Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook
Central, http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 IoT Technical Challenges
and Solutions
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
Copyright © 2016. Artech House. All rights reserved.

For a listing of recent titles in the

Artech House Power Engineering Library,
turn to the back of this book.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 IoT Technical Challenges
and Solutions
Arpan Pal
Balamuralidhar Purushothaman
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the U.S. Library of
Congress.

British Library Cataloguing in Publication Data

A catalog record for this book is available from the British Library.

ISBN-13: 978-1-63081-111-2

Cover design by John Gomes

© 2017 Artech House

All rights reserved. Printed and bound in the United States of

America. No part of this book may be reproduced or utilized in
any form or by any means, electronic or mechanical, including
photocopying, recording, or by any information storage and re-
trieval system, without permission in writing from the publisher.
All terms mentioned in this book that are known to be trade-
marks or service marks have been appropriately capitalized.
Artech House cannot attest to the accuracy of this information.
Use of a term in this book should not be regarded as affecting the
validity of any trademark or service mark.

10 9 8 7 6 5 4 3 2 1
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 To the IoT Research and Innovation community
at Tata Consultancy Services
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Contents

Preface   13

1
Internet of Things Today   15

1.1 Introduction: Key Trends   15

1.2 Application Landscape for IoT   19
1.2.1 IoT for Facilities   20
1.2.2 IoT for Products   20
1.2.3 IoT for Consumers   21
1.2.4 IoT for the Supply Chain   22

1.3 Technologies of IoT   23

1.3.1 Sensor Subsystem   24
1.3.2 Local Sensor Networks   25
1.3.3 Gateway Subsystem   25
1.3.4 Cloud Connectivity Networks   26
1.3.5 Cloud Subsystem   26

1.4 IoT Standardization   28

1.5 Challenges and Open Problems   32
1.5.1 Handling the Scale    32
Copyright © 2016. Artech House. All rights reserved.

1.5.2 Security and Privacy   33

1.5.3 Context-Aware Analytics   34
1.5.4 Affordable Implementation and Deployment   34
1.5.5 Ease and Economy of Development   35

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 8 IoT Technical Challenges and Solutions

1.5.6 Realistic Deployments    35

1.6 Conclusions   36
References    36
Selected Bibliography   38

2
Scalability of Networks and Computing   41

2.1 Introduction   41
2.2 Use Cases and Requirements   42
2.2.1 Smart Transportation   43
2.2.2 Smart Environment   43
2.2.3 Smart Energy   44
2.2.4 Smart Water   44
2.2.5 Smart Security and Surveillance   45
2.2.6 Smart Retail and Logistics   45
2.2.7 Smart Manufacturing   46
2.2.8 Smart Farming   46
2.2.9 Smart Home   46
2.2.10 Smart Health   47

2.3 Application Classification Templates   47

2.4 Communication Technologies for IoT   49
2.4.1 Personal/Local Area Network Technologies   50
2.3.2 Technologies for Low-Power Wide Area Networks
(LPWAN)   53
2.3.3 Cellular Technology for IoT   54
2.4.4 Application-Level Protocols    55

2.5 Scalable Network Architectures for IoT   56

2.5.1 Network Topologies   57
2.5.2 IoT Protocol Design Space   58
Copyright © 2016. Artech House. All rights reserved.

2.5.3 Delay-Tolerant Networks   58

2.5.4 Software-Defined Networking (SDN)   60

2.6 Practical Considerations for Scalable IoT System

Implementation   62
2.6.1 Real-Time and Power Considerations for IoT
Applications   62

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Contents 9

2.6.2 Utilizing the Edge Devices for Computing   64

2.6.3 Need for a Platform for Application Development and
Deployment   65

2.7 Conclusions   67
References    68
Selected Bibliography   68

3
Security and Privacy   73

3.1 IoT Security: A Perspective   73

3.1.1 Business Objectives of Security   75

3.2 IoT Security: Key Requirements   75

3.3 IoT Security Challenges   79
3.3.1 Typical Threats on Various IoT Subsystems   80

3.4 Data Protection   81

3.5 Communication Security   83
3.5.1 Cryptographic Key Management   84

3.6 Identities and Identity Management   86

3.7 Authentication   87
3.8 Access Control   88
3.9 Secure Software Updates   89
3.10 Privacy in IoT Systems   90
3.11 System-Level Security Assessment    92
3.11.1 Risk-Based Security   92
3.11.2 Threat Modeling and Risk Estimation   94
Copyright © 2016. Artech House. All rights reserved.

3.12 IoT Security: Practical Guidelines   100

3.13 Summary   103
References    104
Selected Bibliography   105

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 10 IoT Technical Challenges and Solutions

4
Sensor Informatics and Business Insights   109

4.1 Introduction   109

4.2 Sensor Signal Processing   111
4.2.1 Signal Acquisition and Conditioning    111
4.2.2 Signal Representation   114
4.2.3 Feature Extraction and Inference   116

4.3 Semantic Interpretation of Processed

Information   119
4.3.1 Machine Learning   119
4.3.2 Rule Engine   123
4.3.3 Reasoning   124

4.4 Business Insights from Interpreted Knowledge   126

4.4.1 Visual Analytics   126
4.4.2 Modeling and Simulation   127
4.4.3 Optimization and Planning   127

4.5 Data and Algorithm Marketplaces as New Business

Models   128
References   129
Selected Bibliography   132

5
Mobile Sensing    135

5.1 Introduction   135

5.2 Applications and Use Cases for Mobile Sensing   136
5.2.1 Mobile Sensing for Environmental Monitoring    137
Copyright © 2016. Artech House. All rights reserved.

5.2.2 Mobile Sensing for Emergency Response   137

5.2.3 Collaborative Sensing for Urban Transportation   138
5.2.4 Robots in Healthcare   138
5.2.5 Robotic Telesensing and Operation   138
5.2.6 Aerial Robots for Spatial Intelligence   139

5.3 Technologies and Challenges in Mobile Sensing   143

5.3.1 Smartphone-Based Sensing   143

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Contents 11

5.3.2 Robotic Sensor Networks   147

5.3.3 UAV for Aerial Mapping   149

5.4 Economics of Mobile Sensing   152

5.5 Summary   154
References    154
Selected Bibliography   155

6
Democratizing Analytics:
Analytics as a Service   157

6.1 The Need for IoT Analytics    157

6.2 The Need for Analytics as a Service   161
6.3 Analytics as a Service for Developers: Model-Driven
IoT   165
6.4 An Example of a Model-Driven IoT Framework   168
6.4.1 Domain Concern   168
6.4.2 Development and Orchestration Concern   169
6.4.3 Infrastructure Concern   171

6.5 Summary   172

References   173
Selected Bibliography   174

7
The Real Internet of Things and Beyond   177

7.1 Realistic Internet of Things   177

Copyright © 2016. Artech House. All rights reserved.

7.1.1 Key Contributing Factors to Real IoT   178

7.2 Real IoT Is a Network of Trade-Offs   181

7.2.1 Some of the Common Trade-Offs Encountered in IoT Systems
and Applications   182
7.2.2 Safety on the Machine Floor: An Illustrative
Example   184

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 12 IoT Technical Challenges and Solutions

7.3 Drivers for the Next Wave of IoT   186

7.3.1 Resilient IoT Systems   186
7.3.2 Cognitive IoT   187
7.3.3 Impact of 5G as the Next Wave of Communication
Technology in IoT   188

7.4 Concluding Remarks   190

References    191
Selected Bibliography   191

About the Authors   193

Index   195
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Preface

We started working on the Internet of Things (IoT) in 2010. Its ear-

lier evolution as wireless sensor network was started even earlier.
We conceptualized and started building a horizontal IoT platform
for multiple applications. As we struggled through the process of
building the platform, one thing became clear: IoT is not a mono-
lithic technology stack. It is an abstraction for multiple technolo-
gies, some of which had existed long ago. There were three key
trends that were driving adoption of IoT: the cost of sensor devices
were coming down, the cloud technology was becoming mature,
and the capacity of the Internet was getting better. However, as we
started moving beyond pilot deployments, we realized that it is
not a technology play alone. IoT systems must consider business
goals and return-on-investment (RoI) aspects early in the design of
solutions. This prompted us to look at IoT systems aside from its
hype as real IoT systems focusing on a realistic value proposition
to businesses and social applications. This needs an integrated ap-
proach with abstraction of technology components such as com-
munication, computing, storage, mobility, security and privacy,
business value-add via analytics, and economy of mobile sensing
and automation. This book is an attempt to knit together a prag-
matic approach towards real IoT covering all of these.
In Chapter 1, we outline the IoT as it is perceived and seen
today; we discuss about key trends, IoT application landscape, IoT
Copyright © 2016. Artech House. All rights reserved.

technologies, and standardization and introduce the challenges

and open problems. We devote Chapters 2 and 3 into the core tech-
nologies of IoT with regard to communication, computing, stor-
age, security, and privacy. In Chapter 4, we introduce sensor in-
formatics as the main driving force for IoT for creating business

13

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 14 IoT Technical Challenges and Solutions

value-adds. In Chapter 5, we bring in the concept of mobile sens-

ing, where sensors mounted on platforms that are mobile (like mo-
bile phones, robots, and drones), can bring in economies of scale
during deployment. In Chapter 6, we introduce the concept of au-
tomation of the analytics via analytics as a service, which leads to-
wards democratization of analytics resulting in easing the special-
ized skill requirement for IoT. Finally in Chapter 7, we summarize
and outline the real requirements for a realistic IoT deployment
and try to provide a glimpse of technologies to come in future.
This book would not have been possible without the help
of several people. First and foremost, we want to thank Mr. K.
Ananth Krishnan, chief technology officer of TCS who encouraged
us to deep dive into IoT research. We want to thank various sci-
entists and researchers from the Embedded Systems and Robot-
ics Research Area of TCS Research and Innovation, whose work
provided the backbone of the content of this book. We also thank
the editorial staff of Artech House, whose diligent follow-ups and
feedback kept us on our toes and helped us in delivering the book
on time. Last but not the least, we want to thank our children,
wives, and parents without whose support and inspiration the
writing of this book would not have been possible.
We believe that this book would be useful to a wide audience
including practitioners and people who are in general interested
in exploring a pragmatic approach to IoT. This is not a technology
cookbook that is prescriptive in nature, but a suggestive one dis-
cussing practical considerations towards building real IoT.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 1
Internet of Things Today

1.1 Introduction: Key Trends

The Internet of Things (IoT) is defined as [1]: “the network of

physical objects—devices, vehicles, buildings and other items
which are embedded with electronics, software, sensors, and net-
work connectivity, which enables these objects to collect and ex-
change data.” Forbes defined it as [2]: “a giant network of connect-
ed things (which also includes people)—between people-people,
people-things, and things-things.” A thing, in IoT, can be physical
objects like a bridge, a building, or a transport having sensors like
vibration, temperature, and accelerometer, respectively, or human
beings like a person wearing a smart watch or having a biochip
implant; in IoT, all of them can have an IP address through which
the sensor data can be transferred over a network (including the
Internet). The concept of IoT came into being in 1999. However, the
Copyright © 2016. Artech House. All rights reserved.

first reported IoT-like device was built at Carnegie Melon Univer-

sity in the early 1980s; it was a soft drink machine connected to the
Internet allowing availability to be checked online in real time [3].
Why is IoT being regarded as one of the most disruptive tech-
nologies that will drive business digitization? The answer is

15

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 16 IoT Technical Challenges and Solutions

simple. Until now, the use of information technology in most of

the business verticals were to improve and automate the business
support services and not the core physical business process. IoT’s
ability to sense the physical world and digitize the physical world
context will be bring fundamental change and digitization to the
existing physical business processes across all verticals leading to
either significant reduction in operation cost or improvement in re-
sponse time and quality or bringing in real-time customization in
the products or service offerings or helping in increasing security
and ensuring compliance. For end consumers, IoT has the potential
to change the way that products and services are consumed; there
will be much more real-time understanding of consumer needs
and subsequent personalization thereof. This will be discussed in
further detail with examples in Section 1.2.
The market for IoT is clearly heating up. According to Gartner,
there were scheduled to be 6.4 billion connected “things” in 2016
(an increase of 30% from 2015), with a projected deployment of
20.8 billion things by 2020 (5.5 million new things to be connected
every day) [4]. Gartner has also reported that, in 2016, IoT would
be instrumental in U.S. dollar spending in total services of $235 bil-
lion and in total end-point spending of $1.4 trillion (which is an in-
crease of 22% from 2015). By 2020, the total IoT end-point spending
was projected to rise to $3 trillion. The mix of consumer compared
with business applications in this spending is also interesting; in
2015, the consumer versus business application share in the overall
spending was 35% to 65%. The consumer spending was projected
to increase over the years and, by 2020, the consumer versus busi-
ness application share in the overall spending would be 51% to
49%.
In a recent global IoT trend study by Tata Consultancy Services
[5] involving more than 3,500 executives across small and large
corporations all over the world, some interesting facts about IoT
Copyright © 2016. Artech House. All rights reserved.

trends have come to light:

• Nearly 80% companies surveyed have initiated IoT programs.

On average, such companies had a revenue increase of 16%.
• Nearly half of them either track customers or monitor opera-
tions through mobile applications and IoT technologies.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 17

• The spending in IoT seems to be directly proportional to actual

product offering price of the company. Companies whose prod-
ucts’ prices are more than $10 million will spend an average of
$335 million, while those with product prices of $100 or less will
spend an average of only $39 million. This can be attributed to
the fact that the return on investment (ROI) for IoT infrastruc-
ture and systems are more justified when it is used to improve,
optimize, protect, and monitor high-value products.
• There will be varying degree of IoT adoption across the four
major geographies: North America, Europe, Asia-Pacific, and
Latin America. The IoT spending will be more in North Ameri-
can and European companies (with North American companies
spending 0.45% of revenue, and their European counterparts
spending 0.40%). The Asia-Pacific companies are not going
to lag much behind in spending, with IoT-related investment
amounting to 0.34% of revenue, while the Latin American en-
terprises will spend around 0.23% of revenue on IoT.
• Companies who demonstrated largest revenue increases from
IoT initiatives had some common traits:
• They were able to create substantial value for their end cus-
tomers and not just value for themselves.
• They could deliver that value through new business mod-
els focusing on product and service offerings around the IoT
data.
• All of them leveraged the IoT technology to understand
firsthand their product and service performances and usage
patterns with regard to their end customers.

The trends described above are macro business trends being

observed globally in the IoT space. Next, we will explore slightly
deeper details of the application landscape for IoT in different in-
Copyright © 2016. Artech House. All rights reserved.

dustry verticals.
Some of the recent successful deployments of IoT that demon-
strate the above traits include [6]:

1. Energy monitoring and usage optimization across facilities for

large enterprises (final value add: reduction in energy cost);

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 18 IoT Technical Challenges and Solutions

2. Quality monitoring in the food processing chain (final value

add: ensuring same quality and taste for the end user);
3. Predictive analytics of refrigeration systems (final value add:
reduction in overall maintenance cost and increase of operat-
ing life of the machines);
4. Real-time alerting around wellness and activity for elderly
people at home and factory people in hazardous areas (final
value add: quick intervention with the appropriate medical
help).

All of these have one common factor: all of them needed to

demonstrate that the value-add (economic or otherwise) outscores
the cost of infrastructure investments, thereby justifying the ROI.
Almost every imaginable type of company seems to have an IoT
strategy in 2016 [7]. Here are some of the unlikeliest ones to throw
their hat into the IoT ring: the toymaker Mattel with the connected
Barbie doll [8], the gamemaker Atari with smart home products
[9], the nonprofit browser company Mozilla with an operating sys-
tem for connected devices [10], and the credit card company Visa
with integration of payments inside IoT devices [11]. There are a
lot of innovative products and solutions coming out in the mar-
ket in the IoT space. Their sustainability in the marketplace will
depend again on the value-add that they bring to the end custom-
er; however, it is worthwhile to follow these innovations as some
of them are bound to succeed to build a sustainable business [12,
13]. The interesting ones worth mentioning are automated cook-
ing systems, self-charging and wireless-charging platforms, wa-
ter monitoring systems, blindspot-free rear-view mirrors for cars,
three-dimensional (3-D) printing pens, and wearable monitoring
of human physiology. There are also innovative services and busi-
ness models being created around IoT-enabled products; for ex-
Copyright © 2016. Artech House. All rights reserved.

ample, instead of selling a washing machine, the washing machine

vendor can give it away for free and charge the consumer per wash
based on the time and the load.
Another interesting trend is the emergence of an ecosystem for
IoT. The chip makers like Intel, Qualcomm, TI, ARM, and Arduino
are providing the core sensing and processing technology with low
power consumption and universal connectivity that can be embed-
ded into the “Things.” The equipment vendors like Cisco, Huawei,

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 19

Ericsson, and Nokia are trying to create new network equipment

that address the new challenges posed by IoT, while the large tele-
com operators all over the world are trying to create a value-added
services model around machine to machine (M2M) communica-
tions. Finally, the major software vendors like IBM, Google, Micro-
soft, SAP, Accenture, and TCS are focusing on data handling, data
analytics, and system integration aspects of IoT.

1.2 Application Landscape for IoT

A typical IoT stack is outlined in Figure 1.1. The sensors are put
on physical objects and human beings to sense their context. The
meaningful information from the sensor signals is extracted and
sent over the Internet to the cloud using a gateway device. Fur-
ther analysis of the extracted information is done to understand
the physical events and business insights are derived thereof in
Copyright © 2016. Artech House. All rights reserved.

Figure 1.1 Internet of Things stack.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 20 IoT Technical Challenges and Solutions

different business verticals. These business insights trigger the re-

sponse cycle, which may range from a completely offline process
change to completely real-time actions taken to mitigate the effects
of the underlying physical event. This sense, extract, analyze, and
respond cycle presented here is found to be common to all possible
IoT applications. The final benefits to the business and stakehold-
ers in an IoT system will invariably come from either reduced cost
due to improved operational efficiency via IoT-based monitoring
or better insights from analytics of IoT data creating value and
competitive advantage for the stakeholders.
There are four core business areas in different verticals where
IoT can be applied to generate value through sensing and subse-
quent analysis and response: facility, product, customer, and sup-
ply chain [14]. The type of applications can be monitoring, control,
optimization, or autonomous depending upon the use case. These
are depicted pictorially in Figure 1.2.

1.2.1 IoT for Facilities

Most of the industry verticals that have fixed assets in form of build-
ings, campuses, infrastructure and heavy equipment will benefit
from deploying IoT for its facilities. There can be monitoring ap-
plications like perimeter security through video surveillance, con-
trol applications like building energy management via smart meter
data analytics or predictive maintenance for machines, and optimi-
zation applications like emergency evacuation via people sensing
and localization. Such applications will generate business value in
terms improved security or compliance, lower cost of operation,
lower cost of maintenance, and improved lifespan of assets in al-
most all industry verticals including travel and hospitality, retail,
energy and utilities, banking, industrial manufacturing, farming,
healthcare, and life sciences that usually have significant real es-
Copyright © 2016. Artech House. All rights reserved.

tate or office spaces or infrastructure.

1.2.2 IoT for Products

Using IoT for products can take two forms: it can ensure quality
when the product is built, or it can detect finished product perfor-
mance in the field. The former requires putting sensors into the

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 21

Figure 1.2 IoT application space.

product to monitor its quality as it is manufactured and to reject

or segregate bad quality products and can therefore be regarded
as a monitoring application. The latter requries putting sensors on
products deployed in the field, thereby having a monitoring appli-
cation for generating feedback and future product requirements,
a control application for predictive maintenance of such products
to improve lifespan, and an optimization application in the form
of prescriptive analytics to ensure lower cost of operations and
maintenance. All industry verticals that produce and deploy ma-
chinery or packaged goods in the field can benefit from such IoT
deployments including automotive and transportation, telecom-
munications and media, energy and utilities, environment, retail,
and consumer packaged goods, industrial manufacturing, farming
and healthcare, and life sciences.
Copyright © 2016. Artech House. All rights reserved.

1.2.3 IoT for Consumers

Using the sensors present in smart phones and wearable and oth-
er unobtrusive sensors like a camera, it will be possible to sense,
locate, and understand customers, which in turn, will be able to
provide enterprises with the invaluable knowledge about the cus-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 22 IoT Technical Challenges and Solutions

tomers’ likes and dislikes. This knowledge can be used to create

actionable insights around a customer profile, leading towards the
personalization of the product or service offering and resulting in
cost and experience benefits for the end user. Almost all such ap-
plications can be regarded as control or optimization kinds of ap-
plications. While this kind of possibility is there in each and every
industry vertical with human beings as their ultimate consumers,
there are two industries which will have the potential to be dis-
rupted via IoT-based consumer monitoring. First, in-store and on-
line retail for personalized recommendation-based shopping expe-
rience will become possible using various physical, physiological,
and biological sensors on consumers deployed in form of mobile
phone or wearable. Second, healthcare or life sciences industry
can change from an illness-driven industry to a wellness-oriented
industry providing personalized wellness plans, diagnosis, treat-
ment protocols, medicine, and therapy based on each patient’s
health condition sensed through various physical, physiological
and biological sensors on consumers deployed in form of mobile
phone or wearable or implantable.

1.2.4 IoT for the Supply Chain

A supply chain (SC) is a network of supplier, production centers,

storage, distribution centers, sellers, and buyers. Tracking an item
in the supply chain using sensor technology can provide a near-re-
al-time view, which, in turn, can lead to improve of the efficiency of
the supply chain via advanced optimization techniques. Almost all
industry verticals that have tangible products as offerings typically
have supply chains that can be optimized via IoT deployment. In
particular, consumer packaged goods and energy and utilities in-
dustries are heavily dependent on this supply chain and stand to
be disrupted via suitable deployment of sensors and the associated
Copyright © 2016. Artech House. All rights reserved.

analytics and optimization. Automation in supply chain will also

come via technologies like driverless cars that can be regarded as a
mix of IoT, robotics, and artificial intelligence; it stands to disrupt
the automotive and transportation industry.
Table 1.1 details the IoT application landscape described above.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 23

Table 1.1
IoT Application Landscape
Type of Application Potential
Industry Disruption via
Vertical Facility Product Consumer Supply Chain IoT
Automotive Building or Car monitoring Driving Automation High, through
and transpor- infrastructure and behavior via driverless driverless cars
tation surveillance maintenance monitoring cars and drones
Travel and Building Room Customer Optimization Medium, through
hospitality surveillance monitoring and behavior of operations, personalization
maintenance monitoring recommender
systems
Retail and Building or End-product Customer Optimization High, through
CPG infrastructure sensing and behavior of operations, personalization
surveillance monitoring monitoring Recommender
Systems
Energy and Building or Quality Customer Demand Medium, through
utilities infrastructure monitoring behavior response peak load
surveillance and control: influencing optimization management
electricity and and peak load via customer
water management behavior
influencing
Banking, Building or End-product Customer Optimization Medium, through
insurance, infrastructure sensing and behavior via risk personalization
and financial surveillance monitoring monitoring profiling
services
Industrial Building, End-product Customer Optimization High, through
manufacturing infrastructure, quality control feedback of Operations improved
or equipment and defect monitoring efficiency
surveillance inspection
Farming Building, Produce Customer Optimization Medium, through
infrastructure, sensing and feedback of operations improved
equipment monitoring monitoring efficiency
surveillance
Environment Infrastructure Environmental People health Visualization Medium, through
surveillance sensing and and feedback real-time views
monitoring monitoring of pollution map
Healthcare Building, End-product Patient Optimization High, through
and life infrastructure sensing and prognostics of operations personalized
sciences surveillance monitoring and health risk healthcare
profiling
Telecommu- Building, End-product Customer Recommender Medium, through
nication and infrastructure sensing and feedback systems personalized
Copyright © 2016. Artech House. All rights reserved.

media surveillance monitoring monitoring content recom-

mendation

1.3 Technologies of IoT

The technology stack is IoT is described in Figure 1.3. Architec-

turally, typical IoT system is divided into three subsystems: sen-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 24 IoT Technical Challenges and Solutions

Figure 1.3 IoT technology stock.

sor subsystem, gateway subsystem, and cloud subsystem with the

necessary underlying network connectivity between the subsys-
tems. The sensor subsystem is connected to the gateway subsystem
via local sensor networks. The gateway subsystem is connected to
cloud subsystem via a wide area network like the Internet.
Each of these three subsystems connected via two types of net-
working are described in brief next.

1.3.1 Sensor Subsystem

The sensor subsystem uses connected transducers to covert physi-

cal world stimuli into digitized electrical signals. Once digitized,
these signals can be transported to gateway devices for further
processing via wired or wireless local sensor networks. Signal
conditioning of the analog transducer signal followed by analog-
to-digital (A/D) conversion and subsequent digital signal pro-
Copyright © 2016. Artech House. All rights reserved.

cessing of the digitized data is required to produce good-quality

sensor data under noisy environments. Dynamic range of the
sensor transducer (should accommodate extreme cases of physi-
cal world stimuli), sampling frequency for the sensor digitization
(as per Nyquist Sampling theory, this should be more than twice
the maximum useful frequency of the physical world stimuli), and
minimization of energy consumption (this should be as low as pos-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 25

sible to conserve battery and extend sensor life in the field) are a
few of the important technical considerations for sensor subsystem
deployment. Sensor subsystems are typically computer memory
and power-constrained devices, but advances in semiconductor
technology are making sensor devices more and more powerful
yet miniaturized in this aspect. The sensors can sense environmen-
tal properties like temperature and pressure, physical properties
like location, velocity, acceleration, strain, vibration, contact, and
proximity, and physiological/biological properties like heart rate,
blood pressure, electrocardiogram (ECG), and electroencephalo-
gram (EEG). Advances in science of mechanics, electromagnetics,
acoustics, thermodynamics and optics, chemistry, and biology are
also creating increasingly more new transducers making it pos-
sible to sense newer physical events.

1.3.2 Local Sensor Networks

Local sensor networks carry the sensor data from sensors to a gate-
way device for further processing and transport of the data over
the Internet or other public networks to the cloud. They can have
fixed network topologies like star, ring, bus tree, or mesh networks
or they can be formed in an ad hoc manner. Shared media access
protocols using time division multiple access (TDMA), frequency
division multiple access (FDMA) or code division multiple access
(CDMA) technologies are used on top of the physical network con-
nectivity for seamless transportation of the sensor data. Bluetooth
and Zigbee (discussed in detail in Chapter 2) are the most popular
wireless sensor network technologies, while WiFi also can be used
in some scenarios. Depending upon the use case, the sensors can
be interconnected using wired network also or can be connected
point-to-point to the gateway using serial interfaces like universal
serial bus (USB).
Copyright © 2016. Artech House. All rights reserved.

1.3.3 Gateway Subsystem

Gateway subsystems connect to local sensor networks on one side

and public networks like the Internet on the other side. They typi-
cally operate as a router, gateway, or switch bridging the two dif-
ferent types of physical network and protocol stacks. For exam-
ple, the public network is typically Internet Protocol (IP) enabled,

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 26 IoT Technical Challenges and Solutions

whereas in most of the cases the local sensor network is not. One of
the sensor nodes in local sensor network can become the gateway
or there can be dedicated gateway devices. Because typically gate-
way devices can have more memory and computing power and in
many scenarios are electrically powered, it is possible to execute
some of the high sampling rate sensor signal processing and noise
cancellation algorithms in the gateway itself so that clean data at a
reduced rate goes to the cloud.

1.3.4 Cloud Connectivity Networks

The cloud connectivity networks are typically IP networks; in most

of the cases, this will be the Internet for IoT systems. However,
there may be scenarios where private networks and private clouds
are deployed depending upon the use case requirements. Band-
width, latency, reliability, and security of this network are critical
for viable implementation of these systems.

1.3.5 Cloud Subsystem

The cloud subsystem receives the sensor data over IP, stores them,
and allows analytics to be run on the stored data. The elastic nature
of cloud is needed to cater for uneven demand of processing and
storage emanating from fluctuating nature of the sensor data. In
some cases, the data is processed even before storing; such systems
are known as complex event processing (CEP) systems. The stor-
age database needs to handle huge data coming from sensors and
hence needs to be Big Data-enabled; there may be limited number
of huge files (like video surveillance data) or huge number of small
files (coming from a lot of sensors). The processing and analytics
engine is the software service available on the cloud to derive busi-
ness insights from the sensor data, as outlined earlier.
It is clear that the main value delivered by the IoT technolo-
Copyright © 2016. Artech House. All rights reserved.

gy comes from information extraction and analytics of the sensor

data. In Figure 1.4, we present the technology stack for IoT ana-
lytics, which can be distributed across sensors, gateways, and the
cloud. The raw sensor data at the bottom of the knowledge pyra-
mid needs to be processed to create contextual information trying
to answer questions like who did what, where, and when. It boils
down to summarizing or visualizing the sensor data along with

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 27

Figure 1.4 Technology stack for IoT analytics.

identity, location, and timestamp information. This contextual in-

formation can be applied to build knowledge models of the physi-
cal system. The models can either be built on scientific knowledge
linking how the physical event generated the information or can
be statistical models learned from the data or a hybrid of both.
These models help in having a better understanding of the system
answering questions like why the physical event has happened.
These kinds of insights are extremely useful in business operations
helping in business decision-making and value-add. Finally, a col-
lection of such understandings can create the true wisdom, which
can prescribe what should be done to prevent the physical event (if
it is counterproductive to business) or facilitate the physical event
(if it is aiding business outcome) or run the current operation fast-
er, cheaper, or better.
The knowledge pyramid can be explained using a simple ex-
ample. Let us assume that a building is installed with tempera-
Copyright © 2016. Artech House. All rights reserved.

ture and gas sensors. The data from these sensors form the raw
data layer. The visualization and reporting of these sensor data in
a spatiotemporal map can be the contextual information layer. The
knowledge model can be simple, telling that because there is high
abnormal temperature in certain zones, the building at that part
must have caught fire. The understanding layer can do causal anal-
ysis of the temperature sensor data and gas sensor data and infer
that the fire has been caused by a gas leak and hence appropriate

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 28 IoT Technical Challenges and Solutions

fire extinguishers capable of handling gas fire needs to be sent.

The final wisdom layer would be how to design the building infra-
structure to prevent such fires from happening or minimizing the
number of people affected due to such events.
It is quite obvious that as we climb up the pyramid the business
value of the analyzed data becomes more and more useful. How-
ever, the volume of the data to be analyzed continues to reduce as
we move from information to knowledge to insights to wisdom.

1.4 IoT Standardization

It is quite clear from looking at the diversity of the technologies

and vertical domains, that standardization will become one of the
key elements to the success of IoT. All stakeholders of IoT systems
will need to invest substantially into standardization to bring in
interoperability and avoid vendor lock-in. The areas in which stan-
dardization efforts are necessary can be looked upon from differ-
ent perspectives.
In one view, the standardization can happen in IoT domain ap-
plications (including domain-specific physical infrastructures such
as building, road, and traffic), in software services layer [via ap-
plication programming interfaces (APIs) and software as a service
or (SaaS)], the ICT infrastructure (consisting of cloud, Internet and
connectivity), and edge devices (consisting of sensing devices and
embedded systems). This is pictorially depicted in Figure 1.5.
In another view from interoperation perspective, as outlined
by European Telecommunication Standards Institute (ETSI), IoT
standardization can be regarded as a set of intermixing interopera-
tion requirements in technical, syntactic, semantic, and organiza-
tional level [15]. This is represented in Figure 1.6. The technical in-
teroperation requires standardization in basic communication and
Copyright © 2016. Artech House. All rights reserved.

computing systems for IoT (distributed systems, physical commu-

nication systems, and embedded systems). The syntactic interop-
eration tries to standardize how these different computing systems
talk to each other using standard communication protocols. The
semantic interoperation brings in the requirement of interpreting
the IoT data in a common vocabulary, which can be either technol-
ogy-specific or physical infrastructure-specific or vertical domain-
specific or a combination of all. The organizational interoperation

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 29

Figure 1.5 Services and technology-level standardization for IoT.

Figure 1.6 Different levels of interoperability in IoT.

Copyright © 2016. Artech House. All rights reserved.

deals with the domain view to ensure interoperation between

two different systems in the same vertical domain and also with
a system-of-systems view where two different IoT systems from
two different organizations (e.g., transportation and the healthcare
system of a city) can talk to each other in a common language so
that larger, value-added services can be built using multiple or-
ganizational systems (e.g., healthcare systems and transportation

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 30 IoT Technical Challenges and Solutions

systems talking to each other to find out the fastest route for dis-
patching an ambulance).
Many of the notable standardization organizations all over the
world are working in the area of IoT Standardization under the
umbrella of the International Telecommunication Union–Telecom
(ITU-T) [16]. They include TIA (the U.S. Telecommunications In-
dustry Association) [17], ETSI (European Telecommunications
Standards Institute) [18], CCSA (China Communications Standards
Association) [19], TSDSI (Telecommunication Standards Develop-
ment Society, India) [20], ARIB (Association of Radio Industries
and Business, Japan) [21], 3GPP (Third Generation Partnership
Program), and OMA (Open Mobile Alliance) [22]. There are also
independent (or working under other organizations) alliances,
consortia, or standard-making bodies like IPSO Alliance [23], In-
ternet Engineering Task Force (IETF) [24], and OneM2M [25]. All
these bodies are working to create technical standards around
machine-to-machine (M2M) communications in various areas of
physical layer, data link layer, network layer, transport layer, and
application layer protocols and interfaces [26]. Figure 1.7 shows
the association among different M2M standardization organiza-
tions. These standardization efforts mainly take care of technical
and syntactic interoperability outlined in Figure 1.6.
In the semantic interoperation space, quite a bit of work has
been achieved in Open Geo Spatial Consortium (OGC) [27]. The
Sensor Web Enablement (SWE) and Sensor Mark-up Language
(SensorML) from OGC provides rich semantic description of sen-
sor data. These specifications from OGC are being put into stan-
dardization through the World Wide Web Consortium (W3C) [28]
as an Internet Standard for Spatial Data on the Web [29].
In the space of industry-specific interoperability, the Industrial
Internet Consortium (IIC) is doing a lot of work in terms of creat-
ing industry vertical-specific use cases, test beds, and technology
Copyright © 2016. Artech House. All rights reserved.

requirements around energy, healthcare, manufacturing, public

sector, and transportation verticals [30]. There are also some initia-
tives to standardize the smart city space, which can be thought of
as a large organization consisting of multiple industry verticals by
the International Organization for Standardization (ISO) [31]. ISO
has tried to identify the types of stakeholder and types of ICT sys-
tems of a smart city and has proposed an ecosystem-based domain

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 31

Figure 1.7 Relationship between different standardization

bodies in M2M space.

model. Similar work has also been done by the U.K. standardiza-
tion body called British Standards Institution (BSI) [32] in the area
of smart city vocabularies.
It should be noted that although a lot of effort is being spent on
the technical interoperation part, given the complexity and diver-
sity of IoT systems, it is not possible or prudent to define standards
for each and every layer and component; each vertical and use case
will have different requirements. Rather, the focus in an IoT system
should be to adopt multiple existing standards in each layer and
define the Interoperation between those standards at the syntactic,
semantic, and organizational levels. There are quite a few consor-
tia trying to work in this space [33], notable among them being
Copyright © 2016. Artech House. All rights reserved.

Thread (backing from NEST in the connected home space looking

at the radio layer), Open Interconnect Consortium/IoTivity (back-
ing from Intel looking at the radio layer and upper layer protocols)
[34], AllSeen Alliance/AllJoyn (driven by Qualcomm in connected
home looking at upper layer protocols) [35], ITU-T SG 20 (backed
by ITU-T as the umbrella international standard for IoT and smart
cities/communities [36], IEEE P2413 (complete umbrella of IoT
standards from IEEE supporting 350 existing standards and 110

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 32 IoT Technical Challenges and Solutions

new standards) [37], Apple HomeKit (not a standard, but Apple’s

view of connected home) [38], and Google Brillo (Android for IoT)
[39].

1.5 Challenges and Open Problems

There is no disputing the fact that IoT has the potential to disrupt
every business and blur the boundaries between industry verti-
cals. There is no doubt about its potential impact. However, is this
just a hype or are there some real-life deployments in IoT? It is true
that most of the IoT deployments are happening in a pilot scale
and very few have scaled out beyond pilots. We try to explore the
main challenges in IoT that need to be addressed to convert the
hype around IoT to a practical reality. There seems to be five areas
of concern around which the challenges are emerging: (1) gearing
up the ICT infrastructure to the massive scale of IoT sensors and
data, (2) ensuring security of IoT systems and complying to pri-
vacy requirements for IoT data, (3) context-aware analytics of IoT
data leading towards business insights and value-adds, (4) afford-
able implementation and deployment of IoT system to ensure ROI,
and (5) ease of development of IoT analytics systems [40]. This is
depicted pictorially in Figure 1.8 and explained in detail next.

1.5.1 Handling the Scale

With immensely large number of sensors (potentially in billions)

connecting over the Internet, the scalability of the connecting net-
works to transport the sensor data to cloud, the scalability of the
storage systems to store and retrieve the huge volume of data and
the scalability of the computing infrastructure to analyze such huge
volume of data within the required response time of the applica-
tion; all become important considerations. The storage technology
Copyright © 2016. Artech House. All rights reserved.

has scaled quite well; however, the network and computing scal-
ability, unless addressed properly can always become a bottleneck
for practical implementation of IoT systems, especially under real-
time constraints on the response and energy constraints on the sen-
sor device side. The storage, although capable of handling very
large data, mainly tries to handle a limited number of large and

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 33

Figure 1.8 Challenges in IoT deployments.

very large files. In the IoT context, it is more likely to encounter a

huge number of small files; it is expected that the current storage
technology may need some tweaking to cater to this requirement.
The scalability of networks and computing [41, 42] is discussed in
detail in Chapter 2 with some real-world use cases and examples.

1.5.2 Security and Privacy

Security in IoT systems has a different dimension from the impli-

cation perspective. Because we are dealing with physical systems
Copyright © 2016. Artech House. All rights reserved.

consisting of sensors placed on real infrastructure and human be-

ings, a security lapse in the ensuing ICT system can potentially put
the infrastructure or human lives at risk. It is not only important to
make IoT systems as much secure as possible, given the constraints
of power, computing, and memory that typical embedded edge
devices have, but also it is important to do a proper security risk

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 34 IoT Technical Challenges and Solutions

assessment of potential implications on the physical infrastructure

if a security breach happens. However, because IoT systems sense
contextual information, they can contain sensitive personal data
like location and health. The end user or the infrastructure own-
er should be able to determine or have control over with whom
their data can be shared. If appropriate measures are not imple-
mented, rich personal data falling into the wrong hands can have
catastrophic consequences. In Chapter 3, we discuss in depth the
implications of security and privacy in IoT systems and possible
approaches towards solving the issues.

1.5.3 Context-Aware Analytics

It is important to understand that analytics of IoT systems does not

stop at visualizing sensor data in form of charts and graphs in geo-
tagged maps. In line with Figure 1.4, it is also imperative for IoT
analytics to gain deeper knowledge and insights into the physical
system events (how and why). This means building and validating
models backed by physical science of physics, chemistry, or biol-
ogy or learning statistical models backed by data science. Having
a validated model may allow businesses to predict some outcome
from the IoT system and use that contextual knowledge for human-
intervened action and control. Going one step further, if the system
analytics can automatically infer what actions need to be taken to
have the desired effect in the business process or operation, then a
complete autonomous system can be built that can be regarded as
a giant control system. In Chapter 4, we describe different aspects
of the sensor informatics and business insights needed to achieve
the above objectives.

1.5.4 Affordable Implementation and Deployment

A bulk of the cost of IoT deployment goes into the hardware and
Copyright © 2016. Artech House. All rights reserved.

operational cost of deploying sensors in the field. While many sys-

tems will need fixed, dedicated sensor installations and will have
requisite value-add coming from the data collected from them jus-
tifying their ROI, there will be many cases where such ROI cannot
be justified, nor is there a need for putting fixed, dedicated sen-
sor installations. Novel methods like ad hoc sensor deployment
on mobile platforms like robots and drones or crowdsourcing

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 35

physical world data from the sensors of millions of mobile phones

carried by people need to be regarded as possible options; oth-
erwise, many potential use cases of IoT may not see the light of
the day beyond pilots. In Chapter 5, we introduce mobile ad hoc
sensing technologies and discuss how they can lead to economy of
deployment.

1.5.5 Ease and Economy of Development

Just like open application development platforms and applications
stores had revolutionized the mobile phone market, such open
platforms are also needed for successful adoption of IoT. When
more people start writing value-adding applications on the IoT
data, then only adoption of IoT will rise significantly and will solve
real problems plaguing today’s business and society creating the
necessary impact. Today, most of the IoT systems are closed silos,
and that needs to change. We call it “democratization” of IoT ana-
lytics where different applications can provide different analytics
from the IoT data as a service. This, coupled with automation of
some parts of the analytics using artificial intelligence, can lead
to a cost-effective and easy way to create value from IoT sensor
data. In Chapter 6, we introduce this subject and give it a detailed
treatment.

1.5.6 Realistic Deployments

To realistically deploy IoT-based systems in the field, we need to

consider robustness, user-centricity, choice of the right processes
and business model, and use of the right ecosystems. This is elabo-
rated in detail in Chapter 7, which also outlines how a realistic de-
ployment of IoT is always a trade-off between multiple conflicting
factors, notable among them being direct hardware sensing versus
indirect software sensing, application-specific hardware versus
Copyright © 2016. Artech House. All rights reserved.

general-purpose hardware, security versus user experience, bat-

tery lifetime versus performance, and communication range ver-
sus power versus bandwidth. Further resilient and cognitive IoT
systems in conjunction with fifth generation (5G) wireless commu-
nication systems will pave the way for realistic application deploy-
ments in near future.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 36 IoT Technical Challenges and Solutions

1.6 Conclusions

In this chapter, we first introduced the key macro business trends

that are being observed in the IoT space; this includes creating
value for end customers and new business models through better
understanding of end customers’ requirements. Then we outlined
the main business components in the application landscape that
are affected by IoT; these components include facility, product,
consumer, and supply chain. We covered this from the perspec-
tive of different industry verticals and gave a summary outline of
possible application use cases, including the potentially disrupting
ones. Then we introduced different technologies needed for IoT-
based systems including different architectural components span-
ning sensor devices to cloud systems connected in between via lo-
cal sensor networks, gateway devices, and Internet. Following that,
we introduced the need for standardization and interoperation in
IoT space and discussed how different standardization activities
are trying to address the standardization in terms of technical, syn-
tactic, semantic, and organizational interoperability. Finally, we
identified the real-world challenges in terms of scalability, security,
privacy, affordability, context-awareness, and ease of development
that must be addressed to create and deploy a full-scale IoT system
in the real world. Diving deeper into these challenges, identifying
their issues, and trying to provide feasible and reali�stic solution
approaches to them become the objective for the rest of the book.

References
[1] http://www.webopedia.com/TERM/I/internet_of_things.html.
[2] ������������������������������������������������������������������
http://www.forbes.com/sites/jacobmorgan/2014/05/13/simple-explana-
tion-internet-things-that-anyone-can-understand/#14b285316828.
[3] http://internetofthingsagenda.techtarget.com/definition/
Copyright © 2016. Artech House. All rights reserved.

Internet-of-Things-IoT.
[4] http://www.gartner.com/newsroom/id/3165317.
[5] http://sites.tcs.com/internet-of-things/.
[6] http://www.tcs.com/research/Pages/TCS-Connected-Universe-Platform.
aspx.
[7] http://www.ioti.com/iot-trends-and-analysis/10-most-important-iot-
trends-2016.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 37

[8] https://www.engadget.com/2016/02/15/
barbie-hello-dream-house-iot-voice-commands/.
[9] https://techcrunch.com/2016/06/05/atarinet-of-things/.
[10] https://blog.mozilla.org/blog/2015/12/09/firefox-os-pivot-to-connected-
devices/.
[11] http://www.techworld.com/mobile/internet-of-things-will-be-game-
changing-for-visa-says-director-3629227/.
[12] https://powermore.dell.com/technology/the-10-most-innovative-
consumer-based-internet-of-things-companies/.
[13] http://wonderfulengineering.com/30-innovative-products-you-did-not-
know-exist-but-are-too-awesome-to-miss/.
[14] http://sites.tcs.com/internet-of-things/.
[15] http://www.etsi.org/images/files/ETSIWhitePapers/IOP%20
whitepaper%20Edition%203%20final.pdf.
[16] http://www.itu.int/en/ITU-T/Pages/default.aspx.
[17] http://www.tiaonline.org/standards/procedures/manuals/scope.
cfm#TR50.
[18] http://www.etsi.org/deliver/etsi ts/102600 102699/102690/01.01.01 60/ts
102690v010101p.pdf.
[19] http://www.ccsa.org.cn/english/.
[20] http://www.tsdsi.org/.
[21] www.arib.or.jp/english/.
[22] http://www.openmobilealliance.org/.
[23] http://www.ipso-alliance.org/.
[24] www.ietf.org.
[25] http://www.onem2m.org/.
[26] http://www.iot-a.eu/public/public-documents/d3.1.
[27] www.opengeospatial.org/ogc.
[28] https://www.w3.org/.
Copyright © 2016. Artech House. All rights reserved.

[29] https://www.w3.org/2015/spatial/wiki/Main_Page.
[30] http://www.iiconsortium.org/.
[31] http://www.iso.org/iso/home.html.
[32] http://www.bsigroup.com/en-GB/smart-cities/.
[33] http://techbeacon.com/state-iot-standards-stand-big-shakeout.
[34] https://www.iotivity.org/.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 38 IoT Technical Challenges and Solutions

[35] https://allseenalliance.org/.
[36] http://www.itu.int/en/ITU-T/studygroups/2013-2016/20/Pages/
default.aspx.
[37] https://standards.ieee.org/develop/project/2413.html.
[38] https://developer.apple.com/homekit/.
[39] https://developers.google.com/brillo/.
[40] https://www.computer.org/csdl/mags/it/2015/03/mit2015030002.pdf.
[41] http://blogs.cisco.com/digital/4-key-requirements-to-scale-the-internet-
of-things.
[42] http://electronicdesign.com/communications/understanding-how-iot-
systems-scale-and-evolve.

Selected Bibliography
Atzori, L., A. Iera, and G. Morabito, “The Internet of Things: A Survey,” Computer
Networks, Vol. 54, No. 15, 2010, pp. 2787–2805.
Balamurali, P., P. Misra, and A. Pal, “Software Platforms for Internet of Things
and M2M,” Journal of the Indian Institute of Science, A Multidisciplinary Reviews Jour-
nal, Vol. 93, No. 3, July–September 2013.
Bandyopadhyay, D., and J. Sen, “Internet of Things: Applications and Challenges
in Technology and Standardization,” Wireless Personal Communications, Vol. 58,
No. 1, 2011, pp. 49–69.
Bandyopadhyay, S., P. Balamuralidhar, and A. Pal. “Interoperation Among IoT
Standards,” Journal of ICT Standardization, Vol. 1, No. 2, 2013, pp. 253–270.
Bandyopadhyay, S., et al., “Role of Middleware for Internet of Things: A Study,”
International Journal of Computer Science & Engineering Survey (IJCSES), Vol. 2, No.
3, August 2011.
Biswas, A. R., and R. Giaffreda, “IoT and Cloud Convergence: Opportunities and
Challenges,” 2014 IEEE World Forum on in Internet of Things (WF-IoT), 2014, pp.
375–376.
Botts, M., et al., “OGC® Sensor Web Enablement: Overview and High Level Ar-
Copyright © 2016. Artech House. All rights reserved.

chitecture,” in GeoSensor Networks, New York: Springer, 2008, pp. 175–190.

Botts, M., and A. Robin, “OpenGIS Sensor Model Language (SensorML) Imple-
mentation Specification,” OpenGIS Implementation Specification OGC, 2007, p.
07-000.
BSI Standards Publication, Smart Cities Vocabulary, PAS, Vol. 180, 2014.
Chen, H., R. H. L. Chiang, and V. C. Storey, “Business Intelligence and Analytics:
From Big Data to Big Impact,” MIS Quarterly, Vol. 36, No. 4, 2012, pp. 1165–1188.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Internet of Things Today 39

Cheng, H. -C., and W. -W. Liao, “Establishing a Lifelong Learning Environment

Using IOT and Learning Analytics,” 2012 IEEE 14th International Conference on
Advanced Communication Technology (ICACT), 2012, pp. 1178–1183.
Davenport, T. H., and J. G. Harris, Competing on Analytics: The New Science of Win-
ning, Cambridge, MA: Harvard Business Press, 2007.
Demirkan, H., and D. Delen, “Leveraging the Capabilities of Service-Oriented
Decision Support Systems: Putting Analytics and Big Data in Cloud,” Decision
Support Systems, Vol. 55, No. 1, 2013, pp. 412–421.
Ding, G., L. Wang, and Q. Wu, “Big Data Analytics in Future Internet of Things,”
arXiv preprint arXiv:1311.4112, 2013.
Dinh, N. -T., and Y. Kim, “Potential of Information-Centric Wireless Sensor and
Actor Networking,” 2013 IEEE International Conference on Computing, Management
and Telecommunications (ComManTel), 2013, pp. 163–168.
Emery, D., and R. Hilliard, “Every Architecture Description Needs a Framework:
Expressing Architecture Frameworks Using ISO/IEC 42010,” Joint Working IEEE/
IFIP Conference on Software Architecture & European Conference on Software Architec-
ture 2009 (WICSA/ECSA 2009), 2009, pp. 31–40.
Evans, J. R., and C. H. Lindner, “Business Analytics: The Next Frontier for Deci-
sion Sciences,” Decision Line, Vol. 43, No. 2, 2012, pp. 4–6.
Farooqui, K., L. Logrippo, and J. de Meer, “The ISO Reference Model for Open
Distributed Processing: An Introduction,” Computer Networks and ISDN Systems,
Vol. 27, No. 8, 1995, pp. 1215–1229.
Gubbi, J., et al., “Internet of Things (IoT): A Vision, Architectural Elements, and
Future Directions,” Elsevier Journal on Future Generation Computer Systems, Vol. 29,
2013, pp. 1645–1660.
Haas, P. J., et al., “Data Is Dead... Without What-If Models,” PVLDB, Vol. 4, No.
12, 2011, pp. 1486–1489.
ISO/IEC JTC 1, Smart Cities Preliminary Report, 2014.
Kansal, A., et al., “Senseweb: An Infrastructure for Shared Sensing,” IEEE Multi-
media, Vol. 4, 2007, pp. 8–13.
Kozlov, D., J. Veijalainen, and Y. Ali, “Security and Privacy Threats in IoT Archi-
tectures,” Proc. of 7th Intl. Conf. on Body Area Networks, 2012, pp. 256–262.
Lewis, F. L., “Wireless Sensor Networks,” Smart Environments: Technologies, Proto-
Copyright © 2016. Artech House. All rights reserved.

cols, and Applications, 2004, pp. 11–46.

Li, W., J. Bao, and W. Shen, “Collaborative Wireless Sensor Networks: A Survey,”
2011 IEEE International Conference on Systems, Man, and Cybernetics (SMC), 2011,
pp. 2614–2619.
Ma, H. -D., “Internet of Things: Objectives and Scientific Challenges,” Journal of
Computer Science and Technology, Vol. 26, No. 6, 2011, pp. 919–924.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 40 IoT Technical Challenges and Solutions

Medaglia, C. M., and A. Serbanati, “An Overview of Privacy and Security Issues
in the Internet of Things,” in The Internet of Things, New York: Springer, 2010, pp.
389–395.
Ning, H., and Z. Wang, “Future Internet of Things Architecture: Like Mankind
Neural System or Social Organization Framework?” IEEE Communications Letters,
Vol. 15, No. 4, 2011, pp. 461–463.
Ojo, A., E. Curry, and T. Janowski, “Designing Next Generation Smart City Ini-
tiatives Harnessing Findings and Lessons from a Study of Ten Smart City Pro-
grams,” 22nd European Conf. on Information Systems, Tel Aviv, 2014.
Pal, A., “Internet of Things – from Hype to Reality,” IEEE Computer Society IT
Professional Magazine, May 2015.
Pal, A., et al., “System and Method for Identifying and Analyzing Personal Con-
text of a User,” U.S. Patent Application 14/376,536, filed January 22, 2013.
Parker, L. E., et al., “Distributed Heterogeneous Sensing for Outdoor Multi-Robot
Localization, Mapping, and Path Planning,” in Multi-Robot Systems: From Swarms
to Intelligent Automata, New York: Springer, 2002, pp. 21–30.
Smith, S. W., The Scientist and Engineer’s Guide to Digital Signal Processing,��������
San Di-
ego: California Technical Publishing, 1997.
Suo, H., et al., “Security in the Internet of Things: A Review,” 2012 IEEE Interna-
tional Conference on Computer Science and Electronics Engineering (ICCSEE), Vol. 3,
2012, pp. 648–651.
Tata Consultancy Services Ltd., “Internet of Things: The Complete Reimaginative
Force,” TCS Global Trend Study, July 2015.
Wallace, D. P., Knowledge Management: Historical and Cross-Disciplinary Themes,
Westport, CT: Libraries Unlimited, 2007, pp. 1–14.
Weber, R. H., “Internet of Things–New Security and Privacy Challenges,” Com-
puter Law & Security Review, Vol. 26, No. 1, 2010, pp. 23–30.
Wei, R. E. N., “A Study of Security Architecture and Technical Approaches in In-
ternet of Things,” Netinfo Security, Vol. 5, 2012, p. 025.
Xu, T., J. B. Wendt, and M. Potkonjak, “Security of IoT Systems: Design Chal-
lenges and Opportunities,” Proc. of the 2014 IEEE/ACM International Conference on
Computer-Aided Design, 2014, pp. 417–423.
Zhu, Q., et al., “IoT Gateway: Bridging Wireless Sensor Networks into Internet
Copyright © 2016. Artech House. All rights reserved.

of Things,” 2010 IEEE/IFIP 8th International Conference on Embedded and Ubiquitous

Computing (EUC), 2010, pp. 347–352.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 2
Scalability of Networks and Computing

2.1 Introduction

As outlined in Chapter 1, the Internet of Things (IoT) is all about

putting sensors on physical objects and human beings and con-
necting them to monitor, diagnose, or predict physical states and
events. Different studies predict that there will be 40 to 50 billion
IoT devices connected to the Internet by 2020 [1]. However, are the
network and computing infrastructure scalable enough to handle
the deluge of data that will be churned out by these devices? We
try to explore in depth this infrastructure scalability issue in this
chapter.
From a networking perspective, scalability is needed in every
layer of the communication stack. Here scalability means provid-
ing sufficient bandwidth, capacity, and low latency for transport-
ing the sensor data to handle the IoT application requirements. We
Copyright © 2016. Artech House. All rights reserved.

also need to keep in mind the coverage of the network to imple-

ment cost-effective systems. Hence, choice of the right communi-
cation technology, network topology, and network protocol has
paramount importance in this respect.

41

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 42 IoT Technical Challenges and Solutions

However, value of IoT systems is mainly derived from mean-

ingful analytics of the collected sensor data. The volume of data
generated by sensors being enormous leads us to the computing
scalability problem. Depending upon the application type, the re-
quired analytics need to be computed in real time or in near real
time or offline. The computing infrastructure needs to take cog-
nizance of this varying requirement and implement the required
system accordingly.
In Section 2.2, we present an exhaustive set of possible use cases
for applying IoT across various verticals like transportation, envi-
ronment, energy, water, security surveillance, retail, manufacturing,
agriculture, and healthcare. For each of the application use cases,
the requirements for communication bandwidth, network capac-
ity, latency, network coverage, timeliness of analytics, and compu-
tational complexity are tabulated. The analysis of communication
technologies, network architectures, and computing architectures
in subsequent sections draws extensively from these diverse re-
quirements from diverse use cases. In Section 2.3, we outline dif-
ferent communication technologies for IoT in form of personal,
wide area, or cellular networks and application layer protocols. In
Section 2.4, we will cover different network architectures for IoT
that can be built on top of these communication technologies to
provide scalability. These include various network topologies, pro-
tocol design, delay-tolerant networks, and software-defined net-
works. Subsequently, in Section 2.5, we will discuss some practi-
cal considerations for scalable IoT system deployment in form of
real-time and power considerations for sensor data analytics, and
utilizing the edge device for analytics, service-oriented platforms
for IoT application development.

2.2 Use Cases and Requirements

Copyright © 2016. Artech House. All rights reserved.

IoT can give rise to several interesting use cases, which hold the
potential to add value to the end user. At a broad level, IoT use
cases can be verticalized into various domains [2]. These use cases
demand and pose a lot of requirements on the infrastructure of
network, storage, and computing. One needs to understand these
use cases first before understanding the infrastructural require-
ments from these use cases. Hence, we start with a list of example

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 43

IoT application use cases in different verticals that promise to cre-

ate a disrupting value to the end-user experience or business.

2.2.1 Smart Transportation

2.2.1.1 Smart Parking
Sensor-based parking slot monitoring followed by aggregate park-
ing slot monitoring and demand-based parking price modification
can go a long way to ease parking woes of busy city downtowns.

2.2.1.2 Traffic Congestion Management

Sensor-based localization and tracking of vehicles can provide in-
sights into temporal and spatial patterns of traffic, which, in turn,
can be used for better traffic planning. In addition, such sensing
systems can be used in real time for dynamic congestion control
and signaling.

2.2.2 Smart Environment

2.2.2.1 Pollution Control
Air quality monitoring and sound level monitoring can easily be
done using IoT-based sensing systems. Analysis of such data can
produce pollution heat maps, which can either tell people the areas
to avoid or help authorities to take control action.

2.2.2.2 Waste Management

Urban waste collection until now has been a scheduled regular
process; however, production of waste is dynamic in nature and
hence sensing of waste content in the dustbins and optimal sched-
uling of waste collection vehicles accordingly can lead to a much
more efficient system.
Copyright © 2016. Artech House. All rights reserved.

2.2.2.3 Forest Fire Detection

Forest fires can cause havoc to the environment and can affect the
nonurban, suburban, and urban populaces. Timely detection of
forest fires when they are small can make them much easier and
cheaper to control. Sensor-based systems with analytics for early
detection of forest fires can be an impactful application in many
locations.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 44 IoT Technical Challenges and Solutions

2.2.2.4 Natural Hazard Detection and Prediction

In many areas, due to their geographical locations, landslide and
avalanches are potential life hazards. Similarly, earthquakes and
tsunamis are natural hazardous events. Strategic placement of
wireless sensors coupled with analysis of the aggregate sensor
data can provide early warning for landslide, avalanche, earth-
quake, and tsunami, thereby enabling suitable timely evacuation
procedures.

2.2.3 Smart Energy

2.2.3.1 Grid Monitoring and Control
Continuous monitoring and control of electricity grid parameters
can make the grid smart. A smart grid can be considered to be con-
sisting smart control centers, smart transmission networks, and
smart substations. Sensing critical grid parameters, transmitting
them in real time to a central control station, and creating auto-
matic actionable insights from the grid sensor data can be seen as
the IoT-enabled version of the smart grid.

2.2.3.2 Peak Load Management

With the proliferation of smart meters and smart gateways at
home and buildings that can be regarded as the electricity sen-
sor connected over network, it is now possible to have complex
demand-response analysis create suitable policies for peak load
management.

2.2.4 Smart Water

2.2.4.1 Water Quality Monitoring
Water quality monitoring, both for fresh water and ground water,
is an important aspect that is becoming more pertinent in the cur-
Copyright © 2016. Artech House. All rights reserved.

rent context. Analysis of water content via biological and chemical

sensors can provide impactful health benefits for all.

2.2.4.2 Leakage Detection

The water distribution network of any city is one of the most com-
plex ones and, being typically underground, is quite hard to main-
tain. There is significant amount of waste of water that happens
due to leakages in water distribution pipes. Such leakages also re-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 45

sult in higher power consumption as the pumps need more power

to push water through the leaky network. Hence, suitable place-
ment of water pressure sensors all along the water network and
resultant analytics for leakage detection can be very useful.

2.2.5 Smart Security and Surveillance

2.2.5.1 Perimeter Access Control
Video camera-based surveillance is increasingly proliferating
across cities, neighborhoods, and homes as an effective means for
protection, threat detection and access control. Due to the inher-
ently large size of the video data, the biggest challenge is to create
meaningful analytics on video data with automatic detection and
alerting for security events.

2.2.5.2 Hazardous Scenario Detection and Response

Hazardous scenarios like fire in a building need real-time detec-
tion and analysis triggering emergency evacuation if needed. Vid-
eo surveillance augmented by suitable sensors and associated ana-
lytics can be used create such applications involving real-time alert
generation and optimal evacuation path prediction.

2.2.6 Smart Retail and Logistics

2.2.6.1 Supply Chain Optimization
A supply chain is a network of supplier, production centers, stor-
age, distribution centers, sellers, and buyers, and tracking an item
in the supply chain using sensor technology can provide a near-
real-time view, which, in turn, can lead to improve of the efficiency
of the supply chain via advance optimization techniques.

2.2.6.2 Personalized Shopping

Copyright © 2016. Artech House. All rights reserved.

Sensing, location tracking, and understanding customers via the

sensor present in smartphone and wearable items can provide
retail stores with the invaluable knowledge about the customers’
likes and dislikes. This knowledge can be used to create actionable
insights around customer profiles. This can help to create a person-
alized shopping experience for the end user resulting in cost and
experience benefits.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 46 IoT Technical Challenges and Solutions

2.2.7 Smart Manufacturing

2.2.7.1 Defect Detection and Quality Control
Automated defect detection and quality control in a manufacturing
plant or an assembly line are important applications to efficiently
improve product quality. Camera-driven three-dimensional (3-D)
vision systems followed by necessary analytics can provide such
solutions.

2.2.7.2 Predictive Maintenance and Diagnosis for Machines

Sophisticated and costly machines employed in today’s shop
floor have expensive downtime and maintenance issues. Putting
sensors on such machines (both obtrusive and unobtrusive) and
performing modeling, diagnosis, and prognosis on the collected
sensor data can provide the much-needed, just-in-time predictive
maintenance for such systems.

2.2.8 Smart Farming

2.2.8.1 Produce Quality Monitoring
Smart farms of the future have started employing sensors for plant
monitoring ranging from simple cameras to specialized sensors
like spectroscopes and x-rays. Analytics of such sensor data can re-
veal early diagnosis of plant diseases and also can recommend op-
timal farming processes (watering, insecticide, and so forth) tuned
for a particular plant or crop.

2.2.8.2 Smart Control of the Greenhouse Environment

With the advent of advanced networked sensing, a smart green-
house can be regarded as a closed loop control system, which sens-
es various environmental and plant parameters and maintains the
optimal environment suitable for plant growth.
Copyright © 2016. Artech House. All rights reserved.

2.2.9 Smart Home

2.2.9.1 Smart Home Surveillance Systems
Camera and other home sensor-based security and surveillance
systems for home are coming up rapidly. The key to deployment
of such systems is smart cameras, which can do the video analyt-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 47

ics themselves, so that only abnormal events are reported over the
Internet and video is not sent out over the network.

2.2.9.2 Home Energy and Water Use Optimization

With most of the homes getting fitted with smart or digital meters,
it is possible to monitor in real time the energy and water con-
sumption at home. Analytics of such data can provide appliance-
level consumption, peak load management, and, in general, opti-
mization of overall usage and associated cost.

2.2.10 Smart Health

2.2.10.1 Monitoring or Wellness Monitoring at Home and the Workplace for
Elderly People and People with Disease
Due to s significant raise in the number of elderly people and peo-
ple with chronic diseases, it is becoming more important to impart
sensor-based remote healthcare with a patient at home. Medical
devices, wearable devices, and mobile phones can be used to cre-
ate such systems to provide real-time health views and alerts to the
remote doctor.

2.2.10.2 Predictive Analytics and Disease Prognosis

With the advent of connected sensors in the medical domain via
mobile phones, wearable devices, and other devices, it is becoming
possible to obtain 24-7 health and wellness data from patients. Pre-
dictive analytics on such data can provide capability of immediate
disease diagnosis and future disease prognosis. It can also help in
discovering new data-driven diagnostic procedures helping the
healthcare system to move from current illness-driven reactive sys-
tems to more proactive, wellness-driven systems.
Copyright © 2016. Artech House. All rights reserved.

2.3 Application Classification Templates

One interesting observation from the wide gamut of applications

presented above is the variety in terms of sensors, deployment,
and analytics requirements. The applications space of IoT is too
huge and it is useful to abstract them into a set of application cat-
egories or templates for a systematic analysis.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 48 IoT Technical Challenges and Solutions

Broadly, the application requirements can be classified with

regard to type of application (monitoring, control, optimization).
They can be further subdivided into different classes based on the
number of sensors (small or large), data traffic nature (burst or con-
tinuous), data traffic size (large or small), system latency (real time,
near real time, or offline), coverage area (low, medium, or high)
and analytics complexity (low or high). Tables 2.1, 2.2, and 2.3, re-
spectively, try to classify the three types of applications (monitor-
ing, control, and optimization) with their respective subdivision
under the headings of number of sensors, data traffic nature, data
traffic size, system latency, coverage area, and analytics complex-
ity. These headings mainly depict the application attributes that is
useful for the design of a suitable system architecture.
As obvious from Tables 2.1, 2.2, and 2.3, each application use
case, based on its classification, will pose different IoT infrastruc-
tural requirement in terms of networking, storage, and comput-
ing. Understanding these requirements beforehand and creating
the right architectural design for the IoT system is of paramount
importance for a cost-effective, scalable, successful deployment
of IoT-based system; there is no one-size-fits-all solution. Later in
this chapter, we will delve more deeply into the main technological
components of the IoT architecture.

Table 2.1
Requirements Mapping for Monitoring Applications
Number Data Data
Application Use of Traffic Traffic System Analytics Coverage
Case Sensors Nature Size Latency Complexity Area
Smart parking Large Burst Small Real-time Low Medium
Pollution monitoring Small Burst Small Offline Low High
Forest fire detection Medium Continuous Small Near real time Low Medium
Natural hazard detec- Large Continuous Medium Real time High High
tion and prediction
Copyright © 2016. Artech House. All rights reserved.

Water quality Small Burst Small Offline Medium Low

monitoring
Water leakage Medium Continuous Small Real time High Medium
detection
Perimeter access Small Continuous Large Real time High Low
control
Produce quality Small Continuous Large Offline High Medium
monitoring
Smart home Small Continuous Large Near real time High Low
surveillance Systems

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 49

Table 2.2
Requirements Mapping for Control Applications
Number of Data Traffic Data System Analytics Coverage
Application Use Case Sensors Nature Traffic Size Latency Complexity Area
Traffic congestion Large Continuous Medium Real time Low High
management (traffic
control)
Grid monitoring and Medium Continuous Medium Real time Medium Medium
control
Peak load management Large Continuous Small Offline High High
Hazardous scenario Small Continuous Large Real time High Medium
detection and response
Personalized shopping Large Continuous Medium Offline Medium Low
Defect detection and Small Continuous Large Real time High Low
quality control
Predictive maintenance Small Continuous Large Offline Medium Low
and diagnosis for
machines
Smart control of green- Medium Continuous Medium Near real Medium Low
house environment time
Home energy and Small Burst Small Offline Medium Low
water use optimization
Monitoring or wellness Medium Burst Medium Real time Medium Low
monitoring at home
and workplace of
elderly people or
people with disease
Predictive analytics Medium Burst Medium Offline High High
and disease prognosis

Table 2.3
Requirements Mapping for Optimization Applications
Number Data
Application Use of Data Traffic Traffic System Analytics Coverage
Case Sensors Nature Size Latency Complexity Area
Traffic conges- Large Continuous Medium Offline High High
tion management
(planning)
Copyright © 2016. Artech House. All rights reserved.

Waste management Medium Burst Small Near real time Low Medium
Supply chain Large Burst Small Near real time High High
optimization

2.4 Communication Technologies for IoT

There are several communication technologies applicable for IoT.

Selecting a specific technology from the pool of existing communi-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 50 IoT Technical Challenges and Solutions

cation technologies will depend on various factors. It is important

to take into account the application requirements, communication
requirements, architecture, maturity, and acceptance of the stan-
dard. There are a number of popular access-level technologies used
in IoT systems. As shown in Figure 2.1, they may be categorized
majorly into wireless personal or local area networks (WPAN/
WLAN) and wireless wide area networks (WWAN). Salient fea-
tures of major standards in both the categories are presented next.

2.4.1 Personal/Local Area Network Technologies

One of the key challenges of IoT is the last-meter connectivity.

There are several wireless technologies available that can be con-
sidered to suit specific application requirements. A summary of the
major technologies come under this category is given in Table 2.4.

2.4.1.1 IEEE 802.15.4

IEEE 802.15.4 is a physical layer and the media access control lay-
er standard and several secondary communication protocols has
been designed on top of it. It has several features relevant to stan-
dard IoT networks. The basic link in 802.15.4 supports 10-m range
with a data transfer rate of 250 Kbps. Important features include
Copyright © 2016. Artech House. All rights reserved.

Figure 2.1 High-level view of scalable IoT networks.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 51

Table 2.4
Popular Short- and Medium-Range IoT Communication Technologies
Max
Frequency Max. Data Max. Network
Technologies Standards Body Band Range Rate Power Type
Bluetooth (https:// Bluetooth SIG 2.4-GHz 100m 1-3 1W WPAN
www.bluetooth.org/) ISM Mbps
Bluetooth Smart (BLE) IoT Interconnect 2.4-GHz 35m 1 Mbps 10 mW WPAN
(https://www.blue- ISM
tooth.com/what-is-
bluetooth-technology/
bluetooth-technology-
basics/low-energy)
ZigBee (https://www. IEEE 802.15.4, 2.4-GHz 160m 250 100 Star,
zigbee.org) Zigbee alliance ISM Kbps mW mesh
Wi-Fi (http://www. IEEE 802.11 g/n/ 2.4 GHz, 100m 6 to 780 1W Star,
wi-fi.org/) ac/ad (http:// 5 GHz, 60 Mbps, 6 mesh
www.ieee802. GHz Gbps at
org/11/Reports/ 60 GHz
tgah_update.htm)
Zwave (http://www.z- Zwave 908 MHz 30m 100 1 mW Star,
wave.com/) kbps mesh
ANT+ (https://www. ANT Alliance 2.4 GHz 100m 1 Mbps 1 mW Star,
thisisant.com/) mesh
Rubee (http://www. IEEE 1902.1, IEEE 131 kHz 5m 1.2 Kbps 40 to P2P
rubee.com) 1902.2 (long wave 50 nW
magnetic)

real-time suitability by reservation of guaranteed time slots, col-

lision avoidance through carrier sense multiple access/collision
avoidance (CSMA/CA) and integrated support for secure com-
munications. It incorporates low-power management functions.
There are three possible frequency bands of operation (868, 915,
and 2,450 MHz) from which to choose. Several network topologies
are possible. It should be noted that this standard does not define
a network layer; other protocols have been developed to incorpo-
rate higher-layer integration. There are quite a few standard proto-
cols based on IEEE 802.15.4, prominent among them being ZigBee,
Copyright © 2016. Artech House. All rights reserved.

WirelessHART, and 6LoWPAN.

2.4.1.2 Zigbee
The ZigBee standard is maintained by the ZigBee Alliance. ZigBee
was developed for short-range communication in the order of 10m
to 100m based on IEEE 802.15.4. Its main feature is low energy; the
protocol can maintain very long sleep intervals and low operation

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 52 IoT Technical Challenges and Solutions

duty cycles enabling devices to run on batteries for years. ZigBee

is being used extensively in smart energy, home automation, and
lighting control applications, each of which has a specific ZigBee
profile and certification.

2.4.1.3 6LoWPAN
The full expansion of 6LoWPAN is IPv6 over low-power wireless
personal area networks. The standardization effort is driven by
Internet engineering task force (IETF) with a goal to define an ef-
ficient adaptation layer between the 802.15.4 link layer and trans-
mission control protocol over Internet protocol (TCP/IP) stack. It
includes the 802.15.4 link layer, the IP header compression layer
and a TCP/IP stack. A major advantage of 6LoWPAN is that the
devices running on different heterogeneous physical networks can
communicate with each other over the Internet, which is a very
important requirement from an IoT perspective. For connecting to
an IPV4 network, which is most of the deployed Internet today, it
requires a gateway to take care of IPv6-to-IPv4 conversion.

2.4.1.4 WirelessHART
WirelessHART is the wireless flavor of the Highway Addressable
Remote Transducer (HART) protocol. WirelessHART is also based
on the 2.4-GHz IEEE 802.15.4 physical layer. It uses time division
multiple access (TDMA) with time synchronization for achieving
reliable low-power consumption.

2.4.1.5 Thread
Thread uses IEEE 802.15.4 and 6LowPAN and specifies an IP-
based mesh network protocol. Every Thread-certified device gets
an IPv6 address. On the network layer, Thread supports user data-
gram protocol (UDP) on top of 6LowPAN.
Copyright © 2016. Artech House. All rights reserved.

2.4.1.6 Wi-Fi
The highly popular Wi-Fi technology, based on the IEEE 802.11
standard, is also used for IoT applications. Power consumption
can be an issue for low-power IoT devices. Clever power manage-
ment design can save the situation to an extent. The attraction is
that the IoT application can leverage the existing wireless network

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 53

infrastructure. Also, it can work with common off-the shelf Wi-Fi-

enabled devices such as smart phones.

2.4.1.7 Bluetooth
Bluetooth is another successful wireless technology grown with
the popularity of mobile phones. Bluetooth is mainly used today
as a cable replacement for short range communication. It supports
data throughput up to 2 MBps. Bluetooth is generally used in a
point-to-point or in a star network topology. A master Bluetooth
device can communicate with a maximum of seven devices in a
piconet. The technology is fairly low power but high compared to
Zigbee. Bluetooth Low Energy (BLE) is positioned to fill that gap.

2.4.1.8 Bluetooth Low Energy (BLE)

Bluetooth Low Energy (BLE) (also called Bluetooth Smart or Ver-
sion 4.0+ of the Bluetooth specification) is the low power version
of Bluetooth that was built mainly for the IoT. Theoretically, BLE
can support an unlimited number of devices; however, in practice,
the number of simultaneously connected devices can be between
10 and 20.
There are several other proprietary protocols, many of them
having a strong domain focus are in wide usage. Similar to BLE,
the ANT technology is also an ultralow-power protocol typically
used in body and health-monitoring applications. ANT technol-
ogy specifies the radio interface, while ANT+ covers specifications
also for more complex network topologies. ANT+ technology
has already penetrated into the smartphone market. Some smart-
phones have been launched in the market supporting ANT tech-
nology. RuBee (IEEE 1902.1 and 1902.2 standards) modulates the
magnetic component of its carrier wave at around 131 kHz. It has a
short range of about 5m and can penetrate many materials includ-
Copyright © 2016. Artech House. All rights reserved.

ing steel. Key features of some these technologies are summarized

in Table 2.4.

2.4.2 Technologies for Low-Power Wide Area Networks (LPWAN)

The uses for wide-area IoT technology are predominant in civic

infrastructure systems. It includes parking resource management,
traffic control, utilities monitoring, and environmental monitor-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 54 IoT Technical Challenges and Solutions

ing. There are also agricultural use cases like crop monitoring and
livestock movement monitoring that need wide-area coverage. As-
set monitoring and tracking in transportation and cold chain re-
quire ubiquitous coverage that is statewide, nationwide or world-
wide. Another use case for wide-area monitoring is monitoring
of transportation infrastructures such as rail lines and roadways
need. Consumer applications like health and wellness monitor-
ing can also benefit from having an alternative to cell phones for
their wide-area connectivity. Due to power constrained nature of
the sensing devices, this connectivity needs to low power. Major
technologies applicable for LPWAN are summarized in Table 2.5.

2.4.3 Cellular Technology for IoT

Current cellular 2G, 3G, and 4G technologies are also popular for
IoT applications. They are suitable for applications that require
wide area connectivity, mobility, and areas where main power sup-
ply is available or frequent recharging is possible. They support

Table 2.5
Low-Power Wide Area Networking Technologies for IoT
Standards or
Governing Devices per
Technology Body Frequency Range Data Rate Topology Access Point
Weightless — N&P (Sub- 2–5 km 200 bps– Star Unlimited
(http://www. gigahertz (urban) 100 Kbps, W
weightless. ISM), W (TV (1 Kbps–10
org/) whitespace) Mbps)
LoraWAN LoRa 433, 780, 2.5–15 0.3–50 Kbps Star on 1 million or
(https://www. Alliance 868, 915- km star more
lora-alliance. MHz ISM
org/)
SigFox (http:// SigFox Ultranarrow 30–50 km 100 bps Star 1 million
www.sigfox. band (rural),
com/en/) 3–10 km
(urban)
Copyright © 2016. Artech House. All rights reserved.

WiFi LowPower IEEE Subgigahertz 1 km 150 Star, Tree 8,191

P802.11ah license-ex- (outdoor) Kbps–340
empt bands Mbps
Dash7 (http:// Dash7 433, 868, 2 km 9.6, 56, 167 Star, tree Connectionless,
www.dash7- Alliance 915 MHz Kbps can support
alliance.org/) large numbers
LTE-Cat 0 3GPP R-13 Cellular 2.5–5 km 200 kbps star ≥20,000
UMTS(3G), 3GPP Cellular 27 km or 0.73–56 Star Hundreds per
HSDPA/HSUPA 10 km Mbps cell

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 55

the coexistence of many simultaneously connected devices, the

absence of interference, high reliability, and long range, and they
have the capability to service both low-data rate latency-sensitive
and high-data rate applications on the same infrastructure. The
3GPP LTE profile for Machine Type Communications (MTC) [3] is
specifically designed for IoT. LTE MTC has made battery life, cost,
and coverage optimizations for applications that require ubiqui-
tous coverage, high reliability, and robust security of 4G LTE Ad-
vanced, but does not need the high data rate. It is possible for LTE
MTC to coexist with existing 4G LTE mobile broadband services
(as part of 3GPP release 13). LTE MTC has ~1-MHz narrowband
operation and peak data rates less than 2 Mbps. It reduces the cost
and complexity of devices and can improve coverage for machines
deployed in radio-challenged locations like inside-buildings or
underground.

2.4.4 Application-Level Protocols

Although IP connectivity is not necessarily required to ensure

machine-to-machine communication, many standards, in their
specification, include IP to easily connect things to the Internet.
However, this is not enough to ensure interoperability at the re-
quired IoT application level. Because the communication nodes
in IoT are resource-challenged (memory, central processing unit
(CPU), battery), efficient data transmission mechanisms are need-
ed at IP level; constrained application protocol (CoAP), extensible
messaging and presence protocol (XMPP), representational state
transfer (RESTful) hypertext transfer protocol (HTTP), and mes-
sage queue telemetry transport (MQTT) are relevant application-
level protocols.

2.4.4.1 MQTT (Message Queue Telemetry Transport)

Copyright © 2016. Artech House. All rights reserved.

MQTT [4] is a publish-subscribe messaging system based on a

broker model. The protocol has a very small overhead (~2 bytes
per message) and works under lossy and intermittently connected
networks. MQTT was designed to flow over TCP. MQTT is useful
in large networks of small devices that need monitoring and con-
trolled from a back-end server on the Internet. It is not designed for
device-to-device transfer.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 56 IoT Technical Challenges and Solutions

2.4.4.2 CoAP (Constrained Application Protocol)

CoAP [5] is an application-layer protocol championed by IETF. It
is designed to provide a RESTful application protocol modeled on
HTTP semantics, but with a much smaller footprint. Requests and
responses are exchanged asynchronously using CoAP messages
over User Datagram Protocol (UDP). CoAP binary encodes all the
headers, methods and status codes, thereby reducing the protocol
overhead.

2.4.4.3 XMPP (Extensible Messaging and Presence Protocol)

XMPP [6] is an open XML-based protocol for near-real-time mes-
saging, presence and request-response services using TCP as trans-
port. Using its store and push mechanism, it can store contents
even if the receiving entity is in sleep mode or offline. However,
it needs a persistent TCP connection and lacks an efficient binary
encoding. For these reasons, it has typically not been practical for
low-power and lossy networks, but the recent improvements in
XMPP in form of XEP-0322, XEP-323, and XEP-324 can make it
more suitable for IoT.

2.5 Scalable Network Architectures for IoT

With the multitude of potential applications for IoT, there is a need

for specific architecture to support such diverse uses with such a
large number of sensors. Advanced IT systems successfully used
the client-server model and now cloud services are driving net-
worked service architectures. For IoT applications, a client-server
approach may not be the best model. Referring to Figure 2.1, a
typical IoT deployment would need a set of sensors connected to
the Internet cloud via some gateway device. When a group of sen-
sors may be required to make some decisions locally and trigger
Copyright © 2016. Artech House. All rights reserved.

an instant action, computing at a remote server may not meet the

required response time and computing at the gateway device may
make more sense. However, a fully distributed model may also
have issues, as the cost and technical constraints that the sensors
have will not allow heavy computing and storage resources to be
integrated into the sensors.
For practical purposes, a better solution may lie between the
two extremes, based on a specific IoT application. For example,

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 57

consider a large agriculture sensor network. In a typical large ap-

plication, thousands of sensors can be connected to the cloud. Pro-
viding some computing and storage resources at the IoT WAN
gateway level can enable the system to do some in-network com-
puting to reduce the communication bandwidth requirements. We
can leverage modern WAN orchestration technologies to support
the networking needs of the IoT application.

2.5.1 Network Topologies

There are majorly three network topologies utilized in the IoT:

point-to-point, star, and mesh networking. In a hybrid approach, a
mix of these topologies is also used.

2.5.1.1 Point-to-Point Network

This is a direct connection between two network nodes. A cell
phone connecting to a medical device using Bluetooth is an ex-
ample. The obvious limitation of this topology is its scalability and
coverage. It can support only two nodes and span the communica-
tion range offered by the underlying communication technology.

2.5.1.2 Star Network

Here one central node acts as a common connection point for many
peripheral nodes. The peripheral nodes cannot talk to each other
directly, but can communicate through the central node. This cen-
tral node also takes the role as a gateway to external networks.
Star topology has a few important advantages. The network
performance in terms latency and throughput is consistent and
predictable. Fault detection and isolation in such networks are
easy and thus provide high overall network reliability.
There are also some disadvantages. The central node is a single
point of failure. The scale of the network depends on the capacity
Copyright © 2016. Artech House. All rights reserved.

of the central node. Further, the range is limited to the transmission

range of a single device.

2.5.1.3 Mesh Network

In a mesh network, the nodes can have multiple connectivity paths.
It is a mix of simple nodes and router nodes. Router nodes capture
and disseminate their own data. They also serve as relays for oth-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 58 IoT Technical Challenges and Solutions

er nodes in collaboration with neighbors. One or more nodes in a

mesh network will have gateway roles for providing connectivity
to external world.
As the data path can be in multiple hops, the network range
is not limited to the transmission range of a single node. A mesh
network is scalable and resilient to individual node failures, but at
larger sizes a mesh network becomes too complex to manage. The
network latency can increase.
In practice, an IoT network can be a combination of all these to-
pologies. The majority of sizable applications follow a cluster-tree
or cluster-star topology; clusters are a group of localized IoT devic-
es networked together and controlled by a head node or gateway.
The clusters by themselves may have a start or mesh topology.
These clusters are integrated using a backhaul network or Internet.

2.5.2 IoT Protocol Design Space

Many of the communication technologies and protocols options

from the lists presented earlier can be used to build a scalable IoT
network system. An example depiction of design space for proto-
col stack architectural options using some of the popular protocols
are presented in Figure 2.2. A gateway (GW) element is introduced
to connect non-IP networks to IP networks. Any path in that graph
could be selected to build a workable protocol stack. The choice
will also depend on the application requirements and the underly-
ing communication technology capabilities corresponding to the
nodes selected.

2.5.3 Delay-Tolerant Networks

Many communication protocols require connectivity to be main-

tained for the network communication to happen. Communicating
over intermittently connected networks is a challenge. Some IoT
Copyright © 2016. Artech House. All rights reserved.

applications using mobile nodes face this problems. When there

are many interruptions in connectivity, the source device will need
to keep retrying for a successful retransmission. Similar to existing
store and forward technologies used in current sensor networks,
one can allow nodes on the route to buffer packets and send them
on when connectivity to the next hop has been restored.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 59

Figure 2.2 Design space for IoT networking protocols: example.

Store and forward are often used in sensor networks to allow

packets to travel through the network one hop at a time. Delay-tol-
erant networking (DTN) aims to improve the connectivity between
devices by providing a consistent methodology for managing de-
lays across multiple links. Further DTN can improve the network
efficiency and reliability of communications.
Most of the current work on implementing DTN uses the Bun-
dle Protocol (BP) for the deep-space network use case. The Bundle
Protocol implements an overlay network on top of existing net-
works encapsulating the data within it. This makes it possible to
transmit packets over multiple network types and allows it to use
the mixture of different network stacks and the Internet. However,
one overhead of this is a larger packet header size, which is need-
ed to store information in the encapsulated protocol and networks
Copyright © 2016. Artech House. All rights reserved.

own headers.
BP is a message-based overlay that follows the store, carry,
and forward principle. The BP defines the format of the messages,
called bundles, and the logic layout to process them. Bundles have
a lifetime and will be deleted if the lifetime expires. BP uses a cache
to store the bundles. These bundles are either processed by the
node, if it is the destination or forwarded to other nodes toward

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 60 IoT Technical Challenges and Solutions

the bundle destination. An IoT application can consider using this

protocol in case it has an intermittent connectivity scenario.

2.5.4 Software-Defined Networking (SDN)

In SDN, the upper layers of the networking hardware are replaced

with software. More specifically, SDN separates the data and
control planes of the network; the usually computer-heavy data
plane can still remain in hardware, but the control plane can be
completely software-driven, thereby allowing the network to re-
configure itself in run time. Applications can now see the network
as a virtual entity. Figure 2.3 depicts the SDN architecture. To the
applications and policy engines, the network appears as a single,
logical switch accessible through a set of application program-
ming interfaces (APIs). These APIs implement common network
services such as routing, multicasting, security and access control,
bandwidth management, traffic engineering, quality of service,
processor and storage optimization, energy usage, and all forms
of policy management. The Open Networking Foundation (ONF)
[7] drives the standardization and promotion of open APIs. The
Copyright © 2016. Artech House. All rights reserved.

Figure 2.3 SDN architecture.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 61

open APIs between the SDN control and applications layers make
it possible for business applications to operate on an abstraction of
the network without being tied to the details of the network im-
plementation. SDN makes the networks application-customized
and applications network-capability-aware. As a result, comput-
ing, storage, and network resources can be optimized and both
computer-aware and storage-aware networks can be implemented
easily on top of SDN.
OpenFlow [8] is the first standard communications interface
directed towards an SDN architecture. OpenFlow allows direct ac-
cess to the forwarding plane of network devices such as switches
and routers (depicted as network hardware abstraction in Figure
2.3); it exposes an open interface (depicted as network control inter-
faces in Figure 2.3) for the formation of the required control plane.
Network function virtualization (NFV) is a related technology that
provides a software virtualization layer on top of the variety of
network equipment and devices providing a common interface.
The combination of SDN and NFV can provide a lot of technologi-
cal boost towards the required adaptive network scalability for IoT
systems. These kinds of adaptive architectures would be particu-
larly useful for application scenarios that have widely fluctuating
network traffic. For example, a citywide video surveillance sys-
tem that sends raw video only on detection of anomalous events
can generate very low data traffic when there are no surveillance
events and huge data traffic when some events occur.
With the potential huge scale of network for IoT, current net-
work systems may find it difficult to handle the capacity demands
and real-time needs without compromising on reliability. Hence,
there is need to create adaptive networks on top of existing net-
working technology via addition of in-network storage in network
routers to handle network fluctuations. They need a customized
link-state routing protocol to enable routers to temporarily store
Copyright © 2016. Artech House. All rights reserved.

and/or replicate data in response to detected network problems

and forward them later as per application requirements. This
works rather well for delay-tolerant applications that focus more
on reliability than real timeliness. For IoT, applications handling
critical data can benefit from this. Examples include smart health-
care and smart energy with offline analytics requirement.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 62 IoT Technical Challenges and Solutions

2.6 Practical Considerations for Scalable IoT System Implementation

As depicted in Figure 2.1, a typical IoT deployment architecture

consists of a set of sensors connected via a gateway device to the
cloud. The protocols for connecting sensors to gateway and gate-
way to cloud are shown in Figure 2.3. Application services are de-
ployed both in edge and cloud on top of this sensor data transport
layer. This architecture is outlined in Figure 2.4; the subsequent
discussions generally refer to this architecture. There are several
implementation practices that can affect an IoT system. Their re-
quirements stem from the need for adaptive and configurable
networks, real-time and power considerations for sensor data ana-
lytics, utilizing the edge devices for analytics, and using a service-
oriented platform for application development and deployment.
These are explored in detail next.

2.6.1 Real-Time and Power Considerations for IoT Applications

As outlined in Table 2.1, many of the applications in IoT need ana-

lytics insights to be computed and presented in real time. A typi-
cal data flow of an IoT system involves sensing the physical phe-
nomenon, extracting information from sensed data, analysing the
extracted information to derive insights and undertake suitable
response based on those insights. The insight and response can
Copyright © 2016. Artech House. All rights reserved.

Figure 2.4 IoT deployment architecture.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 63

range from simple descriptive visualization to diagnostic, prog-

nostic, and prescriptive systems.
IoT systems are typically constrained by low configuration sen-
sor data acquisition and gateway devices at the edge of the net-
work with potentially infinite compute capability available in the
cloud. There are some conflicting requirements that typically char-
acterize such systems:

• For the ideal real-time system, all processing of sensor data can
be done at the edge device (sensor aggregators and gateways)
and suitable decision or responsive action taken from there it-
self. However, due to limited computational power and memo-
ry of the edge device, all the required analytics cannot be done
at the edge and hence need to be offloaded to the cloud.
• The less computation is done at the edge device, the more data
needs to be sent to the cloud, which results in higher conges-
tion, higher data transfer cost, and increased system latency,
which, in turn, affects the real-time performance.
• Edge devices needing to send more data to the cloud over
wireless connectivity typically end up consuming more power
resulting in precious battery drain for battery-powered edge
gateways, required in many IoT systems due to deployment
considerations.

A good IoT system design needs to take care of all these con-
flicting requirements. Needless to say, such design is completely
specific to the application use case under consideration. Referring
to Table 2.1, we can suggest the following thumb rules for IoT sys-
tem design consideration:

• If there are real-time requirement for analytics insights, try to

Copyright © 2016. Artech House. All rights reserved.

do as much computation as possible in the edge device. Only

computations requiring aggregate sensor information from
multiple systems should be done on cloud. The partitioning of
the computation between edge and cloud can be formulated
as an optimization problem that tries to balance data flow be-
tween edge and cloud without draining too much battery on
the edge device.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 64 IoT Technical Challenges and Solutions

• For electrically powered edge device with high-bandwidth,

low-cost network, a lot of computation can be offloaded to the
cloud provided that the real-time requirements are not strin-
gent enough compared to the available network latency.
• For small systems with limited data and a battery-powered
gateway, it is advisable to do as much computation as possible
in the edge and send only the insights and results to the cloud.
• For systems not having real-time requirements but having a
high load of computational complexity in analytics, it is better
to do as much computation as possible in the cloud.

Many of these recommendations point towards doing analytics

computation at the edge device. This is discussed in detail next.

2.6.2 Utilizing the Edge Devices for Computing

The routers or switches of the network usually have decent com-

puter power; hence, they can perform customized computations
on the messages flowing through them. This is extremely use-
ful in filtering the data as it moves through the network passing
only the meaningful data packets, thereby reducing the overall
network load. In other words, the analytics intelligence can now
be performed at the network nodes instead of the cloud. This is
particularly useful in reducing network congestion and proving
real-time support. Examples include city traffic control, hazard or
abnormality detection and response, water leakage detection, and
electricity grid monitoring. To implement such systems, there is a
need to first create a basic networking platform that can be soft-
ware-controlled, as both of these will demand adaptation from the
network layer. SDN can provide that basic adaptation framework.
Utilizing the edge device for computing can be an elegant way to
Copyright © 2016. Artech House. All rights reserved.

address both the network and compute scalability problem of IoT

systems. This new paradigm of computing is also referred to as fog
computing. In fog computing, not only are the sensor aggregators
and gateways utilized for computing, but also the switches on the
network are brought into the distributed computing architecture.
It operates on the principle of trying to do as much computation as
possible near the source of the sensor data.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 65

To implement fog computing systems, it is imperative to cre-

ate a virtualized distributed computing infrastructure across the
edge devices and cloud. Each device in the network can run a soft-
ware agent capable of dynamically downloading and executing
analytics code. However there are quite a few challenges in creat-
ing a practically deployable fog computing framework. They stem
mainly from the huge heterogeneity of the edge devices in terms
of computational capability, memory, and battery power and also
from the unpredictable availability of such devices.

2.6.3 Need for a Platform for Application Development and Deployment

The traditional way to develop IoT applications is to build it from

the bottom up for the given vertical application use case. It starts
from sensor integration and moves to sensor networking, sensor
data collection, and sensor data storage finally ending up in sensor
data analytics or visualization, but this kind of bottom-up devel-
opment system may not be a scalable and sustainable for a system
as diverse and as large as typical IoT systems; a desirable approach
is to create horizontal service-oriented platforms. Standardization
organizations like ETSI and OneM2M have already published the
reference architecture of IoT along with desired components of the
platform [9]. The basic features of these reference architectures are
quite similar to the one depicted in Figure 2.4. In Figure 2.5, we
summarize and present the required features and value additions
from such a horizontal IoT platform.
Such a horizontal development platform can help in reduc-
ing development cost and time via software reuse and provide a
cost-effective yet scalable way to crowdsource application devel-
opment. Such platforms can expose application programming in-
terfaces (APIs) for application developers to quickly build applica-
tions relevant to a given use case.
Copyright © 2016. Artech House. All rights reserved.

Such platforms can provide the basis for creating massive IoT
data exchanges connecting sensor providers to data consumers via
application stores. It not only allows IoT data to be shared among
applications, but it also allows it to be shared among different ver-
tical systems creating real intelligent system of systems. Such con-
cepts are extremely useful in designing complex interdependent
systems like smart cities, which rely on intelligent interoperation
between multiple verticals like governance, energy, transportation,

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 66 IoT Technical Challenges and Solutions

Figure 2.5 Horizontal IoT platform features.

and healthcare. Open data encourages transparency and it can gen-

.

erate new products and services, and can promote efficiency due
to better visibility. A horizontal platform for IoT can really provide
the fulcrum for creating such open data systems. Additionally, the
variety of sensor devices, their data syntax and semantics, and
the variety of the communication protocols that they use for con-
Copyright © 2016. Artech House. All rights reserved.

necting to the cloud via the Internet demands platforms that can
provide all possible combinations covering sensor device manage-
ment, sensor data management, sensor data transport, and analyt-
ics services. Providing these implementations in form of a set of
libraries and APIs in the platform can provide the much needed
abstraction layer insulating the application developers from the
tiny details of the underlying IoT sensor data network.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 67

There are quite a few horizontal end-to-end service integration

platforms being built [10, 11]. Such platforms will provide the nec-
essary technology abstraction needed for democratizing the appli-
cation development for IoT by making it rich, cost-effective, and
value-adding.

2.7 Conclusions

The power of IoT is its capability to build a scalable distributed

and efficient communication and computing for the things within
the system. The wide range of application use cases are catego-
rized into three sections. They are monitoring, control, and optimi-
zation. Surveillance applications typically require near-real-time
communications. Computing requirements might include analyt-
ics for event detection, forensic analysis, and insights generation.
This may be handled by the resources in a cloud. Control applica-
tions are likely to have more stringent real-time requirements and
high quality of service (QoS). To achieve the required robustness
and real-time response for control action, some computing might
have to be done at the edge gateway level. A suitable computing or
communication partition may be required. For the third category
of study or planning applications, the system may not have real-
time performance targets. Here data is collected systematically
and communicated periodically. Instead of streaming the data to
the cloud, it may be collected and transmitted as blobs or bundles;
here, edge processing may not be a major requirement.
Following this line of thought, a methodology to arrive at scal-
able network architecture for a specific application or application
category is possible. The first step is to gather all application re-
quirements or characteristics such as application category, number
of nodes (scale), spatial distribution, data traffic pattern, computa-
Copyright © 2016. Artech House. All rights reserved.

tion complexity of data processing, and constraints on devices de-

ployable. These requirement attributes are used to pick a suitable
protocol or technology architecture from the protocol design space
codified as a graph structure in Figure 2.2. This can help an IoT
system designer to get a set of initial solutions to work towards
an optimal architecture. From an overall IoT system perspective,
including the computing aspects the architectural guidelines dis-
cussed in Section 2.4 would be highly useful.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 68 IoT Technical Challenges and Solutions

References
[1] http://www.cisco.com/web/about/ac50/ac207/crc_new/university/
RFP/rfp13078.html.
[2] http://www.libelium.com/top_50_iot_sensor_applications_ranking/.
[3] http://www.3gpp.org/news-events/3gpp-news/1714-lc_mtc.
[4] http://mqtt.org/.
[5] http://coap.technology/.
[6] http://xmpp.org/.
[7] https://www.opennetworking.org/about/onf-overview.
[8] https://www.opennetworking.org/sdn-resources/openflow.
[9] http://www.etsi.org/plugtests/COAP2/Presentations/03_ETSI_M2M_
oneM2M.pdf.
[10] Balamurali P, P. Misra, and A. Pal, “Software Platforms for Internet of Things
and M2M,” Journal of the Indian Institute of Science, A Multidisciplinary Reviews
Journal, Vol. 93, No. 3, July–September 2013.
[11] http://www.tcs.com/about/research/Pages/TCS-Connected-Universe-
Platform.aspx.

Selected Bibliography
Bartram, J., and R. Ballance, (eds.), Water Quality Monitoring: A Practical Guide
to the Design and Implementation of Freshwater Quality Studies and Monitoring Pro-
grams, Boca Raton, FL: CRC Press, 1996.
Bonomi, F., et al., “Fog Computing and Its Role in the Internet of Things,” ACM
Proc. of 1st Edition of the MCC Workshop on Mobile Cloud Computing, August 17,
2012, pp. 13–16.
Decker, C., et al., “Challenges of Using Edge Devices in IoT Computation Grids,”
2013 Intl. Conf. on Parallel and Distributed Systems (ICPADS), December 15, 2013,
pp. 564–569.
Ekbatani, M. K., et al., “Congestion Control in Urban Networks Via Feedback Gat-
Copyright © 2016. Artech House. All rights reserved.

ing,” Procedia-Social and Behavioral Sciences, Vol. 48, 2012, pp. 1599–1610.
Evans, J. R., and C. H. Lindner, “Business Analytics: The Next Frontier for Deci-
sion Sciences,” Decision Line, Vol. 43, No. 2, 2012, pp. 4–6.
Ghose, A., et al., “UbiHeld: Ubiquitous Healthcare Monitoring System for Elderly
and Chronic Patients,” Proc. of the 2013 ACM Conf. on Pervasive and Ubiquitous
Computing, Adjunct Publication, 2013, pp. 1255–1264.
Gonzalez-Miranda, S., et al., “An IoT-Leveraged Information System for Future
Shopping Environments,” IT Convergence Practice, Vol. 1, No. 3, 2013, pp. 49–65.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 69

Graham, S., and D. Wood, “Digitizing Surveillance: Categorization, Space, In-

equality,” Critical Social Policy, Vol. 23, No. 2, 2003, pp. 227–248.
Gubbi, J., et al., “Internet of Things (IoT): A Vision, Architectural Elements, and
Future Directions,” Elsevier Journal on Future Generation Computer Systems, Vol. 29,
2013, pp. 1645–1660.
Gulliver, J., et al., “Development of an Open-Source Road Traffic Noise Model for
Exposure Assessment,” Environmental Modelling & Software, 2015.
Haas, P. J., et al., “Data Is Dead... Without What-If Models,” PVLDB, Vol. 4, No.
12, 2011, pp. 1486–1489.
Hannan, M. A., et al., “A Review on Technologies and Their Usage in Solid Waste
Monitoring and Management Systems: Issues and Challenges,” Waste Manage-
ment, Vol. 43, 2015, pp. 509–523.
Hawilo, H., et al., “NFV: State of the Art, Challenges, and Implementation in Next
Generation Mobile Networks,” IEEE Network, Vol. 28, No. 6, November 2014, pp.
18–26.
Hengstler, S., et al., “MeshEye: A Hybrid-Resolution Smart Camera Mote for Ap-
plications in Distributed Intelligent Surveillance,” ACM Proc. of 6th Intl. Conf. on
Information Processing in Sensor Networks, 2007, pp. 360–369.
Herzig, D., “Cost-Benefit Model for Smart Items in the Supply Chain,” in The
Internet of Things, Berlin, Germany: Springer, 2008, pp. 155–172.
Holnicki-Szulc, J., P. Kolakowski, and N. Nasher, “Leakage Detection in Water
Networks,” Journal of Intelligent Material Systems and Structures, Vol. 16, No. 3,
2005, pp. 207–219.
Huang, Q., et al., “Intelligent Building Hazard Detection Using Wireless Sensor
Network and Machine Learning Techniques,” Proc. Comput. Civil Eng., 2012, pp.
485–492.
Kuhlbusch, T. A. J., et al., “New Directions: The Future of European Urban Air
Quality Monitoring,” Atmospheric Environment, Vol. 87, 2014, pp. 258–260.
Latvakoski, J., et al., “Towards Horizontal Architecture for Autonomic M2M Serv-
ice Networks,” Future Internet, Vol. 6, No. 2, 2014, p. 261.
Li, D., et al., “From Digital Earth to Smart Earth,” Chinese Science Bulletin, Vol. 59,
No. 8, 2014, pp. 722–733.
Li, F., et al., “Smart Transmission Grid: Vision and Framework,” IEEE Trans. on
Copyright © 2016. Artech House. All rights reserved.

Smart Grid, Vol. 1, No. 2, 2010, pp. 168–177.

Liu, Y., et al., “Forest Fire Detection Using Artificial Neural Network Algorithm
Implemented in Wireless Sensor Networks,” ZTE Communications, Vol, 12, 2015.
LTE-M, “Optimizing LTE for Internet of Things,” White paper, Nokia Networks,
http://networks.nokia.com/sites/default/files/document/nokia_lte-m_-_opti-
mizing_lte_for_the_internet_of_things_white_paper.pdf.
Mahmood, A., et al., “A New Scheme for Demand Side Management in Future
Smart Grid Networks,” Procedia Computer Science, Vol. 32, 2014, pp. 477–484.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 70 IoT Technical Challenges and Solutions

Maiti, S., et al., “Historical Data Based Real Time Prediction of Vehicle Arrival
Time,” 17th Intl. IEEE Conf. on Intelligent Transportation Systems (ITSC14), 2014.
Auzias, M., Y. Mahéo, and F. Raimbault, “CoAP over BP for a Delay-Tolerant In-
ternet of Things,” 3rd Intl. Conf. on Future Internet of Things and Cloud (FiCloud),
2015.
Michałkiewicz, A., M. Kujawińska, and K. Stasiewicz. “Digital Holographic Cam-
eras and Data Processing for Remote Monitoring and Measurements of Mechani-
cal Parts,” Opto-Electronics Review, Vol. 16, No. 1, 2008, pp. 68–75.
Mukherjee, A., A. Pal, and P. Misra. “Data Analytics in Ubiquitous Sensor-Based
Health Information Systems,” 2012 IEEE 6th Intl. Conf. on Next Generation Mobile
Applications, Services and Technologies (NGMAST), 2012, pp. 193–198.
Mukherjee, A., et al., “ANGELS for Distributed Analytics in IoT,” 2014 IEEE World
Forum on Internet of Things (WF-IoT), March 6, 2014, pp. 565–570.
Ojo, A., E. Curry, and T. Janowski, “Designing Next Generation Smart City Ini-
tiatives: Harnessing Findings and Lessons from a Study of Ten Smart City Pro-
grams,” 22nd European Conf. on Information Systems, Tel Aviv, 2014.
Omnes, N., et al., “A Programmable and Virtualized Network & IT Infrastructure
for the Internet of Things: How Can NFV & SDN Help for Facing The Upcoming
Challenges?” 2015 IEEE 18th Intl. Conf. on Intelligence in Next Generation Networks
(ICIN), February 2015, pp. 64–69.
Pal, A., “Internet of Things: From Hype to Reality,” IEEE Computer Society IT Pro-
fessional Magazine, May 2015.
Pal, A., et al., “Energy Information Gateway for Home,” 2011 IEEE 2nd Intl. Conf.
on Intelligent Systems, Modelling and Simulation (ISMS), 2011, pp. 235–240.
Polycarpou, E., L. Lambrinos, and E. Protopapadakis, “Smart Parking Solutions
for Urban Areas,” 2013 IEEE 14th Intl. Symp. and Workshops on a World of Wireless,
Mobile and Multimedia Networks (WoWMoM), 2013, pp. 1–6.
Ruiz-Altisent, M., et al., “Sensors for Product Characterization and Quality of
Specialty Crops—A Review,” Computers and Electronics in Agriculture, Vol. 74, No.
2, 2010, pp. 176–194.
Somani, N., et al., “Storage Aware Routing Protocol for Robust and Efficient Serv-
ices in the Future Mobile Internet,” 2012 IEEE Intl. Conf. on Communications (ICC),
2012, pp. 5849–5853.
Stähli, M., et al., “Monitoring and Prediction in Early Warning Systems for Rapid
Copyright © 2016. Artech House. All rights reserved.

Mass Movements,” Natural Hazards and Earth System Science, Vol. 15, No. 4, 2015,
pp. 905–917.
Tennenhouse, D. L., et al., “A Survey of Active Network Research,” IEEE Commu-
nications Magazine, Vol. 35, No. 1, 1997, pp. 80–86.
Caraguay A. L. V., et al., “SDN: Evolution and Opportunities in the Development
IoT Applications,” International Journal of Distributed Sensor Networks, May 4, 2014.
Gazis, V., et al., “A Survey of Technologies for the Internet of Things,” Intl. Conf.
on Wireless Communications and Mobile Computing (IWCMC), Dubrovnik, 2015.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Scalability of Networks and Computing 71

Xiaoli, X., Z. Yunbo, and W. Guoxin, “Design of Intelligent Internet of Things for
Equipment Maintenance,” 2011 IEEE Intl. Conf. on Intelligent Computation Technol-
ogy and Automation (ICICTA), Vol. 2, 2011, pp. 509–511.
Zhao, J. -C., et al., “The Study and Application of the IOT Technology in Agri-
culture,” 2010 3rd IEEE Intl. Conf. on Computer Science and Information Technology
(ICCSIT), Vol. 2, 2010, pp. 462–465.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 3
Security and Privacy

3.1 IoT Security: A Perspective

The power of IoT is in its distributed data gathering using a net-

work of sensors and devices to enable individuals and businesses
to make better decisions. Due to its very connected nature, devices
are exposed to increased potential attacks and other infringements
from the connected world. Attacks such as viruses, malware, tam-
pering, and others can result in industrial failures, business losses
or safety compromises, and other security issues. Adequate de-
ployment of security measures is an important requirement of any
IoT system.
According to International Data Corporation (IDC), by 2017,
90% of organizations that implement the IoT will suffer an IoT-
based breach of back-end information technology (IT) systems [1].
This is an important data point to note and it confirms the current
Copyright © 2016. Artech House. All rights reserved.

concerns IoT developers and users on the lack of confidence re-

garding security and more often privacy.

73

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 74 IoT Technical Challenges and Solutions

Stuxnet [2], BlackEnergy [3], and many other attacks on vehi-

cles, medical devices, and so forth have shown that IoT systems
are highly vulnerable [4]. The attacks on sensors and devices can
impact the system in multiple dimensions. Tampered health infor-
mation or manipulated sensor data can create an improper diag-
nosis and wrong treatments for patients. A denial of service attack
on vehicle control bus can cripple vehicle navigation system and
cause serious accidents. In general, attacks on devices as part of
various control systems in manufacturing and process control can
be a big threat to personal safety as well. Further, it may lead to
heavy losses to businesses and brand reputation. Malicious filter-
ing of critical alarm notification from a temperature sensor device
can unleash a fire hazard. Attacks on devices related to financial
systems are another area of concern. Manipulation of financial
transactions through unauthorized point-of-sale (POS) and mobile
point-of-sale (mPOS) device access can create substantial losses to
banks and retail outlets. Even a mere breach of access control itself
can create a lot of flutter to the reputation of highly visible orga-
nizations such as airports, nuclear installations, and power grid.
Software update-related vulnerabilities of IoT edge devices are
being used to gain unauthorized access to manipulate data and
firmware. This is a big threat to connected vehicles, homes, and
medical devices. Profiles of individuals can be created through un-
lawful access of network and location tracking leading to privacy
concerns. Even analytics attacks can be performed to extract per-
sonal information where apparently unconnected data from vari-
ous user devices can be linked and analyzed. It is very difficult to
detect and mitigate these kinds of attacks.
In this chapter, a pragmatic approach to IoT security and priva-
cy is presented. We begin looking at the objectives of security and
related system requirements and technology challenges. Subse-
quently useful technologies for data security and communication
Copyright © 2016. Artech House. All rights reserved.

security are discussed. Important aspects on identity, authentica-

tion, key management, and software security are reviewed later.
The need for threat analysis and risk modeling to understand and
plan system-level security is motivated next. Then a summary of
practical guidelines for architects, developers, and users of IoT sys-
tem is presented. Some concluding remarks are made at the end of
the chapter.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 75

3.1.1 Business Objectives of Security

A secure IoT system should provide for the required level of confi-
dentiality, integrity, and availability of data and services. Further,
they should ensure safety, reliability, resilience, and usability. These
aspects may get different emphasis from the varied perspectives
of different stakeholders including equipment vendor, business
owner or operator, service provider, and end user. The business
impact of security attacks on the IoT system is at multiple levels. It
includes service availability, system access and control, equipment
damage, data loss, and breach of privacy.
The major value proposition of security to the business is in the
reduction and mitigation of risk, supporting new business models
and improving operation performance. Minimizing the impact of
security measures on user experience is another aspect gaining im-
portance these days. The following are the key dimensions where
a business is impacted by a breach of security:

• Financial: This is the potential financial loss to a business as a

consequence of a lack or breach of security.
• Reputation: Estimated loss of reputation derived from the data
misuse or successful attack.
• Identity: Attacks that lead to disclosure, misuse, or loss of evi-
dence of identity of stake holders.
• Privacy and regulatory: Meeting the privacy and regulatory
requirements of data management. This will also include safety
aspects of operations.
• Availability guarantees: Operational availability of the system
and services under suitable guarantees.

System vulnerabilities need to be analyzed and suitable secu-

Copyright © 2016. Artech House. All rights reserved.

rity countermeasures to be deployed to keep the system secure

against possible threats affecting stated business goals.

3.2 IoT Security: Key Requirements

IoT can be viewed as a convergence of existing information technol-

ogy (IT) and operational technology (OT). At a high level, a typical

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 76 IoT Technical Challenges and Solutions

IoT system can be mapped into end points, gateway, and cloud.
End points are terminal devices such as sensors and actuators.
Gateways include access points and other network elements
that help aggregation at the edge. Cloud infrastructure provides
the required computing, storage, and visualization resources for
the IoT application. As shown in Figure 3.1, each of these com-
ponents has targets and associated vulnerabilities for attacks. The
artifacts to be protected include sensor data, security credentials,
identities of devices and people, configuration, and software. In
addition to this, limited resources at endpoints make them target
for easy overloading attacks on memory, CPU cycles and battery.
The external human agents including users, consumers, develop-
ers, workers, and hackers are predominantly the sources of po-
tential attacks. At times, software agents running on these devices
have also potential for triggering attacks. Efficient access control
Copyright © 2016. Artech House. All rights reserved.

Figure 3.1 IoT system components and typical attack targets.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 77

mechanisms are required as a primary security measure between

these entities. Security policies are required to ensure that they
have to interact according to their expected roles and responsibili-
ties. Enforcement of these policies will require effective authentica-
tion of the entities. Authentication mechanism verifies the identity
of entities by validating some sort of credentials. Authorization is
the process to grant specific access permissions to these authenti-
cated entities.
IoT security encompasses the schemes to ensure the confiden-
tiality, integrity, and availability of the data while it is at rest, in
motion, or in process. Communication protocols often incorporate
mechanisms for message security while in motion. Access control
and encryption can be considered to protect the data at rest. Sensor
devices may hold data for a short time in a buffer before transmit-
ting. Many times, an attack on this may not have a serious impact
for generic applications and one may be tempted to overlook the
confidentiality and integrity measures. However, it is advisable to
incorporate adequate security by default as per the availability of
resources. The protection of data while computing is somewhat
tricky and relatively difficult. Computing in trusted environment
is an option to consider. Computing in an encrypted domain is
another direction to explore where upcoming encryption schemes
such as homomorphic encryption are being developed, but they
are computationally intensive and not yet ready for constrained
devices.
Protecting the software from tampering and external injections
is another major security challenge. Periodic checking of the in-
tegrity of the software is necessary. Attestation is the process by
which a program can authenticate itself. Remote attestation allows
a remote system to check the integrity of another connected sys-
tem. These mechanisms will be useful to verify the identity and
integrity of the program code.
Copyright © 2016. Artech House. All rights reserved.

IoT systems have an additional challenge in protecting the end-

points and gateways from physical tampering or movements. This
will include tampering with sensors, power source, and the whole
device itself. Physical security mechanisms are needed to detect
such attacks and enable the system to take remedial measures.
Many times IoT systems gather and share sensitive information
for various applications. Providing confidentiality to the personal

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 78 IoT Technical Challenges and Solutions

identifiable data is the major scope of privacy. There should be

clarity with the data ownership, intended use, and traceability as
acceptable to the data provider. Figure 3.2 depicts key capabilities
required IoT security.
From a cybersecurity perspective, confidentiality, integrity, and
availability (CIA) are the major pillars of IoT data security among
other factors. Confidentiality in IoT refers to the restricted disclo-
sure of information. Cryptographic protocols play a major role here.
IoT services and devices require integrity mechanisms to prevent
the modification or tampering of the data, whether it is on the de-
vice or during transmission. They must be accessible and available
when called upon by authorized entities. This includes ensuring
that communication channels are protected, and IoT devices and
services are resistant to denial-of-service (DoS) attacks. There can
be serious problem from compromised IoT devices as they can be
used to launch DoS attacks against other connected systems. The
devices must be able to conduct mutual authentication with users,
other devices, and the cloud to block such propagation of attacks.
Communication security involves mutual authentication, traffic
authorization, and transmission with secured data integrity and
data privacy. Additionally, the endpoint devices should also sup-
port secure updates, firmware attestation, and secure boot process.
Copyright © 2016. Artech House. All rights reserved.

Figure 3.2 Key aspects of an IoT system security.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 79

Each endpoint must have the ability to maintain its identity in-
formation. Identity is a token assigned to a single instance to iden-
tify that device uniquely. Manually managing endpoint identity on
individual systems is not practical for large-scale systems. In such
cases, an automated identity management infrastructure should be
in place.

3.3 IoT Security Challenges

Many IoT systems have considerable complexity and diversity

of protocols and technologies. They are inadequately designed to
match their complexity. The time to market pressure to create new
IoT systems and services tend to neglect their security require-
ments. Cryptographic protocols play a major role for providing
confidentially and integrity of data. Due to their resource con-
straints IoT devices tend to implement relatively low-complexity
security primitives. The state-of-the-art cryptographic algorithms
and protocols are not easy to deploy on such constrained devices.
There is a need for lightweight protocols to address this limitation.
Further keeping their software updated with patches bug fixes or
security mitigations is another challenge.
Authentication and authorization are important aspects of IoT
security and there is a lack of robust standards for authentication
and authorization of IoT edge devices. The security level achieved
using multifactor authentication in traditional cybersecurity sys-
tems is not easily extendable to IoT systems. In IoT devices, mul-
tiple credentials may need to be stored in the same device and
this takes away the out-of-band benefit associated with multifac-
tor authentication. Some of the third-party options available for
certificate authentication and identity services are not adequate to
handle device-specific profiles and authorizations.
Copyright © 2016. Artech House. All rights reserved.

There are challenges with monitoring of a huge number of IoT

edge devices for security events. Due to heterogeneity of these de-
vices and lack of standardization, they do not have a unified way
for capturing and reporting security-related events. Adequate au-
dit and logging standards are highly required for IoT components.
Device-level security and reliability alone does not translate
in to a secure system at the global level. The system-level secu-
rity should consider and capture the process of combining and

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 80 IoT Technical Challenges and Solutions

configuring the interactions between devices involved in the IoT

system.
The life cycle of an IoT system has many perspectives based
on the roles of those manufacturing, deploying, operating, manag-
ing, and disposing the system components. Consideration must be
given to protecting IoT devices and data throughout its life cycle.
Data will flow across many administrative boundaries with differ-
ent policies and intents. Managing the data flow across levels is a
challenge.
Many IoT services require cloud platform to support multiple
business domains and users on a single platform. This requires
multitenancy support with secure and lightweight virtualization.
Security standards for platform configurations involving such vir-
tualized IoT platforms supporting multitenancy need to mature.
From a developer perspective, the best practices should pro-
vide guidelines for secure development of systems. However,
there is not much information on best practices available for IoT
security design and development. Familiarity with the available
technology itself is limited among IoT developers and awareness
for secure development needs to be enhanced.

3.3.1 Typical Threats on Various IoT Subsystems

The IoT system spans across multiple subsystems including sen-

sor, devices, gateways, communication network, cloud, and appli-
cations or services. The number of threats possible on the entire
IoT system is very high as the attack surface is too large. A publica-
tion from Open Web Applications and Security Project (OWSAP)
[5] lists the top 10 IoT vulnerabilities. It includes:

• Insecure Web interface;

• Insufficient authentication and authorization;
Copyright © 2016. Artech House. All rights reserved.

• Insecure network services;

• Lack of transport encryption or integrity verification;
• Privacy concerns;
• Insecure cloud interface;
• Insecure mobile interface;
• Insufficient security configurability;

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 81

• Insecure software or firmware;

• Poor physical security.

Table 3.1 gives a list of common threats pertaining to various

IoT subsystems.

3.4 Data Protection

The dynamics of the IoT data fall into three classes. They are data
in use (DIU), data at rest (DAR), and data in motion (DIM). The
data in use pertains to that being used by a running algorithm.
DIU is somewhat tricky because it hard to maintain cryptographic
measures to secure it while processing. This may be handled by
operating system layers that provide a confidential computing en-
vironment for the processing. The protection of such data requires
a trusted environment for the execution of code. The Trusted Ex-
ecution Environment (TEE) [6] provides this capability for use on
various processors such as those from Advanced RISC Machines
(ARM), which has a family of reduced instruction set CPU (RISC).
ARM-based devices leverage technologies such as TrustZone [7]

Table 3.1
Representative List of Threats at Various Subsystem Level
Subsystem Threats
Cloud, Web Account hijacking, data breaches, loss of data, insecure APIs,
applications and denial of service, malicious insiders, cross-site scripting (XSS)
services where client side scripts are injected to Web pages viewed by
others. Cross-Site Request Forgery (CSRF) that forces an end
user to execute unwanted actions on a Web application in
which they are currently authenticated. Broken authentication
and session management, insecure direct object references,
security misconfiguration, missing function level access
control, unvalidated redirects and forwards.
Copyright © 2016. Artech House. All rights reserved.

Network Eavesdropping, data modification, spoofing, stolen credentials

such as user name and password, man in the middle, denial
of service attack, sniffer attack, open ports in firewalls and
routers, misconfigured NAT-Port Mapping Protocol (NAT-PMP)
services, host intrusion and malware, password breach, Wireless
Encryption Protocol (WEP) vulnerability
Devices Lost or stolen devices, improper use of devices, stealing of
data or credentials, data modifications, malware injection;
tampering with firmware or firmware download process

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 82 IoT Technical Challenges and Solutions

for this purpose. Security integrated circuit (IC) chips have high-
quality cryptographic implementations. Random number genera-
tion, certification, and resistance to physical attacks are particular
strengths. Secure microcontroller unit (MCU) is a good proposition
for embedded device security. It would have built-in hardware
crypto modules that can generate various session keys, verify, and
sign using certificates and support data encryption. All the keys
and certificates are securely stored in a secure MCU vault. Embed-
ded microcontrollers may make use of security fuses to provide
security for Flash memory from external manipulations. Secure
operating systems and microkernels are other options to consider.
Separating the data from the metadata and storing data integ-
rity checks in with the metadata are another approach. Here the
data path units can work in its native environment and can check
the integrity as and when needed. More novel approaches includ-
ing homomorphic encryption attempt to allow the processing of
data while still in encrypted state. However, their complexity is too
high to be practical with the current state of the art.
Most of the data protection schemes will involve the applica-
tion of encryption. There are numerous cryptographic primitives
available in hardware or software form. The National Institute of
Standards and Technology (NIST) provide recommendations for
various crypto algorithms and configurations to use for the protec-
tion of sensitive information.
Some commonly used encryption algorithms are listed in Table
3.2.
If a new crypto scheme is developed, then it should undergo
cryptographic algorithm validation. For this purpose, the Crypto-
graphic Module Validation Program (CMVP) and Cryptographic
Algorithm Validation Program (CAVP) from the NIST can be
helpful.
Data is at rest when the device holds the data for a longer time.
Copyright © 2016. Artech House. All rights reserved.

The encryption key used here should be securely stored in a hard-

ware cryptographic module resident in the device. Further, it is
recommended that all secret and private keys, authentication, ac-
cess control, and other security configurations should be secured
sufficiently. Data-in-transit requires various communication secu-
rity measures to be deployed.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 83

Table 3.2
Commonly Used Encryption Algorithms
Crypto Algorithm Comments
Triple DES Symmetric, uses three individual keys of 56 bits each;
hardware implementable
RSA Public key encryption, key size 1,024 to 4,096 bits
Blowfish Symmetric, encrypts in 64-bit blocks, key sizes 32 to
448, very fast except when there is a key change due to
preprocessing delay
Twofish Symmetric, block sizes of 128 bits, key sizes up to 256 bits
AES Symmetric, block size 128, key sizes 128, 192, 256, Most
popular contemporary standard, available with many devices
ECDH and ECDSA Elliptic curve Diffie-Hellman (ECDH) scheme for encryption
key establishment Elliptic Curve Digital Signature algorithm
(ECDSA) for data signing

3.5 Communication Security

Generally communication standards include cryptographic confi-

dentiality, integrity, and authentication algorithms incorporated as
part of the protocol. Most of the IoT applicable wireless communi-
cation standards incorporate some level of security features at the
link level. In many cases, they offer some flexibility to configure
for specific application requirements. Table 3.3 provides a list of

Table 3.3
Salient Security Features of Popular IoT Wireless Communications Standards
Wireless
Communication
Standards Security Aspects
Wi-Fi Wi-Fi Protected Access 2 (WPA2) with AES (IEEE 802.11i), EAP
methods for layer 2 authentication
Bluetooth, BLE Discovery or connect through enquiry or paging, no device
Copyright © 2016. Artech House. All rights reserved.

address privacy, Encryption E0/, BLE has AES-CCM, signed

data, frequent change of address, ECDH for key exchange
Zigbee (802.15.4) Link layer encryption with 128-bit AES; trust center for key
distribution
6LowPAN Secure mode using 802.15.4 link layer encryption; access
control list (ACL)
Weightless Supports authentication, encryption, repudiation
GSM Authentication algorithm (A3), cipher key generating
algorithm (A8) embedded in SIM; Temporary Mobile Subscriber
Identity (TMSI) to address intrusions

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 84 IoT Technical Challenges and Solutions

popular wireless communication standards and their key security

features.
The security requirements and schemes for data communica-
tions at the application level will have a dependency on the data
exchange patterns as well. Major data communication patterns in
IoT applications are request-response and publish-subscribe.
The request-response paradigm is used by several protocols in-
cluding Web Services/simple object access protocol (SOAP), Java
remote method invocation (RMI), constrained application proto-
col (CoAP), remote procedure call over data distribution services
(RPC-over-DDS), and the popular serial communications protocol,
Modbus. The security support of the standard protocols varies in
terms of confidentiality, integrity, and key management features.
Protocols such as MQTT (formerly MQ telemetry transport)
and advance message queuing protocol (AMQP) support the pub-
lish-subscribe mode for data exchange. The primary categories of
threats faced by them are unauthorized subscription, unauthor-
ized publication, tampering and replay, and unauthorized access
to data. The intermediate brokers used by some of these protocols
can be a single point of failure that may affect their availability.
Control plane communications have a focus on the safety and
reliability of the system and the security requirements are com-
paratively high. Here mere enhancement of encryption and au-
thentication may not be sufficient. An additional layer of security
mechanisms such as context-based sanity checks and firewalls and
unidirectional gateways should be considered.

3.5.1 Cryptographic Key Management

In crypto-based security protocols, the key establishment and

management form critical part of cryptographic operations. Gen-
eration and periodic renewal of cryptographic keys or certificates
Copyright © 2016. Artech House. All rights reserved.

are needed when a new communications session is established.

Keys must often be certified by a certifying authority (CA) or de-
rived from other keys.
The public key infrastructure (PKI) model is a proven approach
to a strong key management system. It has the required flexibility
to work in a distributed system. Resource constraints with the IoT
devices would require a lightweight PKI model to be applied to
them. Such a lightweight PKI can be used for distributing device

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 85

keys or identities as a tamperproof root-of-trust token that requires

low computation and storage capability.
A typical X509 certificate is around 800–1,200 bytes. A tailored
variant of the X.509 certificate is used known as Device Verifiable
Certificate (DVC), which is typically 200–300 bytes. DVC supports
mutual authentication, encryption, digital signature, key manage-
ment, secure messaging, and secure updates. There is a wide de-
ployment of such certificates in millions of devices today.
When a device is decommissioned, it is necessary to revoke
a certificate before it expires. In X.509, this is accomplished by
publishing of certificate revocation lists (CRL). CRLs are publicly
available and also digitally signed by the certificate authority. Any
device that wants to authenticate a peer device must check the cur-
rent CRL to check the revocation status. This can also be done us-
ing Online Certificate Status Protocol (OCSP). However, for all IoT
device scenarios, this might not be a very easy process.
There are various other application-level security protocols that
can be used in IoT system scenarios. Specifically, they can be con-
sidered to be implemented in the gateway to the cloud communi-
cation segment. Table 3.4 provides a list of security protocols and
their salient features.
IoT systems need a security policy for protection of IoT data
from sensors to the cloud. Key generation algorithms should
use an effective random number generator as prescribed by the

Table 3.4
Popular Security Protocols and Their Salient Features
Protocol Security Aspects
IKEv2/IPsec Establishes a secure tunnel, X.509 certificates for
authentication, Diffie-Hellman key exchange for shared session
secret, cryptographic key generation using shared secret
TLS/SSL X.509 certificate for authentication, asymmetric keys from
X.509 used for symmetric key exchange, symmetric key used for
Copyright © 2016. Artech House. All rights reserved.

data encryption
DTLS Datagram TLS (based on TLS)
HIP Host Identity Protocol, host identity based on public key
(instead of popular IP address/DNS)
EAP Extensible Authentication Protocol, an authentication
framework supporting multiple authentication methods, works
on the link layer, uses different protocols for EAP messages
transmission, supports key delivery and usage mechanisms
SSH A cryptographic network protocol, uses public key cryptography
for mutual authentication, creates a secure channel for data

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 86 IoT Technical Challenges and Solutions

standards. Periodic key changes, distribution, and destruction

should be driven by well-defined security policies. When the sys-
tem spans multiple stakeholders, data ownership policies and dec-
larations are also required. For this data, elements and associated
attributes need to be identified.

3.6 Identities and Identity Management

The IoT poses the challenge to manage a huge set of identities that
are larger than conventional identity management systems. Here,
in addition to humans, the identities of several other machines and
identities of things are to be managed. In many cases, they are con-
nected intermittently or mobile. Some people refer to this identity
ecosystem as the Identity of Things (Adopt) [8] where there is a
networked relationship between devices, applications or services,
and humans.
An identifier is typically a publicly known dedicated attribute
or name for a device. A device can have more than one identifier.
Sometimes the network addresses are used as identities. However,
that is not a recommended practice. Addresses such as IP deter-
mine the communication endpoint within a certain system, but the
device address may not be permanently associated with the device
and it may change. Thus, it is better to have a different identifier
other than the network address for the device.
There are many types of device identifiers in common use. The
simplest of them are a globally unique identifier (GUID) or a pub-
lic name. However, these by themselves do not provide an authen-
ticated identity for an IoT device and are open for easy spoofing.
Another technique is to use a cryptographic identifier such as
802.1AR device identifications (IDs). A trusted platform module
(TPM) [9] can handle cryptographic device identities that are ro-
Copyright © 2016. Artech House. All rights reserved.

bust to many software and hardware attacks.

ITU-T defines a number of specifications pertaining to object
identifiers (OIDs). OIDs can be used to refer to physical objects,
in the management information base (MIB) used by the Simple
Network Management Protocol (SNMP). Logical objects include
software, services, data and databases, documents, and other digi-
tal objects. Identification of software is another area of consider-
able interest. Approaches to it include software ID tags and the

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 87

Common Platform Enumeration. ITU-T OIDs can be used to refer

to a number of logical objects. The Digital Object Identifier (DOI)
is standardized as ISO 26324:2012 and provides a way of directly
referencing digital objects.

3.7 Authentication

Authentication is one of the important requirements for IoT de-

vices to know and verify each other’s identity to interact in a safe
and secure manner.
Shared key and public key models are two basic models used
for authentication in the Internet domain. Shared keys are shared
secrets known to participants. They can be passwords, personal
identification numbers (PINs), images, biometrics, gestures and
cryptographic keys. The trouble here is that if the key is exposed,
then all the devices and services using that key are vulnerable. In
the IoT, this threat can go to massive proportions due to its scale.
However, PKI has two keys (private and public) and the
scheme is stronger from the security perspective. For some of the
constrained devices, it may be difficult to implement due to its al-
gorithm complexity.
SSL (Secure Sockets Layer) and TLS (Transport Layer Security)
are the most widely used Internet security protocols. Both of them
support PKI and shared-key models to create a secure channel.
OAuth 2.0 [10] and OpenID Connect 1.0 [11] are two standard-
ized frameworks for authentication and authorization in the In-
ternet domain. It is used to explicitly participate in the issuance of
tokens to applications. Further, OpenID Connect has support for
discovery and registration mechanisms that are highly relevant for
IoT. The limitation of OAuth and Connect is that they have only
been bound to HTTP so far. There are ongoing works on binding
Copyright © 2016. Artech House. All rights reserved.

them to other IoT protocols as well.

The authentication mechanisms supported by some of the
prominent IoT protocols are given in Table 3.5.
Although multifactor authentication (MFA) enhances the effec-
tiveness, it is not always feasible to use traditional MFA methods
to support strong authentication of things. Contextual intelligence
[12] formed by real-time contextual clues, in addition to creden-
tials, can be used for authentication. For example, if an anomaly is

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 88 IoT Technical Challenges and Solutions

Table 3.5
Authentication Options for Some of the IoT Protocols
Protocol Authentication Methods Supported
MQTT Username and password are used and are sent in the clear. It is
recommended that TLS be employed when using MQTT.
CoAP Support for multiple authentication options for device-to-device
communication. Can use DTLS additionally for higher-level confidentiality
requirements. Uses preSharedKey, rawPublicKey, and certificate.
XMPP Supports multiple authentication patterns via the Simple Authentication
and Security Layer (SASL). Includes one-way anonymous and mutual
authentication with encrypted passwords, certificates, and other means
through SASL.
DDS The Data Distribution Standard (DDS) from OMG provides endpoint
authentication. Also supports key establishment to perform message
data origin authentication (HMAC). Both digital certificates and
various identity or authorization token types are supported. Uses X.509
certificates (PKI) using RSA and DSA tokens.
HTTP/REST Support of TLS protocol for authentication and confidentiality. For basic
authentication, credentials are passed in the clear and can be used with
TLS. However, a token-based authentication approach such as OAUTH 2 is
recommended.
Zigbee Supports network and application-level authentication through the use
of master key, network, and, optionally, application link keys. Keys are
preshared.
Bluetooth Provides authentication through standard automatic pairing or simple
pairing with a human-in-loop to verification (following a simple Diffie-
Hellman exchange). Supports one-way and mutual authentication options.
For device-device authentication, .secure simple pairing offers Justworks,
Passkey entry, and Out of Box options.
Bluetooth-LE Supports two-factor authentication system. LE Secure Connections pairing
model that combines several of the available association models. In
addition, ECDH is used for key exchange. Unencrypted data authenticated
using Connection Signature Resolving Key (CSRK) Device Identity/Privacy
is via an Identity Resolving Key (IRK).

detected in a sensor data traffic with abrupt changes in the values,

a second factor could be initiated to reauthentication of the de-
vice. This is very much useful for smartphone-based applications
where there are a number of sensors to provide real-time contex-
Copyright © 2016. Artech House. All rights reserved.

tual intelligence.

3.8 Access Control

Access control pertains to the permissions in the usage of resources

and services, assigned to different entities of an IoT system. The
major challenge is in providing access permission in an environ-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 89

ment where human users and machines or devices are authorized

to interact.
Entities seeking access to others may have multiple roles. In
role-based access control (RBAC), users or devices have to prove
their identity; consequently, a session is created and a role is estab-
lished for the user or device to perform authorized tasks. The iden-
tity mechanisms that can be used for authorization were discussed
earlier. However, when there are a large number of devices with
multiple purposes, the resulting role explosion could be difficult
to manage. Policy-based access control or dynamic authorization
management could be helpful here. Attribute-based authorization
provides fine granularity, high flexibility, and rich semantics, and
other beneficial features like partial authentication could be used.
Here sensor attributes like location, sensor type, or actuator cat-
egory can be used to define access categories dynamically and it
could be a support for role-based access control.

3.9 Secure Software Updates

Software or firmware updates are a critical feature to consider for

IoT systems. From an operational point of view, it is highly use-
ful to maintain and advance the lifetime of the connected devic-
es. Manufacturers would use this feature to patch their firmware
bugs as and when they are discovered. The mechanisms deployed
for the support of firmware updates are vulnerable for security
breaches. The upgrade process could be exploited by hackers to
tamper with the device data, operations, or actuations. Further, it
may be used to inject a malware to trigger secondary attacks in the
network. The risk of performing remote updates must be carefully
considered against its value.
Secure update for devices has multiple parts to look at. It has
Copyright © 2016. Artech House. All rights reserved.

to ensure firmware integrity and confidentiality while commu-

nicating, flashing, and booting up the device. Secure update also
involves attesting to the integrity of the update and verifying it
after a secure transmission to the target device. The boot-time and
run-time software are critical for the attestation and verification
process. Some of the data security measures described previous-
ly can be used here as well. Keys used in upgrades can be man-
aged by a third-party certificate authority and updated as needed.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 90 IoT Technical Challenges and Solutions

This mechanism can also be used for the secure update of system
configurations.

3.10 Privacy in IoT Systems

A prevalent definition of privacy is that “the right to select what

personal information about an individual is known to what peo-
ple” [13]. In the IoT paradigm, this means:

1. Awareness of privacy risks posed by smart things and services;

2. Individual control over the collection and processing of per-
sonal information by the devices;
3. Awareness and control of subsequent use and dissemination of
personal information by those entities to any entity outside the
personal control sphere.

Basically, this entails one to assess one’s personal privacy risks

and take suitable action to protect one’s privacy. Also, the user
should be assured that it is enforced beyond his or her immediate
control sphere. The trouble is that the exact scope of the subject’s
personal sphere can differ from situation to situation. Further, it
is still unclear what constitutes the individual’s personal sphere.
Privacy is a deeply social concept and subject to greatly varying
individual perception and requirements. From a practical perspec-
tive, privacy in IoT is around a set of personally identifiable infor-
mation (PII) and the user should be involved in selecting them. In
one view, privacy is providing confidentiality of PII elements as
desired by the user.
There are many challenges in implementing privacy in IoT. In
general, awareness of privacy rights and compromises are low
among the user community. In IoT systems, the data collection
Copyright © 2016. Artech House. All rights reserved.

happens more passively. Due to this, many privacy breaches go

unnoticed. Although there are privacy regulations, many big busi-
nesses do not enforce them as the economy of breaching privacy
regulations are in their favor.
Some of the major threats to privacy are discussed here:

• Identification: This is a threat associating an identifier with an

individual and data about the person. The identifier can be a

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 91

name, address, or pseudonym of any kind. Privacy measures

should include identity protection and protection against iden-
tification. Anonymization and privacy-enhancing identity man-
agement are some techniques that can be considered, although
they may not completely address the challenge in all situations.
• Localization and tracking: This is the threat of finding and
recording a person’s location at multiple levels of granularity
over time. Generally, this data is collected passively. Leaving lo-
cation trails puts the user at risk of identification, location, and
activity. Anonymization, time, and data resolution control are a
few steps that can be considered to address this threat.
• Profiling: This is the threat of compiling information dossiers
from various devices about individuals. This is generally used
to infer interests by correlation with other profiles and data. Ex-
isting approaches to preserve privacy include client-side per-
sonalization, data perturbation, obfuscation, and anonymiza-
tion.
• Device updates and transitions: Retiring old devices can leave
a lot of privacy-sensitive information out of one’s personal con-
trol. Wiping decommissioned devices is one measure that can
be taken to address this.
• Data linkage and analytics: This threat relates with linking var-
ious data items, possibly from multiple sources, such that the
analysis of a combination of them reveals unintended informa-
tion. Suitable access control and anonymization are some of the
techniques that can be applied here.

The following steps are suggested towards a generic approach

for IoT privacy:

• Privacy and security go hand in hand. A privacy and security

Copyright © 2016. Artech House. All rights reserved.

by design approach should be adopted.

• Purpose limitation measures that promote no use beyond pre-
defined purposes should be taken.
• The data minimization approach should be adopted in which
only necessary data is collected, processed, and stored. Consid-
er anonymization or deletion of data after use.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 92 IoT Technical Challenges and Solutions

• Adopt distributed protection models that provide multilayer

security and resilience in a distributed architecture.
• Invest in the use of advanced techniques for anonymization,
pseudonymization techniques, and encryption.

3.11 System-Level Security Assessment

Major value proposition for security is the reduction and mitigation

of risk, enablement for supporting new business models and im-
proving operations. Evaluation of the return on investment (ROI)
of an implementation of a secure system design must be grounded
in a realistic assessment of the threats. It should assess the risks
that they pose that may prevent the system from delivering its
intended business functions. It is possible to go forward with no
security and accept all of the risk. It is also possible to spend an
exorbitant amount of money on security to the point where it no
longer justifies the security gains. A balance must be achieved be-
tween the ROI of the security and its effectiveness. To achieve this,
we need a system-level metric-driven security assessment.
Further, in a dynamic IoT system, new nodes are added and
new software are installed and they are evolutionary in nature. In
such situations, operators use sophisticated tools to discover the
network topology and existing vulnerabilities. However, these
tools do not put vulnerabilities into the context of how they can
be exploited. Similarly, they do not automatically determine the
impact to the system. Because of this reason, several automated
approaches to security analysis have been proposed during the
last decade. Some of these automated approaches define security
metrics over the paths of an attack graph, while others define a
security metric in terms of security risk.
Copyright © 2016. Artech House. All rights reserved.

3.11.1 Risk-Based Security

Risk is usually defined as the expected loss. “Risk is a function of

the likelihood of a given threat-source’s exercising a particular po-
tential vulnerability, and the resulting impact of that adverse event
on the organization” [14].

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 93

The risk computation can be done in two parts: a probability

estimation of a successful attack and an impact estimation. It is
useful to have a computable measure of risk to select an optimum
security scheme by assessing the impact of various possible secu-
rity threats.

Risk = Probability of successful attacks * Impact of attacks

The generic block diagram of system for security risk estima-

tion is given in Figure 3.3.
The system faces number of security threats that exploits the
system vulnerabilities. The threats are realized though a number
of attacks. Structuring of these threats through a threat model is
the first step of the risk analysis methodology. The next step is to
categorize the threats based on their nature of security impact to
the system. Subsequently, a metric for severity ranking is comput-
ed based on the threat category. This helps the analyst to find the
ranking of various system threats and it may be used to develop a
security strategy to focus on most severe threats.
A more detailed computation is possible for risk estimation
through a more refined impact estimation and likelihood estima-
tion for the threats. Here the context of the attack is also consid-
ered for the estimation. Impact depends on what impairment or
loss that it creates for the business goals due to the affected data
or functionality elements. Therefore, it will be beneficial to use a
Copyright © 2016. Artech House. All rights reserved.

Figure 3.3 Generic steps for approach security threat analysis

and risk assessment.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 94 IoT Technical Challenges and Solutions

system model that incorporates the information flow, dependen-

cies, hardware or software mappings, security goals, and business
goals as the basis of the impact estimation. Further, the likelihood
of a threat has a dependency on various factors including system
vulnerabilities, the actor (agent), his or her skill level, the skill level
required for the attack, and motivation. There are qualitative and
quantitative approaches to estimate the risk indicator. Subsequent-
ly, various security countermeasures can be planned to reduce the
risk level and the process can potentially optimize the security
investments.
There are numerous frameworks devised for security risk anal-
ysis, including OCTAVE (Operationally Critical Threat, Asset, and
Vulnerability Evaluation; a security risk assessment framework
developed by Carnegie Mellon’s Computer), ISO 27001/27005
[ISO/IEC 27001 is a security management framework developed
by the International Organization for Standardization (ISO), and
ISO/IEC 27005 is the associated guidance for risk assessment and
risk management], COSO [The Committee of Sponsoring Organi-
zations of the Treadway Commission; has its own standard for en-
terprise risk management (ERM) and risk assessment] and NIST,
which has a family of standards similar to the ISO 27000 series
for risk management, in particular NIST 800-30 and 800-53. While
each of them has a number of strengths, there is weakness in us-
ing specific aspects of IoT systems and their interdependencies in
estimating the risk level.
One common argument against the use of security metrics is
that the input to the metric model is often imprecise. Whatever
the model can compute is not meaningful. However, the numbers
obtained can be used in a comparative sense and that itself is a big
step forward. Refining metrics for better capturing of the proper-
ties of network vulnerabilities is a work in progress.
One simpler and practical approach for threat modeling and
Copyright © 2016. Artech House. All rights reserved.

risk estimation for IoT system security is outlined next.

3.11.2 Threat Modeling and Risk Estimation

According to the definition of the Open Web Application Security

Project (OWASP), threat modeling is a method for analyzing the
security of an application or system. This process consists of the
identification and evaluation of security risks and, based on this

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 95

knowledge, the determination of the mitigation techniques, meth-

ods, or algorithms.
Understanding the system to be analyzed is the first step in
threat analysis process. For that, we need to identify the main as-
sets that need to be protected. Here the term assets are used in
a broad sense that includes data, security credentials, certificates,
keys, and functionality states. This also includes the mapping of the
interaction between assets and interactions with external entities.
The structured representation of these interactions can be through
data flow diagrams (DFDs) that model the entire IoT system under
consideration. Nodes of the DFD will be the information elements
and edges form the interactions or dependencies. An exemplary
system model for the remote monitoring of the chiller system for
climate control is shown in Figure 3.4. It illustrates the data flow
and dependency of the IoT system that uses various sensors moni-
toring the environmental and machine states and the data is used
to form a control strategy to keep the environmental parameters
under specified limits. The attributes of the nodes and edges can
capture the other system features or details that can contribute to
the threat analysis.
The risk analysis process starts with studying the system and
preparing a DFD model. Each of the interactions (involving an edge
to a node) will have several threats. It will be difficult to handle the
innumerous threats pertaining to the system in a structured man-
ner. For this, a threat categorization would be helpful. The STRIDE
model from Microsoft is one approach for threat categorization,
and another is a model for qualitative risk evaluation, DREAD.

3.11.2.1 STRIDE Model

The STRIDE model [15] is a component of the Microsoft Security
Development Lifecycle (SDL) for general threat categorization. It is
an acronym formed from the following threat categories: spoofing,
Copyright © 2016. Artech House. All rights reserved.

tampering, repudiation, information disclosure, denial of service,

and elevation of privileges. The threats that translate to masquer-
ade and to impersonate other users or devices to gain undue access
come under the category of spoofing. Actions that illegally modify
the data fall into the tampering category. Repudiation is a mali-
cious action with the goal of hiding or changing the authorship
information of certain committed actions or transactions. Informa-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 96 IoT Technical Challenges and Solutions

Figure 3.4 Data flow model of a chiller monitoring system.

Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 97

tion disclosure means reading or disseminating the data without

permission. Denial of service is a malicious action that blocks ac-
cess to valid users. A threat aimed to maliciously enhance the ac-
cess privilege to enable unauthorized access come under the eleva-
tion of privilege category.
The threats associated with each interactions represented in the
system DFD model are categorized into one of the six categories of
STRIDE model described above. The next step is to rate the threats
organized under each of the STRIDE categories using a qualitative
risk computation mechanism such as DREAD.

3.11.2.2 DREAD Model

DREAD [16] is a classification scheme for threat rating from Mi-
crosoft. This is an acronym formed from the constituents of the
threat metric. They are: damage, reproducibility, exploitability, af-
fected users, and discoverability. The sum of the ratings from each
of these aspects of the threats forms the final DREAD score.
The amount of damage caused by the threat if exploited is rated
under damage. Reproducibility is a metric to represent the ease
of exploitation of the threat. Exploitability captures the required
skills and tools to exploit the threat. The affected user’s rating is
determined from the number of users who will be affected by the
exploitation of the threat. Discoverability expresses how easy it is
to discover the threat. Each of these aspects is rated from 0 to 10,
with highest security risk being rated at 10.
The overall DREAD score is computed as:

Risk_DREAD = (Damage + Reproducibility + Exploitability

+Affected Users + Discoverability)/5

Once each of the threats identified is rated using DREAD, the

next step is to aggregate this over each STRIDE category for all
Copyright © 2016. Artech House. All rights reserved.

paths in the DFG. Based on the numerical values of the aggregate

risk rating for each threat category, an overall assessment of the ef-
fectiveness of system security can be obtained. Figure 3.5 gives an
outline of a risk estimation process involving STRIDE and DREAD
models along with a DFG-based system model [17].
There are multiple variants with suitable refinements available
for threat modeling and risk estimation usable for IoT systems [18].

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 98 IoT Technical Challenges and Solutions

Figure 3.5 Qualitative risk assessment using STRIDE and DREAD

models.

The determination of the countermeasures and mitigation

techniques is the following step. Generic mitigation aspects for
each threat category are listed in Table 3.6. This can provide guid-
ance to refine the selection of protocols, cryptographic primitives,
security policy, access control mechanisms, and other system
configurations.
Several sophistications in approaches can be brought into the
threat analysis process. Instead of a mere listing of threats, they
can be expanded in terms of attack graphs where a threat is real-
ized through a combination of attacks. The attack graphs model
how multiple vulnerabilities contribute in an attack. The graph
represents system states corresponding to security-related condi-
Copyright © 2016. Artech House. All rights reserved.

tions, such as the existence of vulnerability on a particular host or

the connectivity between different hosts. Vulnerability exploitation
can be modeled as a transition between system states.
The overall threat rating can be refined with a more elaborate
risk assessment approach where the probability of attacks is also
considered. It will depend on various factors including the threat
actors, their skill levels, motivation, ease of attacks, and social and
environmental conditions. Further, the impact aspect of various

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 99

Table 3.6
Threat Categories and General Mitigation Approaches
Threat Type Mitigation Techniques
Spoofing identity Malicious party impersonates a device or a user with an illegally acquired
identity. Some of the mitigation approaches are: stronger authentication,
protecting secrets, and avoiding storage of secrets if possible.
Tampering with Deletion or modification of data. This may be mitigated by: enhanced
data authorization, hashes, message authentication codes, digital signatures,
and tamper-resistant protocol.
Repudiation Denial of committing a data operation or transaction. Some mitigation
measures are: digital signatures that can be verified, timestamps, and
audit trails.
Information Unauthorized disclosure of data to an unintended audience. Some steps
disclosure for mitigation are: strong authorization, encryption of data, protect
security keys, credentials, avoiding storage of secrets.
Denial of service Block the access to data and services even to legitimate users. Some of
the following measures can be considered: authentication, authorization,
filtering based on source, ID, size, and type, throttling, and quality-of-
service measures in the network.
Elevation of Unauthorized enhancement of access level and permissions. Some
privilege applicable measures include: authentication, authorization, access
control, and run with least privilege.

threats can be linked with associated security goals and business

goals.
Another direction is to use more sophisticated system models to
cover multiple aspects of the IoT system integrated into the mod-
el. Typical IoT systems are complex and it integrates IT, OT, and
communication technologies. For the purpose modeling, the entire
system can be abstracted in multiple views. They are information
view, deployment or operation view, applications view, and busi-
ness view. The information view looks at the system in terms of
data generation, data flow, and data transformation. The deploy-
ment or operation view depicts the devices on which the informa-
tion plane is deployed and their configurations and functions for
their operations. Data collected and stored in the information view
Copyright © 2016. Artech House. All rights reserved.

may be used by both internal and external applications, which can

take many forms, from software in a console to an application in a
mobile phone. The business view captures the business objectives
and how the underlying information and operations are contribut-
ing to them. The DFG-based system model described earlier can
be extended with these views and used in the threat analysis for a
more realistic outcome.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 100 IoT Technical Challenges and Solutions

3.12 IoT Security: Practical Guidelines

According to the trusted computing group [19] the critical strate-

gies suggested for IoT architects to address system security are:

1. Assessment of IoT system goals and risks;

2. Management of identity and integrity;
3. Encryption of confidential data;
4. Use of hardware security for critical systems;
5. Protection of resource-limited devices with overlay networks.

For a successful IoT system, the architects should define their

business goals, gauge the security risks to those goals, and develop
appropriate security controls to manage or reduce the risks. Securi-
ty should not be an afterthought. This requires building in security
by design and evaluating risks early in the life cycle. For medium
to large IoT systems, rigorous threat modeling and risk analysis
need to be carried out.
There are a few documentations on useful guidelines for IoT
system developers and users [20–22]. Some of the key guidelines
on practical implementation aspects to help the development, de-
ployment and management of IoT systems are summarized next.

• Access control or passwords:

• Change default passwords shared between devices. Avoid
weak passwords.
• Randomly created passwords using high-quality random
number generators are desirable.
• Consider deactivation of sparsely used or unused features
or services (e.g., virtual private networks (VPN) and remote
administration).
Copyright © 2016. Artech House. All rights reserved.

• Identity and authentication:

• Use system-level threat models to design authentication and
authorization schemes.
• Consider context based secondary authentication such as
smartphone-based techniques (e.g., face recognition, speaker
recognition, fingerprint, gesture).

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 101

• Leverage the security features of standards-based IoT proto-

cols such as CoAP, DDS, and REST. This will help interoper-
able authentication and authorization.
• Common identity mechanisms that can be considered in-
clude cryptographic identity, identity based on PKI, hard-
ware root of trust (HRoT), key pair generated in HRoT, and
private key protection. Also consider automated identity
provisioning, identity management, and remote attestation
of identity.
• The devices containing the sensors and actuators need to au-
thenticate themselves to the gateway devices and the cloud.
Mutual authentication should be performed to avoid fake en-
tities from entering into the system.
• Embedding a hardware security module that can take care
of cryptographic functions should be considered wherever
possible to strengthen the security.
• Establish role or relationship mappings between people and
devices through explicit authorizations
• Communications and networking:
• Review all connectivity configurations to the Internet, in the
context of their vulnerabilities and functional requirements.
Use encrypted communication whenever possible.
• Migrate to IPv6 for scalability and future-proof.
• Choose end-to-end security schemes wherever possible, as-
suming the network to be insecure by default.
• Sharing IoT-related data and access with respective device
manufacturers should be under a well-understood policy.
• Implement a privileged user management system to ensure
that administrators can access and monitor systems and de-
vices. Authorize administrators to disable the device in the
case when an attack is detected.
• Device security:
Copyright © 2016. Artech House. All rights reserved.

• At a minimum level, endpoint security at the device level

include root of trust (RoT), attestation, and crypto algorithm
and secure updates. RoT should be ideally in hardware. Re-
mote attestation is preferred.
• Tamper-resistant approaches could be considered including
automatic memory erasure or remote notification during a
tampering attempt.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 102 IoT Technical Challenges and Solutions

• The span of the security should begin from manufacturing

stage itself and should be tracked in each stage of the life
cycle.
• For easy self-registration, the initial provisioning artifacts
including keys and identities may be resolved at the manu-
facturing stage itself.
• The default configuration for the device should be secure.
Firmware should be locked down for restricted access.
• Hardware protection should be used wherever possible.
The use of secure element (SE) or trusted protection modules
(TPM) devices is recommended.
• All unused ports, general purpose input/output pins (GPIO),
universal asynchronous receiver transmitter (UART), debug
ports (such as joint testing action group (JTAG)) interfaces on
the hardware should be disabled.
• Use preventive measures for physical attacks; ICs should be
protected with appropriate sealing.
• Data security:
• All confidential data at rest should be encrypted. The integ-
rity of all data at rest should be verified before it is used. End-
to-end verification of data integrity from its initial source to
the destination should be considered for data communica-
tions.
• Based on the capability of the devices and the application
requirement of end-to-end confidentiality, the data should be
encrypted. Sometimes the confidentiality scheme supported
by the communication standard would be sufficient.
• Transport layer security (TLS) provides a mechanism for
integrity-verified confidential transmission of data between
two hosts on a network. However, TLS cannot guarantee
end-to-end data integrity when intermediary hosts are in-
volved between the source and the destination. There are
Copyright © 2016. Artech House. All rights reserved.

some workarounds that can be used with trusted interme-

diary systems that extend the chain-of-trust across multiple
hops.
• There are cryptographic schemes that attach verifiable integ-
rity and confidentiality to the data itself. In this approach, the
data is logically encapsulated by and transmitted along with
signatures or secure authentication codes that guarantee data
integrity and encrypted when confidentiality is needed.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 103

• The integrity of an endpoint device must be maintained,

even from authorized administrators and users of the system.
• Monitoring the signatures or file hashes of all software on a
system and comparing that to a trusted source is one effective
way of detecting tampering at the software level. Ensure that
only approved, trusted software or firmware can execute on
a system and that any updates or changes are from a trusted
source. Cryptographically signed code should be preferred.
• Cloud interfaces:
• All cloud interfaces, including cloud-based Web services
and APIs, should be reviewed for security vulnerabilities.
The interfaces should avoid weak passwords, provide ac-
count lockout mechanism, multifactor authentication, and
protection against XSS, SQLi, and CSRF vulnerabilities, and
force the change of the default username and password.
• End-user awareness:
• End users need to be educated to operate and manage the
IoT devices that they handle in a secure manner, including
information about the software updating process and man-
agement, which should available to the user. In case of an
error during the process, the rollback mechanism should be
specified.
• Awareness on vulnerabilities of other devices (smartphone,
wireless network, building automation network) can impact
the device security.
• Be aware on how to safeguard customer data and enable
privacy-related configurations.

Also, focus on these security measures and mitigations may

have some variations for different IoT system stakeholders such
as manufacturer, developer, consumer, and service provider. Some
Copyright © 2016. Artech House. All rights reserved.

general recommendations are presented in the IoT Security Guid-

ance document from OWASP [23].

3.13 Summary

Security and privacy are the top concerns of IoT systems these
days. One of the major reasons behind this concern is the uncer-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 104 IoT Technical Challenges and Solutions

tainty on how the system will be impacted by the actions and be-
haviors of various system components and stakeholders. Many
times, a holistic picture of the system security is not available
to the decision-makers and end users. A systematic approach to
analyze various threats and their potential risks with the help of
a system model would be of use here. There are a large number
of security protocols and algorithms available for consideration.
Only the prominent ones are outlined in this chapter, and there are
many others being worked out by the research community. As the
device technology progresses, the affordable computing and stor-
age capability per joule is also increasing. This enables the use of
Internet protocols directly on end points in many cases. Although
we highlight constraint devices as one of the differentiators for IoT
protocols in general, one should leverage proven Internet technol-
ogies wherever possible to reduce overall cost and time to market.
IoT business is at its initial stages and insights from future opera-
tions will testify and enhance the security and privacy schemes as
it evolves.

References
[1] IDC Futurescape for Internet of Things, December 2014, https://www.idc.
com/getdoc.jsp?containerId=prUS25291514.
[2] http://spectrum.ieee.org/telecom/security/the-real-story-of-stuxnet.
[3] http://www.securityweek.com/blackenergy-group-uses-destructive-plu-
gin-ukraine-attacks.
[4] http://www.wired.com/2015/12/2015-the-year-the-internet-of-things-
got-hacked/.
[5] https://www.owasp.org/index.php/OWASP_Internet_of_Things_Top_
Ten_Project#tab=Top_10_IoT_Vulnerabilities__282014_29.
[6] Trusted Execution Environment, https://en.wikipedia.org/wiki/Trusted_
Copyright © 2016. Artech House. All rights reserved.

execution_environment.
[7] http://www.arm.com/products/processors/technologies/trustzone/.
[8] https://kantarainitiative.org/confluence/display/IDoT/Concepts+of+Ide
ntity+within+the+Internet+of+Things.
[9] http://www.trustedcomputinggroup.org/resources/trusted_platform_
module_tpm_summary.
[10] http://oauth.net/2/.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 105

[11] http://openid.net/connect/.
[12] http://www.iot-now.com/2015/03/26/31426-securing-the-identity-of-
things-idot-for-the-internet-of-things/Contextual knowledge is power.
[13] Westin, A. F., “Privacy and Freedom,” Washington and Lee Law Review, Vol.
25, No. 1, 1968.
[14] Stoneburner, G., A., Goguen, and A. Feringa, “Risk Management Guide for
Information Technology Systems,” NIST Special Publication 800-30, 2002.
[15] STRIDE Model https://msdn.microsoft.com/en-us/library/ee823878
(v=cs.20).aspx.
[16] https://msdn.microsoft.com/en-us/library/ff648644.aspx.
[17] ����������������������������������������������������������������������
http://www.upm.ro/facultati_departamente/stiinte_litere/conferinte/si-
tul_integrare_europeana/Lucrari6/06%20-%20Hunor,%20Haller,%20Sebe-
sety-Pal.pdf.
[18] ���������������������������������������������������������������������
http://engage.securestate.com/assess-your-security-risk-with-secures-
tates-simple-irisk-equation.
[19] Trusted Computing Group, published documents on IoT security www.
trustedcomputinggroup.org.
[20] https://www.nccgroup.trust/uk/about-us/newsroom-and-events/
blogs/2014/april/security-of-things-an-implementers-guide-to-cyber-se-
curity-for-internet-of-things-devices-and-beyond/.
[21] https://downloads.cloudsecurityalliance.org/whitepapers/Security_
Guidance_for_Early_Adopters_of_the_Internet_of_Things.pdf.
[22] Guidance for Securing IoT Using TCG Technology Version 1.0, Trusted
Computing Group, 2015, http://www.trustedcomputinggroup.org/files/
resource_files/CD35B517-1A4B-B294-D0A08D30868AB3D1/TCG_Guid-
ance_for_Securing_IoT_1_0r21.pdf.
[23] https://www.owasp.org/index.php/IoT_Security_Guidance.

Selected Bibliography
Clymer, C., et al., “iRisk Equation,” Whitepaper, http://insurehub.org/sites/de-
fault/files/reports/Final%20Report.pdf.
Copyright © 2016. Artech House. All rights reserved.

Folk, C., et al., Dun & Bradstreet, “The Security Implications of the Internet of
Things,” AFCEA International Cyber Committee, February 2015.
Garcia-Morchon, O., et al., “A Comprehensive and Lightweight Security Architec-
ture to Secure the IoT Throughout the Lifecycle of a Device Based on HIMMO,”
11th Intl. Symp. on Algorithms and Experiments for Wireless Sensor Networks, ALGO-
SENSORS 2015, Patras, Greece, September 17–18, 2015.
Hernan, S., et al., “Threat Modeling Uncover Security Design Flaws Using the
STRIDE Approach,” MSDN Magazine-Louisville, 2006, pp. 68–75.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 106 IoT Technical Challenges and Solutions

Internet of Things, “IoT Governance, Privacy and Security Issues,” Position Pa-
per, European Research Cluster on the Internet of Things, 2015.
Kamongi, P., M. Gomathisankaran, and K. Kavi, “Nemesis: Automated Architec-
ture for Threat Modeling and Risk Assessment, for Cloud Computing,” 2014 ASE
Big Data, Social Informatics, PASSAT, BioMedCom 2014 Conf., Harvard University,
December 14–16, 2014.
Keoh, S. L., S. S. Kumar, and H. Tschofenig, “Securing the Internet of Things: A
Standardization Perspective,” IEEE Internet of Things Journal, Vol. 1, No. 3, June
2014.
Lee, M. -C., “Information Security Risk Analysis Methods and Research Trends:
AHP and Fuzzy Comprehensive Method,” International Journal of Computer Sci-
ence & Information Technology (IJCSIT), Vol. 6, No. 1, February 2014.
Li, Z., and T. Xin, “Threat Modeling and Countermeasures Study for the Internet
of Things,” Journal of Convergence Information Technology, Vol. 8, No. 5, 2013.
Noel, S., et al., “Measuring Security Risk of Networks Using Attack Graphs,” In-
ternational Journal of Next Generation Computing, July 2010.
Nurse, J. R. C., et al., “Smart Insiders: Exploring the Threat from Insiders Using
the Internet-of-Things,” Intl. Workshop on Secure Internet of Things (SIoT), 2015.
Olsson, T., “Assessing Security Risk to a Network Using a Statistical Model of
Attacker Community Competence,” 11th Intl. Conf., ICICS 2009, Beijing, China,
December 14–17, 2009.
Roman, R., P. Najera, and J. Lopez, “Securing the Internet of Things,” Computer,
Vol. 44, No. 9, September 2011, pp. 51–58.
Sicari, S., et al., “Security, Privacy and Trust in Internet of Things: The Road
Ahead,” Elsevier Journal on Computer Networks, Vol. 76, 2015, pp. 146–164.
Sitnikova, E., and M. Asgarkhani, “A Strategic Framework for Managing Internet
Security,” 2014 11th Intl. Conf. on Fuzzy Systems and Knowledge Discovery (FSKD),
2014, pp. 947–955.
Stoneburner, G., A. Goguen, and A. Feringa, “Risk Management Guide for Infor-
mation Technology Systems,” NIST special publication, Vol. 800, No. 30, 2002, pp.
800–830.
Ukil, A., J. Sen, and S. Koilakonda, “Embedded Security for Internet of Things,”
2nd Natl. Conf. on Emerging Trends and Applications in Computer Science (NCETACS),
2011.
Copyright © 2016. Artech House. All rights reserved.

U.S. Department of Homeland Security guide, “Cyber Security Assessments of

Industrial Control Systems,” Washington, D.C., 2010.
Weber, R. H., “Internet of Things: New Security and Privacy Challenges,” Com-
puter Law & Security Review, Vol. 26, No. 1, 2010, pp. 23–30.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Security and Privacy 107

Whitehouse, O., “Security of Things: An Implementers’ Guide to Cyber-Security for

Internet of Things Devices and Beyond,” NCC group, http://www.nccgroup.trust/
globalassets/our-research/uk/whitepapers/2014-04-09_-_security_of_things_-_
an_implementers_guild_to_cyber_security_for_internet_of_things_devices_
and_beyond.pdf.
Zhang, X., et al., “Information Security Risk Management Framework for the
Cloud Computing Environments,” 2010 IEEE 10th Intl. Conf. on Computer and In-
formation Technology (CIT), 2010, pp. 1328–1334.
Zhao, K., and L. Ge, “A Survey on the Internet of Things Security,” 2013 9th IEEE
Intl. Conf. on Computational Intelligence and Security (CIS), 2013, pp. 663–667.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 4
Sensor Informatics and Business Insights

4.1 Introduction

Cambridge Dictionary defines context as [1]: “the situation within

which something exists or happens, and that can help explain it.”
In the IoT context, sensors pick up the manifestation of something
happening (e.g., fire, earthquake, traffic congestion, faults, disease,
likes or dislikes, object location) via their transducers and measure
the physical features (e.g., temperature, vibration, location, speed,
physiological parameters). The role of sensor informatics is to ana-
lyze the collected sensor signal to understand the corresponding
context and from there deduce about instantaneous and dynamic
state of the cyberphysical system (CPS). This understanding of
the situational behavior of the CPS gives insights and knowledge
into the physical phenomenon, which can help businesses in three
ways:
Copyright © 2016. Artech House. All rights reserved.

1. They can help optimize the business operations and run busi-
nesses more efficiently.

109

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 110 IoT Technical Challenges and Solutions

2. They help in a deeper understanding of the physical process-

es and customers by adding new features and services to the
business line.
3. They help modeling the business processes and predicting
future events by removing the risk of business and business
planning.

The whole process of deriving context from sensors and draw-

ing business insights out of it is an extremely complex process with
lot of dependence on the physical context, type of sensors and na-
ture of insights. The whole sensor informatics process can be de-
picted as a knowledge pyramid [2] (as shown in Figure 4.1).
There are three major components of sensor informatics as out-
lined in the data, information, knowledge, and wisdom (DIKW)
pyramid of Figure 4.1:

• Sensor signal processing: It helps in converting raw sensor

data to information.
• Semantic interpretation of processed information: It helps in
converting information to knowledge.
Copyright © 2016. Artech House. All rights reserved.

Figure 4.1 The DIKW pyramid.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 111

• Business insights from interpreted knowledge: It helps in con-

verting knowledge to wisdom.

These are explained in detail next.

4.2 Sensor Signal Processing

As depicted in Figure 4.2, the sensor signal processing workflow

consists of four major blocks: signal acquisition and condition-
ing, signal representation, feature extraction, and inference. Each
of these blocks is explained in detail next with an example appli-
cation use case of activity detection from accelerometer signals in
mobile phones and wearable devices.

4.2.1 Signal Acquisition and Conditioning

4.2.1.1 Sampling
Sampling of the analog sensor signal is the first step for digitiza-
tion. Sampling frequency needs to more than twice the maximum
frequency of the sensor signal (as per the Nyquist sampling theo-
rem [3]) to avoid aliasing [4]. In real sensor systems, it is typical to
have unequal sampling rate as the sampling happens only when
Copyright © 2016. Artech House. All rights reserved.

Figure 4.2 Sensor signal processing workflow.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 112 IoT Technical Challenges and Solutions

the sensor signal changes. Because most sensor signal processing

typically works on equal sampling rate signals, this unequal sam-
pling rate may need to be made uniform using techniques of resa-
mpling and interpolation [5].

4.2.1.2 Matched Filtering

The matched filter is the optimal linear filter for maximizing the
signal-to-noise ratio (SNR) in the presence of additive stochastic
noise [6]. It is typically used to tune the filter to the sensor sig-
nal so that any noise that is outside the signal band is filtered out.
Typically matched filters are designed as lowpass or highpass or
bandpass filter depending upon sensor signal characteristics. Digi-
tal filtering techniques [7] like finite impulse response (FIR) [8] and
infinite impulse response (IIR) [9] are useful in designing suitable
matched filters. FIR filters are easier and stable to design but need
more delay taps to achieve the desired frequency response; how-
ever, the higher number of delay taps can result in higher latency.
IIR filters have a smaller number of taps and hence smaller latency,
but they may result in unstable behavior.

4.2.1.3 Adaptive Filtering

Matched filtering does not provide the necessary performance
when the noise falls in the same frequency band as the desired
signal. Adaptive filtering can help in such scenarios. An adaptive
filter [10] is a system with a linear filter that has a variable param-
eter transfer function where these parameters can be adjusted ac-
cording to an optimization algorithm [11]. An adaptive filter can
be designed using FIR or IIR filters [12]. There are two basic kinds
of adaptive filter: least-mean-square (LMS) and recursive-least-
square (RLS). LMS adaptive filters are used for FIR filter architec-
ture and RLS adaptive filters are used for IIR filter architecture. In
Copyright © 2016. Artech House. All rights reserved.

other words, an adaptive filter can be regarded as a digital filter

that self-adjusts its coefficients to minimize an error function. This
error function (or the cost function) is the distance between the ref-
erence or desired signal and the output of the adaptive filter [13].
Typically, the cost function optimization is done in least-squares
sense, hence the name LMS or RLS [14].

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 113

4.2.1.4 Outlier Detection

An outlier is typically a data point that is statistically distant from
other data points; it typically occurs due to error in the sensor data
that needs to be excluded [15]. Outliers in a sensor signal stream
typically indicate measurement error; they either need to be dis-
carded or subsequent processing blocks in the workflow should
use statistics that are robust to outliers. For the removal of outliers,
the distribution can be modeled as mixture of two distinct sub-
populations, representing correct trials and measurement errors.
A generalized method to detect outliers is to find a set of features
in the signal that can separate between regular values and outliers
[16].

4.2.1.5 Applications of Signal Acquisition and Conditioning

Resampling
Many times, the physical signals have unequal sampling frequen-
cies (due to hardware limitations of reporting sensor data only
when they change) or have higher or lower than necessary sam-
pling frequencies. Typically, all signal informatics algorithms work
on uniformly sampled data; in the case of working with multiple
sensors, there is the need to put all the sensor signals in the same
sampling rate so that they can be fused. Through the use of sam-
pling and interpolation algorithms, signal informatics can help in
acquiring the digitized signal into the system in the right sampling
rate. For example, for accelerometer signals, the hardware is typ-
ically designed to send data only when there is a change in the
value of acceleration, thereby making it a variable sampling rate
system that needs resampling to make it uniform sampling-rate.

Noise Cleaning
Copyright © 2016. Artech House. All rights reserved.

Sensor data, in all practical scenarios, is corrupted by noise due

to ambient and sensor inaccuracies. Efficient matched filtering
techniques and adaptive filtering techniques can be used to clean
this noise. For example, an accelerometer signal used for activity
detection of people can use a matched bandpass filter with a pass-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 114 IoT Technical Challenges and Solutions

band of 0.5 Hz to 4 Hz, as most of the frequencies while standing,

walking, or running occur within this frequency band [17]. Addi-
tionally, adaptive filtering can be used for canceling noise from ac-
celerometer signals [18], which can arise from sensor irregularities
and limitations.

Anomaly Detection
Sometimes, even after cleaning the sensor data, the artifacts of the
unwanted signal remain at an unacceptably high level, thereby
corrupting the analytics insights. In such cases, it is better to throw
away such data than process it. Statistical processing-based anom-
aly detection techniques can be used for such requirements. In the
activity detection use case, this kind of unwanted signal can be
generated if the wearable device or the mobile phone is not firm-
ly held against the body or if there are irregularity in the sensor
electronics.

4.2.2 Signal Representation

Once the time-domain signal is preprocessed and cleaned, it can be

either be processed in the time domain itself or it can be processed
after transforming it to another domain like frequency or wavelet.
Typical signals have a lot of redundancy and hence techniques of
dimensionality reduction needs to be applied to reduce the effec-
tive signal space.

4.2.2.1 Frequency Domain

The Fourier transformation (FT) [19] transforms a signal from the
time domain to a frequency domain. The discrete version of FT is
known as a discrete Fourier transform (DFT) [20]. The most popu-
lar implementation of DFT is fast Fourier transform (FFT) [21]. The
transforms result in an output vector of frequencies ranging from
Copyright © 2016. Artech House. All rights reserved.

low-frequency to high-frequency coefficients. Frequency-domain

representation helps in doing efficient matched filtering to remove
unwanted frequency components in a signal. They are also useful
in finding out multiple repetitive patterns in the signal and the rate
at which the repetition occurs. The inverse discrete Fourier trans-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 115

form (IDFT) or the inverse fast Fourier transform (IFFT) is used to

convert back from the frequency domain to the time domain.

4.2.2.2 Wavelet Domain

Wavelet transforms are a generic type of transform where instead
of using sine and cosine as basic functions as in the Fourier trans-
form, more generic basis functions are used [22]. As in DFT, there is
a discrete version of wavelet transform available: discrete wavelet
transform (DWT). DWT is particularly useful when the frequency
content of the signal becomes time-varying and there is a need to
represent different frequencies in different resolutions [23]. In com-
parison to the Fourier transformation that loses the time informa-
tion of the data and transforms the data globally, DWT preserves
the time dimension and transforms the data locally, which leads to
a faster calculation.

4.2.2.3 Statistical Processing

To handle the large amount of data for processing, statistical pro-
cessing techniques can be applied to the signal to effectively repre-
sent the significant information-bearing part of the signal. We can
do simple statistical processing [24] to represent the time-domain
signal with some of its first- and second-order statistics like mean,
median, mode, variance, and correlation. As another option, the
data is transformed to frequency domain via DFT and DWT [25]
and statistical processing is applied there to effectively represent
the signal sequence. There are also windowed statistical process-
ing techniques like piecewise aggregation approximation (PAA)
[26] and symbolic aggregate approximation (SAX) [27], which can
provide good dimensionality reduction for specific kind of signals
and applications.
Copyright © 2016. Artech House. All rights reserved.

4.2.2.4 Applications of Signal Representation

Compression
There is a large cost associated with sensor data transport and
storage. Hence, there is need for efficient lossless and lossy com-
pression techniques. The transform and statistical processing tech-
niques effectively represent the significant part of the signal, there-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 116 IoT Technical Challenges and Solutions

by providing a potential way to efficiently compress the signal.

Taking the example of the accelerometer-based activity detection,
frequency analysis of the accelerometer shows that there are only a
few frequency bands that are significant; the rest can be removed.

Visualization
Visualization of various signal components can be the first step to-
wards having some insight into the signal behavior, which, in turn,
can lead to efficient signal analytics development. However, due
to the large amount of information available in the signal space,
sometimes it is extremely difficult to manually visualize all parts
of the signal. Transform and statistical processing techniques can
help in breaking down the signal into smaller, nonredundant com-
ponents, which, when visualized in a graph, can provide much
better insights about the signal and underlying events. This can
be the precursor to visual analytics, described later in this chapter.

4.2.3 Feature Extraction and Inference

After preprocessing of the raw data and the dimensionality reduc-

tion, features (e.g., interesting signal properties that correlate with
the physical event captured by a sensor) have to be extracted. Fea-
ture extraction describes the process of extracting representative
features from the sensor data. Inference describes methods that use
the extracted features to gain more information about the data and
infer knowledge from that. They include detection of specific pat-
terns in the signal, identification of such patterns, estimation of
signal features from previous signals, and fusion of multiple fea-
tures from multiple signals. Many times, the feature extraction and
inference needs to be done on-the-fly as the sensor signal flows
in stream; special techniques called stream processing are used in
conjunction with existing algorithms for such systems.
Copyright © 2016. Artech House. All rights reserved.

4.2.3.1 Detection
Statistical signal processing techniques are used detect the pres-
ence or absence of a particular pattern in the sensor signal. It can
be regarded as having a single hypothesis and inferencing the hy-
pothesis to be right or wrong based on certain patterns present
in the sensor signal [28]. Signal detection theory is a very mature

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 117

field that developed around communication and radar systems.

The same techniques can be used in sensor signal informatics. In a
detection problem, there are four possible outcomes: hit (the pat-
tern is present and system detects it), miss (the pattern is present,
but the system misses detection), false alarm (the pattern is absent,
but the system detects it), and correct rejection (the pattern is ab-
sent and the system also does not detect it) [29]. While the hits and
correct rejections are desirable, false alarms and misses are not de-
sirable. The whole detection theory is built around maximizing the
former and minimizing the latter. It usually involves correlating
certain features of the sensor signal with the desired pattern and
then applying variants of thresholding criteria to detect presence
and absence of the desired pattern. Most of the challenges in detec-
tion arise from noisy signals; there are specific techniques like con-
stant false alarm rate (CFAR) [30] and voting (M of N) [31], which
are used to reliably detect under noisy conditions.

4.2.3.2 Identification
Identification can be considered as detection of identity of instanc-
es that works on multiple patterns instead of one pattern. It can be
regarded as having multiple hypothesis and inferencing one of the
hypotheses to be right or wrong based on specific patterns present
in the sensor signal [32]. It also involves correlating certain features
of the sensor signal with the desired pattern, but instead of apply-
ing thresholding techniques as in detection, identification systems
employ pattern recognition techniques [33].

4.2.3.3 Estimation
Signal estimation can be thought of inferring value of unknown
states observed by the sensors under noisy conditions [34]. While
detection or identification can be regarded as proving a discrete
Copyright © 2016. Artech House. All rights reserved.

set of hypotheses as right or wrong, estimation techniques can be

regarded as a continuous set of hypotheses that are almost always
wrong, with the focus being minimization of the estimation er-
ror. Popular estimation techniques include least square estimation
(LSE) [35], maximum likelihood estimation (MLE) [36], Bayesian
minimum mean squared error (BMSE) [37], and Kalman filtering
(KF) [38].

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 118 IoT Technical Challenges and Solutions

4.2.3.4 Fusion
Sensor information fusion can be thought of as techniques and
tools for exploiting the synergy in the information acquired from
multiple sensors [39]. Integration of multisensory information can
help in making better inferences about the physical event. Multi-
sensor data fusion can result in better accuracy and noise perfor-
mance exploiting the sensor redundancy and sensor correlation.
The fusion can happen at the data level (mix all raw sensor signals
together), feature level (extract relevant features from each sensor
signal and then fuse), or decision or inference level (infer from each
sensor signal and then fuse the inference through techniques like
majority voting). Challenges on multisensor fusion mainly emerge
around different noise distribution of different sensor signals and
their nonstationarity [40].

4.2.3.5 Stream Processing

A lot of applications require that the above kind of feature extrac-
tion and inferencing (detection, identification, recognition, estima-
tion) happen on live data streams as the live sensor data samples
become available from the acquisition system. Such systems have
specific requirements like handling missing or delayed samples
and making algorithms low computational complexity to handle
low-latency, high-sampling rate systems [41]. When multisensor
data streams are handled together, the synchronization of the dif-
ferent sensor signals becomes an additional challenge. Stream pro-
cessing systems typically need specialized software architecture
[42] around in-memory processing, low-latency processing, data
pipelines, and scheduling [43]. They also need streaming versions
of algorithms instead of traditional batch versions to make them
low computational complexity and low memory [44].
Copyright © 2016. Artech House. All rights reserved.

4.2.3.6 Applications of Feature Extraction and Inference

Pattern Recognition
This is the detection of specific patterns or discovery of recurrent
patterns that have causation with physical events captured by the
sensor. For example, the accelerometer data from a smart phone
may have different specific patterns in the x, y, and z-axes when

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 119

the user is stationary versus walking versus running; these can be

used to detect activity of the user.

Prediction
Once a clean sensor data is available, in many use cases, the main
requirement from informatics becomes modeling of the events to
predict them in the future. The estimation techniques outlined
above can provide the necessary statistical models (LSE, ML,
BMSE, and so forth). The Kalman filtering techniques addition-
ally allow for state-space modeling through state transition ma-
trices, which make it easy to model time series that are governed
by physical equations (like rectilinear motion). For example, in the
accelerometer-based activity detection use case, use of Kalman fil-
ter in conjunction with the activity detector can predict the user’s
motion, thereby making it useful for indoor localization kind of
applications.

4.3 Semantic Interpretation of Processed Information

Once the data from sensors are cleaned up and processed to gener-
ate information, the next important step becomes converting the
processed information into knowledge. The critical part of this
step is to interpret the information in a semantic manner that corre-
lates the information to physical events and maps the cyber world
findings into interpretations in the physical world. As depicted in
Figure 4.3, the workflow of semantic interpretation has three ma-
jor components: machine learning, rule engine, and reasoning. It
should be noted that the machine learning can be regarded as part-
ly belonging to inference, described previously. However, the out-
come of machine learning always provides a powerful semantic
interpretation of the data; hence, we thought it prudent to describe
Copyright © 2016. Artech House. All rights reserved.

machine learning here. All the three components are described in

detail next.

4.3.1 Machine Learning

Machine learning has evolved from the pattern recognition and

computational learning theory in artificial intelligence [45]. Ma-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 120 IoT Technical Challenges and Solutions

Figure 4.3 Semantic interpretation workflow.

chine learning explores the study and construction of algorithms

that can learn from and make predictions on data. Such algorithms
operate by learning a model from example inputs to make data-
driven predictions or decisions expressed as outputs on new set
of data; they do not follow strictly static program instructions. The
key feature is the ability to build models from example inputs;
this allows the application developer to semantically describe the
example input data mapped to physical events. This semantic
knowledge is used by machine learning algorithms to learn and
build models, which can be used to classify, cluster, and predict
physical events. Machine learning can be broadly classified into
four models: supervised, unsupervised, semisupervised, and rein-
forcement [46]. These are described next.

4.3.1.1 Supervised Learning

In supervised learning, the machine is given example inputs with
mapping to desired annotated outputs (also known as training
data); the machine tries to learn a general rule that maps inputs to
Copyright © 2016. Artech House. All rights reserved.

outputs. There are supervised learning algorithms [47] like sup-

port vector machine (SVM) classification, Naïve-Bayes classifica-
tion, least-squares regression, and Ada-boost gradient boosting,
which learns patterns from training data to predict the values of
the label on additional unlabeled data; this can be used to classify
events based on sensor data or predict events.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 121

4.3.1.2 Unsupervised Learning

In unsupervised learning, no labels are given to the output; the
algorithm needs to find structure in its input. Unsupervised learn-
ing can be used to discover hidden patterns in data or finding rel-
evant features. Popular unsupervised learning algorithms include
decision trees, self-organizing maps, nearest-neighbor mapping,
k-means clustering, and singular value decomposition. These al-
gorithms can be used to segment, recommend, and identify data
outliers.

4.3.1.3 Semisupervised Learning

Semisupervised learning lies in between supervised and unsuper-
vised learning, where the example inputs with desired mapping to
annotated outputs are there for some of the data but not for all. It
typically uses a small amount of labeled data with a large amount
of unlabeled data.

4.3.1.4 Reinforced Learning

Here the learning happens continuously in a dynamic environ-
ment; the system needs to automatically judge its closeness to the
goal and there is not explicit way to evaluate the closeness to the
goal. Reinforcement learning uses trial and error to discover which
actions yield the greatest rewards. One example of reinforced learn-
ing is learning to play a game by playing against an opponent.
Machine learning uses many of the same algorithms and tech-
niques as data mining, but there is a fundamental difference be-
tween the two: while data mining discovers previously unknown
patterns and knowledge, machine learning is used to learn from
known patterns and knowledge and apply them to other data for
making decisions. In other words, while data mining uses known
properties of data to discover unknown properties of the same
Copyright © 2016. Artech House. All rights reserved.

data, machine learning uses known property of data to find other

known properties with a new instance of similar data. Machine
learning is better suited for IoT analytics because there is always
knowledge and sensor data available on known physical events
and sensors are always producing new data points.
The increased computing power has also promoted a new class
of algorithms called neural networks that loosely model the way

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 122 IoT Technical Challenges and Solutions

our brain works with many layers of networked decision-making.

These are called deep neural networks [48]. Advances in artifi-
cial intelligence (AI) have given rise to artificial neural networks
(ANN), which can potentially model any kind of relationship
within a data set. Deep multilayered ANNs (also popularly known
as deep learning [49]) use special types of neural networks to learn
complicated patterns in large amounts of data. They need a lot of
computing power and recent advances in the computing systems
have contributed greatly in proliferation of deep learning. Deep
learning has shown lot of promise in image data; there needs to be
further development in the field to apply deep learning effectively
in sensor data. It should be noted that deep learning needs a huge
dataset to learn and big data techniques need to be used to handle
such huge data.

4.3.1.5 Applications of Machine Learning

Clustering and Classification

The unsupervised method of segregating sensor data into multiple
clusters without any a priori knowledge of data is known as clus-
tering. This is useful when the sensor data produced by specific
events have distinct patterns as compared to normal events. When
there are known labeled data available depicting various physi-
cal events, supervised learning can be used to learn the rules of
classification into different events; this learned model can be used
to classify new sensor data and identify the physical events from
them. In the activity detection use case, machine learning on accel-
erometer data can be used to classify different activities like sitting,
standing, lying down, walking, and running.

Regression and Prediction

Labeled sensor data can be used to create numerical regression
Copyright © 2016. Artech House. All rights reserved.

models between physical events (inputs) and sensor data (out-

puts). This numerical model can be used to predict future events
from the captured sensor data. Accelerometer-based motion esti-
mation can be fitted into a regression model to predict locations of
users.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 123

4.3.2 Rule Engine

A rule engine is a software system that runs a combination of rule

rules in an execution environment. The rules can come from reg-
ulations, policies, or domain knowledge. Rule engines typically
support rules, facts, exclusion, preconditions, and other function-
alities. Rule engine software provides the ability to register, define,
classify, and manage all the rules [50]. In a nutshell, rule engines
allow for putting logic combining detected events from sensor
data through processing blocks described previously to include
or exclude a higher-level event. Most rule engines support a data
abstraction layer representing the business entities and relation-
ships supporting different file formats for the model. Rules can be
written in standard programming languages or custom languages
or architectures like Business Process Execution Language (BPEL)
[51].
There are three types of rules supported by rule engines [52].

4.3.2.1 Inference Rules

Inference are used to represent system state behaviors of the type
IF <condition> THEN <action>. They can be regarded as a com-
plex chain of if-then-else statements operating on inferences ob-
tained from sensor data. The rules can be executed either when the
applications invoke them or when one of the attributes of the rule
decision (inference from sensor data processing) is changed. These
kinds of rule engines are typically stateless.

4.3.2.2 Reaction Rules

The reactive rule engines go one step further; after detecting an
event via inference rules, they process event patterns to decide
what action to take and then invoke relevant action tasks. These
kinds of rule engines are typically state-driven.
Copyright © 2016. Artech House. All rights reserved.

4.3.2.3 Goal-Driven Rules

The inference rules and reaction rules are often referred to as for-
ward-chaining rules as they process information as they come in to
take decision. There exists another class of rules called backward-
chaining rules; here the rules try to resolve facts to fit a particular

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 124 IoT Technical Challenges and Solutions

goal. In other words, they try to find out if something exists based
on existing information. They are also known as goal-driven rules.
Backward chaining evaluates different conditions simultaneously
to ascertain which one is closest to the desired goal.

4.3.2.4 Applications of Rule Engines

Inferencing
Once a set of information has been extracted from the sensor data,
inferencing rules allow processing of that to draw higher-level in-
ference regarding the physical event affecting the sensor data. For
example, rule engine-based inferencing can be used on accelerom-
eter-based activity detection to detect micro-activities like sitting
or lying down while stationary, slow or brisk while walking, and
jogging or fast running while running.

Complex Event Processing

There is another class of applications called complex event pro-
cessing (CEP) [53], which uses reaction rules to effectively map a
complex logic from detected events to desired action. CEP is typi-
cally performed in real time and often employs stream process-
ing, described previously, before applying the rules. In the activity
detection use case, CEP can be used to create alerts on user abnor-
malities; for example, the alert can be generated if the user is about
to fall.

4.3.3 Reasoning

A reasoning system can generate conclusions from available

knowledge using logic. They are an integral part of artificial in-
telligence and knowledge-based systems [54]. Reasoning systems
can either be in interactive mode with human-in-loop intervention
Copyright © 2016. Artech House. All rights reserved.

or in batch mode that is fully automated. The first practical ap-

plication of automated reasoning was expert systems focusing on
domain-specific problems like medical diagnosis or machine fault
diagnosis. We can regard reasoning as the use of existing knowl-
edge to draw conclusions and create explanations [55]. There are
three kinds of reasoning systems as outlined next.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 125

4.3.3.1 Deductive
Deductive reasoning is the process of reasoning from one or more
propositions or facts to reach a logically certain conclusion. Deduc-
tive reasoning is a top-down logic in which a conclusion is reached
by assertion of a general rule leading towards a guaranteed spe-
cific conclusion. For example, if x = 10 and y = 11, we can deduce
that x + y = 21.

4.3.3.2 Inductive
Inductive reasoning starts with specific observations and tries
to move towards a generalized most likely conclusion; it weighs
the accumulated evidence to reach the most likely outcome. Here
the premises are viewed as supplying the necessary evidence for
the likely conclusion. The conclusion of an inductive argument is
probable, based upon the evidence given. An example of induc-
tive reasoning is scientific research-gathering evidence, identify-
ing patterns, and forming a hypothesis or theory to explain the
observation.

4.3.3.3 Abductive
Abductive tries to infer a theory from an observation; it tries to find
the simplest and most likely explanation that justifies the observa-
tion. In abductive reasoning also, the premises do not guarantee
the conclusion. Abductive reasoning usually uses an incomplete
set of observations to find the likeliest possible explanation for the
set. The abductive reasoning process usually needs creativity and
intuition, thereby making it most difficult to implement in com-
puter systems. An example of abductive reasoning is a medical di-
agnosis in which, given a set of symptoms and signs, doctors try to
determine the most likely cause (disease).
Copyright © 2016. Artech House. All rights reserved.

4.3.3.4 Applications of Reasoning Systems Mostly Involving

Repair and Diagnosis Systems

Diagnosis
Diagnosis of fault in machines or disease in human beings is com-
plex hybrid process that includes straightforward deduction (if x
is the symptom, then y must be the cause), induction (if x is the
symptom, then y is the most likely cause among n possible op-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 126 IoT Technical Challenges and Solutions

tions), and abduction (intuitively start with y to the most likely

cause and show that it indeed creates x symptom).

Expert Systems
Expert systems can be regarded as automated diagnostic systems
in which rules and reasoners with domain knowledge are embed-
ded into the system and the system tries to recommend a possible
diagnosis or a set of diagnoses to the doctor or mechanic. It can
help experts like doctors and mechanics to make quick diagnostic
decisions.

4.4 Business Insights from Interpreted Knowledge

Drawing business insights from the interpreted knowledge that

comes out of the signal informatics chain is the most important
step to create value in the IoT application space. This is elaborated
upon in Chapter 6; however, it is worthwhile to look at the kind of
business insights that analytics on processed sensor information
can bring. We describe three important analytics methods next;
these are also elaborated upon in Figure 4.4.

4.4.1 Visual Analytics

Visual analytics [56] is the science of analytical reasoning facilitated

by interactive visual interfaces. It can be termed as the science and
Copyright © 2016. Artech House. All rights reserved.

Figure 4.4 Business insights workflow.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 127

the art of the analysis of overwhelming amounts of disparate, con-

flicting, and dynamic information that can emanate from sensors
[57]. Visual analytics can be seen as an integral approach combin-
ing visualization, human interpretation, and data analysis. Visual
analytics need to take care of human factors, including the areas of
cognition and perception. From a technology perspective, visual
analytics relate to the areas of information visualization, computer
graphics, and data analysis (information retrieval, data manage-
ment and knowledge representation, and data mining) and usu-
ally is a complex interaction between visualization, data, analytics,
and knowledge models with human-in-loop.

4.4.2 Modeling and Simulation

The information and knowledge extracted from sensor data not

only cannot be used to detect or predict events, but they also can be
used to create new models. Modeling and simulation use physical,
mathematical, or logical models of a system as a basis for simula-
tions; it is a mean to develop data as a basis for decision-making. It
helps to get information about how something will behave in real-
ity without actually testing it in real life. Modeling and simulation
can be used for what-if analysis during the design phases of sys-
tems. It can support early evaluation and optimization of designs
and ongoing verification as changes occur in complex IoT-enabled
systems [58].

4.4.3 Optimization and Planning

Finally, the biggest business value-add from the information and

knowledge derived from sensor informatics is achieved when it
helps in designing better business processes in terms of cost, ef-
ficiency, or features, thereby directly affecting the return on invest-
ment (ROI). Optimization and planning have been developed and
Copyright © 2016. Artech House. All rights reserved.

matured in the standard operation research (OR) [59] field for a

long time; it can be applied easily to information and knowledge
extracted from the sensor data via sensor informatics. It uses math-
ematical modeling, statistical analysis, mathematical optimization,
and graph-based analytics to arrive at optimal or near-optimal so-
lutions to complex decision-making problems, thereby helping in
the creation of optimal planning tools. The most obvious applica-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 128 IoT Technical Challenges and Solutions

tion of IoT-based optimization and planning is in the supply chain

[60] and logistics or, in general, any system that encompasses the
flow of objects in a network.

4.5 Data and Algorithm Marketplaces as New Business Models

The most important aspect of any IoT system is the data being col-
lected. However, traditionally data will be in silos of individual
verticals or organizations. To deliver value, the data from different
sources and systems need to be aggregated together to gain far
greater insight. In this regard, this is the big push towards open
data, especially governments and smart cities all over the world
providing data sets to the public from their departments through
Web sites like data.gov. In many cases, this is not real-time data but
a slightly older data. This data needs to be complete, primary, time-
ly, accessible, machine-processible, nondiscriminatory, nonpropri-
etary, and license-free. Open data encourages transparency and
can generate new products and services and promote efficiency
due to better visibility. However, at this point, beyond smart cities
and government data, there is very little open data available in IoT
space for enterprises. It is obvious that such data cannot be made
available for free to everybody; hence, there is a case for creating a
data marketplace [61]. In a data marketplace [62], application de-
velopers need to buy data to create value-added applications on
it. When the business value created (in terms of operational cost
reduction or process optimization or improvement in efficiency)
is worth more than the data cost (which will be the usual case for
value creation via sensor informatics and analytics), it creates a vi-
able and scalable business model that is democratized. However,
ensuring the security and maintaining the privacy of such data are
important concerns that need to be addressed; in Chapter 3, we
Copyright © 2016. Artech House. All rights reserved.

explored some of the security and privacy issues.

Another interesting aspect to address for creating scalable val-
ue-added IoT systems is creation of an analytics algorithm reposi-
tory. All the people having the right domain knowledge about the
business problem and hence being in a positon to conceptualize
the right value-added business application may not have techni-
cal knowledge about the required sensor informatics and analytic
algorithms required to draw the right insights from the sensor data

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 129

for creating such an application. However, such algorithms can be

crowdsourced from a group of algorithm experts and stored in a
repository [63] that is available as working and executable algo-
rithm code to the application developers [64]. To create a sustain-
able model, such a repository can also be created in a marketplace
framework so that the algorithm providers have some incentive
and reward [65]. We explore this concept of analytics as a service in
detail in Chapter 6. However, as is true for any such crowdsourced
system, trust and reputation of the algorithm provider become im-
portant factors.
Data marketplace, along with the algorithm marketplace, has
the potential to create an economic and sustainable model of IoT
analytics-based application development that capitalizes on open-
ness and reuse to deliver real value and hence can address the big-
gest challenge plaguing the IoT ecosystem: that of creating killer
applications that justifies the ROI.
In summary, in this chapter, we covered the different aspects
of the most important part in the IoT system: sensor signal infor-
matics. We have described how sensor informatics is important
to understand the context and state of physical systems and have
introduced three major sections of signal informatics: sensor sig-
nal processing (which is about acquisition, clean up, and represen-
tation of sensor signals to convert it into information), semantic
interpretation of sensor information (which is about relating and
interpreting sensor information into physical system knowledge),
and business insights from interpreted knowledge (which is about
creating actionable insights from the knowledge already interpret-
ed from sensor signals). Finally, we outlined the concept of data
and algorithm marketplace as the future architecture for crowd-
sourced sensor informatics and discussed the challenges thereof.
Sensor informatics can be regarded as the brain of an IoT system
and hence responsible for all the intelligence and smartness usu-
Copyright © 2016. Artech House. All rights reserved.

ally associated with IoT.

References
[1] http://dictionary.cambridge.org/dictionary/english/context.
[2] http://inls151f14.web.unc.edu/files/2014/08/rowleydikw.pdf.
[3] https://www.cs.cf.ac.uk/Dave/Multimedia/node149.html.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 130 IoT Technical Challenges and Solutions

[4] http://redwood.berkeley.edu/bruno/npb261/aliasing.pdf.
[5] http://in.mathworks.com/help/signal/ref/resample.
html?s_tid=gn_loc_drop.
[6] http://ee.eng.usm.my/eeacad/mandeep/EEE436/chp%203.pdf.
[7] http://learn.mikroe.com/ebooks/digitalfilterdesign/chapter/
basic-concepts-of-digital-filtering/.
[8] http://dspguru.com/dsp/faqs/fir/basics.
[9] http://dspguru.com/dsp/faqs/iir/basics.
[10] http://www.ece.mcmaster.ca/faculty/reilly/coe4tl4/adaptive%20
filters%20Scot%20Douglas.PDF.
[11] http://in.mathworks.com/help/dsp/ug/overview-of-adaptive-filters-
and-applications.html.
[12] http://www.ee.lamar.edu/gleb/adsp/Lecture%2007%20-%20
Adaptive%20filtering.pdf.
[13] https://www.lsv.uni-saarland.de/fileadmin/teaching/dsp/ss15/
DSP2016/ADF_intro.pdf.
[14] https://courseware.ee.calpoly.edu/~fdepiero/fdepiero_dsp_notes/
notes%20%26%20materials/Adaptive_Filtering.pdf.
[15] http://www.itl.nist.gov/div898/handbook/eda/section3/eda35h.htm.
[16] http://eprints.whiterose.ac.uk/767/1/.
[17] https://arxiv.org/ftp/arxiv/papers/1107/1107.4414.pdf.
[18] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3270843/.
[19] http://www.thefouriertransform.com/.
[20] http://www.robots.ox.ac.uk/~sjrob/Teaching/SP/l7.pdf.
[21] http://www.dspguide.com/ch12/2.htm.
[22] http://gwyddion.net/documentation/user-guide-en/wavelet-transform.
html.
[23] http://www.cse.iitm.ac.in/~vplab/courses/CV_DIP/PDF/Lect-wavelet_
filt.pdf.
Copyright © 2016. Artech House. All rights reserved.

[24] https://reference.wolfram.com/language/guide/Statistics.html.
[25] http://www.ncbi.nlm.nih.gov/pubmed/12762451.
[26] http://jmotif.github.io/sax-vsm_site/morea/algorithm/PAA.html.
[27] https://cs.gmu.edu/~jessica/sax.htm.
[28] http://www.ece.uic.edu/~devroye/courses/ECE531/lectures/b2c3.pdf.
[29] http://www-psych.stanford.edu/~lera/psych115s/notes/signal/.
[30] http://www.ece.iisc.ernet.in/~cmurthy/E1_244/Slides/Rohling.pdf.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 131

[31] http://ece.wpi.edu/radarcourse/Radar%202010%20PDFs/Radar%20
2009%20A_6%20Detection%20of%20Signals%20in%20Noise.pdf.
[32] http://research.microsoft.com/en-us/um/redmond/events/
cognetsummit/presentations/stancil.pdf.
[33] http://www.sis.pitt.edu/spring/patterns/node11.html.
[34] http://www.ece.uic.edu/~devroye/courses/ECE531/lectures/intro.pdf.
[35] http://stat.ethz.ch/~geer/bsa199_o.pdf.
[36] http://times.cs.uiuc.edu/course/410/note/mle.pdf.
[37] http://www.lx.it.pt/~mtf/learning/Bayes_lecture_notes.pdf.
[38] http://stanford.edu/class/ee363/lectures/kf.pdf.
[39] https://www.ieee.li/pdf/viewgraphs/multisensor_data_fusion.pdf.
[40] http://preserve.lehigh.edu/cgi/viewcontent.
cgi?article=2318&context=etd.
[41] http://cs.brown.edu/~ugur/8rulesSigRec.pdf.
[42] http://edoc-intranet.epfl.ch/files/content/sites/lts4/files/frossard/
publications/pdfs/icassp06.pdf.
[43] http://db.csail.mit.edu/pubs/cidr07.pdf.
[44] http://www.nist.gov/smartspace/downloads/ICME08_paper.pdf.
[45] http://alex.smola.org/drafts/thebook.pdf.
[46] http://www.sas.com/it_it/insights/analytics/machine-learning.html.
[47] http://www.cs.huji.ac.il/~shais/UnderstandingMachineLearning/
understanding-machine-learning-theory-algorithms.pdf.
[48] http://neuralnetworksanddeeplearning.com/.
[49] http://deeplearning.net/.
[50] https://en.wikipedia.org/wiki/Business_rules_engine.
[51] https://www.oasis-open.org/committees/tc_home.php?wg_
abbrev=wsbpel.
[52] http://www.brcommunity.com/b410.php.
Copyright © 2016. Artech House. All rights reserved.

[53] http://www.complexevents.com/about-real-time-intelligence-and-cep/.
[54] https://www.butte.edu/departments/cas/tipsheets/thinking/reasoning.
html.
[55] http://research.microsoft.com/pubs/192773/tr-2011-02-08.pdf.
[56] http://www.visual-analytics.eu/faq/.
[57] http://vis.pnnl.gov/pdf/RD_Agenda_VisualAnalytics.pdf.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 132 IoT Technical Challenges and Solutions

[58] http://www.automationworld.com/all/internet-things-and-importance-
modeling-and-simulation.
[59] http://mathworld.wolfram.com/OperationsResearch.html.
[60] http://web.mit.edu/~sgraves/www/papers/Graves_Kletter_Hetzel.pdf.
[61] http://radar.oreilly.com/2012/03/data-markets-survey.html.
[62] http://www.bigdataexchange.com/homepage/buyersellers/.
[63] https://www.datamapper.com/algorithms.
[64] https://algorithmia.com/.
[65] http://www.gartner.com/smarterwithgartner/the-algorithm-economy-
will-start-a-huge-wave-of-innovation/.

Selected Bibliography
Ackoff, R. L., “From Data to Wisdom,” Journal of Applied Systems Analysis, Vol. 16,
1989, pp. 3–9.
Allen, J., “Photoplethysmography and Its Application in Clinical Physiological
Measurement,” Physiol. Meas., Vol. 28, 2007, pp. R1–R39.
Arel, I., D. C. Rose, and T. P. Karnowski, “Deep Machine Learning: A New Fron-
tier in Artificial Intelligence Research [Research Frontier],” IEEE Computational
Intelligence Magazine, Vol. 5, No. 4, 2010, pp. 13–18.
Arora, A., et al. “Exscal: Elements of an Extreme Scale Wireless Sensor Network,”
Proc. 11th IEEE Intl. Conf. on Embedded and Real-Time Computing Systems and Ap-
plications, 2005, pp. 102–108.
Chandel, V., et al., “AcTrak: Unobtrusive Activity Detection and Step Counting
Using Smartphones,” in Mobile and Ubiquitous Systems: Computing, Networking,
and Services, New York: Springer, 2013, pp. 447–459.
Coutaz, J., et al., “Context Is Key,” Communications of the ACM, Vol. 48, No. 3, 2005,
pp. 49–53.
Daubechies, I., “The Wavelet Transform, Time-Frequency Localization and Signal
Analysis,” IEEE Trans. on Information Theory, Vol. 36, No. 5, 1990, pp. 961–1005.
Derler, P., E. A. Lee, and A. Sangiovanni Vincentelli, “Modeling Cyber–Physical
Copyright © 2016. Artech House. All rights reserved.

Systems,” Proc. of the IEEE, Vol. 100, No. 1, 2012, pp. 13–28.
Douglas, S. C., “Introduction to Adaptive Filters,” in Digital Signal Processing
Handbook, V. K. Madisetti and D. B. Williams, (eds.), Boca Raton, FL: CRC Press,
1999.
Friedman-Hill, E. “Jess, the Rule Engine for the Java Platform,” 2008.
Frigo, M., and S. G. Johnson, “FFTW: An Adaptive Software Architecture for the
FFT,” Proc. of the 1998 IEEE Intl. Conf. on Acoustics, Speech and Signal Processing, Vol.
3, 1998, pp. 1381–1384.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Sensor Informatics and Business Insights 133

Goodwin, G. C., and K. S. Sin, “Adaptive Filtering Prediction and Control,” Cou-
rier Corporation, 2014.
Graves, S. C., D. B. Kletter, and W. B. Hetzel. “A Dynamic Model for Requirements
Planning with Application to Supply Chain Optimization,” Operations Research,
Vol. 46, No. 3, Supplement 3, 1998, pp. S35–S49.
Grubbs, F. E., “Procedures for Detecting Outlying Observations in Samples,” Tech-
nometrics, Vol. 11, No. 1, February 1969, pp. 1–21.
Gubbi, J., et al., “Internet of Things (IoT): A Vision, Architectural Elements, and
Future Directions,” Future Generation Computer Systems, Vol. 29, No. 7, 2013, pp.
1645–1660.
Hall, D. L., and J. Llinas, “An Introduction to Multisensor Data Fusion,” Proc. of
the IEEE, Vol. 85, No. 1, 1997, pp. 6–23.
Haykin, S., Adaptive Filter Theory, Upper Saddle River, NJ: Prentice Hall 1996.
Honig, M. L., and D. G. Messerschmitt, Adaptive Filters, Structures, Algorithms and
Applications, Boston, MA: Kluwer Academic Publishers, 1984.
Jenkins, W. K., et al., Advanced Concepts in Adaptive Signal Processing, Boston, MA:
Kluwer Academic Publishers, 1996.
Kay, S. M., Fundamentals of Statistical Signal Processing, Vol. 1: Estimation Theory,
Upper Saddle River, NJ: Prentice Hall, 1993.
Kay, S. M., Fundamentals of Statistical Signal Processing, Vol. 2: Detection Theory, Up-
per Saddle River, NJ: Prentice Hall, 1998.
Kay, S. M., Fundamentals of Statistical Signal Processing: Practical Algorithm Develop-
ment, Vol. 3, Boston, MA: Pearson Education, 2013.
Keogh, E. J., and M. J Pazzani. “Scaling Up Dynamic Time Warping for Datamin-
ing Applications,” Proc. of 6th ACM SIGKDD Intl. Conf. on Knowledge Discovery and
Data Mining, 2000, pp. 285–289.
Kielman, J., and J. Thomas, (Guest Eds.), “Special Issue: Foundations and Fron-
tiers of Visual Analytics,” Information Visualization, Vol. 8, No. 4, Winter 2009, pp.
239–314.
Lertniphonphun, W., and J. H. McClellan. “Unified Design Algorithm for Com-
plex FIR and IIR Filters,” Proc. on 2001 IEEE Intl. Conf. on Acoustics, Speech, and
Signal Processing, (ICASSP’01), Vol. 6, 2001, pp. 3801–3804.
Li, C., and G. Miklau, “Pricing Aggregate Queries in a Data Marketplace,” Web-
DB, 2012, pp. 19–24.
Copyright © 2016. Artech House. All rights reserved.

Liggins II, M., D. Hall, and J. Llinas, (eds.), Handbook of Multisensor Data Fusion:
Theory and Practice, Boca Raton, FL: CRC Press, 2008.
Lin, J., et al., “A Symbolic Representation of Time Series, with Implications for
Streaming Algorithms,” Proc. of 8th ACM SIGMOD Workshop on Research Issues in
Data Mining and Knowledge Discovery, 2003, pp. 2–11.
Maddala, G. S., “Outliers,” in Introduction to Econometrics, 2nd ed., New York:
MacMillan, 1992, pp. 88–96.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 134 IoT Technical Challenges and Solutions

Magnani, L., Abduction, Reason, and Science: Processes of Discovery and Explanation,
New York: Kluwer Academic/Plenum Publishers, 2001.
Mannila, H., “Data Mining: Machine Learning, Statistics, and Databases,” IEEE
Computer Society Intl. Conf. Scientific and Statistical Database Management, 1996.
Manolakis, D. G., V. K. Ingle, and S. M. Kogon, Statistical and Adaptive Signal Pro-
cessing: Spectral Estimation, Signal Modeling, Adaptive Filtering, and Array Process-
ing, Norwood, MA: Artech House, 2005.
Menychtas, A., et al., “A Marketplace Framework for Trading Cloud-Based Ser-
vices,” in Economics of Grids, Clouds, Systems, and Services, New York: Springer,
2011, pp. 76–89.
Ojo, A., E. Curry, and T. Janowski, “Designing Next Generation Smart City Ini-
tiatives: Harnessing Findings and Lessons from a Study of Ten Smart City Pro-
grams,” 22nd European Conf. on Information Systems, Tel Aviv, 2014.
Pal, A., et al., “A Robust Heart Rate Detection Using Smart-Phone Video,” Proc. of
3rd ACM MobiHoc Workshop on Pervasive Wireless Healthcare, 2013, pp. 43–48.
Poor, V. H., An Introduction to Signal Detection and Estimation, 2nd ed., New York:
Springer, 1998.
Proctor, M., “Drools: A Rule Engine for Complex Event Processing,” in Applica-
tions of Graph Transformations with Industrial Relevance, New York: Springer, 2011,
pp. 2-2.
Russell, S., and P. Norvig, Artificial Intelligence: A Modern Approach, 2nd ed., Upper
Saddle River, NJ: Prentice Hall, 2003.
Skolnik, M. I., “Introduction to Radar,” in Radar Handbook, 2nd ed., New York:
McGraw Hill, 1962.
Sokwoo, R., “Design and Analysis of Artifact-Resistive Finger Photoplethysmo-
graphic Sensors for Vital Sign Monitoring,” Ph.D. dissertation, MIT, 2000.
Suh, J., et al., “A Performance Analysis of PIM, Stream Processing, and Tiled Pro-
cessing on Memory-Intensive Signal Processing Kernels,” ACM SIGARCH Com-
puter Architecture News, Vol. 31, No. 2, 2003, pp. 410–421.
Sun, Y., et al., “Advances on Data, Information, and Knowledge in the Internet of
Things,” Personal and Ubiquitous Computing, Vol. 18, No. 8, 2014, p. 1793.
Thomas, J. J., and K. A. Cook, “Illuminating the Path: The R&D Agenda for Visual
Analytics,” National Visualization and Analytics Center, 2005, p. 4.
Vickers, J., “The Problem of Induction,” Stanford Encyclopedia of Philosophy, June
Copyright © 2016. Artech House. All rights reserved.

21, 2010.
Hodge, V. J., and J. Austin, “A Survey of Outlier Detection Methodologies,” Arti-
ficial Intelligence Review, Vol. 22, No. 2, 2004, pp. 85–126.
Wang, H., D. Estrin, and L. Girod, “Preprocessing in a Tiered Sensor Network for
Habitat Monitoring,” EURASIP Journal on Advances in Signal Processing, Vol. 4,
2003, pp. 392–401.
Wu, J., et al., “Apparatus and Method for Providing a Data Marketplace,” U.S.
Patent Application 11/750,993, filed May 18, 2007.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 5
Mobile Sensing

5.1 Introduction

Mobile sensing is a class of sensing architecture in which sensors

mounted on a moving platform or gateway is used in a space over
time to collaboratively monitor physical and environmental condi-
tions. Mobility enables sensor nodes to target and track objects and
phenomena that are either moving or spatially distributed. The
mobile platform can be people carrying personal devices, trans-
port vehicles, robots, and drones. In addition to being the eyes and
ears of the IoT, these platforms provide hands and legs to IoT. They
can be efficiently used for observing large geospatial systems, civil
infrastructures, vehicles, and environmental events such as wind,
pollution, and chemical clouds. Along with static sensors, these
mobile sensing systems enable a seamless coverage at the required
time resolutions as required by the application.
Copyright © 2016. Artech House. All rights reserved.

Each platform has its own sweet spot in the economic spatio-
temporal coverage as depicted in Figure 5.1. Satellites provide
a coarse view from space that can provide a large coverage but
has a low time resolution (i.e., periodicity of views available is in
days). Views from unmanned aerial vehicles (UAVs) provide a

135

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 136 IoT Technical Challenges and Solutions

Figure 5.1 Mobile sensing: sweet spot for different mobile-

sensing platforms.

much better observation quality with improved time resolution;

however, the practical coverage possible is less than that of satel-
lites. The data collection by a person carrying a personal or hand-
held device can be effectively used for applications that can work
with data collection at a lower time resolution and coverage. At the
same time, fixed sensors can provide a high time resolution, but
their coverage is relatively limited. Each sensing platform has its
own sweet spot for economic spatiotemporal coverage as shown
in Figure 5.1. However, exceptions are possible from this generic
view on the data collection behaviors and platform. The coverage
of sensor system can be increased by using a network of fixed and
mobile sensors. Here the economics of sensing will play a role in
deciding the right platform and architecture, and we will have a
brief discussion on this aspect at the end of this chapter.

5.2 Applications and Use Cases for Mobile Sensing

As mentioned above, popular platforms that can be used for mo-

Copyright © 2016. Artech House. All rights reserved.

bile sensing are robots, UAVs, and smartphones. Various sensors

mounted on these platforms can be mobilized to the location of in-
terest as and when required. Specifically, for applications in which
the phenomenon of interest does not require a dedicated sensor,
sampling periodicity is low and spatially distributed can substan-
tially benefit from a mobile-sensing approach. There are several

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 137

use cases of mobile sensing, and some of the representative classes

are discussed in the following sections.

5.2.1 Mobile Sensing for Environmental Monitoring

Preventive health researchers are interested in gathering data

to build a fine-grained model of environmental factors affecting
pollution-induced diseases. Sensors mounted on private or pub-
lic transport along with static sensors deployed at strategic loca-
tions can monitor temperature, humidity, carbon monoxide, and
so forth. A synthesized model from this data is highly helpful for
giving guidance and warnings to people with asthma, fine-particle
allergies, and so forth. One such example is the pollution hotspots
being monitored with use of smartphones [1]. Smartphones are
used to collect the ultrafine particle density data from the sensors
mounted on cars of volunteers. In-car GPS is used for position
approximation. Volunteers get discounts or waivers on their cell
phone charges. This information is offered to news media for dis-
semination and raising public awareness.
Another example is on the monitoring of deserts. Dynamics
of deserts are a serious concern for people living in and around
them. The Tumbleweed [2] is a robot that can roll across the desert,
gathering information about desertification. It can operate auton-
omously for years and can travel thousands of kilometers using
only the power of the wind.

5.2.2 Mobile Sensing for Emergency Response

First responders involved in indoor emergency situations, such as

urban fire, are often challenged by dangerous and dynamic op-
erating environments with little access to situational information.
This situational information, such as the extent and severity of an
indoor fire, can enhance rescue planning and reduce risk to per-
Copyright © 2016. Artech House. All rights reserved.

sonnel. Traditional predeployed sensors are either absent or often

sparsely distributed and inadequate to provide required informa-
tion. Also, fixed deployment of sensors to cover large infrastruc-
ture are relatively expensive to install and maintain. Controlled
mobility of robots having the capability to move per need can pro-
vide substantial help to first responders.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 138 IoT Technical Challenges and Solutions

5.2.3 Collaborative Sensing for Urban Transportation

For collection of urban traffic information, which is a highly dis-

tributed phenomenon, different methods have been traditionally
used. This includes surveillance cameras, loop detectors on the
road, and Doppler radars. Achieving wide coverage using this
fixed infrastructure is an expensive and nonflexible approach. Mo-
bile sensing is an economic alternative or complementary scheme
to be used here.
Probe vehicles such as taxis and buses could collect traffic data
in a city. For example, they could periodically report their speed
and location to a central server over cellular communication net-
work. Additionally, location information revealed by using their
smartphones can also be a source of traffic estimation.

5.2.4 Robots in Healthcare

Robots can have the ability to perform tasks that humans find dif-
ficult, monotonous, or inefficient to do [3]. They can assist in home
care support functions such as asset location, can move patients,
and can communicate with doctors in an emergency. Robotic pets
are becoming popular for wellness and theraputic purposes. In a
home care scenario, they may assist in sharing the location of med-
ication, objects, and people and may also help in navigation. They
need to be autonomous in terms of both mobility and decision-
making abilities. This requires systems for the data collection, pro-
cessing, storing, and use of information for the autonomous ser-
vices. Further medical robots are being used for several minimally
invasive surgical procedures.

5.2.5 Robotic Telesensing and Operation

Teleoperation means controlling or doing work at a distance.

Copyright © 2016. Artech House. All rights reserved.

Telerobotics based teleoperation has a wide use in many different

fields including military or defense, space, underwater, offshore
exploration, and healthcare. It can aid in repairs and exploration
in difficult-to-reach areas such as space, mines, and inaccessible
areas. Also, it is useful in assisting unskilled staff in the field for
expert operations. Remote-operated vehicles (ROVs) are increas-
ingly being used in disaster management, mining, and underwater

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 139

operations replacing humans. In general, ROVs find a wide range

of applications including surveys, oceanography, and inspection.
Doctors can use tele-operated medical robots for examining a re-
mote patient and perform surgical procedures. Tele-operation is
often found to be enhanced with augmented reality (AR) interfaces
in which a remote person is provided with near real-life visualiza-
tion of the scene acquired by the robots. To enhance the operator
perception of the mobile robot workplace, haptic interfaces can be
used. They offer feedback or tactitle feedback, giving operators a
better sense of the remote environment. This will help in enhanc-
ing the efficiency and robustness of interactive tele-operation.

5.2.6 Aerial Robots for Spatial Intelligence

A growing interest is shown for using unmanned aerial vehicles

(UAVs) for inspection and other remote-sensing applications. UAV
has been traditionally used in several military applications. A
wide spectrum of civilian applications is emerging, including pub-
lic area monitoring, infrastructure inspection, intelligent farming,
and aerial mapping. The low-cost, small UAVs, generally called
drones, have the potential for several industrial applications and
public services. In spite of severe restrictions and regulations, ap-
plications of drones have increased exponentially over the past
few years.
In precision agriculture, UAVs allow farmers to gather real-time
data on crops, detect crop diseases as early as possible, estimate
yield, detect stress due to inadequate water and nutrition, and
make better decisions about using fertilizers, herbicides, and pes-
ticides. There are applications reported in which drones are used
for pesticide spraying. Animal tracking is another task well suited
to the capabilities of drones. They can also help to protect crops
from the attack of wild animals. In forestry, the use of unmanned
Copyright © 2016. Artech House. All rights reserved.

aerial systems in spotting and mapping forest fires is getting atten-

tion. Further, the estimation and classification of tree resources and
wildlife in forests are also compelling applications of drones. Envi-
ronmental protection is an area in which UAVs can contribute a lot.
The three-dimensional (3-D) surface mapping of the landscape
is useful for designing and planning structures. Aerial mapping
of the building at various stages of construction can provide

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 140 IoT Technical Challenges and Solutions

useful information regarding the progress and compliance of the

construction. Further, for post-disaster investigation of damaged
buildings and utility structures, UAV can be a powerful tool for
support. Table 5.1 lists various industry verticals that can benefit
from UAV-based sensing.

5.2.6.1 Utility Asset Inspection Using UAV [4]

Utility systems such as power lines, oil and gas pipelines, and
water supply lines require periodic inspections. The inspection
involves assessing the entire structure to identify and locate anom-
alies such as cracking, snapping, shifting, chipping, surface deg-
radation, and weed encroachment. For power lines and pipelines,
there is a need for checking their rights of way for any intrusions
or infringements. Also, there is a need to look for signs of leakage,
such as dead or discolored vegetation, presence of liquid on the
ground, dirt, rusting or unusual fog, or suspended particles. Such
inspections are also relevant for railway lines.
Manual utility inspection often involves difficult and danger-
ous conditions. The unsafe conditions faced by the field workers
include working at high heights or working with high-voltage
assets, biological hazards, and large bodies of water or heat. Em-
ploying a drone to inspect assets through near-field remote sens-
ing lets field staff stay at a safe distance, reducing the likelihood of
injury or incapacitation. Moreover, such inspection methodology

Table 5.1
Some of the Prominent UAV Applications for Different Verticals
Domains UAV Applications
Utilities Pipelines (oil, gas, water), power lines, wind farm, solar farm
Transportation Road traffic, railways, water transport
Government Coastal surveillance, public events, pollution, forests, smart city,
heritage sites, land use, and law and order
Copyright © 2016. Artech House. All rights reserved.

Construction Site survey, planning, development, building management

Insurance Inspection, damage assessment, risk monitoring, emergency care
Healthcare Emergency care, remote monitoring or assistance
Retail Outlet planning, shop shelf monitoring
Mining Asset monitoring, mapping, disaster management
Telecom Telecom tower inspection, radio frequency (RF) signal mapping,
temporary cells

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 141

can provide higher-quality, more accurate, and usable data, as and

when needed. It helps in improving efficiency and reducing costs.
It supports a variety of visualization formats including high-den-
sity images, 3-D views, annotated video summaries, and various
geographical information system (GIS) layers.

5.2.6.2 UAV-Based Aircraft Inspection

Hailstorms, lightning strikes, and even normal wear and tear cause
damage to aircraft fuselage. Timely monitoring and remediation
are required to avoid millions of dollars in loss of revenue. Speed-
ing up the inspection of aircrafts by using drones as part of the
Integrated Vehicle Health Management System (IVHMS) can be of
tremendous value towards improving aircraft utilization.

5.2.6.3 UAV for Urban Inspection

Drones can also be effectively deployed for inspection of com-
munication towers, wind turbines, solar farms, buildings, and
roadways, as well as unidentified objects including suspected ex-
plosives. Intelligent infrastructure is a key focus for smart cities.
Effective sensing methodologies to monitor the state of the infra-
structure are of high interest there. In addition to visual cameras,
infrared, multispectral, and hyperspectral sensors are often used
to gather data for detecting various changes and damages at high
precision. UAVs can also play a key role in monitoring and surveil-
lance strategies of smart cities.

5.2.6.4 UAV for Insurance Business

Drones can also be a potential tool for insurance companies to
evaluate risk and collect digital evidences for evaluating claims.
A standardized inspection process to cover various asset catego-
ries that are of interest to insurance business, such as crops, build-
Copyright © 2016. Artech House. All rights reserved.

ings, and vehicles could be formulated with drones. Drones can

be quickly deployed to collect and record digital evidence of asset
damages in case of a disaster or emergency. This can open up new
business models in insurance.
Figure 5.2 shows various enterprise sensing applications of
UAV.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 142 IoT Technical Challenges and Solutions

Figure 5.2 UAV-based sensing: potential business applications.

Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 143

5.3 Technologies and Challenges in Mobile Sensing

There are several issues and challenges related to technologies,

methodologies and business models in mobile sensing. In this
section, we discuss those aspects related to mobile sensing using
smart phones, robots, and UAVs.

5.3.1 Smartphone-Based Sensing

Today, smartphones serve as integral mobile devices for person-

al communication and the computer devices of choice. It comes
with a set of embedded sensors, such as an accelerometer, Global
Positioning System (GPS), compass, gyroscope, microphone, and
camera. Individually and collectively, these sensors can provide a
rich set of contextual information. Bluetooth and Wi-Fi radios also
provide information around the surrounding environment. They
are enabling several new applications across a wide variety of do-
mains, such as healthcare, wellness, social networks, safety, envi-
ronmental monitoring, and transportation.
Many more new sensors such as a barometer and a heart rate
sensor are being added to smartphones. The primary objectives of
these sensors were for mobile applications to improve their per-
formance, user interfaces, and other functionalities. However, they
also enable a significant opportunity to gather data about people
and their environments. For example, the physical movements,
actions, and gestures of a person can be decoded by processing
accelerometer data from the smartphone being carried. Distinct
patterns within the accelerometer data can reveal different activi-
ties such as running, walking, standing, or counting the number of
steps taken or estimating the distance walked. Further, microphon-
escan capture the ambient sound that can be analyzed in order to
determine the ambient noise level and person’s location context.
Copyright © 2016. Artech House. All rights reserved.

This is a key input for several wellness applications.

In addition to this, there are several sensors that can be exter-
nally integrated or connected to the smartphone to extend its sens-
ing capabilities. A large number of medical or healthcare sensors
that can be connected over Bluetooth fall in this category: devices
for measuring blood pressure, glucose level, heart rate, respiration,
and various biomarkers in blood and perspiration, to mention a
few.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 144 IoT Technical Challenges and Solutions

5.3.1.1 Mobile Phone Sensing: Different Scales

Mobile phone sensing applications can be categorized at differ-
ent scales into personal sensing, community sensing, and public
sensing.

• Personal sensing: Applications focus on monitoring the differ-

ent aspects of the personal life of individuals and collect infor-
mation about their activities. The sensor data for such applica-
tion is private or may be shared only with highly trustworthy
entities. For instance, healthcare applications share data with
medical professionals to provide necessary support in case of
emergency.
• Community sensing: Applications make use of collaborative
sharing within a group of phone users who are interested in
common goals or a purpose. In community sensing, systems
collect and share information among phone users, their friends,
and their social groups, which can promote interaction among
people and improve the efficiency of organizing social activi-
ties.
• Public sensing: Applications collect rich and heterogeneous
data from a large number of people, which can be mined for a
variety of purposes to aid decision processes. Phone users’ par-
ticipation is the heart of all public mobile-sensing applications
in which data can be shared and can return collective benefits.

The popularity of smartphones created a new trend for crowd-

sensing networks (CSNs), which is a promising platform for com-
munity-driven participatory sensing. People can use their smart-
phones or sensor-equipped devices (including vehicles) to form
a crowd-sensing network to enable collaborative sensing. This
voluntarily collects data for several valuable applications such as
Copyright © 2016. Artech House. All rights reserved.

environmental monitoring and protection, traffic jam reporting,

citizen journalism, and tourist queries.

5.3.1.2 Users’ Participation in Mobile Phone Sensing

The participation of the general population in data collection and
analysis has led to a sensing paradigm shift known as crowdsourc-
ing. Mobile crowdsourcing can be seen as a distributed approach

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 145

to solve a complex problem leveraging the participation of a large

group of individuals with mobile devices. Smartphone-based sens-
ing can be classified into two sensing paradigms, namely, partici-
patory sensing and opportunistic sensing.
Participatory sensing requires active participation with specific
user-initiated actions for collecting the data such as snapping a pic-
ture or recording a voice or a manual data entry. In participatory
sensing, the phone user has complete control regarding the con-
tribution of the sensed data, which makes it more meaningful and
less personal but places a considerable burden on the phone users.
MobiShop [5] is a people-centric participatory sensing application,
which shares prices of products among users. The users use their
mobile phones’ camera to capture a digital image of store receipts.
Opportunistic sensing shifts the burden of tasks from the phone
users to a background sensing system and applications can collect
data without active participation of the user. Google’s Android op-
erating system does several such user context data collections in
the background.
Along with the benefits of opportunistic sensing, the opportu-
nistic model introduces many challenges. As users do not directly
control the data sharing, privacy of the users remains a major con-
cern. Another challenge is design of incentive schemes, which can
motivate users to contribute data in sensing system.

5.3.1.3 Some Challenges and Trends in Collaborative Sensing

Although the wide availability of sensors offers very interesting
opportunities for collaborative sensing, there are also challenges
that need to be tackled.

• Encouraging cooperation from and among the users: Suitable

incentive or reward schemes should be deployed for encour-
aging the participants. In a platform-centric model, the system
Copyright © 2016. Artech House. All rights reserved.

itself provides a reward to the participating users. However,

in a user-centric model, the incentives are based on auctions
in which users demand a specific reward in exchange for their
sensing services. The use of gamification to engage users is an-
other model to consider.
• Management of the data quality: Guaranteeing or even esti-
mating the quality of data collected using opportunistic sens-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 146 IoT Technical Challenges and Solutions

ing is a major difficulty. There are certain reputation estimation

techniques in the literature that estimate the expected level of
cooperation and the reliability of the different contributors.
Data quality indicators such as the accuracy, completeness, and
freshness can be quantified with this information. A number of
data quality metrics for sensor feeds have been in use includ-
ing a metric called the quality of contributed service to charac-
terize the quality and timeliness of a sensed quantity obtained
through participatory sensing.
• Spatial crowdsourcing: In certain cases, it is required from the
contributors to collect data from specific areas or locations, for
reasons such as to influence a person or a mobile-sensing plat-
form such as a vehicle to slightly detour from his or her or its
route to collect data. At the same time, this detour suggestion
should also be beneficial to them, in terms of less congestion,
better ride quality, and so forth. Also, there can be a location-
specific dynamic incentive scheme for data collection.
• Semantic data management: The use of semantic techniques,
such as ontologies and reasoners can provide significant ben-
efits. Among them, we can highlight enabling the interoper-
ability among different devices and enriching sensor data with
higher-level information, which could be potentially queried in
a flexible and semantic way.
• Automation of data collection: The effort involved from users
to collect and collaborate for data collection is another factor to
consider. Well-designed mobile applications with unobtrusive
interfaces can alleviate this problem to a great extent. The com-
plete engagement model including data collection, authentica-
tion, and incentive mechanism can be automated to achieve the
best user engagement.
• Management of trust and privacy: Collaborative sensing often
Copyright © 2016. Artech House. All rights reserved.

involves the collection of data that has some privacy concerns.

Specific privacy protection mechanism should be incorporated
throughout the life cycle of the data. Similarly, trust manage-
ment is a fundamental aspect of collaborative sensing, partic-
ularly if humans actively provide information. For example,
malicious users could provide false information to disturb the
system or to gain a competitive advantage. Suitable authentica-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 147

tion mechanism for the participants should be incorporated in

the system.
• Data ownership: Clarity on the ownership, purpose, and legal-
ity of the data and the process is another issue to be addressed.
Transparency on these aspects should be provided to all the
stakeholders.
• Battery and communication cost anxiety: Will the mobile sens-
ing application deplete the battery energy faster? Will it cause
increased data charges? The sensor data acquisition, process-
ing, and communication should be as energy-efficient as pos-
sible to alleviate the anxiety of the user.

5.3.2 Robotic Sensor Networks

As discussed earlier, a number of applications have been support-

ed by robotic sensor networks. Major applications are in environ-
mental monitoring and in search and rescue. Search-and-rescue
systems should quickly and accurately map search space and lo-
cate the victims. It should also maintain communication with hu-
man responders. It should also be capable to handle a dynamic
and potentially hostile environment. Robotic ad hoc networks
having robust communication, sensing, computing, and localiza-
tion capability are required to be used for this purpose. It should
autonomously conduct target tracking with minimal support of
localization capabilities.

5.3.2.1 Multirobot Sensing and Coverage Issues

Robots can be considered as active, mobile, and autonomous data
processing units operating in a network. Robots use embedded
sensing capabilities for their control and task execution. For sens-
ing applications involving a larger spatial coverage, the use of
Copyright © 2016. Artech House. All rights reserved.

multirobots holds many advantages. Multiple robots can leverage

their redundancy and parallelism to increase the robustness, flex-
ibility, and coverage. Multiple robots can also localize themselves
more efficiently when they have different sensor capabilities. The
advantage of faster coverage is relevant for many applications like
search and rescue, intrusion detection, sensor deployment, har-
vesting, and mine clearing.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 148 IoT Technical Challenges and Solutions

In a multirobot system, there are several techniques that can be

used for planning the coverage. Using the informative approach,
each robot is given a local identity and is in the communication
range of the other robots. This enables them to spread out in a
coordinated manner. However, in a molecular approach, no lo-
cal identification is made for the robots and they do not perform
any directed communication as well. Here trajectories are formed
with each robot moving in a direction away from all its immediate
sensed neighbors. Ant societies and their stigmergic communica-
tion are the motivation for another trajectory planning approach.
In mobile-sensing networks, potential field techniques are also
used for tasks like local navigation and obstacle avoidance. This
is used for cases where mobile robots are to be arranged in an un-
known environment. Fields are constructed such that each node is
repelled by other nodes. Also, obstacles in the environment force
the network to spread. A virtual force-based algorithm can be ef-
fectively used by combining repulsive forces and attractive forces
to navigate the robots through virtual motion paths.

5.3.2.2 Localization for Robots

For robot localization, there are several choices of technologies in-
cluding GPS, active and passive beacons, odometer, and sonar. For
indoor applications, the problem is challenging because GPS can-
not be used. In such cases, various sensor-assisted methods are to
be explored. In one approach, a sensor-driven local map is gener-
ated and is correlated with a global map. Here no prior knowledge
of the environment is assumed. Sensor data is used to construct the
global map dynamically. A set of feasible poses are generated as
local maps. The algorithm then selects the best fit from the feasible
poses that gives the location. Simultaneous location and mapping
(SLAM) fall in this category.
Dead reckoning can be used as a secondary source for refining
Copyright © 2016. Artech House. All rights reserved.

the above location estimation. The Monte Carlo method, combined

with the grid-based Markov localization, is another approach for
robot positioning. It can globally localize and can deal with am-
biguities. A panoramic image-based model for robot localization
is another approach. In urban environments, a possible localiza-
tion method is by using scene fingerprinting of images taken by

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 149

cameras. Here feature correspondence between images taken by

camera with their reference images are used to match and locate.
Humanoid robots are getting popular for variety of applica-
tions such as assistive robotics and personal robotics. A vision-
based approach for localization and navigation is often followed
in such applications. Scene understanding and SLAM are used by
the humanoid robot to maneuver in the interactive environment.

5.3.3 UAV for Aerial Mapping

The majority of the industrial applications of UAV is in aerial map-

ping. Although satellites are capable of providing high-quality aer-
ial images, UAVs have their sweet spots in low-cost, high-resolu-
tion mapping free from cloud interferences as and when required.
Aerial images can be used to generate several maps including or-
thorectified (geographically accurate) two-dimensional maps, el-
evation models, thermal maps, and 3-D maps or models. They can
be made with sufficient accuracy for the precise measurement of
various objects of interest using photogrammetry techniques.
Two-dimensional maps are still the most commonly used prod-
uct from UAV-captured images. A series of overlapping aerial pho-
tographs can be combined into a single panoramic image. For an
ortho-mosaic creation, a series of overlapping aerial photographs
is geometrically corrected (orthorectified) and combined in a uni-
form scale. Orthorectified photos can be used to produce GIS-com-
patible maps for various GIS applications including archaeological
applications, construction, and cadastral surveying.
The 3-D visualizations and volume calculations from a set of
aerial images are possible through a 3-D reconstruction process.
For this, a sequence of images of the terrain with at least 80% over-
lap is required for the 3-D reconstruction algorithm.
Other data products from UAV-collected imagery include digi-
Copyright © 2016. Artech House. All rights reserved.

tal elevation models (DEMs), Normalized Difference Vegetation

Index (NDVI) (vegetation) maps, and thermal maps. DEMs are
topographical maps and they represent the underlying terrain. The
detailed surface features such as buildings, vegetation, and other
man-made aspects are not captured in DEM. Hidden features with
cavities like buildings cannot be adequately represented in DEM.
NDVI maps are majorly used for agricultural applications. It
uses visual and near-infrared images to compute NDVI. NDVI

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 150 IoT Technical Challenges and Solutions

indicates how green (vegetated) a terrain is based on the amount of

infrared light reflected by living plants. Thermal maps are useful
for applications such as detecting structural damages, groundwa-
ter discharges, and hidden archaeological assets.
Many image processing, computer vision, and geometrical pro-
cessing techniques are available for a rich set of applications of
UAV-based sensor data. For example, in a building construction
scenario, the aerial sensing can provide several useful insights on
the status and progress of the construction process. Table 5.2 gives
a list of use cases of UAV for construction monitoring and com-
puter vision techniques for addressing them.

5.3.3.1 Challenges of UAV-Based Infrastructure Inspections [6]

While the potential benefits are high, enterprises are not yet ready
to deploy inspection drones in their regular operations. Regulatory
restrictions are one of the major hurdles; most regulators (such as
the Federal Aviation Administration in the United States and the
Swedish Transportation Agency in Sweden) have restricted drone
usage.

Table 5.2
UAV-Driven Computer Vision Methods for Construction and Building Monitoring
Use Cases Techniques or Approaches
Progress monitoring Leveraging spatial and temporal information in four
dimensions (4-D) for monitoring work-in-progress,
appearance-based reasoning about the progress
Site monitoring 4-D visualization of digital surface model (DSM), geometry-
based change detection, integrating aerial images with
augmented reality
Building inspection 3-D model-based damage estimation, edge detection for
identifying cracks on building façades and roofs
Building 3-D model, extracting building contours for measurement
measurement
Copyright © 2016. Artech House. All rights reserved.

Surveying Georeferencing using time-stamped GPS data

Structural damage Image segmentation and object classification for damage
assessment estimation, machine learning-based classification of
damaged buildings using feature sets obtained from feature
extraction and transformation in images
Road assessment Image-based 3-D reconstruction, feature extraction through
image filtering, analyzing the structure and dimension of
road surface, feature extraction and Orthophoto mapping
Geo-hazard Orthophoto mapping and visual interpretation to inspect
investigations geologic hazards along oil and gas pipelines

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 151

Drones pose concerns around safety, privacy, and interference

with air traffic. Potential injury due to faulty operation is the ma-
jor cause of safety concerns; this needs to be addressed with ro-
bust machines that have fail-safe capabilities. This will require the
flying machine to have improved path planning, navigation, and
control, with obstacle avoidance features. This also calls for auto-
nomic navigation, so that drones are not limited by the skill level
of the human operator. The autopilot system should have the abil-
ity to enforce compliance to regulations. Human operators should
be able to define the inspection scope intuitively to the autonomic
system, which manages flight planning, safety planning, and pos-
sibly even payload configuration.
The drone operating system should also take care of the envi-
ronmental and operational conditions in planning and navigating
the drone. Failure of a drone may lead to loss of the equipment,
which calls for low-cost hardware, including sensors.
Today, it takes a team of skilled personnel to operate a drone for
asset inspection. A qualified pilot is required to fly the machine, a
domain expert can suggest the view orientations, and a technician
needs to ensure the quality of data acquisition. This need for several
experienced experts raises the cost of the program. The autonomic
platform should be sophisticated enough to allow a semiskilled
professional to conduct the session in a semiautonomic manner,
and we should move to fully autonomic mode in the future.
Should there be an air traffic controller (ATC) for drones? This
may not be necessary, but monitoring organization will be re-
quired to coordinate drone operations, so as to address compliance
and conflicts. Standards are needed for the operation of drones in
common air space; initiatives such as the Drone Safety Council are
working toward this.
Handling the Big Data gathered by drones is another problem
that needs attention. While there are many image and video pro-
Copyright © 2016. Artech House. All rights reserved.

cessing tools for visualization of aerial data, the fact is that end con-
sumers are not interested in just collecting and visualizing the data.
They are more interested in actionable information extracted from
the mission. Automatic object recognition, object-based measure-
ments, and change or anomaly detection are major requirements.
This requires a data processing pipeline that can work autono-
mously without manual intervention and provides information

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 152 IoT Technical Challenges and Solutions

services. Approaches from cognitive robotics and deep learning

could add substantial impact here. The huge volumes of streamed
data would need real-time processing (distributed across onboard,
ground station, and cloud computing). Integrating this data with
other sources such as remote sensing satellites, smartphones, and
other sensors can offer higher-quality insights.
Privacy is a major social concern with regard to drones. Howev-
er, for the inspection applications described above, the possibility
of sensitive data being captured is relatively low. This concern can
be further alleviated by the use of image obfuscation techniques to
suppress sensitive regions before moving the content to secondary
storage.
The promise of inspection drones is appealing. More novel use
cases are emerging, and the industry needs to prepare for effective
use of these powerful devices.

5.4 Economics of Mobile Sensing

In IoT, the decision to physically connect a thing will depend on

the value realizable from the resultant information and cost of
connecting it. Cost encompasses the design, implementation, op-
eration, and maintenance of the IoT system. Beyond that, it also
has to factor the security, privacy, and overall risk of connecting it.
We use the term “Thinganomics” as the economics of connecting
things and it is an important aspect to be considered in choosing
and architecting IoT applications. Mobile-sensing approaches are
an important contributor to the economy of sensing.
Coverage is an important metric for the sensing applications
and it can be used as one of the measures for quality of service.
It basically addresses the detection of a phenomenon or event
of interest in spatio-temporal space for a given application. The
Copyright © 2016. Artech House. All rights reserved.

phenomenon under observation has a temporal behavior and it

needs to be detected and responded within a specified time du-
ration desired by the application. For example, a fire alarm sys-
tem will require the fire to be detected and responded to within
seconds of its occurrence. In another application, an air pollution
alert system may expect the detection of air pollution and response
to be achieved in couple of hours. In the fire detection case, the
coverage is desired to be achieved using a fixed sensor system in

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 153

which the sensors continuously monitor the event at a high sam-

pling rate, whereas in the pollution monitoring application could
benefit from mobile sensing in which a mobile sensor can achieve
a higher coverage due to its mobility and at the same time meet
the time resolution requirement. Figure 5.3 outlines the relation-
ship between sampling, response requirement, and occurrence. It
also outlines how the cost of mobile sensors goes down compared
to fixed sensors as the coverage requirement goes up; this is also
explained mathematically below.
Let Cd be the desired coverage, Ns is the number of sensor nodes,
ms is the mobility rate of the sensor in meters per second, Cs is the
coverage of individual sensor, Tp is the duration of the phenom-
enon to be observed, and Tr is the maximum delay tolerance for the
detection of the phenomenon or event.
The total number of sensor nodes required for a uniform cover-
age Cd desired by the application can be estimated as

Cd
Ns =
Cs (1 + ms ) * min(Tp , Tr )

From the above expression, it is evident that the number of sen-

sors required for a desired coverage is inversely proportional to
the mobility, persistence of the phenomenon, and delay tolerance
Copyright © 2016. Artech House. All rights reserved.

Figure 5.3 Economics of mobile sensing.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 154 IoT Technical Challenges and Solutions

in detection. For a fixed sensor system, the mobility is zero and we

can see that requires the maximum number of sensors. The cost of
sensing has capital cost pc (for a single sensor unit/platform) and
operational cost P0 · a′ is a multiplying factor that differentiates the
operational cost dependency on number of units for fixed sensors
versus mobile sensors.

Ps = ( pc + a ′p0 ) * N s

The cost of the sensing will be roughly proportional to the number

of sensor units. For mobile sensing, typically the sensor units (in-
cluding platform) are individually substantially costlier compared
to fixed sensor nodes. However, for a fixed sensing case, the num-
bers required are substantially high, although they are cheaper.
These factors to be taken into account for a specific application sce-
nario to arrive at an optimum sensing scheme. Finally, the optimal
architecture may involve only fixed sensors or only mobile sensors
or a mix of both.

5.5 Summary

Mobile sensing is an important aspect to consider for economic

sensing architectures for various applications, especially for large-
scale spatially distributed systems. Mobile platforms such as
smartphones, vehicles, robots, and drones can be used as mobile
gateways for data collection. Several representative use cases for
each of the sensing modalities have been discussed highlighting
the challenges and practical directions to address them. Crowd-
sourcing of sensing enables participatory and collaborative sensing
involving people. It can potentially achieve very large-scale cover-
age. However, additional challenges related to human behaviors,
Copyright © 2016. Artech House. All rights reserved.

privacy, and social factors have to be addressed. The emergence of

drones as a sensor platform is disruptive and is predicted to be a
highly impactful platform for diverse mobile-sensing applications.

References
[1] http://www.ijcaonline.org/archives/volume83/number10/14488-2798.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Mobile Sensing 155

[2] http://techxplore.com/news/2014-02-data-gathering-tumbleweed-robot-
desertification.html.
[3] http://www.werobot2015.org/wp-content/uploads/2015/04/Simshaw-
Hauser-Terry-Cummings-Regulating-Healthcare-Robots.pdf.
[4] http://sites.tcs.com/blogs/research-and-innovation/
promise-inspection-drones-enterprises.
[5] https://www.researchgate.net/publication/229013654_MobiShop_Using_
Mobile_Phones_for_Sharing_Consumer_Pricing_Information.
[6] http://sites.tcs.com/blogs/research-and-innovation/
challenges-adopting-drones-inspection-utility-assets.

Selected Bibliography
Aberer, K., et al., “OpenSense: Open Community Driven Sensing of Environ-
ment,” Proc. IWGS ’10 of the ACM SIGSPATIAL Intl. Workshop on GeoStreaming,
2010.
Brezmes, T., J. L. Gorricho, and J. Cotrina, “Activity Recognition from Accelerom-
eter Data on a Mobile Phone,” Proc. of 10th Intl. Work-Conference on Artificial Neural
Networks: Part II: Distributed Computing, Artificial Intelligence, Bioinformatics, Soft
Computing, and Ambient Assisted Living, 2009.
Burke, J., et al., “Participatory Sensing,” Proc. ACM Sen-Sys Workshop, World-Sen-
sor-Web, 2006.
Campbell, A. T., et al., “People-Centric Urban Sensing,” 2nd ACM WICON, 2006,
p. 18.
Changkun, J., et al., “Economics of Peer-to-Peer Mobile Crowdsensing,” 2015
IEEE Global Communications Conference (GLOBECOM), 2015.
Das, T., et al., “Prism: Platform for Remote Sensing Using Smartphones,” Proc. 8th
ACM MobiSys, 2010.
Fossel, J., et al., “OctoSLAM: A 3D Mapping Approach to Situational Awareness
of Unmanned Aerial Vehicles,” Proc. of Intl. Conf. on Unmanned Aircraft Systems
(ICUAS), Atlanta, GA, May 28–31, 2013, pp. 179–188.
Guo, B., et al., “From Participatory Sensing to Mobile Crowd Sensing,” Proc. of
Copyright © 2016. Artech House. All rights reserved.

IEEE PERCOM Workshops, March 2014, pp. 593–598.

Ilarri, S., O. Wolfson, and T. Delot, “Collaborative Sensing for Urban Transporta-
tion,” Bulletin of the IEEE Computer Society Technical Committee on Data Engineering,
2014.
Jayaraman, P. P., et al., “MOSDEN: A Scalable Mobile Collaborative Platform for
Opportunistic Sensing Applications,” Transactions on Collaborative Computing, Is-
sue 1, 2014.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 156 IoT Technical Challenges and Solutions

Kakaes, K., et al., “Drones and Aerial Observation: New Technologies for Proper-
ty Rights, Human Rights and Global Development: A Primer,” July 2015, http://
drones.newamerica.org.
Krause, A., et al., “Toward Community Sensing,” Proc.7th ACM/IEEE IPSN, 2008,
pp. 481–492.
Lane, N. D., et al., “A Survey of Mobile Phone Sensing,” IEEE Communications
Magazine, September 2010.
Li, Q., “Security, Privacy & Incentive Provision for Mobile Crowd Sensing Sys-
tems,” Doctoral Dissertation, Pennsylvania State University, University Park, PA,
2013.
Máthé, K., and Lucian Buşoniu, “Vision and Control for UAVs: A Survey of Gen-
eral Methods and of Inexpensive Platforms for Infrastructure Inspection,” Sen-
sors, Vol. 15, No. 7, June 25, 2015.
Pal, A., et al., “Road Condition Monitoring and Alert Application,” Pervasive Com-
puting and Communications Workshops (PERCOM Workshops), 2012.
Simshaw, D., et al., “Regulating Healthcare Robots: Maximizing Opportunities
While Minimizing Risks,” Richmond Journal of Law and Technology, Vol. 22, No. 3,
2016, http://jolt.richmond.edu/v22i2/article3.pdf.
Whitehead, K., and C. H. Hugenholtz, “Remote Sensing of the Environment with
Small Unmanned Aircraft Systems (UASs), Part 1: A Review of Progress and
Challenges,” J. Unmanned Veh. Syst., Vol. 2, 2014, pp. 69–85.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 6
Democratizing Analytics: Analytics as a Service

6.1 The Need for IoT Analytics

In the IoT context, there are physical objects in the real world (e.g.,
buildings, cars, roads, environment, human bodies), which, when
perturbed by external stimuli in form of physical events (e.g., fire,
earthquake, traffic congestion, faults, disease), generate specific
sensor signals from specific features of interests (e.g., temperature,
vibration, location, speed, physiological parameters). Sensors are
the transducers placed on the physical objects that can observe the
property change of the feature of interests and provide that infor-
mation in the form of sensor signals. Sensor informatics, covered
in Chapter 4, extracts information from the sensor signal. The role
of analytics is to ingest this information and provide intelligence
from it. The whole process is outlined in Figure 6.1.
The objective of any IoT analytics application is the derivation
Copyright © 2016. Artech House. All rights reserved.

of physical world contexts and building of models and insights

from the collected sensor data (Figure 1.4 from Chapter 1) with
the main purpose of deriving intelligence from observed sensor
information. The intelligence can be derived in different ways
providing different levels of insights depending upon application

157

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 158 IoT Technical Challenges and Solutions

Figure 6.1 Sensor-driven analytics.

requirements. Depending upon the kind of contextual questions

that we try to answer in the analytics system, there can be four
broad types of analytics: descriptive, diagnostic, predictive, and
prescriptive. We have redrawn Figure 1.4 as Figure 6.2, in which
the mapping of these four types of analytics is clearly articulated
in the reference of the knowledge pyramid.
In Table 6.1, we map a few IoT analytics use cases with different
types of analytics and the type of questions that they try to an-
swer. Descriptive analytics can be described as the analytics to find
Copyright © 2016. Artech House. All rights reserved.

Figure 6.2 Different types of analytics.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 159

Table 6.1
Different Types of Analytics
Contextual Question Analytics Type IoT Analytics Case
Who, What, Where, When? Descriptive Smart parking, pollution monitoring
How did it happen? Diagnostic Hazardous scenario detection, water
leakage detection, perimeter access
control, elderly people monitoring:
fall or inactivity detection
Why did it happen? What is Predictive Electricity grid monitoring and
expected to happen? control, monitoring for people with
chronic disease, alert generation,
predictive analytics and disease
prognosis for health, predictive
maintenance of machines
What should be done to Prescriptive Traffic congestion management,
make it happen or not electricity peak load management,
happen? hazardous scenario response:
emergency evacuation

contextual information in collected sensor data trying to answer

questions like who, what, when, and where. It can be used in many
cases where data visualization in different contextual dimensions
of identity, location, and time can be enough to provide the nec-
essary insights. Typical examples are visualization of parking lot
vacancy using occupancy sensors or pollution level of a city from
various pollution sensors located at various locations of the city,
both being visualized as an overlay of geo-physical map. Howev-
er, given that a specific stimulus causes specific observation on the
sensor, diagnostic analytics can be defined as the inverse problem
of finding the stimuli given the observation. If the model generat-
ing the observation from stimuli is known, the inverse model can
be easily found out. Such analytics typically tries to answer how a
particular event occurred. Examples of diagnostic analytics are to
find out location, time, and subject of specific events from indirect
measurement of sensors; it can be intrusion detection in campus
Copyright © 2016. Artech House. All rights reserved.

surveillance, leakage location detection in a water supply, or fall or

inactivity detection monitoring for elderly people at home. Predic-
tive analytics is another form of analytics used for predicting the
future state of the physical system given the current and historical
sensor data. It usually needs knowledge and reasoning of why a
particular phenomenon has occurred to predict the future behav-
ior. It usually tries to answer questions like why a particular event
happened and from there try to predict what is expected to happen

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 160 IoT Technical Challenges and Solutions

in future. Typical examples are finding out faults in electricity grids

or screening people for likely disease risks or finding and predict-
ing faults in factory machines. Finally, prescriptive analytics can
be regarded as doing deterministic or stochastic optimization on
the overall system with historical sensor data as input to prevent
or aid future event occurrences. Prescriptive analytics is also very
useful for performing what-if analysis on a given physical system
to discover the best possible prescription. It can try to answer ques-
tions like what should be done prevent an unwanted event from
happening. Typical examples include real-time traffic congestion
management using vehicle location data, demand-response con-
trol of electricity to shave off-peak load demand, or planning of
emergency evacuation routes in real time avoiding congestion.
As an example to dive deeper, let us take the use case of pre-
dictive maintenance of machines. The health of machines can be
sensed via vibration, thermal and acoustic sensors. Each of these
sensors generates one-dimensional and two-dimensional (2-D)
time series (2-D in case of images, like thermal images); this time
series is also known as sensor signals. In the given use case, it may
be the vibration sensor signal generated from an accelerometer
sensor along with a 2-D video of the thermal profile of the ma-
chines. A simple machine learning-based diagnostic analytics on
this needs the following steps:

• Clean up the signal and remove all undesired artifacts (includes

filtering of the sensor signal for noise removal and removal of
chunks of time series altogether where the noise cannot be can-
celled effectively).
• Label the cleaned signal in time with information on when the
machine was “good” and when the machine went “bad.”
• Find out by painstaking causal correlation analysis which
Copyright © 2016. Artech House. All rights reserved.

features of the cleaned signal has causal relationship with the

“good” and “bad” condition of the machine; this is also called
labeling. Knowledge of the physics of the machine operation
can also help in this.
• Use the collected labeled sensor signals to train a machine
learning-based classifier. The trained classifier provides the
model mapping the physical problems in the machine to pat-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 161

terns in the sensor signal. This model can be improved with the
scientific knowledge of design and operation of the machine.
• Use the learned and designed model to diagnose from the pat-
terns of the sensor signal whether the machine has gone “bad”
or which parts of the machine have gone “bad.” This is known
as diagnostic analytics.
• Going one step further, if sensor signal can be collected over
the lifespan of a machine gradually going from a good state to
a bad state, a regression-type predictive model can be built to
predict the failure of similar other machines. This can lead to
timely intervention through predictive maintenance reducing
machine downtime and cost of repair.

As obvious in the above example, the analytics process is quite

complex requiring knowledge from multiple stakeholders that in-
cludes programmers, data scientists, mechanical engineers, physi-
cists, and machine maintenance experts. Hence, there is need for
such analytics to be abstracted out as a service with concerns and
knowledge of each stakeholder taken into account. We discuss this
in detail in the next section.

6.2 The Need for Analytics as a Service

Any typical information technology (IT) development involves

three stakeholders: business people (who provide the require-
ments), developers (who architect and write the code), and opera-
tions people (who deploy the system and maintain it). In Agile de-
velopment process parlance [1], this is also known as Biz-Dev-Ops
[2, 3]. An efficient plan for IT software development always in-
volves continuous interaction of the three stakeholders. However,
Copyright © 2016. Artech House. All rights reserved.

the development of IoT analytics applications becomes a bit more

complex. This typically involves a fourth entity other than the three
standard Biz-Dev-Ops people, an entity who understands the why
and how of physical world or the physicists, chemists, and biolo-
gists of the world (jointly referred to as scientists). We collectively
call this new development paradigm Biz-Sci-Dev-Ops: the four
stakeholders and their interdependency are depicted in Figure 6.3.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 162 IoT Technical Challenges and Solutions

Figure 6.3 The four stakeholders for IoT analytics development.

We take another example of building an IoT-enabled remote

healthcare application where physiological sensor data on wear-
ables worn by patients are collected 24 hours per day, 7 days per
week and analytics is done that data provide real-time alerts on
abnormal conditions and historical insights on the patient’s health.
Here the Biz role is typically played by healthcare providers focus-
ing on requirement generation. The Sci role is played by doctors
and biomedical specialists. In the Dev role, we have standard IT
programmers and architects who not only create a system that col-
lects sensor data in real time and sends it to the cloud, but also
write the analytics application to generate real-time anomaly alerts
and historical insights based on the algorithms and models pro-
vided by the Sci experts. Finally, we have the IT infrastructure ex-
Copyright © 2016. Artech House. All rights reserved.

perts in the Ops role who take the applications and systems de-
veloped by the Dev teams and deploy it in sensor network cum
cloud-based IoT infrastructure (Figures 2.1 and 2.4). For the ma-
chine predictive maintenance example given earlier, the Biz person
is the factory maintenance person, the Sci person is the physicist
or mechanical engineer who was involved in the machine design

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 163

or the person who understands the machine design. The Dev and
Ops roles remain similar to the healthcare example mapping to
programmers and application developers and IT infrastructure ex-
perts, respectively.
Traditionally, IoT applications are built bottom-up as per verti-
cal requirements starting with sensor integration followed by sen-
sor data collection using sensor networking, storing the collected
data, and finally analyzing the stored data and streamed data to
draw actionable insights. However, instead of taking the bottom-
up vertical approach for application development, from the cost of
development, ease of development, and code reuse perspective, it
is beneficial to have a horizontal, platform-driven approach. Many
of the existing IoT service platforms support features like user
management, resource provisioning, application life-cycle man-
agement, device management and configuration, and connectiv-
ity service provisioning and management. Another such platform
[TCS Connected Universe Platform (TCUP) [4]] tries to address a
few of the above concerns by providing an integrated application
development platform covering device management, data storage
and management, an application programming interface (API)
-based application development framework and a distributed ap-
plication deployment framework.
What is primarily missing in the above frameworks is a service
layer abstracting analytics to a higher level, thereby making it easy
to use. The study of IoT analytics problems (all of which is handled
in detail in Chapter 4) clearly indicates that the developers of IoT
analytics systems typically spend maximum of their development
time in few specific tasks:

1. Clean up the sensor signal from unwanted artifacts; this is also

known as anomaly detection.
2. Understand which features of the collected sensor signal (or
Copyright © 2016. Artech House. All rights reserved.

image or video) causally correlates with the observed physi-

cal event; this is known as feature engineering. It can also be
termed as dimensionality reduction.
3. Remove the redundancy from the sensor signal while preserv-
ing the desired events; this is also known as compression.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 164 IoT Technical Challenges and Solutions

What is not mentioned in Chapter 4 is the practical fact that

anomaly detection, feature engineering, and compression remain
the key challenges in any IoT analytics development. The main
problem here for these challenges lie with choosing the right al-
gorithms from multitude of algorithms available, each working
better for a specific type of sensor signal. Hence, at the very least,
the analytics as a service should provide sensor-agnostic anomaly
detection as a service, sensor-agnostic feature engineering as a ser-
vice and sensor-agnostic compression as a service. Once the sen-
sor data has been cleaned, all redundancies are removed, and the
right features are extracted, there still remains the challenge for
modeling the data to draw the right insights from it; this typically
involves combination of science-based modeling and data-based
modeling (also known as machine learning), which we can call
modeling as a service. Here we are trying to highlight the need for
automation services in implementation of the systems described in
Chapter 4 given their complexity and multidisciplinary nature of
such systems that can make the implementation of such systems
easier for developers who are not analytics algorithm experts.
These services partially address the issues related to ease of
analytics application development like code reusability, need for
multiple skills in domain, analytics, sensors, or programming or
visibility of data across applications. However, to address these is-
sues fully, we need a more holistic and top-down approach. One
way to address all the relevant issues is through the concept of
separation of concerns among different stakeholders; this has been
prevalent in the area of model-driven development (MDD) [5, 6],
and recently it was also applied in the context of IoT to provide
a framework to specify the requirements at different levels [7].
However, these frameworks do not address issues like analytics
algorithm reuse, distributed execution of analytics, generation of
analytics, and reasoning workflows, a common ontology and se-
Copyright © 2016. Artech House. All rights reserved.

mantics for sensor data needed for a full-fledged IoT system. The
need for a distributed computing environment to deploy and ex-
ecute the analytics across multiple computing entities distributed
in space while still meeting timing requirements adds interesting
new dimensions to the analytics deployment challenges. This be-
comes further challenging with the advent of multicore processors

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 165

[8] and accelerators like graphical processing units (GPUs) [9] on

single computing platforms.
Hence, a model-driven development (MDD) approach that can
abstract out the metamodel of the IoT system and automate much
of the analytics application development and deployment process
allowing the application developer to focus only on application re-
quirement specific concerns will be of high interest to the industry.
We discuss this in detail in the next section.

6.3 Analytics as a Service for Developers: Model-Driven IoT

There has been quite a bit of ongoing work on MDD systems on the
implementation side. One example of such effort is from an organi-
zation called OASIS (Advancing Open Standards for the Informa-
tion Society) [10]. They have brought a new standard metamodel
for IT services “The Topology and Orchestration Specification for
Cloud Applications” (TOSCA) [11] for improving portability of
cloud applications in heterogeneous application environments.
Here the structure of a service is defined by the topology template
with a directed graph, which, in turn, consists of node templates
and relationship templates. TOSCA allows defining complex
workflows that can be used for the management process of cre-
ating, deploying, and terminating services. It specifies an Exten-
sible Markup Language (XML)-based syntax for describing the
basic data and components [12], while existing business process
modeling languages such as Business Process Execution Language
(BPEL) [13] and Business Process Model and Notation (BPMN)
[14] are used to describe plans.
An open source implementation and development framework
is available in OpenTOSCA [15]. The architecture [16] supported
in OpenTOSCA is outlined in Figure 6.4. It has descriptors for end
Copyright © 2016. Artech House. All rights reserved.

points, instances, and models along with modeling tools. It has

a planning engine and a controller engine to orchestrate the ser-
vices, an implementation artifact engine to execute the services,
all available through suitable APIs. Such frameworks, while hav-
ing most of the features for cloud-based application development
platforms, generally lack support for sensor-specific systems and
analytics required in an IoT scenario. They definitely support
Dev and Ops and, to some extent, requirements of the business

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 166 IoT Technical Challenges and Solutions

Figure 6.4 Open TOSCA architecture.

processes (Biz), but they complete lack support of the Sci for IoT
analytics applications development. Their support from Biz, Dev,
and Ops also lacks the support of sensors, sensor integration to
the cloud, and cloud-based sensor device management. However,
these drawbacks can be addressed through addition of sensor in-
tegration and management functionalities in the IoT platform and
through knowledge-driven architectures as outlined in Figure 6.5.
Such knowledge-driven architectures can support modeling of
knowledge of the different expert stakeholders described in Fig-
ure 6.3, thereby allowing separation of their concerns. An applica-
tion developer can take advantage of the knowledge of the other
stakeholders encoded in the system to quickly develop analytics
applications.
The different knowledge bases described in Figure 6.5 is out-
lined next.

• Biz knowledge base: Application developers are mainly con-

Copyright © 2016. Artech House. All rights reserved.

cerned about solving a specific set of domain problems. They are

expected to have good programming skill but limited knowl-
edge about the domain applications and the kind of sensors to
be deployed towards those requirements. The domain-specific
knowledge base that is populated by domain experts intends
to bridge this gap by providing knowledge like mapping be-
tween physical phenomena and sensor observation, mapping
between sensor application and sensor technology.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 167

Figure 6.5 Knowledge base to support different stakeholders.

• Sci knowledge base: The Sci knowledge base consisting of a

repository of analytics algorithms is contributed mainly by the
algorithm writers and scientists. It not only contains the archive
of algorithm executables in form of libraries, it also contains
metadata about algorithms detailing their application areas,
limitations, performance parameters, accuracy, CPU complex-
ity, and memory load. This knowledge base can be used by the
Copyright © 2016. Artech House. All rights reserved.

application developers to query and look for specific algorithms

suitable for their application.
• Dev knowledge base: In addition to the basic development
support available in cloud-based systems like TOSCA, the Dev
knowledge base can provide details on sensor communication
interfaces and protocols, device drivers, and sensor data models.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 168 IoT Technical Challenges and Solutions

• Ops knowledge base: After the development of sensor analyt-

ics applications for the operations, people need to deploy them.
Usually, part of the applications runs on edge or gateway de-
vices collecting sensor data and part of the applications runs
on the cloud. In addition to what is available in existing frame-
works, the Ops knowledge base needs to collect information
about the computing, memory, and communication capabilities
of the available gateway devices, available gateway-to-cloud
communication channels, storage capacity of the cloud, and de-
tails of available hardware in infrastructure of the cloud and
map them to the information available from the Sci knowledge
base.

6.4 An Example of a Model-Driven IoT Framework

An IoT analytics application can have multitude design concerns:

domain information, development and orchestration, and infra-
structure (sensors, devices, networks, and cloud). This is depicted
in Figure 6.6.
Figure 6.6 also shows how these three different design concerns
map to the four different knowledge bases described earlier. For
example, domain information needs both Biz and Sci knowledge
bases, development and orchestration needs both Sci and Dev
knowledge bases, and finally infrastructure needs Ops knowledge
bases.

6.4.1 Domain Concern

Domain concern can be viewed as a set of data flows from sensor

to sinks (actuator, database, reports, visualizer) traversing many
Copyright © 2016. Artech House. All rights reserved.

computing operations that transforms the data to various informa-

tion elements. This information flow can be modeled as a directed
graph with nodes as computing modules that computes the desig-
nated information element. The edges indicate the input/output
dependency relation. This also serves as a semantic model for the
IoT application. It can also map to Biz knowledge base through
a suitable semantic mapping. Figure 6.7 depicts the flow graph
corresponding to an example vehicle telematics service where we

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 169

Figure 6.6 IoT service model.

clearly show how the information flows from sensors to feature of

interests to context to insights finally leading towards value-added
application.
It not only uses the Biz knowledge base, but also the Sci knowl-
edge base to infer context, insights, and value from the sensor
data. The nodes requires a specification of the computing model
or algorithm to be used for computing the designated information
element(s); this model comes from the scientific knowledge about
the physical event and can be specified as a node property.

6.4.2 Development and Orchestration Concern

This view captures the information to develop analytics applica-

tion via interconnecting various analytics algorithm blocks, build
Copyright © 2016. Artech House. All rights reserved.

the corresponding executable software modules, and test them for

deployment in the target devices. Also, it can specify the opera-
tional behaviors of the entire system that need to orchestrate dur-
ing the operation. There is need to have an underlying metamodel
to represent all the dimensions of an IoT service that can provide a
framework to build and deploy the application from its specifica-
tions that includes communications, temporal behavior of the sys-
tem, and business process orchestrations. It can use TOSCA-like

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 170 IoT Technical Challenges and Solutions

Figure 6.7 Example information flow graph (vehicle telematics).

Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 171

basic framework on top of which this service orchestration layer

can be added. The analytics flow suggested by the Sci knowledge
base can be made available in form of an algorithm repository
from which the complete analytics workflow can be created. The
workflow created at this step is then deployed on the test devices
for validation. There could be standard reusable computing blocks
that can be used across applications or there can be application
specific algorithms. The analytics can also include rule engines uti-
lizing the knowledge bases and can be done offline or in real time
using complex event processing.

6.4.3 Infrastructure Concern

In a typical IoT application, sensing and computing may span mul-

tiple devices and platforms involving sensor device, gateways, and
cloud platforms. We can view this as partitioning of the computing
flow graph into subgraphs and binding suitable computing devic-
es to execute them. Specific device with its detailed specifications
can be selected from the infrastructure knowledge base and the re-
lated node property set to depict the compute specifications (CPU,
clock, number of cores, GPU availability, memory). The executa-
bles created as part of the workflow during service development
and orchestration needs to be mapped and matched into the right
(and available) infrastructural choice. Once the computing opera-
tions are matched and mapped, the communication mechanism
between these modules needs to be specified. The modules within
same computing environment can use standard parameter passing
mechanisms like API, messaging, or interprocess communication
primitives. Communication across devices needs external inter-
faces and protocols such as USB [17], Bluetooth [18], Zigbee [19],
Wi-Fi [20], and 2G/3G [21]. Further, the nature of data exchange
may follow high-level syntactical models like REST, Pub-Sub, and
Copyright © 2016. Artech House. All rights reserved.

Proxy. These were described in detail in as described in Chapter

2. The edges in the flow graph model can be used to capture the
communication interface details and the infrastructure knowledge
base can be used to check the compatibility and consistency. It is
obvious that the partitioning of workflow between sensor nodes,
edge gateways, and the cloud depend on a number of factors like
hardware configuration of these systems, complexity of the algo-
rithms to be run and real-time requirement from the applications.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 172 IoT Technical Challenges and Solutions

Edge computing mechanisms described in Section 2.5 can be used

to deploy and operate such systems. In most of the application use
case scenarios, this partitioning of computing can be done stati-
cally between sensor nodes, edge gateways, and the cloud. How-
ever, in case of applications with strict real-time requirements and
dynamically changing environments like robotic navigation, there
may be a need to dynamically partition and allocate workflow sub-
tasks to different components of the infrastructure mapping their
capabilities and also keeping in mind the power and communica-
tion considerations.

6.5 Summary

In this chapter, we first established the need for IoT analytics citing
some example use cases. We took an example use case of predic-
tive maintenance of machines to elaborate on the complexity of the
analytics development process. Then we elaborated the need for
analytics as a service for IoT, taking an example of health and well-
ness via physiological sensing and suggested how services like
anomaly detection, feature engineering, and compression, if done
in a sensor-agnostic way, can really add value to the IoT analytics
developer. We also introduced the need for learning models from
the cleaned up sensor data using the hybrid knowledge of physi-
cal sciences and data sciences there by establishing the multidisci-
plinary skill requirement for IoT analytics. This, when combined
with the typical IoT analytics development project requirement of
involving multiple stakeholders in form of Biz-Sci-Dev-Ops, point
towards MDD as a possible approach towards separating and ad-
dressing the concerns of each of the stakeholders. We introduced
knowledge models for different these stakeholders of IoT analytics
and identified four different knowledge models for four different
Copyright © 2016. Artech House. All rights reserved.

stakeholders and three different stakeholder concerns around that.

For illustration, we described an example framework for applica-
tion development for IoT platforms.
It should be noted that the concept of MDD-based analytics as
a service specific to IoT is a recent one. It is possible to address
this by creating knowledge models for various stakeholders and
separate out concerns of different stakeholders in IoT application

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 173

development through use of models and knowledge bases. This is

extremely important to improve the ease of application develop-
ment and democratizing analytics for IoT. Such democratization
would lead towards widespread adoption and deployment of IoT
analytics across multiple applications areas.

References
[1] http://agilemethodology.org/.
[2] https://blog.appdynamics.com/devops/
new-modern-family-business-development-operations-bizdevops/.
[3] ����������������������������������������������������������������
http://www.tcs.com/resources/newsletters/Pages/Biz-DevOps-Enter-
prise-Agility.aspx.
[4] http://www.tcs.com/about/research/Pages/TCS-Connected-Universe-
Platform.aspx.
[5] http://www.agilemodeling.com/essays/amdd.htm.
[6] https://msdn.microsoft.com/en-us/library/aa964145.aspx.
[7] http://www.w3.org/2014/02/wot/papers/prehofer.pdf.
[8] https://software.intel.com/en-us/articles/
frequently-asked-questions-intel-multi-core-processor-architecture.
[9] http://www.nvidia.com/object/what-is-gpu-computing.html.
[10] https://www.oasis-open.org/org.
[11] https://www.oasis-open.org/committees/tc_home.php?wg_abbrev=tosca.
[12] http://www.w3schools.com/xml/.
[13] http://searchsoa.techtarget.com/tutorial/BPEL-tutorial.
[14] http://www.bpmn.org/.
[15] http://www.iaas.uni-stuttgart.de/OpenTOSCA/indexE.php.
[16] http://www.iaas.uni-stuttgart.de/OpenTOSCA/container_architecture.
php.
Copyright © 2016. Artech House. All rights reserved.

[17] http://www.usb.org/home.
[18] https://www.bluetooth.com/.
[19] http://www.zigbee.org/.
[20] http://www.wi-fi.org/.
[21] http://www.3gpp.org/.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 174 IoT Technical Challenges and Solutions

Selected Bibliography
Balamurali P, P. Misra, and A. Pal, “Software Platforms for Internet of Things and
M2M,” Journal of the Indian Institute of Science, A Multidisciplinary Reviews Journal,
Vol. 93, No. 3, July–September 2013.
Bandyopadhyay, S., et al., “Demo: IAS: Information Analytics for Sensors,” Proc.
of the 13th ACM Conf. on Embedded Networked Sensor Systems, 2015, pp. 469–470.
Bonomi, F., et al., “Fog Computing and Its Role in the Internet of Things,” AMC
Proc. of the 1st Edition of the MCC Workshop on Mobile Cloud Computing, MCC ’12,
New York, NY, 2012, pp. 13–16.
Chattopadhyay, T., et al., “Automated Workflow Formation for IoT Analytics: A
Case Study,” IoTaaS, IoT360, Rome, Italy, October 2015.
Dasgupta, R., and S. Dey, “A Comprehensive Sensor Taxonomy and Semantic
Knowledge Representation: Energy Meter Use Case,” 7th Intl. Conf. on Sensing
Technology, 2013.
Dey, S., et al., “A Semantic Sensor Network (SSN) Ontology Based Tool for Se-
mantic Exploration of Sensor,” Semantic Web Challenge Competition ISWC, 2014.
Ghose, A., et al., “Design Insights for a Mobile Based Sensor Application Frame-
work: For Aiding Platform Independent Algorithm Design,” 14th Intl. Conf. on
Information Processing in Sensor Networks (IPSN), 2015.
Gubbi, J., et al., “Internet of Things (IoT): A Vision, Architectural Elements, and
Future Directions,” Elsevier Journal on Future Generation Computer Systems, Vol. 29,
2013, pp. 1645–1660.
Jaiswal, D., et al., “Demo: A Smart Framework for IoT Analytic Workflow Devel-
opment,” Proc. of 13th ACM Conf. on Embedded Networked Sensor Systems, 2015, pp.
455–456.
Köhler, M., D. Wörner, and F. Wortmann, “Platforms for the Internet of
Things: An Analysis of Existing Solutions,” http://cocoa.ethz.ch/down-
loads/2014/02/1682_20140212%20-%20Bocse.pdf.
Li, F., et al., “Towards Automated IoT Application Deployment by a Cloud-Based
Approach,” IEEE 6th Intl. Conf. on Service-Oriented Computing and Applications
(SOCA), 2013.
Mistry, S., et al., “P2P-Based Service Distribution over Distributed Resources,”
2015 IEEE 29th Intl. Conf. on Advanced Information Networking and Applications
Copyright © 2016. Artech House. All rights reserved.

(AINA), 2015, pp. 515–520.

Mukherjee, A., et al., “ANGELS for Distributed Analytics in IoT,” 2014 IEEE World
Forum on Internet of Things (WF-IoT), 2014, pp. 565–570.
Pal, A., P. Balamuralidhar, and A. Mukherjee, “Model-Driven Development for
Internet of Things: Towards Easing the Concerns of Application Developers,” Io-
TaaS 2014, IOT 360, Rome, October 2014.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Democratizing Analytics: Analytics as a Service 175

Patel, P., B. Morin, and S. Chaudhary, “A Model-Driven Development Framework

for Developing Sense-Compute-Control Applications,” MoSEMInA ’14, May 31,
2014.
Patel, P., et al., “Enabling High-Level Application Development in the Internet of
Things,” in Sensor Systems and Software, New York: Springer, 2013, pp. 111–126.
Patel, P., and D. Cassou, “Enabling High-Level Application Development for the
Internet of Things,” Journal of Systems and Software, Vol. 103, 2015, pp. 62–84.
Paul, H. S., et al., “Compute on the Go: A Case of Mobile-Cloud Collaborative
Computing Under Mobility,” in Adaptive Resource Management and Scheduling for
Cloud Computing, New York: Springer, 2015, pp. 78–90.
Shahnawaz, A. et al., “Mobile Sensing Framework for Task Partitioning Between
Cloud and Edge Device for Improved Performance,” ISCC, 2016.
Soukaras, D., et al., “IoT Suite: A Tool Suite for Prototyping Internet of Things Ap-
plications,” 4th Intl. Workshop on Computing and Networking for Internet of Things
(ComNet-IoT), co-located with 16th Intl. Conf. on Distributed Computing and Network-
ing (ICDCN), 2015, p. 6.
Ukil, A., et al., “Adaptive Sensor Data Compression in IoT Systems: Sensor Data
Analytics Based Approach,” 2015 IEEE International Conference on Acoustics, Speech
and Signal Processing (ICASSP), 2015, pp. 5515–5519.
Ukil, A., S. Bandyopadhyay, and A. Pal, “IoT Data Compression: Sensor-Agnostic
Approach,” IEEE Data Compression Conference (DCC), 2015, pp. 303–312.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 7
The Real Internet of Things and Beyond

7.1 Realistic Internet of Things

In the midst of excitement and hype, there are several practical

deployments of IoT happening today delivering realistic business
benefits. Examples include fleet monitoring, building monitoring,
customer intelligence in retail stores, telemedicine, navigation,
and tracking. The potential user community is becoming aware
of this technology and what it can offer. Some restaurant chains
use remote monitoring of cooler and freezer temperatures to avoid
thousands of dollars in potential spoilage. Retail stores see the
increased level of customer engagement that a beacon platform
provides, providing customers valuable brand information and
generating buzz and increased sales. There is an interesting case
study on how an IoT-based energy management system delivered
the return on investment (ROI) in 18 months [1]. Smart vending
Copyright © 2016. Artech House. All rights reserved.

allows machines to monitor their own inventory levels, calling

in for deliveries only when needed and with a list of what items
are needed, saving delivery drivers valuable time. Fleet and as-
set tracking provides not only for dispatching more efficiently but
also monitoring driver behavior, which can also save on the cost of

177

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 178 IoT Technical Challenges and Solutions

insurance and provide management with valuable information for

driver training. Monitoring and processing of evaporation credits
for cooling towers and landscape irrigation can save many dollars
in water and sewer bills.
However, is it a great success story for all IoT applications? The
truth is the count of failed IoT projects has outnumbered successful
projects by a huge margin [2]. There may be several reasons for the
failures and any emerging technology may face this challenge on
its travel to maturity. In this context, it may be appropriate to ex-
amine the concept of a real Internet of Things, by which we mean
an IoT solution with an associated business model that provides
ROIs on realistic business applications. Even for an application
focusing on social impact, it should have a business model to en-
sure its sustainability. For a data-gathering application, it means
return of invesetment (ROI), the ratio of business value of the data
collected, and the cost of acquiring it should be comfortably high.
There are several factors contributing to the success and failures of
IoT projects and we will discuss a few of them next.

7.1.1 Key Contributing Factors to Real IoT

The challenges with IoT go beyond making and connecting de-

vices that work. The integrated product and services need to work
seamlessly, almost invisible to an end user. As Mark Weiser sug-
gests in his ubiquity paradigm, we need machines that fit the hu-
man environment instead of forcing humans to enter theirs [3]. The
service needs to meet needs, easily integrate into daily life or the
industrial process, and has to enhance the user’s life or the busi-
ness process. This requires the reliability and robustness of all the
components of the integrated solution. Sometimes the system is
over engineered to deliver the application goals. The system ar-
chitecture should be pruned and tuned to the business goals and
Copyright © 2016. Artech House. All rights reserved.

deployment context.
The following are several factors contributing to a real IoT.

7.1.1.1 Robust Devices and Connectivity

Devices that work reliably in the deployment environment that do
not shut down or restart due to failures or bugs are key to any IoT
application. Impact and consequences of a buggy device may vary
from a consumer application to an industrial application. Also, the

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 The Real Internet of Things and Beyond 179

Figure 7.1 Major contributing factors to real IoT.

devices should enable system level security in terms of identity,

authentication, and tamper-proofing.
Another aspect to consider is the choices that one has to make in
terms of technologies and standards for a cost-effective IoT system.
There are many IoT standards at various maturity levels address-
ing various aspects of the system. One should expect to change
the architecture as the technology evolves. The curse of choices for
almost every component also makes the system design difficult.
There are multitudes of silicon vendors, device platforms, energy
sources, communication protocols, and backend software plat-
forms from which to choose. One should also recognize that there
is a trade-off between robustness and cost effectiveness, and this
should be addressed within the larger context of the application.
For large-scale deployments, remote management is a key re-
quirement and it contributes to system robustness as well. Ap-
proaches such as aggregate programming should be used to
Copyright © 2016. Artech House. All rights reserved.

simplify the design, creation, and maintenance of complex IoT

software systems. Here, the basic unit is no longer a single device,
but instead a cooperating collection of devices: details of behavior
and position and number of devices are largely abstracted away.
This can be accomplished by a layered approach to programming
complex services, building on foundational work on the composi-
tion of distributed systems and then on general mechanisms for
robust and adaptive coordination.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 180 IoT Technical Challenges and Solutions

7.1.1.2 Functionality and Processes

From a functional perspective, IoT systems in general are expected
to collect sensor data, transport it to processing and decision-mak-
ing locations, and enable the implementation of actionable insights.
Real-time analytics will be a requirement involving business in-
telligence to machine learning, data mining, predictive analytics,
condition monitoring, and visualizations. This often involves com-
plex processes and possibly a model-driven approach would help
to address the complexity. For example, business process model-
ing (BPM) is an established technique for modeling and executing
complex processes in enterprises. The enterprise adoption of IoT
technologies could be accelerated if these techniques are adapted
to the requirements of real-world IoT.

7.1.2.3 Keep the End User in Focus

User experience is a key factor for the success of IoT applications.
It is important to keep the user at the center to validate the IoT
product’s value to the user. Mere technology “wow” factors can-
not ensure the longevity of the device and application. The value
proposition should consider social context and expectations as
well. For example, driving behavior monitoring for a usage-based
insurance premium had mixed acceptance with personal vehicles.
At the same time, it had a taste of success in fleet-monitoring ap-
plications. Here for the same solution, the end user and the context
were different. In the first case, the car owner is the end user who
has the choice of enrolling in the scheme, and there is a shadow of
privacy concern. For the second use case, the fleet manager is the
end user and he or she has the value proposition and control to
implement the solution.

7.1.2.4 Appropriate Business Model

Copyright © 2016. Artech House. All rights reserved.

The business model is a plan for the successful operation of a busi-

ness, identifying sources of revenue, the intended customer base,
products, and details of financing. A strong business model is the
key to sustain any IoT application, and IoT has the potential to en-
able new business models. The World Economic Forum outlined
four phases of adoption of industrial IoT applications in their re-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 The Real Internet of Things and Beyond 181

port [4]. They are operational efficiency, new products and services,
outcome economy, and autonomous and pull economy. In the ini-
tial phase, focus can be given to improving operational efficiency
through various means such as improved asset utilization, opera-
tional cost reduction and worker productivity. In the second phase,
new products and services can be focused. This may involve new
software services, data monetization, and payment models such as
pay per use. Outcome economy will focus on systems by forming
and leveraging new connected ecosystems, platform-enabled mar-
ketplaces, and new payment processes. Finally, autonomous and
pull economy is leveraged by continuous demand sensing, end-to-
end automation, resource optimization, and waste reduction.
In each of these phases, there is a potential to explore appro-
priate new business models. Remote monitoring and maintenance
are an upcoming service and business model enabled by IoT. New
payment models that will enable charging for usage and service
quality levels are promising. “TotalCare” aerospace service from
Rolls Royce is an example of a business model applicable to IoT
where the payment mechanism for aircraft engines is $/engine fly-
ing hours. Supporting such a service requires extensive sensing,
monitoring, analytics, and prediction [5].

7.1.2.5 Develop an Ecosystem

For an IoT application to succeed, there should be an ecosystem
to support in terms of technology awareness, skill set, developer
community, and third-party services. Platform-based offerings
should consider providing easy access to third-party developers.
Opensourcing is being increasingly adopted for some of the hori-
zontal technologies those are hard to crack, and it may result in
a robust solution with the contribution from the diversity of the
community.
Copyright © 2016. Artech House. All rights reserved.

7.2 Real IoT Is a Network of Trade-Offs

There are several trade-offs to be made in IoT systems to be eco-

nomically and functionally viable and also efficient. IoT networks
in general have a highly distributed architecture. One of the major

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 182 IoT Technical Challenges and Solutions

underlying trade-offs is between computing, communication, and

storage across the network. How much data can be processed local-
ly so that the communication requirement can reduced? How can
the computations be restricted locally so that the device lifetime is
enhanced with the reduction of battery power consumption? How
long can the data be buffered so that the computing and commu-
nication can be scheduled in bursts to reduce the overall energy
consumption? A centralized cloud-based architecture in which the
sensor nodes directly communicate to the cloud offers easy man-
ageability of the network. However, the centralized architecture
can stress the communication network due to the transportation
of whole sensor data to the cloud. In a distributed computing ar-
chitecture, in-network computing is possible and data reduction
can happen at each intermediate node, enabling reduced overall
communication cost. Managing and updating distributed comput-
ing elements can be complex. This will lead to architectural trade-
offs between centralized versus hierarchical architecture and edge
computing versus cloud computing.

7.2.1 Some of the Common Trade-Offs Encountered in IoT Systems and Applications
7.2.1.1 Hard Sensing Versus Soft Sensing
Appropriate hard sensors can be used for every phenomenon or
feature to be monitored. In some cases, there is a possibility of us-
ing analytics and inference to reduce the number of hard sensors
required. While these soft sensors that are implemented in soft-
ware will help to reduce the sensor cost, it may increase the soft-
ware complexity and cost of software and computing platform de-
pending on where it is implemented. A suitable balance should be
made between hard sensing and soft sensing.

7.2.1.2 Application-Specific Devices Versus General-Purpose Device Platforms

Copyright © 2016. Artech House. All rights reserved.

Should we use application-specific optimized devices or general-

purpose device platforms such as Arduino and smartphone? Wear-
ables are often application-specific devices. Application-specific
devices offer a spectacular performance; they are often costly and
require a large scale-up to break even. However, general-purpose
devices will have a large user base and experience available to de-
velop solutions more quickly and cheaply, but they may not be

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 The Real Internet of Things and Beyond 183

the optimum platform to give the required performance and user

experience.

7.2.1.3 Security Versus User Experience

Security is a top concern of IoT systems. Due to the large attack
surface and evolving cyber-physical threats, the security schemes
required to be implemented often gets complex. This may likely
to affect overall user experience of the solution due to the added
interactions and processes required for the security scheme. The
same goes with privacy as well.

7.2.1.4 Battery Lifetime Versus Performance

High performance and quality often demand additional hardware
and software complexity with enhanced resources. Energy is a ma-
jor resource of importance to IoT devices and they need to have a
strategy to manage energy resources efficiently to maximize their
lifetime. Managing frequent battery replacement is a painful exer-
cise affecting user experience as well. This leads to need of a trade-
off between multiple energy-related system design parameters
and energy optimization is a major agenda in IoT system design.

7.2.1.5 Communication Range, Power, and Bandwidth

For transporting the data from source to sink, there are multiple
candidate communications standards and architectures possible.
A multihop communication with short-range links are one option.
However, one may use a single-hop, long-range link. For the same
bandwidth, short-range links require lesser power compared to
long-range wireless links. While multihop provides a flexibility
and redundancy, it may result in increased delay and may require
additional relay nodes. Narrow, long-frequency bands can offer
higher communication range at a low power (for example, low
Copyright © 2016. Artech House. All rights reserved.

power wide area network (LPWAN)), whereas higher-frequency

standards with wider bandwidths have higher power consump-
tion and a lesser range. There is a trade-off between frequency
bands, bandwidth, transmit power, and hop count in the IoT net-
work and a careful choice must be made to get a solution with
optimum performance.
An illustrative example of an IoT solution is discussed next.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 184 IoT Technical Challenges and Solutions

7.2.2 Safety on the Machine Floor: An Illustrative Example

Here we discuss an IoT application in a steel plant where the safe-

ty of human workers and the systems around them are focused.
There are crane operators in the plant who move iron ore and coal
into furnaces for steel making. There are others as well who assist
in the associated processes. It is obvious that they are working in
very dangerous and adverse conditions and their safety and well-
ness is of high value to them and the plant operation. The objective
of the IoT solution is to provide situation awareness on the state of
each worker and generate appropriate notifications, advice, and
guidance. This will lead to reduction on losses due to accidents,
disruptions, and low productivity and, at the same time, will im-
prove the safety and availability of the worker services.
Some of the states and events of interest with respect to a worker
are fall detection, incapacitation, and stress level. Use of wearables
that have accelerometers and photoplethysmograms (PPGs) could
be part of the solution. There can also be other modalities, such
as electrocardiograms (ECGs) and electroencephalograms (EEGs).
The wearable device can be based on a helmet, wrist, footwear, eye-
wear, and headband. From a human factors perspective, the device
should be nonobtrusive and much less intrusive. It should merge
with the normal operation and work style. The device would be
battery-powered; however, it should stay long enough to avoid
any frequent charge depletion.
The next issue is how the device can communicate to a cloud-
based application server, which will take care of the monitoring
process. The 2G/3G cellular connection is not a feasible one due
to power requirements. Can Zigbee, Wi-Fi, or Bluetooth offer a so-
lution? Yes, but they have a short range and a gateway device is
required to connect them to the cloud using a wireless wide area
network (WAN) or cellular network. Then where will the gateway
Copyright © 2016. Artech House. All rights reserved.

device be placed? Installing it in the operator cabin could be ac-

ceptable because each worker operates the crane from that cubicle
integrated with the crane system, although it is moving frequently.
The gateway device could be powered by the power supply avail-
able in the cabin. Alternately, can any of the LPWAN standards
be of use so that individual gateway device for each cabin could
be avoided? Instead, a single local gateway is sufficient to connect

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 The Real Internet of Things and Beyond 185

all the wearable devices of the workers, but are they highly band-
width constrained and will they be sufficient for the communica-
tion requirements?
Communication depends on where the data is required and
where it is processed. The sensor data processing pipeline includes
data preprocessing, filtering, motion artifact removal, parameter
computations, and event detection. One option is to implement the
whole pipeline in the wearable device itself and only the events
need to be communicated to the cloud. This will have an impact
on the computing, memory, and energy required for the wearable
device and may lead to a short battery life. However, there is an
option to distribute the computation between the wearable device,
the gateway, and the cloud. In this approach, the preprocessing
and other low-intensity computations resulting in a data reduction
can be planned to execute in the wearable device. The reduced data
can be further processed at the gateway level or cloud. If there is
a good connectivity available between the gateway and the cloud,
then it is advisable to do it in the cloud for ease of maintenance and
security.
Camera-based sensing is an alternate or complementary ap-
proach for the situation monitoring. There the computation re-
quirements are comparatively high, but it has the potential to
gather additional information. Here also one has to go through the
above analysis process to evaluate the solution architecture.
The exercise is not yet over. We need to look at the business
model as well. Who will build, operate, and maintain the solution
and how is the cash flow? One option is that the steel plant can
get this system implemented by an external vendor and they take
care of the operation and maintenance as well. This will involve a
one-time contract fee with an annual maintenance charge. Another
payment option is to charge per worker on an annual basis. Here
then data and services may be hosted by the vendor. Alternately,
Copyright © 2016. Artech House. All rights reserved.

this may be hosted in a cloud under the control of the steel plant.
There is a room for bringing in insurance companies as a stake-
holder in this business. Here the steel plant may engage with the
insurance company for covering damages related to safety and se-
curity. The insurance company, in turn, can engage a vendor for
deploying the above IoT solution supporting a feature set deter-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 186 IoT Technical Challenges and Solutions

Figure 7.2 Resilient IoT and related concepts.

mined from a risk analysis. There are many more possibilities for
the suitable business model.

7.3 Drivers for the Next Wave of IoT

The next wave of IoT will most likely to focus on robustness and
resilience by leveraging distributed embedded intelligence. A brief
conceptual overview of two major aspects, namely, resilient IoT
systems and cognitive IoT systems, is given next.

7.3.1 Resilient IoT Systems

Resilience is the capacity of a system to survive, adapt, and grow in

the face of unforeseen changes, even catastrophic incidents. A re-
silient system is characterized by their capability to resist external
disturbances and internal failures, ability to recover, enter stable
state(s), and adapt its structure and behavior to constant change.
Resilience is a desired feature for IoT systems having dynamics
and an element of control function involved in the operation. The
Copyright © 2016. Artech House. All rights reserved.

dynamism can be from various component failures such as sensor

failure, communication failure, or device failure. It can also come
from device mobility and changes in the ambience. Impact is more
felt in complex IoT systems such as smart grid, process control,
robotic operations, vehicular monitoring, and remote healthcare.
A useful way of looking into resilience would be to study the
resilience of complex systems. The aspects such as diversity and
adaptation are relevant in this context. Highly dependable systems

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 The Real Internet of Things and Beyond 187

such as avionics use diversity where diverse control systems (elec-

tronics, fluid, and mechanical control) are used to implement tri-
ple level redundancies. In a similar manner, IoT systems can also
leverage diversity in terms of sensor modalities, communication
paths, algorithms, and computing devices for robustness. Adapta-
tion is another aspect that IoT systems can use to modify itself to
address always changing and slowly evolving environments.
Resilience is also related to reliability, dependability, and trust-
worthiness. Reliability looks at design faults and operational fail-
ures to ensure functionality. Dependability is a measure of a sys-
tem’s availability, reliability, and maintainability. Another related
concept is trustworthiness in which a trustworthy system incorpo-
rates resilience, reliability, safety, security, and privacy.

7.3.2 Cognitive IoT

The concept of cognitive IoT explores the integration of human

cognitive process into system design. It embodies a cognitive cycle
that integrates the key processes of sensing, analyzing, learning,
deciding, and acting. It is envisaged that, in cognitive systems,
general objects behave as agents and interact with the physical
environment and/or social networks, with minimum human in-
tervention. Rather than being explicitly programmed, cognitive
computing learns and reasons from its interactions with humans
and from its experiences with its environment, enabling them to
keep pace with the volume, complexity, and unpredictability of
information generated by the IoT. The advantages are multifold:
saving people’s time and effort, increasing resource efficiency, and
enhancing service provisioning, to just name a few.
There are two aspects of cognitive IoT. In the first, the cognitive
approach is used to make the IoT system autonomous. It can en-
able the IoT with self-managing and self-healing capabilities and
Copyright © 2016. Artech House. All rights reserved.

resilience. The second aspect is that how IoT can be used to make
systems cognitive.
There are several aspects of cognitive systems that will be rel-
evant for IoT.

1. Deep human engagement: Cognitive systems enable enhanced

human interactions with people. Towards this, IoT systems are
used to collect human context by sensing speech tone, senti-

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 188 IoT Technical Challenges and Solutions

ment, emotional state, environmental conditions, and nature

of a person’s relationships. Acquiring and processing five
senses can enable applications with enhanced user experience.
Incorporation of tactical sensing can enable remote activities
through realistic haptics. Through this continuous learning,
businesses can offer customers deeper engagements and be-
come more natural, anticipatory, and emotionally appealing.
2. Extended expertise: Cognitive systems can be designed to
gather and learn domain knowledge. This expert system can
be of help as a companion to enhance the workforce’s perfor-
mance. Such systems can both understand and teach complex
expertise. This enables a faster learning curve for workforce.
3. Cognitive products and services: New classes of products and
services that sense, reason, and learn about their users and
their contexts can be enabled with cognitive IoT. It allows for
continuous improvement and adaptation and for augmenta-
tion of capabilities not previously imagined. This is already
being used by automobiles, service robots, and medical de-
vices, among others.
4. Cognitive processes and operations: Making business pro-
cesses cognitive can make them self-adaptive with enhanced
awareness of workflows, context, and environments. The con-
tinuous learning process from the experience gained through
operations can be transformed into knowledge. Leveraging
this, it can provide ability for better forecasting and operation-
al effectiveness.

7.3.3 Impact of 5G as the Next Wave of Communication Technology in IoT

The fifth generation (5G) of mobile networks is an upcoming key

technology towards the next wave of cellular communications.
Copyright © 2016. Artech House. All rights reserved.

Specific support for IoT has been envisaged to include ubiquitous

coverage, signaling efficiency, and the ability to support several
hundreds of thousands of simultaneous connections for massive
sensor deployments. While technology like Zigbee and WiFi will
still be low-cost options for certain IoT industries, 5G will have the
potential as a massive cellular networking infrastructure to provide
an overarching connectivity to futuristic IoT applications. Some of

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 The Real Internet of Things and Beyond 189

the applications such as real-time tactile communication for re-

mote haptics where transporting a realistic five senses experience
to a remote location require the capability of 5G communications.
What today’s IoT applications are looking from a wide area
network are ubiquitous coverage, scalability, coexistence, man-
aged quality of service (QoS), and end-to-end security. These are
being addressed by two recent standards as part of 3GPP Release
13. They are eMTC (enhanced machine-type communications) and
NB-IoT (narrowband IoT). eMTC is named as Cat-M1 and it sup-
ports variable rate up to 1-Mbps, 1.4-MHz narrow band. NB-IoT
is known as Cat-NB1 supporting tens of kilobits per second in the
200-kHz narrow band. They promise to reduce device complexity,
multiyear battery life, and large coverage. It can achieve 1 million
connections per square kilometer, using multiples of a 200-kHz
bandwidth. Also, they work under the standard long-term evolu-
tion (LTE) infrastructure. Beyond this, 3GPP is also working on a
new 5G air interface. It is expected to bring support for massive
IoT with higher capability and efficiency. It is expected to support
time-critical applications including communication and control in-
volving robots and drones. IoT systems and services integrated to
the standard cellular telecom network will have a strong positive
impact to its proliferation and business opportunities.
One of the major hurdles for new wireless communication tech-
nology is the nonavailability of the radio frequency (RF) spectrum
in the traditional range. Higher bands like millimeter waves are
being targeted for enhanced capacity, but they have the disadvan-
tage of higher attenuation in the medium and thus reduced range.
However, for IoT devices, this may not be a limitation and may
enable smaller device form factors.
Application of software-defined networks (SDN) and network
function virtualization (NFV) along with 5G can make a strong im-
pact on future distributed enterprise networks. Many vertical sec-
Copyright © 2016. Artech House. All rights reserved.

tors will require different types of services from the 5G network.

They will have different power budgets, bandwidth requirements,
and other connectivity requirements. The required flexibility and
elasticity can be supported by advanced network virtualization.
Virtualization will be a vital capability with 5G to support di-
versity of IoT requirements. The concept of network slicing is an
approach where the available network bandwidth can be sliced

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 190 IoT Technical Challenges and Solutions

into multiple units based on the speed, availability, capacity, and

coverage requirements. The best way to implement these network
slices will be via virtualization to use service provider SDN, NFV,
and network orchestration. In each of these cases, the SDN con-
troller will configure and build network slices that include several
service chains. The network services themselves are virtualized as
virtual network functions (VNFs) that allow operators to set up
services rapidly and scale them in response to network and service
demands.

7.4 Concluding Remarks

In this book, we introduced IoT in Chapter 1 and outlined the cur-

rent state of the art in terms of application landscape and technolo-
gy. In Chapter 1, we also discussed the need for standardization in
IoT and the challenges and open problems in IoT. Then, in Chapter
2, we discussed scalability of networks and computing and with
help from example use cases and showed how it becomes an im-
portant component for IoT deployment. In Chapter 3, we touch
upon one of the major issues for IoT, namely, security and privacy,
cover important topics like threat analysis, data protection, key
management, identity management, authentication, access con-
trol, and privacy issues and suggest addressing practical deploy-
ment issues. In Chapter 4, we cover sensor informatics and busi-
ness insights, which can be thought of as the brain of IoT spanning
sensor signal processing to gather information from sensor data,
semantic interpretation of the processed Information to convert it
to knowledge, and business insights from interpreted knowledge.
In Chapter 5, we introduced the concept of mobile sensing (i.e.,
sensors mounted on mobile platforms including drones and ro-
bots). We also discuss the various application use cases, technol-
Copyright © 2016. Artech House. All rights reserved.

ogy challenges and the economy or business case for such sensing
and analytics systems. In Chapter 6, we addressed the application
developer concerns by introducing the concept of analytics as a
service on top of the sensor informatics stack outlined in Chapter
4. Finally, in this chapter, we address the real issues that need to be
tackled for successful deployment of IoT applications.
While IoT gets a lot of attention with academia and industry,
there is lot to be desired on the success rate of real-life IoT projects.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 The Real Internet of Things and Beyond 191

With the concept of real Internet of Things, we discussed various

aspects of an IoT solution that provides ROI on realistic business
applications. To make the solution cost-effective, there is a large
number dimensions for design trade-offs to be made. Pruning and
tuning of functionalities and processes are to be made. A paradigm
shift from anytime, anywhere to whenever, wherever as required
is needed. Communications will benefit from edge computing
where processing is done near to the data origin. User centricity
in the design is of paramount importance from a user experience
perspective. Ways to build resilience into the solutions are of high
priority for trustworthy systems in industrial applications. Cogni-
tive computing is a promising approach to explore for building
autonomous capabilities in future IoT systems.

References
[1] http://www.sustainablebrands.com/news_and_views/ict_big_data/bart_
king/case_study_18month_roi_north_face_connected_ems_retail_stores.
[2] http://cio.economictimes.indiatimes.com/news/internet-of-things/8-out-
of-10-iot-projects-fail-even-before-they-are-launched/52448887.
[3] Weiser, M., “The Computer for the 21st Century,” Scientific American, Sep-
tember, 1991, pp. 94-104; reprinted in IEEE Pervasive Computing, January-
March 2002.
[4] ����������������������������������������������������������
http://www3.weforum.org/docs/WEFUSA_IndustrialInternet_Re-
port2015.pdf.
[5] http://www.rolls-royce.com/pro-and-services/civil-aerospace/services/
services-catalogue/totalcare.aspx.

Selected Bibliography
4G Americas Cellular Technologies, “Enabling the Internet of Things,” November
2015, http://www.4gamericas.org/files/6014/4683/4670/4G_Americas_Cellu-
Copyright © 2016. Artech House. All rights reserved.

lar_Technologies_Enabling_the_IoT_White_Paper_-_November_2015.pdf.
Athreya, A. P., B. DeBruhl, and P. Tague, “Designing for Self-Configuration and
Self-Adaptation in the Internet of Things,” 9th Intl. Conf. on Collaborative Comput-
ing: Networking Applications and Worksharing (Collabcom 2013), 2013.
Athreya, P., and P. Tague, “Network Self-Organization in the Internet of Things,”
10th Annual IEEE Communications Society Conference on Sense, Mesh, and Ad Hoc
Communications and Networks (SECON), 24–27 June 2013.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 192 IoT Technical Challenges and Solutions

Atzori, A. Iera, and G. Morabito, “The Internet of Things: A Survey,” Computer

Networks, Vol. 54, No. 15, 2010, pp. 2787–2805.
Bao, F., and I. -R. Chen, “Trust Management for the Internet of Things and Its
Application to Service Composition,” World of Wireless, Mobile and Multimedia Net-
works (WoWMoM), 2012.
Beal, J., D. Pianini, and M. Viroli, “Aggregate Programming for the Internet of
Things,” Computer, Vol.48, No. 9, September 2015, pp. 22–30.
Delic, K. A., “On Resilience of IoT Systems: The Internet of Things,” Magazine
Ubiquity, Volume 2016, Issue February, February 2016.
Saha, D., and A. Mukherjee, “Pervasive Computing: A Paradigm for the 21st Cen-
tury Computer,” IEEE Computer Society, March 2003.
Wu, Q., et al., “Cognitive Internet of Things: A New Paradigm Beyond Connec-
tion,” IEEE Journal of Internet of Things, Vol. 1, No. 2, 2014, pp. 129–143.
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 About the Authors

Arpan Pal received his B.Tech and M.Tech from the Indian Insti-
tute of Technology, Kharagpur, India, in electronics and telecom-
munications in 1990 and 1993, respectively, followed by a PhD.
from Aalborg University, Denmark.
He has more than 24 years of experience in the areas of sig-
nal processing, communication, and real-time embedded systems.
Currently, he works for Tata Consultancy Services (TCS), where, as
principal scientist, he heads the Embedded Systems and Robotics
Research Area. His areas of interest include mobile phone, camera
and biosensing, physiological sensing, machine to machine (M2M)
communications and Internet of Things-based platforms and sys-
tems. He had earlier worked with the Defense Research and Devel-
opment Organization (DRDO) of the Indian government working
on missile seeker signal processing and he had also worked with
Macmet Interactive Technologies, leading their real-time systems
group in the area of interactive TV and set-top boxes.
Dr. Pal currently has more than 85 publications in reputed
journals and conferences along with a couple of book chapters. He
has also filed for more than 75 patents and has 15 patents that have
been granted. He is an associate editor for reputed journals like
ACM Transactions on Embedded Computing Systems and IEEE Trans-
actions on Emerging Topics in Computing and IT Professional Magazine
from the IEEE Computer Society. He is in the program committee
Copyright © 2016. Artech House. All rights reserved.

of various eminent conferences and is a senior member of the IEEE.

P. Balamuralidhar received a B.Tech degree in electrical en-
gineering from T.K.M. Engineering College, Kerala University, in
1985 and M.Tech from Indian Institute of Technology Kanpur in

193

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 194 IoT Technical Challenges and Solutions

1987. Subsequently, he received his PhD from Aalborg University,

Denmark.
He has more than 28 years of research experience in the areas
of signal processing, communications, and embedded systems. He
has more than 100 research publications in various journals and
conference proceedings and authored six book chapters. He has 20
patent grants in multiple technology domains. His research inter-
ests include areas of sensor informatics, Internet of Things, mobile-
sensing systems, and cognitive computing and vision systems.
Dr. Balamuralidhar works for Tata Consultancy Services
(TCS) as a principal scientist and heads TCS Research & Innovation
Lab at Bangalore. He had earlier worked for Sasken Communica-
tion Systems working on 3G wireless communications reference
designs. Prior to that, he was a scientist at the Society for Applied
Microwave Electronics Engineering and Research (SAMEER) and
worked in the area of radar signal processing and systems. He is
a senior member of the IEEE and a member of the Association for
Computing Machinery (ACM).
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Index

6LoWPAN, 52 example model-driven framework,

168–72
A need, as a service, 161–65
Abductive reasoning, 125 need for, 157–61
Access control, 88–89, 100 objective of, 157–58
Ada-boost gradient boosting, 120 predictive, 18, 47
Adaptive filtering, 112 real-time, 63
Aerial mapping sensor-driven, 158
NDVI maps, 149–50 as a service, 157–73
three-dimensional visualizations, as a service for developers, 165–68
149 summary, 172–73
two-dimensional maps, 149 types of, 158, 159
See also Unmanned aerial vehicles visual, 126–27
(UAVs) Anomaly detection, 114
Affordability, 34–35 Application classification templates,
Aircraft inspection, 141 47–49
Analog-to-digital (A/D) conversion, Application landscape, 19–23
24 consumers, 21–22
Copyright © 2016. Artech House. All rights reserved.

Analytics facilities, 20
context-aware, 34 illustrated, 23
data linkage, 91 industry vertical, 23
development stakeholders, 162 products, 20–21
diagnostic, 160–61 supply chain, 22
Application space, 21

195

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 196 IoT Technical Challenges and Solutions

Artificial neural networks (ANNs), 122 security and privacy, 33–34, 73–104
Attestation, 101
Authentication, 87–88 Cloud
in IoT security, 79 connectivity networks, 26
multifactor (MFA), 87 infrastructure, 76
options, 88 interfaces, 103
practical guidelines, 100–101 subsystem, 26–28
SSL (Secure Sockets Layer), 87 Clustering and classification, 122
TLS (Transport Layer Security), 87 CoAP, 56
Authorization, 79 Code division multiple access
(CDMA), 25
B Cognitive IoT, 187–88
Battery lifetime versus performance, Collaborative sensing, 138, 145–47
183 Communication, range, power, and
Bayesian minimum squared error bandwidth trade-offs, 183
(BMSE), 117 Communication security
Biz-Sci-Dev-Ops, 161–63, 166–68 cryptographic key management,
Biz knowledge base, 166 84–86
Dev knowledge base, 167 features of wireless standards, 83
Ops knowledge base, 168 overview, 83–84
Sci knowledge base, 167 practical guidelines, 101
Bluetooth, 53 protocols, 84
Bluetooth Low Energy (BLE), 53 requirements, 84
Bundle Protocol (BP), 59 Communication technologies
Business insights, 111, 126–28 application-level protocols, 55–56
modeling and simulation, 127 cellular, 54–55
optimization and planning, 127–28 fifth generation (5G), 188–90
visual analytics, 126–27 for low power wide area networks
workflow, 126 (LPWAN), 53–54
Business model, 180–81 overview, 49–50
personal/local area network, 50–53
C Community sensing, 144
Complex event processing (CEP)
Carrier sense multiple access/collision
systems, 26, 124
advance (CSMA/CA), 50–51
Copyright © 2016. Artech House. All rights reserved.

Compression, 115–16
Cellular technology, 54–55
Concluding remarks, 190–91
Challenges in deployments
Confidentiality, integrity, and
affordability, 34–35
availability (CIA), 78
context-aware analytics, 34
Consumers, 21–22
ease and economy, 35
Context-aware analytics, 34
realistic deployments, 35
Core business areas, 20
scalability, 32–33, 41–67

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Index 197

Crypto algorithm, 101 E

Cryptographic Algorithm Validation Ease and economy of deployments, 35
Program (CAVP), 82 Economics, of mobile sensing, 152–54
Cryptographic key management, Ecosystem, 86, 181
84–86 Edge devices, 64–65
Cryptographic Module Validation Emergency response, mobile sensing
Program (CMVP), 82 for, 137
Cyberphysical system (CPS), 109 Encryption algorithms, 83
Endpoints, 77, 79
D End-user awareness, 103
Data, information, knowledge, and Energy
wisdom (DIKW) pyramid, home, 47
110 monitoring, 17
Data and algorithm market places, smart, 44
128–29 Environment, smart, 43–44
Data at rest (DAR), 81 Environmental monitoring, 137
Data flow diagrams (DFDs), 94, 96 Equipment vendors, 18–19
Data in motion (DIM), 81 Estimation, 117
Data in use (DIU), 81 European Telecommunication
Data linkage analytics, 91 Standards Institute (ETSI),
Data protection, 81–83, 102–3 28, 65
Deductive reasoning, 125 Expert systems, 126
Delay-tolerant networks, 58–60
Denial-of-service (DoS) attacks, 78 F
Deployment architecture, 62 Facilities, 20
Detection, 116–17 Farming, smart, 46
Device-level security, 79–80, 101–2 Feature extraction and inference
Device updates and transitions, 91 applications of, 118–19
Device Verifiable Certificate (DVC), 84 detection, 116–17
Diagnosis, 125–26 estimation, 117
Diagnostic analytics, 160–61 fusion, 118
DIKW pyramid, 110 identification, 117
Discrete wavelet transform (DWT), overview, 116
115 stream processing, 118
Copyright © 2016. Artech House. All rights reserved.

Disease prognosis, 47 Fifth generation (5G) technology,

DREAD model 188–90
defined, 97 Finite impulse response (FIR), 112
qualitative risk assessment with, 98 Forest fire detection, 43
score computation, 97

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 198 IoT Technical Challenges and Solutions

Frequency division multiple access Internet engineering task force (IETF),

(FDMA), 25 51–52
Frequency domain, 114–15 Internet of Things (IoT)
Fusion, 118 adoption across geographies, 17
analytics, 157–73
G application landscape, 19–23
Gateways, 76, 77 challenges and open problems,
Gateway subsystem, 25–26 32–35
Goal-driven rules, 123–24 cognitive, 187–88
Grid monitoring and control, 44 communication technologies for,
49–56
H for consumers, 21–22
defined, 13, 15
Hard sensing versus soft sensing, 182
deployment architecture, 62
Hazardous scenario detection and
deployments, 17–18
response, 45
as disruptive technology, 15–16
Health, smart, 47
drivers for next wave, 186–90
Healthcare, robots in, 138
ecosystem, 18–19
Home, smart, 46–47
for end consumer, 16
for facilities, 20
I
fifth generation (5G) impact on,
Identification, 90–91, 117 188–90
Identities interoperability, 29
ecosystem, 86 key trends, 15–19
management of, 86–87 market for, 16
practical guidelines, 100–101 for products, 20–21
IEEE 802.15.4, 50–51 protocol design space, 58
Inductive reasoning, 125 realistic, 177–86
Inference rules, 123 revenue increases from, 17
Inferencing, 124 scalable network architectures for,
Infinite impulse response (IIR), 112 56–61
Information flow graph, 170 services and technology-level
Information technology (IT) software standardization, 29
development, 161 spending in, 17
Copyright © 2016. Artech House. All rights reserved.

Infrastructure inspections, 150–52 stack, 19

Insurance business, UAVs for, 141 standardization, 28–32
International Organization for for supply chain, 22
Standardization (ISO), 30 system considerations, 13
International Telecommunication Union technologies of, 23–28
- Telecom (ITU-T), 30, 31 trend study, 16–17

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Index 199

use cases and requirements, 42–47 Maximum likelihood estimation

See also IoT systems (MLE), 117
Interoperability, 29 Mesh network, 57–58
IoT systems Microsoft Security Development
complexity and diversity, 79 Lifecycle (SDL), 95
components and attack targets, 76 Mobile phone sensing
confidentiality, 77–78 community sensing, 144
life cycle, 80 overview, 143–44
resilient, 186–87 personal sensing, 144
security, 73–104 public sensing, 144
trade-offs, 181–86 scales, 144
Mobile sensing, 135–54
K aerial robots for spatial intelligence,
Kalman filtering (KF), 117 139–40
aircraft inspection, 141
L applications and use cases for,
136–42
Least-mean-square (LMS), 112
challenges and trends in, 145–47
Least square estimation (LSE), 117
collaborative, 138
Least-squares regression, 120
cost of, 154
Localization
coverage, 152–53
for robots, 148–49
economics of, 152–54
threat, 91
for emergency response, 137
Local sensor networks, 25
for environmental monitoring, 137
Low configuration sensor data
insurance business, 141
acquisition, 63
introduction to, 135–36
Low power wide area networks
platform sweet spot, 136
(LPWAN), 53–54
robotic sensing networks, 147–49
robotic telesensing and operation,
M
138–39
Machine learning robots in healthcare, 138
applications of, 122 smartphone-based sensing, 143–47
evolution of, 119–20 summary, 154
reinforced learning, 120–21 technologies and challenges,
Copyright © 2016. Artech House. All rights reserved.

supervised learning, 120 143–52

unsupervised learning, 120 UAV for aerial mapping, 149–52
Machine-to-machine (M2M) urban inspection, 141
communications, 30, 31 urban transportation, 138
Manufacturing, smart, 46 users’ participation in, 144–45
Matched filtering, 112 utility asset inspection, 140–41

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 200 IoT Technical Challenges and Solutions

Model-driven development (MDD), Outcome economy, 181

164, 165 Outlier detection, 113
Model-driven framework
development and orchestration P
concern, 169–71 Passwords, 100
domain concern, 168–69 Pattern recognition, 118–19
example information flow graph, Peak load management, 44
170 Perimeter access control, 45
illustrated, 169 Personalized shopping, 45
infrastructure concern, 171–72 Personal/local area network
Modeling and simulation, 127 technologies, 50–53
MQTT, 55 Personal sensing, 144
Multifactor authentication (MFA), 87 Piecewise aggregation approximation
Multirobot sensing, 147–48 (PAA), 115
Platforms
N for application development/
Naïve-Bayes classification, 120 deployment, 65–67
National hazard detection and application-specific versus general-
prediction, 43 purpose, 182–83
Network function virtualization horizontal development, 65, 66, 67
(NFV), 61 Point-to-point network, 57
Network topologies, 57–58 Pollution control, 43
Noise cleaning, 113–14 Prediction, 119
Normalized Difference Vegetation Predictive analytics, 18, 47
Index (NDVI) maps, 149–50 Predictive maintenance, 46
Privacy
O defined, 90
OASIS (Advancing Open Standards drones and, 151
for the Information Society), generic approach to, 91–92
165 management of, 146–47
OCTAVE (Operationally Critical threats to, 90–91
Threat, Asset, and See also Security
Vulnerability Evaluation), 94 Produce quality monitoring, 46
OneM2M, 65 Products, 20–21
Copyright © 2016. Artech House. All rights reserved.

Online Certificate Status Protocol Profiling, 91

(OCSP), 84 Protocol design space, 58
Open Geo Spatial Consortium (OGC), Public key infrastructure (PKI), 84
30 Public sensing, 144
OpenTOSCA, 165, 166
Optimization and planning, 127–28 Q
Organization, this book, 13–14 Quality monitoring, 18

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Index 201

Quality of service (QoS), 67, 189 multirobot sensing and coverage

issues, 147–48
R overview, 147
Reaction rules, 123 Robotic telesensing and operation,
Real IoT 138–39
business model, 180–81 Robots
ecosystem, 181 in healthcare, 138
end user and, 180 localization for, 148–49
functionality and processes, 180 multirobot sensing, 147–48
key contributing factors, 178–81 for spatial intelligence, 139–40
overview, 177–78 Role-based access control (RBAC), 89
robust devices and connectivity, Root of trust (RoT), 101
178–79
trade-offs, 181–86 S
Realistic deployments, 35 Safety
Real-time alerting, 18 on machine flow, 184–86
Real-time analytics, 63 UAVs and, 151
Reasoning systems Sampling, 111–12
abductive, 125 Scalability, 41–67
applications of, 125–26 application classification templates,
deductive, 125 47–49
inductive, 125 as challenge, 32–33
overview, 124 conclusions, 67
Recursive-least-square (RLS), 112 introduction to, 41–42
Regression and prediction, 122 need for, 41
Reinforced learning, 120–21 network architectures, 56–61
Remote-operated vehicles (ROVs), of networks and computing, 41–67
138–39 practical considerations, 62–67
Requirements mapping use cases and requirements and,
for control applications, 49 42–47
for monitoring applications, 48 Scalable network architectures
for optimization applications, 49 delay-tolerant networks, 58–60
Resampling, 113 network topologies, 57–58
Resilient IoT systems, 186–87 overview, 56–57
Copyright © 2016. Artech House. All rights reserved.

Retail and logistics, smart, 45 protocol design space, 58

Return on investment (ROI), 13, 34, software-defined networking
127, 191 (SDN), 60–61
Risk-based security, 92–94 Scalable system implementation
Risk estimation, 94–99 application development/
Robotic sensor networks deployment platform, 65–67
localization for robots, 148–49 edge devices utilization, 64–65

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 202 IoT Technical Challenges and Solutions

Scalable system implementation Sensor signal processing, 110, 111–19

(continued) feature extraction and inference,
real-time and power 116–19
considerations, 62–64 signal acquisition and conditioning,
Security, 33–34 111–14
access control, 88–89 signal representation, 114–16
attacks and, 74 workflow, 111
authentication, 79, 87–88 Sensor subsystem, 24–25
authorization, 79 Services and technology-level
business objectives of, 75 standardization, 29
challenges, 79–81 Signal acquisition/conditioning
communication, 83–86 adaptive filtering, 112
confidentiality, integrity, and applications of, 113–14
availability (CIA), 78 matched filtering, 112
data, 81–83, 102–3 outlier detection, 113
device-level, 79–80, 101–2 sampling, 111–12
identities and identity Signal representation
management, 86–87 applications of, 115–16
key requirements, 75–79 frequency domain, 114–15
perspective, 73–75 statistical processing, 115
physical mechanisms, 77 wavelet domain, 115
policies, 77 Signal-to-noise ratio (SNR), 112
practical guidelines, 100–103 Smart energy, 44
protocols, 85 Smart environment, 43–44
risk-based, 92–94 Smart farming, 46
schemes, 77 Smart health, 47
smart, 45 Smart home, 46–47
software updates, 89–90 Smart manufacturing, 46
statistics-based decomposition, 73 Smartphone-based sensing. See Mobile
subsystem threats and, 80–81 phone sensing
summary, 103–4 Smart retail and logistics, 45
system-level assessment, 92–99 Smart security and surveillance, 45
user experience versus, 183 Smart transportation, 43
Semantic data management, 146 Smart water, 44–45
Copyright © 2016. Artech House. All rights reserved.

Semantic interpretation, 110, 119–26 Software-defined networks (SDN),

machine learning, 119–22 60–61, 189
overview, 119 Software updates, secure, 89–90
reasoning, 124–26 Spatial crowdsourcing, 146
rule engine, 123–24 Spatial intelligence, 139–40
workflow, 120 SSL (Secure Sockets Layer), 87
Sensor-driven analytics, 158 Standardization, 28–32

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 Index 203

Star network, 57 for IoT analytics, 27

Statistical processing, 115 Thread, 52
Stream processing, 118 Threat analysis, 93
STRIDE model, 95–97, 99 Threat modeling, 94–99
Subsystems Time division multiple access
cloud, 26–28 (TDMA), 25
defined, 23–24 TLS (Transport Layer Security), 87, 102
gateway, 25–26 Topology and Orchestration
sensor, 24–25 Specification for Cloud
threats to, 80–81 Applications (TOSCA), 165
Supervised learning, 120 Tracking threat, 91
Supply chain (SC) Trade-offs
defined, 22 application-specific versus general-
IoT for, 22 purpose platforms, 182–83
optimization, 45 battery lifetime versus
Support vector machine (SVM) performance, 183
classification, 120 communication, range, power, and
Symbolic aggregate approximation bandwidth, 183
(SAX), 115 hard sensing versus soft sensing,
System-level security assessment 182
DREAD model, 97–99 overview, 181–82
overview, 92 security versus user experience, 183
risk-based security, 92–94 Traffic congestion management, 43
STRIDE model, 95–97 Transmission control protocol over
threat modeling and risk Internet protocol (TCP/IP)
assessment, 94–99 stack, 52
Transportation, smart, 43
T Trust, management of, 146–47
TCS Connected Universe Platform Trusted Execution Environment (TEE),
(TCUP), 163 81
Technologies
cloud connectivity networks, 26 U
cloud subsystem, 26–28 Unmanned aerial vehicles (UAVs),
communication, 49–56 135–36
Copyright © 2016. Artech House. All rights reserved.

gateway subsystem, 25–26 for aerial mapping, 149–52

local sensor networks, 25 aircraft inspection with, 141
overview, 23–24 ATC and, 151
sensor subsystem, 24–25 infrastructure inspections, 150–52
Technology stack for insurance business, 141
defined, 23–24 potential business applications, 142
illustrated, 23 privacy and, 151

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.
 204 IoT Technical Challenges and Solutions

Unmanned aerial vehicles (continued) V

safety and, 151 Value-added services model, 19
for spatial intelligence, 139–40 Virtualization, 189–90
urban inspection with, 141 Visual analytics, 126–27
utility asset inspection with, 140–41 Visualization, 116
See also Mobile sensing
Unsupervised learning, 120 W
Urban inspection, 141
Waste management, 43
Urban transportation, collaborative
Water
sensing for, 138
leakage protection, 44–45
Use cases
optimization use, 47
mobile sensing, 136–42
quality monitoring, 44
smart energy, 44
smart, 44–45
smart environment, 43–44
Wavelet domain, 115
smart farming, 46
Wellness monitoring, 47
smart health, 47
Wi-Fi, 52–53
smart home, 46–47
WirelessHART, 52
smart manufacturing, 46
smart retail and logistics, 45
X
smart security and surveillance, 45
smart transportation, 43 XMPP, 56
smart water, 44–45
Utility asset inspection, 140–41 Z
ZigBee, 51–52
Copyright © 2016. Artech House. All rights reserved.

Pal, Arpan, and Balamuralidhar Purushothaman. IoT Technical Challenges and Solutions, Artech House, 2016. ProQuest Ebook Central,
http://ebookcentral.proquest.com/lib/binghamton/detail.action?docID=4845595.
Created from binghamton on 2025-05-14 01:10:15.