classified as providing critical (e.g., inten-
sive care patients) and noncritical (e.g., temperature monitoring in hallways) information,
both share a common infrastructure that belongs to the institution. The most common
topology is built around a wired backbone and a mixed wireless/wired capillarity. To
guarantee a minimal quality of service to each data flow, data sources must be classi-
fied and prioritized accordingly, considering the maximum allowed bandwidth leaving
part of the links available to hold peak bandwidth for event-based traffic. Regarding user
mobility to locate patients and professionals inside the building, for example, wireless
infrastructure and indoor-location services shall be also considered, taking into account
emergency situations.
Nowadays, the IoT is a mature technology being applied globally, and in this kind of
real applications environments, optimization is based on standardized communication
algorithms that can interoperate among nodes. The industry claims for efficient solutions
but not at the expense of changing every single deployed device. As a consequence, evo-
lutionary solutions should arise which should take into account legacy device compat-
ibility until the remaining nodes are replaced. So, applied optimization can be done on
top of current standard systems, which provide freedom to the implementer in different
ways, for example, with the use of objective functions for RPL (Thubert, 2012) routing or
by using scheduling functions (Palattella and Accettura, 2015) on 6P resource allocation
protocols. Another advantage of standardization is to simplify performance comparison
between protocols and design improvements, as they run on the same reference platform
(Thubert, 2012).
Finally, another approach for practical performance evaluation is the concept of Lean
Sensing (Martinez et al., 2015; Martínez et al., 2015). This idea describes IoT design tak-
ing into account energy consumption and network life from the point of view of energy
harvesting and establishes a number of limits based on real-field measurements. Another
relevant topic of lean sensing is data-compression performance evaluation, showing the
ranges in which data compression is efficient for transmission and raw-data transmission
is still a valid solution.
The joint effort from both academic and development communities for more than
25 years of evolution since the inception of the first ideas on WSNs has created a solid
base where the open problems have also changed; security risks, scalability, management,
applications, connectivity, enhanced networks, and energy constraints have become the
most relevant drivers of the new wave.
21.5 Data Management and Intelligence

Whatever level of efficiency we use to connect all the devices for sharing data and mutual
communication, the vital part of this aspect is to extract the requisite information from
these data. The dimensions of the data generated being humongous, big data mechanisms
count inevitable for managing these data efficiently and making sense of it by extracting
significant information out of these data. This implies, computer and data scientists who
write such computational algorithms have to work closely with clinicians to accurately
quantify various health conditions and risk factors (Piwek et al., 2016).
Hospital information systems are an integral part of hospitals these days. However,
this system primarily focuses on managing the patient-related records, but the analysis of
these data is not profusely used directly in medical treatment. IoT in health care primarily
intends to bring the data analysis directly beneficial in terms of health care and predictive
treatment as well (Figure 21.5).
The data generated in the hospital are multifarious. The variety of the data linked
to a patient makes it difficult to manage using any single traditional-data manage-
ment scheme. However, broadly classifying the data based on its source, the entire data
linked to a patient can be classified as personal data (supplied by the patient), machine-
generated data (automated medical reports, etc.), and data from the doctor or some-
one related to the treatment. In the common scenario, personal data of a patient along
with other administrative information of the hospital linked to the patient (e.g., billing,
Personal information Doctor’s reports
Machine-generated data
Billing
Prescriptions
Attending doctors
Patients’ data
Administration and business Artificial intelligence for
analytics medical analytics
FIGURE 21.5
Medical intelligence versus business intelligence in a typical hospital data-management scenario.
attending doctors) are mostly in structured formats and are easily handled using tra-
ditional database-management systems. But medical data (mostly generated from the
devices) are more complex and cannot be structured or generalized easily because of
its heterogeneous nature and immense volume (containing images, videos, and other
reports). IoT puts thrust on analysis of this highly heterogeneous and voluminous med-
ical data because the key of comprehensive and predictive health care lies in these data.
Hence, the role of big data analytics becomes inevitable to analyze and make sense of
this medical data.
21.5.1 Intelligent Wearables and Machine Learning

Wearables could provide a platform for at-home management of long-term chronic condi-
tions (Piwek et al., 2016). Though analysis of the medical data seems useful in obtaining
interesting information regarding a patient, it covers only the first half of the system. To
make the most of this information, this needs to be fed to the machine so that it can act
accordingly. The task of making the system intelligent by learning from the humongous
data generated invokes the need of machine learning and artificial intelligence. Intelligent
machines change the efficiency of entire health care system dramatically as the machine-
generated reports could be analyzed easily by the intelligent machines with a minimum
human intervention. For example, deploying an advanced intelligent image-processing
system to analyze videos for PD-affected patients could enhance the efficiency to a large
extent. Some phases of treatment varies a lot depending on patients, and analysis of life-
style data stands very vital in this aspect. Apart from learning from the data generated at
the hospital, the system needs to learn from the lifestyle data, remotely generated at the
patient’s end by the IoT-based wearable devices.
The rapid development of hardware such as biosensors has overshadowed the slower
development of the software needed to manage the enormous amount of data that these
biosensors can generate. Mobile technologies will go well beyond providing data; physi-
cians already know how to treat (e.g., BP at an office visit) and instead will provide brand
new-data streams (e.g., continuous BP during daily activities) that will require tremen-
dous bioinformatics capabilities to eventually understand. The capacity for technologi-
cal advances in the way we are able to process large amounts of data using predictive
analytics and network biology will usher in the era of personalized medicine and the
ability to predict clinically significant events well before they occur (Walsh et al., 2014).
A lot of global players have already started advancing in this field of using artificial intel-
ligence in health care. Google has launched its Google DeepMind Health project to mine
the data of medical records for providing better and faster health services. IBM Watson
launched its special program for oncologists to provide clinicians evidence-based treat-
ment options. Watson for oncology has an advanced ability to analyze the meaning and
context of structured and unstructured data in clinical notes and reports that may be criti-
cal to selecting a treatment pathway. Then, by combining attributes from the patient’s file
with clinical expertise, external research, and data, the program identifies potential treat-
ment plans for a patient. Moreover, advancements are ongoing in developing cognitive
assistants to doctors in simplifying their task of analysis and decision making. IBM’s medi-
cal sieve is an ambitious long-term exploratory project to build a next generation cognitive
assistant with analytical, reasoning capabilities, and a wide range of clinical knowledge.
Medical sieve is qualified to assist in clinical decision making in radiology and cardiology.
To spot and detect problems faster and in a more reliable way, the cognitive health assistant
can analyze radiology images. This leads to a future where radiologists can only look at
the most complicated cases in which human supervision is useful (Mesco, 2016).
In this perspective, IoT enables the seamless platform of interaction between the medical
devices. However, in the quest of making the system intelligent, the approach can be three
fold—mining the medical records, designing the treatment plans, and cognitive assistance.
21.5.2 In-Device-Automated Reporting and Analysis

To extend the reach of health care, it is essential to make sense of the lifestyle data of a per-
son. It helps in either ways, for predictive measures and for supporting a specific treatment.
The concept of telehealth—particularly the application of smartphones in health care—is
extremely broad and covers all specialties of medicine, from the family doctor taking care
of chronic conditions such as diabetes and heart disease, to the cardiologist, ophthalmolo-
gist, dermatologist, and psychiatrist (Krohn et al., 2016). Wearable systems are used in areas
ranging from telehealth care, telemedicine, telecare, telehomecare, eHealth, p-health, m-Health, assis-
tive technology, or gerontechnology (Chan et al., 2012). Data transmission via wireless-body
communication networks enables patient monitoring by health care providers that can be
alerted as soon as a dangerous event occurs which can be taken care of by the medical
personnel. If cardiac monitoring is not regular, arrhythmias can occur unpredictably with-
out warning, causing fatal damage. Critical abnormalities of cardiac rhythm such as atrial
fibrillation can be detected, recorded, and rapidly treated (Andreu-Perez et al., 2015).
Considering the amount of data generated by the IoT-based wearable devices, it is practi-
cally beyond a human scope to single-handedly monitor and analyze the lifestyle data. To
facilitate this process, minimal human intervention is obvious, in which the IoT-devices
would be self-sufficient in recording the lifestyle data and forwarding for analysis in the
cloud or at the edge. This counts a special case of intelligence that is based on a speci-
fied logic; the lifestyle data can be analyzed to generate summarized reports at intervals.
These reports can be analyzed by the machine itself or by a doctor. However, it is possible
to embed the logic into the machine that can analyze the report and will report an alarm,
only when the data seem beyond the safe range. In practical cases of health care, which
is largely dependent on lifestyle data, automated analysis of lifestyle data counts very
efficient. Due to its continuous analysis, many diseases at its very inception get detected
because of its significant change in this lifestyle data.
21.6 Conclusion
IoT in health care has made possible to extend the reaches of health care beyond the
domains of a hospital. The digitization of health care can eventually help us build a
markedly improved physician–patient relationship, allowing greater time for interac-
tion when a patient requires the care of a physician (Steinhubl and Topol, 2015). As the
era of big data per individual comes into play, with terabytes of biological, anatomic,
physiological, and environmental data becoming fully integrated, humans will no
longer be capable of processing the information. This requires synergistic interaction
between man and machine, which ultimately will transform medicine into a digi-
tized data science and, in the process, markedly improve health care (Steinhubl and
Topol, 2015).
The rapid development of the hardware and software involved in the new generation
of wireless cardiac-monitoring devices has outpaced the real-world validation; moreover,
large-scale, pragmatic studies are needed to validate the enormous amounts of data gener-
ated from these monitors. Ongoing clinical trials will be critical to determine the safety,
efficacy, and cost effectiveness of this new technology relative to conventional methods of
monitoring patients. Moreover, a much greater understanding of individual variability in
the acceptance, engagement, and sustainability of these technologies and the most appro-
priate balance of patient and provider involvement are critically important areas of study
(Walsh et al., 2014).
Security (Zhao and Ge, 2013) is a devil in wild, and IoT is no stranger to this world. As in
any other connected device, there is a need to attack this problem in an integral way, pro-
tecting data from the source to the destination, and protecting devices from unauthorized
access and tampering. This is not only a networking but also an IoT architectural problem,
with many current solutions, but with more open problems to deal with.
However, the future direction of deployment of IoT in health care opens up several
challenges, and aspects such as security risks, scalability, management, applications,
interoperability, connectivity, enhanced networks, and energy constraints have become
the most relevant drivers of the new wave.
References
Andreu-Perez, J., Leff, D. R., Ip, H. M. D., and Yang, G. Z. (2015). From wearable sensors to smart
implants-toward pervasive and personalized healthcare. IEEE Transactions on Bio-Medical
Engineering, 62(12), 2750–2762.
Armentano, R. and Kun, L. (2014). Multidisciplinary, holistic and patient specific approach to follow
up elderly adults. Health and Technology, 4(2), 95–100.
Baig, M. M., Gholam Hosseini, H., and Connolly, M. J. (2015). Mobile healthcare applications: System
design review, critical issues and challenges. Australasian Physical & Engineering Sciences in
Medicine, 38(1), 23–38.
Bain, P. G. (2007). Tremor. Parkinsonism & Related Disorders, 13, 369–374.
Budzianowska, A. and Honczarenko, K. (2008). Assessment of rest tremor in Parkinson’s disease.
Neurologia i Neurochirurgia Polska—Journal, 42(1), 12–21.
Chan, M., Estéve, D., Fourniols, J. Y., Escriba, C., and Campo, E. (2012). Smart wearable systems:
Current status and future challenges. Artificial Intelligence in Medicine, 56(3), 137–156.
Chen, C., Knoll, A., Wichmann, H. E., and Horsch, A. (2013). A review of three-layer wireless body
sensor network systems in healthcare for continuous monitoring. Journal of Modern Internet of
Things, 2(3), 24–34.
Da Xu, L., He, W., and Li, S. (2014). Internet of things in industries: A survey. IEEE Transactions on
Industrial Informatics, 10(4), 2233–2243.
DeLong, M. R. and Wichmann, T. (2007). Circuit and circuit disorders of the basal ganglia. Archives
of Neurology, 64(1), 20–24.
Fierro, G., Silveira, F., and Armentano, R. (2016). Central blood pressure monitoring method oriented
to wearable devices. Health and Technology 6(3). Retrieved from https://www.researchgate.net/
publication/309417698_Central_blood_pressure_monitoring_method_oriented_to_wearable_
devices
Foerster, F., Smeja, M., and Fahrenberg, J. (1999). Detection of posture and motion by accelerometry:
A validation study in ambulatory monitoring. Computers in Human Behavior, 15(5), 571–586.
Forman, D. E., Rich, M. W., Alexander, K. P., Zieman, S., Maurer, M. S., Najjar, S. S., Cleveland, J. C.
et al. (2011). Cardiac care for older adults time for a new paradigm. Journal of the American
College of Cardiology, 57(18), 1801–1810.
Ishaq, I., Carels, D., Teklemariam, G. K., Hoebeke, J., Abeele, F. V. D., Poorter, E. D., Moerman, I. et al.
(2013). IETF standardization in the field of the Internet of things (IoT): A survey. Journal of
Sensor and Actuator Networks, 2(2), 235–287.
Jankovic, J. (2008). Parkinson’s disease: Clinical features and diagnosis. Journal of Neurology,
Neurosurgery, and Psychiatry, 79(4), 368–376.
Kakria, P., Tripathi, N. K., and Kitipawang, P. (2015). A real-time health monitoring system for remote
cardiac patients using smartphone and wearable sensors. International Journal of Telemedicine
and Applications, 2015, e373474.
Krohn, R., Metcalf, D., and Salber, P. (Eds.) (2016). Health-e Everything: Wearables and the Internet of
Things for Health. Merritt Island, FL: DM2 Research and Design.
Laurent, S., Cockcroft, J., Van Bortel, L., Boutouyrie, P., Giannattasio, C., Hayoz, D., Struijker-Boudier, H.
(2006). Expert consensus document on arterial stiffness: Methodological issues and clinical
applications. European Heart Journal, 27(21), 2588–2605.
Lorincz, K., Chen, B. R., Challen, G. W., Chowdhury, A. R., Patel, S., Bonato, P., and Welsh, M. (2009).
Mercury: A wearable sensor network platform for high-fidelity motion analysis. Proceedings of
the 7th ACM Conference on Embedded Networked Sensor Systems-SenSys'09, 183–196.
Mainetti, L., Patrono, L., and Vilei, A. (2011). Evolution of wireless sensor networks towards the
Internet of things: A survey. In Software, Telecommunications and Computer Networks (SoftCOM),
2011 19th International Conference (pp. 1–6). IEEE.
Marsden, C. D. (1994). Parkinson’s disease. Journal of Neurology, Neurosurgery and Pstchiatry, 57(6),
672–781.
Martínez, B., Vilajosana, I., and MontÓn, M. (2015). Exploiting Spatio-Temporal Correlations for Energy
Management Policies. Universitat Autònoma de Barcelona. Departament de Microelectrònica i
Sistemes Electrònics. Barcelona, Spain.
Martinez, B., Vilajosana, X., Vilajosana, I., and Dohler, M. (2015). Lean sensing: Exploiting contextual
information for most energy-efficient sensing. IEEE Transactions on Industrial Informatics, 11(5),
1156–1165.
Mesco, B. (2016). Artificial Intelligence will Redesign Healthcare. Retrieved from http://medicalfuturist.
com/artificial-intelligence-will-redesign-healthcare/
Mukkamala, R., Hahn, J.-O., Inan, O. T., Mestha, L. K., Kim, C.-S., TÖreyin, H., and Kyal, S. (2015).
Towards ubiquitous blood pressure monitoring via pulse transit time: Theory and practice.
IEEE Transactions on Bio-Medical Engineering, 62(8), 1879–1901.
Nichols, W., O’Rourke, M., and Vlachopoulos, C. (Eds.) (2011). McDonald's Blood Flow in Arteries, Sixth
Edition: Theoretical, Experimental and Clinical Principles (6th ed.). Boca Raton, FL: CRC Press.
Noh, S., Park, K. S., Chung, T. J., Yoon, H. N., Yoon, C., Hyun, E., and Kim, H. C. (2014). Ferroelectret
film-based patch-type sensor for continuous blood pressure monitoring. The Institution of
Engineering and Technology, 50(3), 143–144.
Palattella, M. R. and Accettura, N. (2015). 6TiSCH D. Dujovne, Ed. Internet-Draft Universidad Diego
Portales Intended status: Informational LA. Grieco Expires: September 8, 2015 Politecnico di
Bari. University of California Berkeley.
Pasluosta, C. F., Barth, J., Gassner, H., Klucken, J., and Eskoifer, B. M. (2015). Pull test estimation
in Parkinson’s disease patients using wearable sensor technology. Conference Proceedings 37th
Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC),
2015, pp. 3109–3112.
Pastorino, M., Arredondo, M. T., Cancela, J., and Guillen, S. (2013). Wearable sensor network for health
monitoring: The case of Parkinson’s disease. Journal of Physics: Conference Series, 450(1), 1–6.
Patel, S., Hughes, R., Huggins, N., Standaert, D., Growdon, J., Dy, J., and Bonato, P. (2008). Using
wearable sensors to predict the severity of symptoms and motor complications in late stage
Parkinson’s disease. Conference Proceedings 30th Annual International IEEE Engineering in
Medicine and Biology Society 2008, pp. 3686–3689.
Piwek, L., Ellis, D. A., Andrews, S., and Joinson, A. (2016). The rise of consumer health wearables:
Promises and barriers. PLOS Medicine, 13(2), e1001953.
Rajaraman, V., Jack, D., Adamovich, S. V., Hening, W., Sage, J., and Poizner, H. (2000). A novel quan-
titative method for 3D measurement of Parkinsonian tremor. Clinical Neurophysiology, 111,
338–343.
Romero, L. E., Chatterjee, P., and Armentano, R. L. (2016). An IoT approach for integration of compu-
tational intelligence and wearable sensors for Parkinson’s disease diagnosis and monitoring.
Health and Technology, 6(3), 167–172.
Salarian, A., Russmann H., Wider, C., Burkhard, P. R., Vingerhoets, F. J., and Aminian, K. (2007).
Quantification of tremor and bradykinesia in Parkinson’s disease using a novel ambulatory
monitoring system. IEEE Transactions on Biomedical Engineering, 54(2), 313–322.
Sethi, K. (2008). Levodopa unresponsive symptoms in Parkinson disease. Movement Disorders, 23(3),
521–533.
Steinhubl, S. R. and Topol, E. J. (2015). Moving from digitalization to digitization in cardiovascular
care. Why is it important, and what could it mean for patients and providers? Journal of the
American College of Cardiology, 66(13), 1489–1496.
Thubert, P. (2012). Objective Function Zero for the Routing Protocol for Low-Power and Lossy Networks
(RPL), RFC 6552. Retrieved from http://www.ietf.org/rfc/rfc6552.txt
Walsh, J. A., Topol, E. J., and Steinhubl, S. R. (2014). Novel wireless devices for cardiac monitoring.
Circulation, 130(7), 573–581.
Zhang, G., Gao, M., Xu, D., Olivier, N. B., and Mukkamala, R. (2011). Pulse arrival time is not an
adequate surrogate for pulse transit time as a marker of blood pressure. Journal of Applied
Physiology, 111(6), 1681–1686.
Zhao, K. and Ge, L. (2013). A survey on the Internet of things security. In Computational Intelligence
and Security (CIS), 2013 9th International Conference, pp. 663–667. IEEE.
Index
Note: Page numbers followed by f and t refer to figures and tables, respectively.
3G technology, 127 data flow, 177–179, 178f, 179f

5G technology, 127 federation, 179–180
6LoWPAN. See IPv6 over Low-Power Wireless Amazon Simple Storage Service (S3), 108
Personal Area Networks (6LoWPAN) Ambient assisted living (AAL), 105
Ambient temperature, 29
American Medical Informatics Association, 383
A
Analog cameras, 253
AAL. See Ambient assisted living (AAL) Analog-to-digital converter, 122
ABI Research, 332 Analytics-as-a-service (AaaS) model, 332
Absolute humidity, 124 Android-based data recording application,
Abstraction layer, 177 252, 252f
Accelerometers, 125 AODV routing protocol. See Ad Hoc
Access control, 82, 94 On-Demand Distance Vector
Access point (AP), 54, 136 (AODV) routing protocol
Accident APIs. See Application programming interfaces
density, 267–268 (APIs)
hotspots, traditional method, 262–264, 263f Apple HomeKit, 328
sliding window method, 262–264 Apple push notification service (APNS), 187
two-dimensional coordinates, 264 Application layer, 227, 395
location identification, 251–253 SD-IoT, 6–9
accurate data, IoT, 252–253 Application programming interfaces (APIs),
GPS, 251 7, 175
prevention. See Prevention of accident Area calculation method, 268f
Accuracy, 34 Automatic service composition, 154
i-hospital infection tracing system, 366
Acknowledgment-based mode, IEEE
B
802.15.4e, 58
ACS. See Adaptive cognitive system (ACS) Babolat company, 327
Active RFID tags, 353 Background Subtraction module, 156
Adaptation layer below IPv6 importance, Bad-mouthing attack, 236
51–52, 52f Ballot-stuffing attack, 236
Adaptive cognitive system (ACS), 138–139, BAN. See Body area network (BAN)
139f, 140f Battery-free IoB, 277–280
Adaptive data rate (ADR) technique, 66 data mitigation, 278–279
Ad Hoc On-Demand Distance Vector (AODV) energy harvesting, 279–280
routing protocol, 70, 135 radio optimization, 279
Ad Hoc topology, nodes connection, 130f routing protocols, 279
ADR. See Adverse drug reaction (ADR) sleep/wake strategy, 277–278
Advanced and Adaptive Network Technology Battery-powered wireless devices, 51
(ANT), 274, 276–277 Bayesian model, 230–231
Adverse drug reaction (ADR), 105 BDI (belief, desire, intention) model, 296,
Agent-oriented architecture, 298 298, 300
ALMANAC platform, 176–180 taklif system, seven pillars, 298–299, 299f
architecture, 176–177, 176f Beaglebone, 212
cloud APIs, 187f Bicycle-sharing system, 281
457
458 Index
Bicycle WSNs (BWSNs), 273–274 Clean-slate solution, 13

general configuration, 274, 274f Client, 398
limitations, 275 Cloud-based
Bidirectional end devices, 65 APIs, 176
Big data big data analytics, 332
analysis, 11–12, 11f module, 110
analytics, 285, 422 Cloud computing, 100, 107
mining system, 438, 438f and IoT, 9–12
systems, 331 big data analysis, 11–12
Bike-sharing systems, 281 cloud of things, 10–11
BitLock, 283 MCC, 107–109
BLE. See Bluetooth Low Energy (BLE) schemes, 285
Blind spoofing, 87 Cloud federation, 179
Blood pressure sensor, 102 Cloud of Things (CoT), 10–11
Bluetooth (BT), 124, 284 Cloud security and IoT, 95
encounters, 358 Cloud service, 246
wireless protocol, 274, 276 models, 331–332
Bluetooth 4.2, 57 Cloud technology, 261, 330–331
Bluetooth Low Energy (BLE), 5, 56–57, 56f, big data ecosystem, 332, 333f
285, 414 Cloud-WBAN-IoT integration, 109–113
Bluetooth Smart, 56, 285 advantages, 111
Body area network (BAN), 398f challenges, 112–113
BOnSAI, 153 communications, 112
Border Gateway Protocol (BGP), 134 heterogeneity, 113
Border router, 398 performance, 113
Bridge with smart sensors, 143f scalability and flexibility, 112
Broker, 83 security and privacy, 112–113
Brought-in connectivity, 247 for remote health care monitoring, 110f
BT. See Bluetooth (BT) Cluster heads (CHs), 399
Business intelligence (BI), 332 Clustering algorithms, 265
Business models, 326–327 CoAP. See Constrained application protocol
BWSNs. See Bicycle WSNs (BWSNs) (CoAP)
Coffee file system, 392
Cognition selfish layer, 140
C
Cognitive and opportunistic RPL (CORPL), 73
Camera Cognitive Network (CN) architecture, 140
images, 269 Cognitive Processor Module (CPMod),
systems, 262 139–141
Capacitive sensor, 131 Cognitive radio, 140
Capillary solutions, 4 Communication
CAPital EXpenditure (CAPEX), 285 intervehicle, 248–249
Carrier Sensing Multiple Access with Collision IoT design, 249
Avoidance (CSMA-CA), 60 stack, 393–395, 393f
Caterpillar company, 327 technologies, 4–6
Cellular network, 247 cellular technologies and IoT, 6
Channel access mechanisms, 60 short-range communications, 5–6, 6t
Channel hopping, 59 between vehicles, 248
CHI. See Consumer Health Informatics (CHI) Communication bandwidth (BW), 123
Citi Bike, 281 Community health care (CH), 106
Citizen-centric application, 193–198 Component-based software, 310–311
City-management system, 295 Component-based system, 310
Cityscoot, 281 Component engineering, 33
C language, 393 Computational model, 296
Index 459
Computer Control layer, SD-IoT, 8

block diagram, 22f Controlled delamination material
languages, 133 (CDM), 105
Concept drift, 422 Control model, 300
Connected bicycles, WSN, 273–286 Cooja simulator with program, 395–396
background problem, 275–277, 276t components, 392
battery-free IoB, 277–280 levels, 392
data mitigation, 278–279 Coordinated case, in-network caching, 16
energy harvesting, 279–280 Coordinator, 101
radio optimization, 279 ZigBee, 60
routing protocols, 279 CORPL. See Cognitive and opportunistic RPL
sleep/wake strategy, 277–278 (CORPL)
IoB, potential opportunities, 283–286 Corporate frameworks, 90
big data analytics, 285 Cost-effective solution, 53
energy-efficient transmission algorithm, CoT. See Cloud of Things (CoT)
284–285 Coverage, i-hospital infection tracing
energy-harvesting radios, 285 system, 366
high-rate 5G Backhaul, 284 CPMod. See Cognitive Processor Module
network security challenge, 286 (CPMod)
wireless backhaul, 283–284 CRC. See Cyclic Redundancy Checks (CRC)
needs, 280–283 Cryptographic algorithm, 286
cyclist safety driven, 282 Cryptography system, 142
economic value chain, 281–282 Customer relationship management
ITS, 281 (CRM), 329
smart city driven, 280–281 Cyber-city system, 296
technology driven, 282–283 knowledge tree, 311, 312f
overview, 273–274 Cyber-physical cloud computing, 12
Connected Streets project, 280 Cyber-physical systems (CPSs), 12, 321
Connectivity approach for IoT, 127–130, 128t, Cyclic Redundancy Checks (CRC), 59
129f, 130f
Constrained application protocol (CoAP), 392
D
smart parking, 398–399
URI, 399 Dangers—car attacks, IoT, 84, 84f
Consumer Health Informatics (CHI), 383 Darwin model, 310–311
Consumer Health Informatics Working Data
Group of the International Medical acquisition, 130
Informatics Association, 383 authentication, 82
Contactless technology, 213, 213f collection phase, WBAN-IoT integration
Contact tracing process, 351, 359, 359f module, 109–110
Content-based naming, 13–15 formatting, 395
Contiki OS, 391–403 link layer, 395
architecture, 393 management, 138–142
case studies, 395–403 framework, 177
Cooja simulator with program, and intelligence, 450–452
395–396 mitigation, 278–279
PH-based secured data aggregation, security requirements, 89
399–403, 400f, 402f sensing and acquisition, 130–133, 131f,
RPL with network coding, 396–398 132f, 133f
smart parking, CoAP, 398–399 tampering, 87–88
communication stack, 393f, 394–395 throughput, 53
features, 392 transmission, 133–134, 133f, 398
overview, 391–392, 393f Data-aggregation schemes, 399
Contract manufacturer (CM), 29 Database Management System (DBMS), 110
460 Index
Data-mining, 265 necessities, 36–39

algorithm, 268 design for support, 37–38
approach-finding accident hotspots, environmental compliance, 39
265–268 global standards compliance, 36
accident density, 267–268 product accuracy, 3
clustering algorithms, 265 product interoperability, 38
DBSCAN, 265–267 product reliability, 37
Data-transfer session, 400 overview, 22–26
DBSCAN algorithm. See Density-Based Spatial parameters and life cycle activities
Clustering of Applications with Noise consumer, industry, and medicine, 46t
(DBSCAN) algorithm power supply and printed circuit
Decay factor, 231 boards, 43t
Decision Maker (DM) service, 156 product repair, support, and
Decryption, 401 disposal, 45t
Defend against IP spoofing, 95 system software and manufacturing
Defensive patents, 42 process, 44t
DEFINE framework, 21–46 Denial of service (DoS), 286
domain characteristics, 26–29 Density, 267
operating environment, 26–27, 27t Density-Based Spatial Clustering of
product cost, 27–28, 28t Applications with Noise (DBSCAN)
product development process, 28 algorithm, 265–267
product life, 28, 28t accident hotspots, 266f
product support requirement, 29 definitions, 266
product usage, 27, 27t with noise
enablers, 41–42 algorithm, 265
intellectual property and patents, 42 method, 265–267
OEM/ODM components, 41–42 parameters, 265
operating system used, 41 Derivation rule, 303
reference hardware platform, 40 Destination advertisement object (DAO), 72
reference software platform, 40–41 Destination Oriented Directed Acyclic Graph
tools and libraries, 42 (DODAG), 72
environment, 29–31 Deterministic computational approach, 230
dust ingress, 30 Device
electrical power, 31 drivers, 393
fluid ingress, 30–31 with network interface, 343
humidity, 30 Dew/frost point, 124
pressure, 30 Dielectric, 131
temperature, 29–30 Differential binary phase shift keying
functionality, 31–34 (DBPSK), 69
manufacturability, 32 Digital signature, 401
product interconnect and interface, 32 Digraph, 137
product performance, 32–33 Direct encounters, 358
product safety, 33 Direct Sequence Spread Spectrum (DSSS)
product sustenance, 33–34 techniques, 58
usability and ergonomics, 32 Discriminative behavior attack, 235–236
implementation, 34–36 Disease spread modeling, 359–360
development process, 34 DODAG. See Destination Oriented Directed
product documentation and control, 36 Acyclic Graph (DODAG)
product manufacturing process, 34–35 DODAG information object (DIO), 72
product support process, 35–36 DODAG information solicitation (DIS), 72
product V&V process, 35 Domain model, 300
key elements, 25 Downlink, 70
Index 461
Drive-thru Internet, 247 End-user module, cloud-WBAN-IoT integration,

Dutch InnoCross project, 281 110–111
Energy
harvesting, 279–280
E
radios, 285
e-bike, 281 scan, 61
EC2. See Elastic Compute Cloud (EC2) Energy–efficiency ratio, 125
eCall system, 246 Energy-efficient
E-cardiology systems, 414 techniques in WSN, 277, 278f
Eclipse modeling framework (EMF) transmission algorithm, 284–285
meta-model, 317 Enhanced distributed channel access (EDCA)
Ecocenter location activity, 198 mechanisms, 56
Economic value chain, 281–282 Enterprise Integration Pattern (EIP), 338
ED. See Emergency department (ED) for data handling in IoT, 341–342
EDCA. See Enhanced distributed channel asynchronous data architecture,
access (EDCA) mechanisms 341–342
eHealth, 382–384 need for integration, 341
EHR. See Electronic health record (EHR) patterns approach, 342
EIP. See Enterprise Integration Pattern (EIP) Enterprise resource planning (ERP), 329
Elastic Compute Cloud (EC2), 108 Enterprise Service Bus (ESB), 337–339
Electrical power, 31 mapping to integration core principles, 339
Electrocardiogram (ECG), 101–102 primary function, 339
monitoring, 414 Ergonomics, 32
Electronic health record (EHR), 382 ERP. See Enterprise resource planning (ERP)
Electronic manufacturing services, 34 ESB. See Enterprise Service Bus (ESB)
Electronic medical record, 382 Event-condition-action (ECA) model,
Electronic products domain, 26 301, 303
Embedded Secure Element (eSE), 206 Event driven approach, 234
Embedded system, block diagram, 23f E-waste reduction, 39
Emergency department (ED), 373, 376 Exclusive-OR (XOR) gates, 137
overcrowding, 376–377, 382 Expected Transmissions (ETX) values, 73
PTV, 376–380
feasibility study, 384
F
indicators for, 384, 384t
interventions for optimization of, Faraday’s Law of Induction, 203
380–384 Federal Aviation Administration (FAA), 28
Encounter infection probability (EncP), 365 Filter by wastetype, 197–198
Encounter side, i-hospital Filtering rules, 255, 256f
communication range, 358 Finite state machines (FSM), 137
high-level framework, 352–355 Firewall, 94
network architectures, 354 Firmware, 23
sensing devices, 353–354 Fluid ingress, 30–31
infection range, 358 Food and drug administration (FDA), 28
infection tracing framework, 356–359 Frame-based mode, IEEE 802.15.4e, 58
collection and processing, 356, 358 Frame structure and power saving, IEEE
direct/indirect, 358 802.15.4e, 58
human–object, 358 Framework, 24
range, 358 Frequency-hopping spread spectrum
sensing range, 358 schemes, 277
EncP. See Encounter infection probability (EncP) Frequency Shift Keying (FSK), 62
Encryption, 96, 399 Fuzzy logic (FL) approach, 138
End-devices, ZigBee, 60 Fuzzy models, 231
462 Index
G Health-IoT, 104
HealthKit platform, 328
G.9959, 68–69
Heidelberger Druckmaschinen, 327
Gadget-based access control systems, 94
Hierarchical naming, 14
Gateways, 63, 65
for smart city application, 14f
Gauss area formula, 267
High-rate 5G backhaul, 284
Geolocation data, 327
Highway addressable remote transducer
Geometric routing, 135
(HART), 62
Global Positioning System (GPS), 127, 251
Highway safety, 243–269
Glucose sensor, 102
accident prevention, 245–250
Google Street View, 254
available services, 245–247
Google Wallet, 213
hacking, 250
GPS. See Global Positioning System (GPS)
Internet access in vehicle, 247–249
GPS-based navigation systems, 245
personal data theft, 250
Green line, smart cities, 144
road accident data
Gyroscopes, 125
analyzing, 261–269
recording, 250–261
Hi triage, 380–382
H
HomeID, 67
H24 medical assistance, 382 Hotspot candidates, 268–269
Hacking, 250 Hövding, 282
Handlebar-mounted device, 281 Human–object encounter, 358
Hard security, 224 Humidity, 30, 124
HART. See Highway addressable remote Hybrid business models, 328
transducer (HART)
Hash algorithms, 14
I
Health care
challenges ICN. See Information-centric networking (ICN)
interoperability, 412–413 IEEE 802.11ah, 54–56, 55t
lack of government support, 413 IEEE 802.15.4e, 58–60, 59f
safety of patients, 413 IETF. See Internet Engineering Task
scalability, 412 Force (IETF)
security and personal privacy, 413 i-hospital, 351
implications, 407 infection tracing framework, 356–368, 357f
medical devices and sensors, 408 application goal, 356
mobile system access, 407–408 coverage, 366
telemedications, 408–410, 409f encounters side, 356–359
wireless patient monitoring, 407 system evaluation, 365–368
IoT accomplishments tracing side. See Tracing, side of infection
aging family member monitoring, tracing framework
415–416 nodes tracking and high-level framework,
electrocardiogram monitoring, 414 352–356
glucose level sensing, 414 access points, 353
medication management, 415 application side, 355–356
rehabilitation system, 414–415 encounter sensing and collection side,
technologies, 416 352–355, 352f
wheelchair management, 415 filters and aggregation, 354–355
IoT and, 385 network architectures, 354
reality IoT technologies sending information to server, 355
connectivity standards, 411 sensing devices, 353–354
graphical user interface, 411 server side, 355
integrated precision-analog capabilities, 411 iMedPack, 105
low power operations, 410–411 Implanted sensor nodes, 397
Index 463
Indirect encounter, 358 evolution phases, 104f

Inductive sensor, 131 of insecure things, 83–85
operation, 132f IoT security countermeasures, 85
Industrial, scientific, and medical (ISM) band, new IoT threats and attacks, 83–84, 85f
54, 284 wireless backhaul, 283–284
Infection tracing, 352, 355 Internet Engineering Task Force (IETF), 51–52,
i-hospital, framework, 356–368, 357f 392, 394
application goal, 356 Internet of Bicycle (IoB), 277, 285
coverage, 366 battery-free, 277–280
encounters side, 356–359 data mitigation, 278–279
system evaluation, 365–368 energy harvesting, 279–280
tracing side. See Tracing, side of infection radio optimization, 279
tracing framework routing protocols, 279
Infection transmission, 349 sleep/wake strategy, 277–278
Infectious period, 359 potential opportunities, 283–286
Information big data analytics, 285
context, 304 energy-efficient transmission algorithm,
disclosure, 88 284–285
distribution, 15–16 energy-harvesting radios, 285
leakage, 81f high-rate 5G Backhaul, 284
models, 303 network security challenge, 286
Information and communication technologies wireless backhaul, 283–284
(ICTs), 406 Internet of Things (IoT), 3, 22–23, 151, 174,
Information-centric networking (ICN), 4, 12–17 221–222, 296, 321–323, 337, 391
content-based naming, 13–15 benefits, 323, 324f
information distribution, 15–16 in cardiovascular diseases and aging, 442
in-network caching, 16–17, 17f blood pressure and stiffness,
naming scheme, 14 442–444
Information technology (IT), 329 monitoring aging, 444
Infrared data association (IrDA), 70 characteristics, 222
Infrastructure-as-a-service (IaaS), 331 and challenges, 222–223
In-network caching, 16–17, 17f security, 86f
Input component, component-based cloud computing and, 9–12
software, 310 big data analysis, 11–12
Integrated circuits (ICs), 122 cloud of things, 10–11
Intellectual property and patents, 42 communications technologies, 4–6
Intelligent city, 296 cellular technologies and, 6
Intelligent hospitals, 350 short-range communications, 5–6
Intelligent navigation systems, 245 complex applications composition in,
Intelligent traffic system (ITS), urban 164–169
clusters, 145 dangers—car attacks, 84f
Intelligent transportation systems (ITSs), 248, data challenges, 343–344
281, 425 ecosystem, 321–333
Intelligent web, 221 technologies supporting development,
Inter-big data processing, 11 330, 330f
Inter-IC connection, 127 end-point communications in, 342f
Internet, 337 evolution of, 81f
access in vehicle, 247–249 and health care, 385
cellular network, 247 hype, 274
communication between vehicles, 248 ICN, 12–17
intervehicle communication, 248–249, 249f content-based naming, 13–15
Wi-Fi networks, 247 information distribution, 15–16
control message protocol, 395 in-network caching, 16–17, 17f
464 Index
Internet of Things (IoT) (Continued) Internet protocol (IP), 50, 136

issues in, 80–83 spoofing, 87
communication, 82 Intervehicle communication, 248–249, 249f
security, 82 IoT design, 249
layered architecture, 223f Interventions for PTV ED optimization,
monetization opportunities, 323, 324f 380–384
in neuro-cognitive systems, 445 eHealth, 382–384
treating Parkinson’s disease, 446 CHI, 383
wearable systems for Parkinson’s disease, EHR, 382
446–448 telemedicine, 382–383
opportunities, types, 323, 325f VHT, 383
paradigm, 244–246 H24 medical assistance, 382
potential economic impact, 326f Hi triage, 380–382
privacy, 224 Intra-big data processing, 11
safer, 94–95 Investment recovery, 384
SD-IoT, 6–9 IoB. See Internet of Bicycle (IoB)
architecture and system design, 8–9, 8f IoT. See Internet of Things (IoT)
benefits of SDNs, 7 IoT-A. See Internet of Things Architecture
cloud computing transition, 9 (IoT-A) ontology
security, 224 IoT-based
challenges, 85–93 ADR, 105
characteristics, 86f health care systems, 406
implementation in, 83 medical devices and sensors, 408
issues, 95–97 IoT-based health care, 103–107
layers of, 93f challenges, 106–107
security methods, 93–94 communication resources, 107
access control, 94 computational resources, 106
authenticity, 93 IoT and cloud computing, 107
firewalling and IPS, 94 storage resources, 106
redesigns and fixes, 93–94 services, 105–106
secure booting, 94 AAL, 105
service, 152 ADR, 105
flow control in, 343–344, 343f CH, 106
and TCP/IP stack, 50, 51t IoT-management system, 296
technological infrastructure, 328–329, 329f IP. See Internet protocol (IP)
technologies to, 338f IPv6 over Low-Power Wireless Personal Area
threats in, 86–89 Networks (6LoWPAN), 393–394
blind spoofing, 87 features, 395
data tampering, 87–88 protocol, 5–6
information disclosure, 88 stack, 394f, 395
man in middle attacks, 85, 87 IrDA. See Infrared data association (IrDA)
network intrusions, 88 ISM band. See Industrial, scientific, and medical
non-blind spoofing, 87 (ISM) band
repudiablility, 88 Italian Society of Emergency Medicine, 382
service attacks denial, 87 Italy, 371
social engineering attacks, 88–89 health care system, 372
spoofing attacks, 86–87 at local level, 372
trust, 224 at national level, 372
universe Lazio, 373
NFC in, 210–213 demographic situation, 373f
overview of, 208–210 ED, 373, 376
Internet of Things Architecture (IoT-A) foreign inhabitant’s distribution, 374f
ontology, 157 inhabitants age distribution, 374f
Index 465
PTV, 376 Machine learning-based models, 231–232

black and grey code tags, 377f Machine-to-machine (M2M), 125
catchment area of patients, 377 communication, 421
ED, 376–380 systems, 50
first patient evaluation, 376 Manual service composition, 154, 161–162,
hospitalizations per patients, 378f 166–167
regions of, 372f Master/slave architecture, 67
sustainability scheme, 375 MCC. See Mobile cloud computing (MCC)
ITSs. See Intelligent transportation systems (ITSs) technology
ITU-T G.9959 transceiver, 68f MDD. See Model-driven development (MDD)
Means-end reasoning process, 298
Mean Time Between Failure (MTBF), 37
J
Mechatronics items, 32
Java emitter templates (JET), 317 Medium access control (MAC), 50, 393
Java programming language, 308 amendment, 58
frames, 59
layer protocols
K
BLE, 56–57, 56f
KM4City, 175 challenges, 53–54
Knowledge G.9959, 68–69
base, structure, 304–306, 305f IEEE 802.11ah, 54–56, 55t
representation and, 305, 306f IEEE 802.15.4e, 58–60, 59f
tree LoRaWAN, 64–66, 65f, 66f
components, 315, 316f and network layers at physical layer,
cyber-city system, 311, 312f 52–53
and rule-based system, 301, 302f weightless, 69, 70t
WirelessHART, 62–64, 63f, 64f
ZigBee and ZigBee low-energy smart
L
above IEEE 802.15.4, 60–62
Latent period, 359 Z-Wave, 67–68, 67f
Lazio, 373 Message-oriented Middleware (MoM),
Light Dependent Resistor (LDR), 124 155–156, 156f
sensor, characteristic curve, 132f, 133f Message Queue Telemetry Transport (MQTT),
Linear discriminate analysis (LDA), 232 83, 395
Link layer state machine, state diagram, 57f protocol, 156
Location checker application, 253, 253f Metaanalyses, PTV ED, 381
Logical Link Control Protocol (LLCP), 204 Metadata, 177
Long-range wide area network (LoRaWAN), M-Flag, 69
64–66, 65f, 66f MICA motes sensors, 127
Long term evolution (LTE), 283 features, 128t
LoRaWAN. See Long-range wide area network Micro-IP (uIP), 391, 394
(LoRaWAN) MicroSD, 206
Loss in free space, 127–129 Middleware
Low-Energy Adaptive Clustering Hierarchy, 284 architecture, 341
Low-Power Wide-Area Networks (LPWANs), software, 9
51, 64, 125, 283 Minimal Service Model (MSM), 153
Low-Rate Wireless Personal Area Networks Mining ubiquitous data streams for IoT
(LR-WPANs), 58 Big data fundamentals
concept drift, 423–424
traditional and dynamic data, 423
M
ubiquitous data stream mining, 424–425
MAC. See Medium access control (MAC) Big data mining system, 425
Machine-code instruction level, 392 conceptual architecture, 425f
466 Index
Mining ubiquitous data streams Near-Field Communication Interface and

for IoT, (Continued) Protocol (NFCIP-1), 204
ITSs, 425 NetID, 67–68
UDM, 424–425 Network
algorithms, 427–435, 428f coding technique, 396, 397f
preprocessing, 425–427 formation, IEEE802.15.4e, 59
M-IoT (m-healthy things), 414 intrusions, 88
MirrorLink technology, 247 layer, 227, 395
Mn* selection method. See Mostly named* CORPL, 73
(Mn*) selection method encapsulation protocols, 73–74, 75f
Mobile protocols, 70–75
application, 185 routing protocols, 71–73
crowdsensing, 12 RPL, 71–73
system access, 407–408 manager, 63
Mobile cloud computing (MCC) technology, security
107–109 challenge, 286
Mobile Cloud for Assisting Healthcare requirements, 89
(MoCAsH), 108 service, 13
Mobile health (m-Health), 411, 414 Networking for medical IoT devices,
Mobile Network Operator (MNO), 206 448–450
Mobosens U.S. research project, 175 Networking subsystem, 177
MoCAsH. See Mobile Cloud for Assisting New comer/whitewashing attack, 237
Healthcare (MoCAsH) NFC. See Near-field communication (NFC)
Model-driven development (MDD), 340 NFC-based payment systems, 213–217
Modern product life cycle, 24f NFC Data Exchange Format (NDEF), 206
MoM. See Message-oriented Middleware (MoM) NFCIP-1. See Near-Field Communication
Mostly named* (Mn*) selection method, 364–365 Interface and Protocol (NFCIP-1)
MQTT. See Message Queue Telemetry Node(s), 65, 73, 131
Transport (MQTT) capture, 286, 82
MSM. See Minimal Service Model (MSM) interface technologies, 125–134
Multiple Points of Vulnerability, 85–86 connectivity approach for IoT, 127–130
Multipoint-to-point communications, 71 data sensing and acquisition,
Multithreading techniques, 308 130–133
My Water application, 187, 188f, 189f data transmission, 133–134
NodeID, 68
Noise
N
algorithm, 265
Name-based routing (NBR), 13 method, 265–267
Name resolution system (NRS), 15 Non-blind spoofing, 87
Naming scheme, IoT, 14 Northbound interfaces, 8
Nanotechnology, 122
NDEF. See NFC Data Exchange Format (NDEF)
O
Near-field communication (NFC), 70, 105, 124,
203–218 Objective Function (OF), 72
in IoT universe, 210–213 Offloading techniques, 108
application, 215–216 On–off attack, 235
operating mode, 203–207 OpenBike project, 285
card emulation mode, 205–206, 206f Open IoT project, 175
NDEF stack description, 206, 207f Open System Interconnection (OSI) layers, 126
peer-to-peer mode, 204–205, 205f OPerating EXpenditure (OPEX), 285
reader/writer mode, 204 Original Equipment Manufacturer/Original
security, 206–207 Device Manufacturer (OEM/ODM)
stack holders in, 215t components, 41–42
Index 467
OSI Reference Model, 64 hacking, 250

Output component, component-based Internet access in vehicle, 247–249
software, 310 cellular network, 247
communication between vehicles, 248
intervehicle communication, 248–249, 249f
P
Wi-Fi networks, 247
Pacific Control company, 327 personal data theft, 250
PAN ID, 61 Privacy homomorphism (PH) technique, 396, 399
Parallel plate capacitor, 131 Privacy principles, IoT technologies, 89–93
Passive IR (PIR) sensors, 124 at granular level, 91–93
Passive RFID tags, 353 organization’s perspective, 90–91
Payload data, 136 Probabilistic computation approach, 230
Payment-related applications, 206 Process component, component-based
Personal data theft, 250 software, 310
Personalized preventative health Product
coaches, 385 interconnect and interface, 32
Person counting service, 152 performance, 32–33
PH-based secured data aggregation, 399–403, safety, 33
400f, 402f sustenance, 33–34
Photoelectric sensors, 132 Product-as-a-service, 327–328
Photo-resistor, 132 Product Lifecycle (PLC), 24, 24f
PH technique. See Privacy homomorphism (PH) Product lifecycle management (PLM), 329
technique Protothreads, 392
Physical infrastructure layer, PTV. See Policlinico Tor Vergata (PTV)
SD-IoT, 8 Public key infrastructure (PKI), 95
Physical layer (PHY), 52, 123, 227 Public services (PSn), 295, 313
amendment, 58 Pull test, 445
IEEE 802.15.4 PHY, 395
Platform-as-a-service (PaaS), 331, 333
Q
service, 110
Point-to-multipoint communications, 71 Quality of service (QoS), 5, 53, 284
Point-to-point communications, 71
Policlinico Tor Vergata (PTV), 376
R
black and grey code tags, 377f
catchment area of patients, 377 Radio
ED, 376–380 communication, 275
feasibility study, 384 optimization, 279
indicators for, 384, 384t Radio access network (RAN), 284
interventions for optimization of, Radio-frequency identification (RFID), 105, 110,
380–384 124, 126, 181, 350, 353
first patient evaluation, 376 readers, 353
hospitalizations per patients, 378f tags, 126
PoloTech Shirt, 327 technology, 203
PORSCE II, 154–155 Random selection method, 364
Power-by-the-hour model, 327 Rank position, 72
Prevention of accident, 245–250 Rapid7 exploration, 83
available services, 245–247 Raspberry-Pi, 212
eCall, 246 Rayleigh fading, 129
intelligent navigation systems, 245 Rayleigh Probability Density Function, 129
monitoring vehicle status, 246 RDF. See Resource description framework (RDF)
self-driving cars, 246 REACH. See Registration, Evaluation,
voice recognition, 246 Authorization, and Restriction of
warning messages, 245–246 Chemicals (REACH)
468 Index
Receive mode, IEEE 802.15.4e, 58 Routing Protocol for Low-Power and Lossy
Reconfiguration strategy, self-adaptive model, Networks (RPL), 71–73, 392, 396
306–310 BAN, 397–398, 398f
Recyclable waste, 181 network coding, 396–397, 397f
Reduced function device (RFD), 5, 58 thermal aware, 398
Reduction of Hazardous Substance (RoHS), 39 RPL. See Routing Protocol for Low-Power and
Reference data session, 400 Lossy Networks (RPL)
Reference hardware platform, 40
Reference software platform, 40–41
S
Registration, Evaluation, Authorization,
and Restriction of Chemicals Safer IoT, 94–95
(REACH), 39 SAS. See Self-adaptive systems (SAS)
Regulated Industry product, 34 Schoberer Rad Messtechnik device, 274
Rehabilitation system, 414–415 SD-IoT. See Software-defined IoT (SD-IoT)
Relative Humidity, 124 SDNs. See Software-defined networks (SDNs)
Remote health monitoring, 104 Search by keyword functionality, 197
Repudiation attack, 237 Security
Resource description framework (RDF), 175 ALMANAC, framework, 177
Retouch photos, IoT services, 254–255, 254f IoT, 93–94, 224
RFID. See Radio-frequency identification (RFID) access control, 94
RFID-passive tag, 353 authenticity, 93
Rime communication stack, 394 challenges, 85–93
Road accident data characteristics, 86f
analyzing, 261–269 firewalling and IPS, 94
accident hotspots, traditional method, implementation in, 83
262–264, 263f issues, 95–97
data-mining approach-finding accident layers of, 93f
hotspots, 265–268 redesigns and fixes, 93–94
hotspot candidates, 268–269 secure booting, 94
IoT sensors, data, 261 and personal privacy, 413
sensor and data types, 262 Segmentation and Reassembly (SAR) layer, 69
attributes, 255, 255t Self-Adaptive Cyber-City System, 295–316
details, 256f–258f modeling of system, 311–314, 313f
recording, 250–261 and seven pillars of life, 295–296
accident location identification, system configuration, 314–316
251–253 Self-Adaptive Model, 300–311
accident scene, photos, 253–255 component-based software, 310–311
scene investigation work, 250–251 knowledge base, structure, 304–306, 305f
TMC, 255–261 representation and, 305, 306f
road interval, 259f reconfiguration strategy, 306–310
Road-to-vehicle communication, 248 rule-based system, 301, 303
Robotics, 246 knowledge tree and, 301, 302f
Rolls Royce, 327 rule representation, 301–304
Router model, 304, 304f
6LoWPAN, 74 Self-adaptive systems (SAS), 296, 316
ZigBee, 60 framework, 300, 301f
Routing, 134–137, 138f system-development model, 316, 317f
current challenges, 134–136 Self-driving cars, 246
protocols, 279 Self-promoting attack, 236
TCNet architecture, 136–137, 136f, 138f Semantic representation framework,
Routing over low power and lossy networks ALMANAC, 177
(ROLL), 71 Semantic Sensor Network (SSN) ontology, 157
Index 469
Semantic web services, 153, 155 Sliding window method, 262–264

Semaphores, 144 one-dimensional, 263f
Semiautomatic service composition, 154, two-dimensional, 264f
162–164, 167–169, 168f Smart and connected communities (SCC), 4
Semicentralized architecture, 356 Smart cities through IoT, 173–200
Sensing as a Service (SenaaS), 10 ALMANAC platform, 176–180
Sensitivity (SV), i-hospital infection tracing architecture, 176–177
system, 366 data flow, 177–179, 179f
Sensor(s), 5, 122–124 federation, 179–180, 180f
and applications types, 124–125 applications, 180–190
accelerometers, 125 waste management, 181–185, 190–198
distance, 124 water management, 185–190, 198–200
gas and smoke, 124 experimentation activities, 190–200
gyroscopes, 125 overview, 173–176
humidity, 124 Smart city, 280–281, 296
level, 125 with WSN, 144–145, 146f, 147f
luminosity, 124–125 Smart city platforms (SCPs), 175
presence, 124 Smart city resource abstraction layer
speed, 124 (SCRAL), 177
temperature, 124 Smart grids, 142
basic structure in blocks, 122f Smart meeting room
and data types, 262 ontology, 152, 157–159
inside vehicles, 262 resource model, 157, 158f
internal structure, wireless sensor node, 123f service model, 159, 159f
platform, SD-IoT, 8 SYNAISTHISI, 156
Serial Peripheral Interface (SPI) protocol, 212 Smart parking, CoAP, 398–399
Server, 398 Smart Pedal, 282
Service Smartphones, 112
attacks denial, 87 Smart Santander project, 175
catalog, 311, 312t, 313 Smart Utility Networks (SUN) Task Group, 58
cyber-city system, 314, 314f SOA. See Service-oriented architecture (SOA)
composition, 152 Social engineering attacks, 88–89
approaches, 160–164 Software adaptable network layer, 140
manual, 154, 161–162 Software-as-a-service (SaaS) model, 331, 110
semantic input/output matchmaking, Software components configuration,
160–161 314–315, 315f
semiautomatic, 154, 162–164 Software-defined IoT (SD-IoT), 6–9
request, 154, 162 architecture and system design, 8–9, 8f
Service level agreement (SLA), 313 application layer, 8–9
Service-oriented architecture (SOA), 339–340 control layer, 8
Session hijacking, 87 physical infrastructure layer, 8
Set-top box operation, 23 benefits of SDNs, 7
SF. See Spreading Factor (SF) cloud computing transition, 9
Shoelace formula, 267 Software-defined networks (SDNs), 7
Sicslowpan module, 395 Solos smart cycling glasses, 282
Sigmoid scaling technique, 426 Southbound interface, 8
Sink node, 136, 233 Space distribution, 107
Skyhook Wireless company, 327 Spaghetti code, 338
Skylock, 283 SPARQL query, 161, 163
Slandering attack, 236 SPA services, IoT, 152, 155, 163
Sleep mode, IEEE 802.15.4e, 58 Specificity (SP), i-hospital infection tracing
Sleep/wake strategy, 277–278 system, 366
470 Index
Speed measuring sensors, 262 windows, infection, 362–363

Speedometers, 269 trace-back process, 362–363
Spoofing attacks, 86–87 trace forward, 363
Spreading Factor (SF), 64 Traffic
Spread spectrum technology, 66 measuring sensors, 262, 268
SSN. See Semantic Sensor Network (SSN) ontology safety, 248
Stochastic algorithms, 230 Traffic Information Center (TIC), 255
Storage manager (SM), 178 Traffic Message Channel (TMC), 255–261
Structural Health Monitoring (SHM), 142 binary-coded, 260
Susceptible-Infected (SI) model, 360 road accident, 260, 260t
Susceptible-Infected-Recovered (SEIR) model, smart device, 261
360, 366 Traitors, 235
Susceptible-Infected-Susceptible (SIS) model, 360 Transducers, 124, 131
Sustainable development, 373 Transmission control protocol/internet protocol
Sybil attack, 237 (TCP/IP), 391
SYNAISTHISI, 152, 155–156, 161, 164 Transmission phase, WBAN-IoT integration
Systematic supply chain process, 34 module, 110
System-development model, 316, 317f Transmission power control (TPC), 279
Transmit mode, IEEE 802.15.4e, 58
Transport layer, 395
T
Trellis Coded Network (TCNet) approach, 134
Taklif system, seven pillars, 297, 297t architecture, 136–137, 136f, 138f
BDI model, 298–299, 299f label, 136
rules, 299–300 Trilayer security, IoT architecture, 85f
TAOM4E tools, 314 True negatives (TN), i-hospital infection tracing
TCNet. See Trellis Coded Network (TCNet) system, 366
approach True positives (TP), i-hospital infection tracing
Technical service (Tsn), 313 system, 366
Telecardiology, 410 Trust dissemination, 233
Telemedications, 408–410, 409f Trusted Execution Environment (TEE), 206
Telemedicine, 382–383 Trusted Service Manager (TSM), 206
Temperature sensors, 268 Trust management in IoT, 221–236
Tesla Motors, 246 attacks against, 234–237, 235f
Tesla, Nikola, 3 based on identity management, 237
ThingWorx platform, 328 based on reputation management, 234–236
Third-Generation Partnership Project, 283 information gathering, 227–230
Time distribution, 107 scaling of information, 229–230
Time-division multiple access (TDMA), 62 source of information, 229
Timing attack, 82 trust parameters, 228–229
Tiny OS, 127 properties of trust, 225–226
TMC. See Traffic Message Channel (TMC) TMS, 227–234
TMS. See Trust-management system (TMS) trust
Touch&Travel, 211 computation, 230–233
Tracing dissemination, 233
back process, 350–351, 360–363, 361f, 361t update and maintenance, 233–234
side of infection tracing framework, 359–365 Turin pilot, 191t
approaches, 364–365
back, 360–361, 361f, 361t
U
disease spread modeling, 359–360
forward, 361, 361f, 362t Ubiquitous data stream mining (UDM), 422,
infection tracing, 360–365 424–425
parameters, 359 algorithms, 427–435, 428f
tracing windows, 362–363 applications for IoT
Index 471
public safety, 437 WBAN-IoT integration module, 109–110

remote health monitoring, 436–437 WBANs. See Wireless Body Area Network
smart vehicles, 435–436 (WBANs)
future challenges, 437–438 Weather sensors, 268–269
preprocessing, 425–427 Weightless technology, 69, 70t
Uncoordinated case, in-network caching, 16 Well-established Test and Calibration
Underground ecological islands (UEIs), 181 process, 33
in ALMANAC pilot, 181, 183f Wheelchair management, 415
Uniform resource identifier (URI), 399 Wide-Band Code-Division Multiple Access, 127
Uniform Resource Name (URN) system, 208 Wi-Fi. See Wireless-Fidelity (Wi-Fi)
Universal Integrated Circuit Card (UICC), 206 Wired Drone Docking system, 83
Universal mobile telecommunications system Wireless access networks, 4
(UMTS), 283 Wireless Body Area Network (WBANs), 101–103
Uplink, 70 challenges
Urban Water project, 175 data transmission capability, 102–103
U.S. Department of Transport study energy consumption categories, 103
(USDOT), 282 and IoT technology, 103
User datagram protocol (UDP), 395 node characteristics, 102
resources limitation, 103
network size, 101f
V
Wireless-Fidelity (Wi-Fi), 124–125
Vehicles tracking systems by trellis networks, 247
clusters, 146f technology, 274, 276
Vehicle-to-vehicle communication, 248 WirelessHART, 62–64, 63f, 64f
VeloCitta project, 281 Wireless local area network (WLAN), 141
Virtual Health care teams (VHT), 383 Wireless patient monitoring, 407
Virtualization layer, 177 Wireless power transfer (WPT), 280
Virtual machines (VMs), 107 Wireless rural area network (WRAN), 283
Voice recognition, 246 Wireless sensor networks (WSNs), 58, 121,
208, 273
energy-efficient techniques, 277, 278f
W
header, 136
WasteApp, 185, 186f Wireless transmission technologies, 126t
Waste bin activity, 198 WLAN. See Wireless local area network
Waste Electronics and Electrical Equipment (WLAN)
(WEEE) directive, 39 World Wide Web (WWW), 103
Waste management, 181–185, 190–198 WSNs. See Wireless sensor networks (WSNs)
citizen-centric application, 193–198
citizen engagement, 184
Z
deployment activities, 181–182
smart waste pilot in Turin, 181 ZigBee
technical details, 182–184 network components, 61f
Waterflow meter designing, 31 protocol, 5
Water management, 185–190, 198–200 wireless protocol, 274, 276–277
city management and communication and ZigBee low-energy smart above IEEE
application, 188–190 802.15.4, 60–62
mobile iOS application (My Water), 187–188 formation procedure, 61–62
Water-resistant products, 38 network and management, 60–61
Waze, 249 Z-Wave, 67–68, 67f

The Internet of Things - Foundation For Smart Cities, Ehealth, and Ubiquitous Computing (PDFDrive) PDF

Uploaded by

Copyright:

Available Formats

The Internet of Things - Foundation For Smart Cities, Ehealth, and Ubiquitous Computing (PDFDrive) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Internet of Things - Foundation For Smart Cities, Ehealth, and Ubiquitous Computing (PDFDrive) PDF

Uploaded by

Copyright:

Available Formats

The Internet of Things

© 2018 by Taylor & Francis Group, LLC

No claim to original U.S. Government works

Printed on acid-free paper

International Standard Book Number-13: 978-1-4987-8902-8 (Hardback)

Library of Congress Cataloging‑in‑Publication Data

Names: Armentano, Ricardo, editor.

Visit the Taylor & Francis Web site at

and the CRC Press Web site at

Section I Introduction to the Internet of Things:

2. DEFINE—An Architecture Framework for Designing IoT Systems ......................... 21

5. Integration of Cloud Computing and IoT: Opportunities and Challenges .............. 99

Section II Frameworks for the Internet of Things:

7. Semantic Complex Service Composition within an IoT Ecosystem ........................ 151

8. Enabling Smart Cities through IoT: The ALMANAC Way........................................ 173

9. NFC in IoT-Based Payment Architecture....................................................................... 203

10. Trust Management in IoT ................................................................................................. 221

Section III Interdisciplinary Aspects of the Internet of Things

11. Highway Safety of Smart Cities ...................................................................................... 243

12. Connected Bicycles: Potential Research Opportunities in Wireless

13. Self-Adaptive Cyber-City System ................................................................................... 295

14. The Role of Big Data, Cloud Computing, and Mobile Technologies

Section IV Future Research, Scope, and Case Studies of the

17. Emergency Department, Sustainability, and eHealth: A Proposal to

20. Mining Ubiquitous Data Streams for IoT ..................................................................... 421

21. IoT toward Efficient Analysis of Aging, Cardiometabolic, and

Index ............................................................................................................................................. 457

• Introduction to the Internet of Things: Definition and Basic Foundation

Robin Singh Bhadoria

Ganesh Chandra Deka

Ricardo Armentano works as a distinguished professor at

Robin Singh Bhadoria earned his PhD from Indian Institute

Parag Chatterjee works as a researcher at the UTN–GIBIO

Ganesh Chandra Deka works as the deputy director of train-

Jemal Abawajy Kelechi Anabi

Mimonah Al Qathrady Bidyut K. Bhattacharyya

Alexandre Alapetite Zoran Bojkovic

Dario Bonino Nitesh Funde

Meera Dhabu Thomas Gilbert

Joyeeta Goswami Diogo Ferreira Lima Filho

Rahul Jichkar Sandro Mancinelli

Jacopo Maria Legramante Mohit Mittal

Laura Morciano A. Pravin Renold

Rosdiadee Nordin Luis Romero

Miguel Arjona Ramírez S.A. Srinivasa Moorthy

Iping Supriana S. Vijayakumar

Kridanto Surendro Archana Vimal

Sándor Szénási Marcel Stefan Wagner

Grigorios Tzortzis Janusz Wielki

Zoran Bojkovic, Bojan Bakmaz, and Miodrag Bakmaz

1.2 IoT-Enabling Communications Technologies

1.2.1 Short-Range Communications Technologies

Frequency band (MHz) 868/915/2,400 2,400 900

1.2.2 Cellular Technologies and IoT

1.3 Software-Defined IoT

1.3.1 Benefits of Software-Defined Networking

1.3.2 Software-Defined IoT Architecture and System Design

Smart city Smart transportation eHealth