IJACSA - Volume 3 No. 8, August 2012

THE SCIENCE AND INFORMATION ORGANIZATION
www. t hes ai . or g | i nf o@t hes ai . or g

(IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 3, No.8, 2012
(i)
www.ijacsa.thesai.org
Editorial Preface
From the Desk of Managing Editor
IJACSA seems to have a cult following and was a humungous success during 2011. We at The Science and Information
Organization are pleased to present the August 2012 Issue of IJACSA.
While it took the radio 38 years and the television a short 13 years, it took the World Wide Web only 4 years to reach 50
million users. This shows the richness of the pace at which the computer science moves. As 2012 progresses, we seem to
be set for the rapid and intricate ramifications of new technology advancements.
With this issue we wish to reach out to a much larger number with an expectation that more and more researchers get
interested in our mission of sharing wisdom. The Organization is committed to introduce to the research audience
exactly what they are looking for and that is unique and novel. Guided by this mission, we continuously look for ways to
collaborate with other educational institutions worldwide.
Well, as Steve Jobs once said, Innovation has nothing to do with how many R&D dollars you have, its about the people
you have. At IJACSA we believe in spreading the subject knowledge with effectiveness in all classes of audience.
Nevertheless, the promise of increased engagement requires that we consider how this might be accomplished,
delivering up-to-date and authoritative coverage of advanced computer science and applications.
Throughout our archives, new ideas and technologies have been welcomed, carefully critiqued, and discarded or
accepted by qualified reviewers and associate editors. Our efforts to improve the quality of the articles published and
expand their reach to the interested audience will continue, and these efforts will require critical minds and careful
consideration to assess the quality, relevance, and readability of individual articles.
To summarise, the journal has offered its readership thought provoking theoretical, philosophical, and empirical ideas
from some of the finest minds worldwide. We thank all our readers for their continued support and goodwill for IJACSA.
We will keep you posted on updates about the new programmes launched in collaboration.
We would like to remind you that the success of our journal depends directly on the number of quality articles submitted
for review. Accordingly, we would like to request your participation by submitting quality manuscripts for review and
encouraging your colleagues to submit quality manuscripts for review. One of the great benefits we can provide to our
prospective authors is the mentoring nature of our review process. IJACSA provides authors with high quality, helpful
reviews that are shaped to assist authors in improving their manuscripts.
We regularly conduct surveys and receive extensive feedback which we take very seriously. We beseech valuable
suggestions of all our readers for improving our publication.
Thank you for Sharing Wisdom!

Managing Editor
IJACSA
Volume 3 Issue 8 August 2012
ISSN 2156-5570 (Online)
ISSN 2158-107X (Print)
2012 The Science and Information (SAI) Organization
Vol. 3, No.8, 2012
(ii)
Associate Editors
Dr. Zuqing Zhu
Service Provider Technology Group of Cisco Systems, San Jose
Domain of Research: Research and development of wideband access routers for hybrid
fibre-coaxial (HFC) cable networks and passive optical networks (PON)
Dr. Ka Lok Man
Department of Computer Science and Software Engineering at the Xi'an Jiaotong-
Liverpool University, China
Domain of Research: Design, analysis and tools for integrated circuits and systems;
formal methods; process algebras; real-time, hybrid systems and physical cyber
systems; communication and wireless sensor networks.
Dr. Sasan Adibi
Technical Staff Member of Advanced Research, Research In Motion (RIM), Canada
Domain of Research: Security of wireless systems, Quality of Service (QoS), Ad-Hoc
Networks, e-Health and m-Health (Mobile Health)
Dr. Sikha Bagui
Associate Professor in the Department of Computer Science at the University of West
Florida,
Domain of Research: Database and Data Mining.
Dr. T. V. Prasad
Dean, Lingaya's University, India
Domain of Research: Bioinformatics, Natural Language Processing, Image Processing,
Expert Systems, Robotics
Dr. Bremananth R
Research Fellow, Nanyang Technological University, Singapore
Domain of Research: Acoustic Holography, Pattern Recognition, Computer Vision, Image
Processing, Biometrics, Multimedia and Soft Computing
Vol. 3, No.8, 2012
(iii)
Reviewer Board Members

A Kathirvel
Karpaga Vinayaka College of Engineering and
Technology, India
A.V. Senthil Kumar
Hindusthan College of Arts and Science
Abbas Karimi
I.A.U_Arak Branch (Faculty Member) & Universiti
Putra Malaysia
Abdel-Hameed A. Badawy
University of Maryland
Abdul Wahid
Gautam Buddha University
Abdul Hannan
Vivekanand College
Abdul Khader Jilani Saudagar
Al-Imam Muhammad Ibn Saud Islamic University
Abdur Rashid Khan
Gomal Unversity
Aderemi A. Atayero
Covenant University
Ahmed Boutejdar
Dr. Ahmed Nabih Zaki Rashed
Menoufia University, Egypt
Ajantha Herath
University of Fiji
Ahmed Sabah AL-Jumaili
Ahlia University
Akbar Hossain
Albert Alexander
Kongu Engineering College,India
Prof. Alcinia Zita Sampaio
Technical University of Lisbon
Amit Verma
Rayat & Bahra Engineering College, India
Ammar Mohammed Ammar
Department of Computer Science, University of
Koblenz-Landau
Anand Nayyar
KCL Institute of Management and Technology,
Jalandhar
Anirban Sarkar
National Institute of Technology, Durgapur, India
Arash Habibi Lashakri
University Technology Malaysia (UTM), Malaysia
Aris Skander
Constantine University
Ashraf Mohammed Iqbal
Dalhousie University and Capital Health
Asoke Nath
St. Xaviers College, India
Aung Kyaw Oo
Defence Services Academy
B R SARATH KUMAR
Lenora College of Engineering, India
Babatunde Opeoluwa Akinkunmi
University of Ibadan
Badre Bossoufi
University of Liege
Balakrushna Tripathy
VIT University
Basil Hamed
Islamic University of Gaza
Bharat Bhushan Agarwal
I.F.T.M.UNIVERSITY
Bharti Waman Gawali
Department of Computer Science &
information
Bremananth Ramachandran
School of EEE, Nanyang Technological University
Brij Gupta
University of New Brunswick
Dr.C.Suresh Gnana Dhas
Park College of Engineering and Technology,
India
Mr. Chakresh kumar
Manav Rachna International University, India
Chandra Mouli P.V.S.S.R
VIT University, India
Chandrashekhar Meshram
Chhattisgarh Swami Vivekananda Technical
University
Chi-Hua Chen
National Chiao-Tung University
Constantin POPESCU
Department of Mathematics and Computer
Science, University of Oradea
Prof. D. S. R. Murthy
SNIST, India.
Dana PETCU
West University of Timisoara
David Greenhalgh
University of Strathclyde
Vol. 3, No.8, 2012
(iv)
Deepak Garg
Thapar University.
Prof. Dhananjay R.Kalbande
Sardar Patel Institute of Technology, India
Dhirendra Mishra
SVKM's NMIMS University, India
Divya Prakash Shrivastava
EL JABAL AL GARBI UNIVERSITY, ZAWIA
Dragana Becejski-Vujaklija
University of Belgrade, Faculty of organizational
sciences
Firkhan Ali Hamid Ali
UTHM
Fokrul Alom Mazarbhuiya
King Khalid University
Fu-Chien Kao
Da-Y eh University
G. Sreedhar
Rashtriya Sanskrit University
Gaurav Kumar
Manav Bharti University, Solan Himachal
Pradesh
Ghalem Belalem
University of Oran (Es Senia)
Gufran Ahmad Ansari
Qassim University
Hadj Hamma Tadjine
IAV GmbH
Hanumanthappa.J
University of Mangalore, India
Hesham G. Ibrahim
Chemical Engineering Department, Al-Mergheb
University, Al-Khoms City
Dr. Himanshu Aggarwal
Punjabi University, India
Huda K. AL-Jobori
Ahlia University
Dr. Jamaiah Haji Yahaya
Northern University of Malaysia (UUM), Malaysia
Jasvir Singh
Communication Signal Processing Research Lab
Jatinderkumar R. Saini
S.P.College of Engineering, Gujarat
Prof. Joe-Sam Chou
Nanhua University, Taiwan
Dr. Juan Jos Martnez Castillo
Yacambu University, Venezuela
Dr. Jui-Pin Yang
Shih Chien University, Taiwan
Jyoti Chaudhary
high performance computing research lab
K Ramani
K.S.Rangasamy College of Technology,
Tiruchengode
K V.L.N.Acharyulu
Bapatla Engineering college
K. PRASADH
METS SCHOOL OF ENGINEERING
Ka Lok Man
Xian Jiaotong-Liverpool University (XJTLU)
Dr. Kamal Shah
St. Francis Institute of Technology, India
Kanak Saxena
S.A.TECHNOLOGICAL INSTITUTE
Kashif Nisar
Universiti Utara Malaysia
Kayhan Zrar Ghafoor
University Technology Malaysia
Kodge B. G.
S. V. College, India
Kohei Arai
Saga University
Kunal Patel
Ingenuity Systems, USA
Labib Francis Gergis
Misr Academy for Engineering and Technology
Lai Khin Wee
Technischen Universitt Ilmenau, Germany
Latha Parthiban
SSN College of Engineering, Kalavakkam
Lazar Stosic
College for professional studies educators,
Aleksinac
Mr. Lijian Sun
Chinese Academy of Surveying and Mapping,
China
Long Chen
Qualcomm Incorporated
M.V.Raghavendra
Swathi Institute of Technology & Sciences, India.
M. Tariq Banday
University of Kashmir
Madjid Khalilian
Islamic Azad University
Mahesh Chandra
B.I.T, India
Mahmoud M. A. Abd Ellatif
Mansoura University
Manas deep
Masters in Cyber Law & Information Security
Manpreet Singh Manna
Vol. 3, No.8, 2012
(v)
SLIET University, Govt. of India
Manuj Darbari
BBD University
Marcellin Julius NKENLIFACK
University of Dschang
Md. Masud Rana
Khunla University of Engineering & Technology,
Bangladesh
Md. Zia Ur Rahman
Narasaraopeta Engg. College, Narasaraopeta
Messaouda AZZOUZI
Ziane AChour University of Djelfa
Dr. Michael Watts
University of Adelaide, Australia
Milena Bogdanovic
University of Nis, Teacher Training Faculty in
Vranje
Miroslav Baca
University of Zagreb, Faculty of organization and
informatics / Center for biomet
Mohamed Ali Mahjoub
Preparatory Institute of Engineer of Monastir
Mohammad Talib
University of Botswana, Gaborone
Mohammad Ali Badamchizadeh
University of Tabriz
Mohammed Ali Hussain
Sri Sai Madhavi Institute of Science &
Technology
Mohd Helmy Abd Wahab
Universiti Tun Hussein Onn Malaysia
Mohd Nazri Ismail
University of Kuala Lumpur (UniKL)
Mona Elshinawy
Howard University
Monji Kherallah
University of Sfax
Mourad Amad
Laboratory LAMOS, Bejaia University
Mueen Uddin
Universiti Teknologi Malaysia UTM
Dr. Murugesan N
Government Arts College (Autonomous), India
N Ch.Sriman Narayana Iyengar
VIT University
Natarajan Subramanyam
PES Institute of Technology
Neeraj Bhargava
MDS University
Nitin S. Choubey
Mukesh Patel School of Technology
Management & Eng
Noura Aknin
Abdelamlek Essaadi
Pankaj Gupta
Microsoft Corporation
Paresh V Virparia
Sardar Patel University
Dr. Poonam Garg
Institute of Management Technology,
Ghaziabad
Prabhat K Mahanti
UNIVERSITY OF NEW BRUNSWICK
Pradip Jawandhiya
Jawaharlal Darda Institute of Engineering &
Techno
Rachid Saadane
EE departement EHTP
Raj Gaurang Tiwari
AZAD Institute of Engineering and Technology
Rajesh Kumar
National University of Singapore
Rajesh K Shukla
Sagar Institute of Research & Technology-
Excellence, India
Dr. Rajiv Dharaskar
GH Raisoni College of Engineering, India
Prof. Rakesh. L
Vijetha Institute of Technology, India
Prof. Rashid Sheikh
Acropolis Institute of Technology and Research,
India
Ravi Prakash
University of Mumbai
Reshmy Krishnan
Muscat College affiliated to stirling University.U
Rongrong Ji
Columbia University
Ronny Mardiyanto
Institut Teknologi Sepuluh Nopember
Ruchika Malhotra
Delhi Technoogical University
Sachin Kumar Agrawal
University of Limerick
Dr.Sagarmay Deb
University Lecturer, Central Queensland
University, Australia
Said Ghoniemy
Taif University
Saleh Ali K. AlOmari
Universiti Sains Malaysia
Vol. 3, No.8, 2012
(vi)
Samarjeet Borah
Dept. of CSE, Sikkim Manipal University
Dr. Sana'a Wafa Al-Sayegh
University College of Applied Sciences UCAS-
Palestine
Santosh Kumar
Graphic Era University, India
Sasan Adibi
Research In Motion (RIM)
Saurabh Pal
VBS Purvanchal University, Jaunpur
Saurabh Dutta
Dr. B. C. Roy Engineering College, Durgapur
Sebastian Marius Rosu
Special Telecommunications Service
Sergio Andre Ferreira
Portuguese Catholic University
Seyed Hamidreza Mohades Kasaei
University of Isfahan
Shahanawaj Ahamad
The University of Al-Kharj
Shaidah Jusoh
University of West Florida
Shriram Vasudevan
Sikha Bagui
Zarqa University
Sivakumar Poruran
SKP ENGINEERING COLLEGE
Slim BEN SAOUD
Dr. Smita Rajpal
ITM University
Suhas J Manangi
Microsoft
SUKUMAR SENTHILKUMAR
Universiti Sains Malaysia
Sumazly Sulaiman
Institute of Space Science (ANGKASA), Universiti
Kebangsaan Malaysia
Sunil Taneja
Smt. Aruna Asaf Ali Government Post Graduate
College, India
Dr. Suresh Sankaranarayanan
University of West Indies, Kingston, Jamaica
T C. Manjunath
HKBK College of Engg
T C.Manjunath
Visvesvaraya Tech. University
T V Narayana Rao
Hyderabad Institute of Technology and
Management
T. V. Prasad
Lingaya's University
Taiwo Ayodele
Lingaya's University
Totok R. Biyanto
Infonetmedia/University of Portsmouth
Varun Kumar
Institute of Technology and Management, India
Vellanki Uma Kanta Sastry
SreeNidhi Institute of Science and Technology
(SNIST), Hyderabad, India.
Venkatesh Jaganathan
Vijay Harishchandra
Vinayak Bairagi
Sinhgad Academy of engineering, India
Vishal Bhatnagar
AIACT&R, Govt. of NCT of Delhi
Vitus S.W. Lam
The University of Hong Kong
Vuda Sreenivasarao
St.Marys college of Engineering & Technology,
Hyderabad, India
Wei Wei
Wichian Sittiprapaporn
Mahasarakham University
Xiaojing Xiang
AT&T Labs
Y Srinivas
GITAM University
Yilun Shang
University of Texas at San Antonio
Mr.Zhao Zhang
City University of Hong Kong, Kowloon, Hong
Kong
Zhixin Chen
ILX Lightwave Corporation
Zuqing Zhu
University of Science and Technology of China
Vol. 3, No.8, 2012
(vi)
CONTENTS
Paper 1: Instruction Design Model for Self-Paced ICT System E-Learning in an Organization
Authors: Ridi Ferdiana, Obert Hoseanto
PAGE 1 7

Paper 2: An Enhanced MPLS-TE for Transferring Multimedia packets
Authors: Abdellah Jamali, Najib Naja, Driss El Ouadghiri
PAGE 8 13

Paper 3: A New Algorithm for Data Compression Optimization
Authors: I Made Agus Dwi Suarjaya
PAGE 14 17

Paper 4: Monte Carlo Based Non-Linear Mixture Model of Earth Observation Satellite Imagery Pixel Data
Authors: Kohei Arai
PAGE 18 22

Paper 5: A Modified Feistel Cipher Involving Substitution, Shifting of rows, mixing of columns, XOR operation with a Key
and Shuffling
Authors: V.U.K Sastry, K. Anup Kumar
PAGE 23 29

Paper 6: Automatic Association of Strahlers Order and Attributes with the Drainage System
Authors: Mohan P. Pradhan, M. K. Ghose, Yash R. Kharka
PAGE 30 34

Paper 7: Performance model to predict overall defect density
Authors: Dr. J. Venkatesh, Mr. Priyesh Cherurveettil, Mrs. Thenmozhi. S, Dr. Balasubramanie. P
PAGE 35 38

Paper 8: Spontaneous-braking and lane-changing effect on traffic congestion using cellular automata model applied to
the two-lane traffic
Authors: Kohei Arai, Steven Ray Sentinuwo
PAGE 39 47

Paper 9: Enhancing eHealth Information Systems for chronic diseases remote monitoring systems
Authors: Amir HAJJAM
PAGE 48 53

Paper 10: E-commerce Smartphone Application
Authors: Dr. Abdullah Saleh Alqahtani, Robert Goodwin
PAGE 54 59

Paper 11: SW-SDF Based Personal Privacy with QIDB-Anonymization Method
Authors: Kiran P, Dr Kavya N P
PAGE 60 66

Vol. 3, No.8, 2012
(vii)
Paper 12: Integration of data mining within a Strategic Knowledge Management framework
Authors: Sanaz Moayer, Scott Gardner
PAGE 67 72

Paper 13: Managing Changes in Citizen-Centric Healthcare Service Platform using High Level Petri Net
Authors: Sabri MTIBAA, Moncef TAGINA
PAGE 73 81

Paper 14: Software Architecture- Evolution and Evaluation
Authors: S.Roselin Mary, Dr.Paul Rodrigues
PAGE 82 88

Paper 15: A hybrid Evolutionary Functional Link Artificial Neural Network for Data mining and Classification
Authors: Faissal MILI, Manel HAMDI
PAGE 89 95

Paper 16: Automatic Aircraft Target Recognition by ISAR Image Processing based on Neural Classifier
Authors: F. Benedetto, F. Riganti Fulginei, A. Laudani, G. Albanese
PAGE 96 103

Paper 17: An Effective Identification of Species from DNA Sequence: A Classification Technique by Integrating DM and
ANN
Authors: Sathish Kumar S, Dr.N.Duraipandian
PAGE 104 114

Paper 18: Brainstorming 2.0: Toward collaborative tool based on social networks
Authors: MohamedChrayah, Kamal Eddine El Kadiri, Boubker Sbihi, Noura Aknin
PAGE 115 120

Paper 19: A Review On Cognitive Mismatch Between Computer and Information Technology And Physicians.
Authors: Fozia Anwar, Dr. Suziah Sulaiman, Dr. P.D.D.Dominic
PAGE 121 124

Paper 20: Techniques to improve the GPS precision
Authors: Nelson Acosta, Juan Toloza
PAGE 125 130

Paper 21: M-Commerce service systems implementation
Authors: Dr.Asmahan Altaher
PAGE 131 136

Paper 22: Clone Detection Using DIFF Algorithm For Aspect Mining
Authors: Rowyda Mohammed Abd El-Aziz, Amal Elsayed Aboutabl, Mostafa-Sami Mostafa
PAGE 137 140

Paper 23: On the Projection Matrices Influence in the Classification of Compressed Sensed ECG Signals
Authors: Monica Fira, Liviu Goras, Liviu Goras, Nicolae Cleju, Constantin Barabasa
PAGE 141 145

Vol. 3, No.8, 2012
(viii)
Paper 24: An Approach of Improving Students Academic Performance by using K-means clustering algorithm and
Decision tree
Authors: Md. Hedayetul Islam Shovon, Mahfuza Haque
PAGE 146 149

Paper 25: Prevention and Detection of Financial Statement Fraud An Implementation of Data Mining Framework
Authors: Rajan Gupta, Nasib Singh Gill
PAGE 150 156

Paper 26: Review of Remote Terminal Unit (RTU) and Gateways for Digital Oilfield delpoyments
Authors: Francis Enejo Idachaba, Ayobami Ogunrinde
PAGE 157 160
Vol. 3, No.8, 2012
1 | P a g e
Instruction Design Model for Self-Paced ICT System
E-Learning in an Organization

Ridi Ferdiana
Electrical Engineering and IT Department
Universitas Gadjah Mada
Yogyakarta, Indonesia
Obert Hoseanto
Partners in Learning Department
Microsoft Indonesia
Jakarta, Indonesia

Abstract Adopting an Information Communication and
Technology (ICT) system in an organization is somewhat
challenging. User diversity, heavy workload, and different skill
gap make the ICT adoption process slower. This research starts
from a condition that a conventional ICT learning through short
workshop and guidance book is not working well. This research
proposes a model called ICT instruction design model (ICT-
IDM). This model provides fast track learning through
integration between multimedia learning and self-paced hands-on
E-learning. Through this case study, we discovered that the
proposed model provides 27% rapid learning adoption rather
than conventional learning model.
Keywords- ICT System Adoption; Learning Model; Multimedia
Learning; E-learning; Instruction Design Model; Learning Plan.
I. INTRODUCTION
Good ICT system is not only about good software or a
good hardware but it also need skilled users. After an ICT
system is developed, it needs additional time for the system to
be used by the user. The implementation phase in software
engineering should make sure that the user feels comfortable
to use the system.
The implementation phase in ICT system is done by doing
several socialization activities such as training, hands-on
workshop, coaching or even giving a grant for the users who
use the system correctly. Tsui and Karam [16] mention several
characteristics of good ICT system implementation which are:
Readability: The software can be easily read and
understood by the programmers.
Maintainability: The code can be easily modified and
maintained.
Performance: All other things being equal, the
implementation should produce code that performs as
fast as possible.
Traceability: All code elements should correspond to a
design element. Code can be traced back to design.
Correctness: The implementation should do what it is
intended to do.
Completeness: All of the system requirements are met.
However, these good implementation characteristics are
only eligible when the users are digital literate and can use the
system well. If the users lack digital literacy or unable to use
the system, the good implementation only happened within the
software team and not in the users perspective.
An organization invests an ICT system to be used by the
employees. It will not be a matter if the user of the ICT system
is only a few (less than 10 users). However, the problem arises
when the ICT system should be adopted for tens or hundreds
of users.
This research starts from a real implementation problem in
an organization called Alpha. Alpha is a government
organization that works to enhance mathematic educators in
Indonesia. Alpha implements an ICT collaboration system
called Live@edu. Live@edu is a collaboration system
developed by Microsoft for education institutions. It contains
software which is:
Email software.
Online storage software.
Instant messaging software.
Personal Information Management software.
Office Productivity software.
The software has more than five features. According to the
Live@edu services page at http://bit.ly/liveateduservices , it
shows that the services have 21 primary features. Additionally,
half of the features are essential for business solution in the
organization. These essential features should be acknowledged
and used by the 50 users within two weeks.
The implementation team has done socialization activities
through face-to-face discussion, full-day workshop, and
delivered a user manual. The socialization activities received a
good response with satisfaction index 8.5 (of scale 10).
However, the organization management did not see a usage
progress of the system and found several issues which are:
User manual exists but more than 93% people does not
read the manual or try by themself.
User feels comfortable with the legacy system or
without the system at all.
Vol. 3, No.8, 2012
2 | P a g e
User has little reason to use the system since they are
too busy with the others job.
Different user skill makes the adoption sluggish for the
entire organization.
User has not much time to learn and to explore the
system when they are inside of working hours.
Based on these issues, this research will make an effort to
improve the ICT system adoption by doing a several activities
such as:
Engage the user personally using self-paced
multimedia learning.
Creating an Instruction design model (IDM) for the
organization.
Creating several learning plan for several different
people based in their existing skill.
In this research, we found that our proposed model can
improve the learning curve of the user. It shows that the
Live@edu usages increase by 27 % after we applied the
Instructions Design Model (IDM) through multimedia
learning. In the next Section, we will discuss about the
previous researches that related with our findings.
I. PREVIOUS RESEARCHES
Carliner [5] proposes an instructions design model (IDM)
process that contains several phases that are definition phase,
design phase, development phase, and implementation phase.
These phases are designed to ensure that the e-learning is well
tested and can be used independently. Table I shows the phase
purposes.
TABLE I. IDM PURPOSES
Phase Purposes
Definition Conducting a need analysis
Settings goal
Preparing the need of analysis report
Design Choosing the form of E-learning
Developing the learning strategy
Establishing guidelines
Development Drafting the E-learning program
Receiving feedback of E-learning
Revising the E-learning
Implementation and
Maintenance
Producing and distributing the E-
learning program
Promoting the E-learning
Maintaining the E-learning program

In Carliner research, it is found that the IDM provides a
sufficient guidance to build generic E-learning system.
Generic E-learning is usually used for schools and universities.
It contains curriculum, course contents, practices, and course
evaluation and student profile. However, the IDM model is
rarely used in an organization. This is because the company
needs is different with the needs of the academic world.
Company focused e-learning as a tool to increase not only the
knowledge but also their productivity.
An implementation of IDM model in organization is done
by OBrien and Hall [8] by constructing a model called
Training Need Analysis (TNA). It is shown that before an
organization creates the e-learning they should enable
companies to identify areas where their employees require
training. The research provides generic TNA tools that are
dedicated for SMEs allowing them to identify training
requirements and assisting them to specify their own e-
learning content in a structure. The tool implicitly shows that
to create an effective e-learning, it needs to align with the
organization need through modeling rule analysis and design
e-learning content.
The research about modeling rule analysis and design in E-
learning content is done by Kim and Choi [10]. In the
research, it shows that the usage a SCORM model as content
aggregation and design model that can be the foundation to
design and develop learning management system and contents.
The development of learning management and e-learning
focused in several initiatives that are related in content, type of
e-learning and deliverable model. Table II display the previous
researches that related with E-learning development that
related with an organization.
TABLE II. RELEVANT RESEARCH IN E-LEARNING DEVELOPMENT IN AN
ORGANIZATION
Authors Research Topic
Teo and Gay [6] A prototype system that performs a subset of
functions (learner profiling, knowledge
visualization, and learning route mapping) is
being developed
Stephenson, et al [9] Building a specific e-learning for an
organization called KaryoLab. KaryoLab
contains background, tutorial, practice, and
assessment
Schiaffino, et al [14] an adaptive hypermedia system that uses the
adaptability concept with the aim of providing
the same content to different students groups
Ferretti, et al [13] Building an E-learning system called We-
Lcome. We-LCoME aim is to mashup
compound multimedia potentials with the so
called collective intelligence which the new
Web 2.0 has revealed. Final e-learning media
by using SMIL
Moller, et al [7] An integrative concept for information and
communication technology (ICT) supported
education in modeling and simulation (M&S).
The implementation of the M&S program
uses ICT as an indispensable part of the
modern education system.

Based on the previous researches that described in Table
II, it shows that the development E-learning in an organization
should focus on:
The specific need of the user in an organization. The e-
learning initiative is done by creating custom
application that delivers specific learning experiences.
Building the system with multimedia and collaboration
standard such as web 2.0 or SMIL.
Profiling the user based on their skill, knowledge gap,
and learning plan.
Vol. 3, No.8, 2012
3 | P a g e
The last step on the IDM model is the implementation and
maintenance phase of E-learning, also called as an adoption
phase. This phase is critical path of the E-learning adoption.
Table III Shows The Related Research That Focuses In E-
Learning Adoption.
TABLE III. RELEVANT RESEARCH IN E-LEARNING ADOPTION
Authors Research Topic
Mesomela and
Villiers [12]
The study that covers usability evaluation or
an examination of the user interface of an e-
learning product and usage analysis of an e-
learning application designed to support
learning in a cognitive domain
Luojus and Vilkki
[13]
The pedagogical starting point for developing
instruction in digital media was to produce
new competence. The outcome of the
development work was a teaching model that
follows the user-driven design process, with
the aim of providing students with the ability
to act as developers of product development
and innovation processes in their fields
Bang [19] Integrate Multimedia Digital AP Server to
provide immediate self-help and
strengthening materials to students so that
students can focus on the part they needed to
be strengthened among extensive pool of
books, thus helping learners to enhance the
fun and efficiency of learning
Mehlenbacher, et al
[3]
The research reviews 300 e-learning journal
and one of the related point of view is about
how Web Based Interface provide sufficient
interface to learn effectively

Based on the research that described in Table III, it shows
that the adoption of the E-learning depends on:
The timeframe of the adoption. It is shown that the
quick adoption make the participant should learn more
intensively. The technology such as web provides a
flexible way to learn.
The usage of a skill that delivered in E-learning. It is
shown that a skill that related with the daily work of an
organization lead to a quick adoption on the
organization E-learning.
Contextual and learning plan selection. Several
researches show that effective e learning need align
with content design based on the user need.
It is found that the implementation E-learning for an
organization to learn something or to adopt anything is
different with the implementation E-learning in the university
or academic learning. Based on the previous researches, it
shows that the E-learning in organization should have unique
approach to achieve the benefits.
II. THEORETICAL BACKGROUND
As mentioned before, the purposes of the research is to
create a learning plan for any organization who wants to adopt
ICT system through E-learning model. Based on previous
researches, it shows various approaches to learn and adopt E-
learning in an organization. In this research, it is chosen that
the research will use the IDM model. The IDM model is
chosen because:
IDM model is mature enough and has been proven in
academic and industry. [1] [4] [5] [11]
IDM model provides complete engineering process for
E-learning lifecycle.
IDM can be pattern matched with a software
development lifecycle (SDLC). It has similar phases
such as requirement, analysis-design, development, and
deployment [17]. Table IV shows the pattern matching
between IDM phase and SDLC.
TABLE IV. IDM AND SDLC PATTERN MATCHING
IDM Phases SDLC Phases
(V&V)
Similarity
Definition Requirements
Engineering
Providing a detail step to
refine the purposes, to
select the technology, to
collect the need of a
system
Design Design Providing a several action
to design the system
Development Development Modification, construct,
and installing the system
Implementation
and Maintenance
Testing and
Implementation
User acceptance testing
and system socialization.

Table IV shows that both IDM and SDLC has several
similarities. The similarity leads the research to create a
research phases that combined both phases. The great
combination focuses in the last phase that is implementation.
In the implementation phase, the SDLC phase focuses in
User Acceptance Test (UAT). UAT in software engineering
provides several activities that dedicated for the properness of
the created system with the user. The proposed system should
meet its business requirements and to provide confidence that
the system works correctly and is usable before it is formally
"delivered" [18]. UAT should also address the testing of
system documentation.
The IDM also bring the UAT in the first class testing with
addition system socialization. When the proposed system
meets the business need, the next challenge is to adopt the
system to the existing environment. In this step, IDM provides
step to promote E-learning. Based on that fact, we strongly
believe that the IDM and SDLC give sufficient lifecycle to
adopt ICT system. Both phases provide complete lifecycle in
ICT development and adoption. In this research, we focus in
the adoption only because several reasons such as:
The system is already built by a software vendor.
Therefore, it has no need to do full lifecycle step in
SDLC. It just needs focus in IDM phase.
The system is a common system for ICT
Communication. In this research, the system is a full
suite of communication platform such as email, online
storage, and PIM software. Therefore, it will no need
special training or long workshop.
The adopted system has a previous similar system.
Therefore, some user feels comfortable with the
Vol. 3, No.8, 2012
4 | P a g e
existing system and need a good reason to use the new
system.
Based on that reasons, it will start the research by creating
adoption plan, constructing the learning model, and evaluating
the learning model.
III. ADOPTION PLANNING
The adoption plan starts with the implementation of the
system. Since, the ICT system is Software as a Services
solution the implementation simply as activating the services,
enrolling the user and testing the services availability. Figure 1
show the research phase that designed based on the
combination of both.

Figure 1. Research Phase
After doing implementation in the new system, the
research starts from conventional adoption strategies which are
the distribution of documentation and guidance. At this point,
the organization stakeholder creates a memo to the employees
to use the system. The memo enriches with manual and user
guide for the system usage. In this case, the research prepares
the instrumentation system to evaluate the usage of the system.
Fortunately, the proposed system has out of the box-reporting
feature called account trend. Account trend covers provisioned
and active account. Provisioned account is the numbers of
enrollment account. Active account is the numbers of active
user that use the services. By dividing the active account with
the provisioned account the research declares the active usage
index. The active usage index will have a value between 0.00
1.00. The closer to the 1.00 the better value to the for active
usage index. Figure 2 shows the reporting features in the
proposed system.

Figure 2. Active ID Excerpt Sample
The active usage index will be the main evidence for the
research result. The active usage index will be calculated
manually on this research. The active usage index will be
measured before and after the treatment. As mentioned before,
the treatment step is started by creating a learning model that
described in Section V. The learning model implementation
report will be discussed in Section VI.
IV. CONTSRUCTING A LEARNING MODEL
In this research, it is assumed that the learning model is an
approach to deliver good e-learning experience. It covers
behavior of user, content learning plan, and socialization
technique. With a scope that the main focus of the training is
quickly adopt a new ICT system. The proposed learning model
is constructed by doing several activities such as:
Observing to the organization and see what they use
and like in the existing system.
Meeting with the stakeholder what they think and they
hope with the new system.
Classifying the learning item based on the need of the
organization
Creating the learning plan based on the system features
classification.
Creating learning content based on the learning plan.
Distributing learning content based on the discussion
within the stakeholder.
The observation activity covers a set of action that engages
between implementation team with the client. In this step, it is
done by using formal discussion and quick pool. Formal
discussion did a quick observation about daily activity of the
employee organization. Fifty employees join the session. The
research does patterns matching between their daily activity
with the proposed ICT system. Table V shows the system
features and the usage scenarios that related with the
organization business process. The system features is ranked
by the result of quick pool.
TABLE V. SYSTEM FEATURES AND USAGE SCENARIOS PATTERNS
MATCHING
System Features Usage Scenarios Priority
Email Business Communication, Internal
memo, business letter
High
Calendar internal meeting schedule High
Task Assignment memo from supervisor
to emplooye
Medium
Online Storage Storing digital multimedia content
such as document and e-learning
content
Medium
Public Address
Book
Organization contact repository Medium
Instant Messaging Quick chat Low

The priority is concluded from a quick pool in the
discussion session. Each participant will select the priority
(high medium and low). If 70% participants select the same
priority, the feature will have the selected priority. It is shown
that email and calendar are the most demanded features in the
organization. The result of the quick pool and discussion is
act Research Plan
Research
Start
System Implementation
and Deployment
User guide and
Documentation
Distribution
Usage Monitoring Start
Creating Learning Plan
Based on IDM
Creating Prototyping
based on IDM Plan
Evaluation IDM plan and
prototyping
Comparing The System
Usage
Implementation, usage
analysis, and pattern
Research
Fi ni sh
Vol. 3, No.8, 2012
5 | P a g e
summarized and presented to the stakeholders in stakeholder
gathering session.
Stakeholder gathering session establishes a shared vision
between implementation team and stakeholder. In this
research, the stakeholder gathering sessions discuss several
items that are:
An initial report that is obtained from the observation
session.
A feedback sharing session that is come from the
stakeholder about the new system.
The current activity and problem that might be solved
by the new system
Short discussion about the challenge that might be or
already happen when adopting the new system.
The milestone plan that will be adopted to implement
new system.
Table VI summarizes the result of the gathering session
with the stakeholder. It shows the problems, approaches, and
the milestone when the problems and approaches is executed.
TABLE VI. MILESTONE, PROBLEM, AND APPROACH PLANNING
Milestone Problem Addressed Approaches
Piloting
Phase
Not all full time
employees already
enrolled to the system
Mass registration to the
system
Employees do test drive
the new system without
manual
User guide and quick
reference card is distributed
using online storage channel
The user guide and quick
manual wont be read
during the business of
the employee
Building a short tutorial
rather than user guide
Employees are confused
to import their existing
work into the new
system
Building a tutorial how to
migrate the current work
into the new system
Adopting
Phase
The new system
adoption should be also
adopted by the client and
customer.
Enrolling and distributing
key access of the system
The impossibility to do
socialization and
workshop for more than
1000 clients and
customers
Building an e-learning
portal that can be accessed
online

Table VI shows that the learning model should covers
three main approaches for learning which are building short
tutorial for daily usage, building short tutorial for migrating a
system, and building a learning portal for remote consumer.
The three activities contents are structured by the team in an
internal meeting.
The internal meeting classifies the main structure of the
leaning model. The structure is composed by considering
several items which are:
The employees have several information assets in the
existing system. Therefore, the first learning structure
should cover a migration strategies and guidelines.
The employees should know the main part of the
application that related with their daily business
activity. Therefore, after the migration the employee
should learn the system features based on priority that
already described in Table V.
The remote users such as client and consumer should
learn how to use e-learning portal that will be deployed
as an approach to make the new system could be used
by the entire organization.
Figure 3 shows the learning plan that will be used as a
framework for the learning content. Migration learning
focuses to cover any actions that should be done to migrate
from existing to new system. Essential learning phase covers
the essential topic that have to be mastered to use the system.
Enhancement learning phase is a continuous phase that will be
updated regularly. The enhancement learning phase will cover
additional topic that not covered by the essential learning
phase. In this step, the implementation learning phase will be
executed in three months.

Figure 3. Learning Phases For Implementing ICT System
Each phase will have several contents based on the
discussion within stakeholder and the 50 main users. Table VII
describes the learning content in every phase.
TABLE VII. MILESTONE, PROBLEM, AND APPROACH PLANNING
Learning
Phase
Learning Topic
Migration
Learning
Migrating the address book of the existing system
Archiving the old email and conversation
Archiving and converting the document to prepare
the online storage
Essential
Learning
Basic email configuring with Outlook Web
Access (OWA)
Downloading and managing email through
Outlook client
PIM management using OWA
Instant messaging in OWA
Office productivity using Office Web Apps
Online storage and archiving with SkyDrive
Enhancem
ent
Learning
Accessing an email in a limited connection
Sharing and managing the online storage in
SkyDrive
Accessing and connecting other system through
OWA
Others topic will be discussed on demand
After the creation of the learning plan, implementation
team to create an ad-hoc team that focuses to develop the
learning content. The team is composed three persons namely
instruction design mentor and two talents. The instruction
design mentor creates the tutorial scenarios. The tutorial itself
works as a demo driven tutorial. After the scenario is created,
the design mentor will propose the scenario to be recorded by
the talents. The tutorial output is composed as three main
outputs that are slide deck presentation, recorded video, and
Migration
Learning
phase
Essential
Learning
phase
Enhancement
Learning
phase
Vol. 3, No.8, 2012
6 | P a g e
demo script document. The outputs will be uploaded to the E-
learning portal. The implementation of the E-learning content
and the socialization of e-learning portal will be described in
Section VI.
V. LEARNING MODEL IMPLEMENTATION
The learning model implementation is aligned with the
ICT implementation phase. The implementation model is done
through two phases which are internal phase and external
phase. Internal phase focuses in implementation activity in
internal organization. It contains 100 employees that work
permanently onsite in the organization. The external phase
focuses implementation in external organization such as
customer, alumni, community, and remote workers. It reaches
1500 persons that are separated geographically around Java,
Sumatra, and Kalimantan.
Learning implementation in internal phase is done through
blended learning. The blended learning contains an onsite
workshop and self-paced online learning. Onsite workshop
contains any material that related with essential learning that
described in Table VII. On the other hand, the online learning
is done by using organization online storage that stored in the
new system. The online learning works as follow:
The learning contents are uploaded into an online
storage that parts of the new system. There are seven
video lessons that uploaded and ready to view as self-
paced e-learning.
The learning contents link is distributed by the ICT
supervisor in the organization. The link is distributed
through a new email system.
The learning content has a playlist as a step-by-step
recommendation to learn the essential features of the
system. Figure 4 shows the video playlist.

Figure 4. Video Playlist For Self-Paced E-Learning
The internal phase implementation model enhances the
usages of the system by 27% for the first month. After that, the
increment of the system usage is between 20-25% and it is
reached to use by the entire employees in the fourth months.
The external phase implementation is started two months
after the internal phase is started. The external phase focuses
in three main activities which are:
Building and configuring online E-learning that can be
accessed through the web. Figure 5 shows the E-online
learning that exposed to the external users.
Collaborating and engaging the local community leader
to learn and to acknowledge the others peers about
online E-learning.
The online E-learning contains the entire E-learning
topic that is described in Table VII.

Figure 5. Online E-Learning For External Implementation Phase
The external phase implementation increases the usages of
the system by 5% for the first week. The thousand numbers of
users that separated in geographically make the
implementation slower. Therefore, the further implementation
and the usage of the new system are much depends with the
local community leader. Therefore, the external phase
implementation is still executed as a part of regular activity in
the organization.
Table VIII shows the overall adoption that reach 27%
average improvement using the proposed method. The
percentage is calculated by counting the system usage in four
months. The calculation compares between before and after
adopting the ICT-IDM method. The percentage is calculated
from the number of active user divides by the total user. The
implementation phase in the table shows the total users based
on external and internal phase for four months.
TABLE VIII. USAGE TABLES
System Features Initial Phase
(Before
Adopting The
ITC-IDM)
Implementation
phase (After
Adopting The
ICT-IDM)
Improvement
Usage
Email 13% 76% 63%
Calendar 2% 23% 21%
Task 0% 21% 21%
Online Storage 5% 38% 33%
Public Address
Book
3% 16% 13%
Instant Messaging 0% 13% 13%
Improvement
average
27%
VI. DISCUSSION AND FURTHER WORK
In this research, it is started with believe that a new ICT
system that developed is not used optimally. Therefore, the
research creates an initiative to do further ICT system
Vol. 3, No.8, 2012
7 | P a g e
socialization by adopting e-learning initiative using IDM
approaches. The research discovers several techniques and
approaches to increase the new ICT system adoption such as
follows.
Closer Communication to the users and the stakeholder.
This technique creates a clear view for the bottleneck and
the real problem that exist in the ICT system
implementation.
Incremental adoption plan. This technique is done by
doing implementation through three milestone phases.
Each phase addresses the specific problem and
approaches.
Problem driven learning development. This technique is
done by creating a learning development through several
learning phases namely migration learning, essential
learning, and enhancement learning. These learning
phases use instruction design model
Multimedia e-learning development. This technique is
done by adopting IDM through multimedia e-learning
content. Video, demo script and slide deck. The contents
are uploaded into online storage to reach the users. The
multimedia e-learning provides self-paced e-learning to
decrease the skill gap between users by providing flexible
way to learn the system.
Iterative learning implementation. This technique is done
by using two phases of the implementation which are
external and internal phase. Both phases use iterative and
continuous improvement model to increase the usage and
the system adoption.
The techniques still has future work since the ICT-IDM
only increase the system usage for just 27%. The hardest part
is to socialize to the external organization. Therefore, it needs
further improvement to eliminate geographical problem, huge
numbers of user, and different skills of the users. The global
model of the ICT-IDM adoption should be proposed for the
further work of the research.
ACKNOWLEDGMENT
We thank Partners in learning program that supports this
research initiative. Microsoft Innovation Center UGM people
and the ICT people at P4TK Mathematic organization that
provided a helpful insight and approaches to do this research.
REFERENCES
[1] Allen, Michael W. 2007. Designing Successful e-Learning: Forget What
You Know about Instructional Design and Do Something Interesting,
Michael Allen's e-Learning Library. Pfeiffer.
[2] Bernard R. Gifford and Noel D. Enyedy. 1999. Activity centered design:
towards a theoretical framework for CSCL. In Proceedings of the 1999
conference on Computer support for collaborative learning (CSCL '99),
USA.
[3] Brad Mehlenbacher, Krista Holstein, Brett Gordon, and Khalil
Khammar. 2010. Reviewing the research on distance education and e-
learning. In Proceedings of the 28th ACM International Conference on
Design of Communication (SIGDOC '10). ACM, New York, USA.
[4] Carliner, Saul, and Patti Shank (eds). 2008. The e-Learning Handbook:
Past Promises, Present Challenges. Pfeiffer.
[5] Carliner, Saul. 2002. Designing E-learning. ASTD. USA. 978-1-56286-
332-6.
[6] Chao Boon Teo and Robert Kheng Leng Gay. 2006. A knowledge-
driven model to personalize e-learning. J. Educ. Resour. Comput. 6, 1,
Article 3 (March 2006).
[7] Dietmar P. F. Mller and Hamid Vakilzadian. 2010. Integrating
modeling and simulation into an e-learning environment in engineering
study programs. In Proceedings of the 2010 Conference on Grand
Challenges in Modeling & Simulation (GCMS '10). Society for
Modeling & Simulation International, Vista, CA, 90-97.
[8] Emma O'Brien and Timothy Hall. 2004. Training Needs Analysis: the
first step in authoring e-learning content. In Proceedings of the 2004
ACM symposium on Applied computing (SAC '04). ACM, New York,
NY, USA, 935-939.
[9] JE Stephenson, WB Morris, HG Tempest, DK Griffin, A Mileham, and
AM Payne. 2007. The use of an e-learning constructivist solution in
workplace learning. In Proceedings of the 14th European conference on
Cognitive ergonomics: invent! explore! (ECCE '07).
[10] Jin-Sung Kim and Kwang-Jin Choi. 2009. Modeling rule for analysis
and design of e-learning content. In Proceedings of the 2009
International Conference on Hybrid Information Technology (ICHIT
'09). ACM, New York, NY, USA, 353-360.
[11] Rothwell, William J., and H.C. Kazanas. 2008. Mastering the
Instructional Design Process: A Systematic Approach, Fourth Edition.
Pfeiffer.
[12] S. S. (Thabo) Masemola and M. R. (Ruth) De Villiers. 2006. Towards a
framework for usability testing of interactive e-learning applications in
cognitive domains, illustrated by a case study. In Proceedings of the
2006 annual research conference of the Souuth African institute of
computer scientists and information technologists on IT research in
developing countries (SAICSIT '06), Judith Bishop and Derrick Kourie
(Eds.). South African Institute for Computer Scientists and Information
Technologists, , Republic of South Africa, 187-197.
[13] Satu Luojus and Olli Vilkki. 2008. Development of user-driven research
methods as the starting point for living lab activities. In Proceedings of
the 5th Nordic conference on Human-computer interaction: building
bridges (NordiCHI '08). ACM, New York, USA.

[14] Silvia Schiaffino, Anala Amandi, Isabela Gasparini, and Marcelo S.
Pimenta. 2008. Personalization in e-learning: the adaptive system vs. the
intelligent agent approaches. In Proceedings of the VIII Brazilian
Symposium on Human Factors in Computing System.
[15] Stefano Ferretti, Silvia Mirri, Ludovico Antonio Muratori, Marco
Roccetti, and Paola Salomoni. 2008. E-learning 2.0: you are We-
LCoME!. In Proceedings of the 2008 international cross-disciplinary
conference on Web accessibility (W4A) (W4A '08). ACM, New York,
USA
[16] Tsui, Frank & Karam, Orlando. 2011. Essentials of software
engineering, second edition. Jones and Bartlett Publishers.
[17] Vliet, Hans van. 2008. Software Engineering: Principles and Practice,
Third Edition. John Wiley & Sons.
[18] Watkins, John & Mills, Simon. 2011. Testing it: an off-the-shelf
software testing process, 2nd edition. Cambridge University Press.
[19] Yao Chin-Bang. 2009. Context-aware customization e-learning system
with intelligent on-line examination mechanism. In Proceedings of the
2nd International Conference on Interaction Sciences: Information
Technology, Culture and Human (ICIS '09). ACM, New York, USA.
AUTHORS PROFILE
Ridi Ferdiana. got his doctoral degree at Universitas Gadjah Mada in 2011.
He earned his master degree from the same university in 2006. In his
professional area, he holds several professional certifications such as MCP,
MCTS, MCPD, MCITP and MCT. In his daily research activities he really
enjoys to learn about software engineering, business platform collaboration,
and programming optimization. He can be reached at [email protected].
Obert Hoseanto is the Partners in Learning Manager of Microsoft Indonesia,
based in Jakarta, Indonesia. His research interests are educational technology,
e-learning and teacher education, and can be contacted at
[email protected].

Vol. 3, No.8, 2012

8 | P a g e
An Enhanced MPLS-TE For Transferring
Multimedia packets

Abdellah Jamali
Dept. of Computer Science and
Mathematics
ESTB, Hassan 1er University
Berrechid, Morocco

Najib Naja
Dept. of RIM
Institute of Posts and
Telecommunications
Rabat, Morocco

Driss El Ouadghiri
Dept. of Computer Science and
Mathematics
My Ismail University
Meknes, Morocco

Abstract Multi-Protocol Label Switching is useful in managing
multimedia traffic when some links are too congested; MPLS
Traffic Engineering is a growing implementation in today's
service provider networks. In This paper we propose an
improvement of MPLS-TE called EMPLS-TE, it is based on a
modification of operation of Forwarding Equivalence Class
(FEC) in order to provide the quality of service to stream
multimedia. The performance of the EMPLS-TE is evaluated by
a simulation model under a variety of network conditions. We
also compare its performance with that of unmodified MPLS-TE
and MPLS. We demonstrate how a small change to the MPLS-
TE protocol can lead to significantly improved performance
results. We present a comparative analysis between MPLS,
MPLS-TE and Enhanced MPLS-TE (EMPLS-TE). Our proposed
EMPLS-TE has a performance advantageous for multimedia
applications in their movement in a congested and dense
environment. EMPLS-TE defines paths for network traffic based
on certain quality of service. The simulation study is conducted in
this paper; it is a means to illustrate the benefits of using this
Enhanced MPLS-TE for multimedia applications.
Keywords- Multi-Protocol Label Switching (MPLS); Multi-Protocol
Label Switching Traffic Engineering (MPLS-TE); Forwarding
Equivalence Class (FEC); Quality Of Service (QoS); Simulation.
I. INTRODUCTION
The goal of Traffic Engineering (TE) is to provide QoS to
multimedia packets by reservation of the resources and
optimum resources utilization [9]. Multiprotocol Label
Switching (MPLS) technology [2] allows traffic engineering
and enhances the performance of the existing protocols over
the traditional IPv4 network. The central idea of MPLS is to
attach a short fixed-length label to packets at the ingress router
of the MPLS domain. Packet forwarding then depends on the
tagged label, not on longest address match, as in traditional IP
forwarding. A router placed on the edge of the MPLS domain,
named Label Edge Router (LER) that is associated to a label
on the basis of a Forwarding Equivalence Class (FEC). In the
MPLS network, internal routers that perform swapping and
label-based packet forwarding are called Label Switching
Routers (LSRs) [15].
MPLS TE also extends the MPLS routing capabilities with
support for constraint-based routing. IGPs typically compute
routing information using a single metric. Instead of that
simple approach, constraint-based routing can take into
account more detailed information about network constraints,
and policy resources. MPLS TE extends current link-state
protocols (IS-IS and OSPF) to distribute such information.
There is another approach to provide QoS to multimedia
traffic: DiffServ-aware Traffic Engineering (DS-TE) [7] [6],
by using three signaling protocols in MPLS networks: Label
Distribution Protocol (LDP) [5], Constraint based Routing
LDP (CR-LDP) [3] and Resource Reservation Protocol-Traffic
Engineering (RSVP-TE) [4].
In this paper we focus on our paper presented in [1] and
MPLS-TE as a technology rather used by operators, then we
make an improvement on MPLS-TE and propose EMPLS-TE
(Enhanced MPLS-TE).
In order to provide a good service for transferring
multimedia packets that requires a large flow in the MPLS-TE
networks we make an improvement to the method of
processing speed in the FEC in MPLS-TE.
Rest of paper is organized as below:
Section II defines QoS as services that provide some
combination of high security, high reliability, low packet drop
rate, low delay, and low jitter. The same section reviews the
working of traditional IP, MPLS and MPLS-TE and their
salient features. In section III, we will describe our proposed
enhancement EMPLS-TE and its methods. In section IV, we
will simulate the MPLS, MPLS-TE and EMPLS-TE, and then
we compare it with original MPLS, and with original MPLS-
TE. Section V, will conclude this paper. Routing, MPLS and
MPLS-TE and their salient features. In section III, we will
describe our proposed enhancement EMPLS-TE and its
methods. In section IV, we will simulate the MPLS, MPLS-TE
and EMPLS-TE, and then we compare it with original MPLS,
and with original MPLS-TE. Section V will conclude this
paper.
II. OVERVIEW
A. Internet Quality of Service (QoS)
Originally, the Internet was developed for transferring file
and accessing remote machines. Therefore, the Internet was
not expected to transfer multimedia data at large data rate.
Today, many different types of applications in Internet
demand more secure more reliable and faster services. Both
non-real time and real-time applications require some kinds of
Vol. 3, No.8, 2012

9 | P a g e
QoS, such as high reliability, bounded delay and jitter, and
high security. Therefore, I would like to define QoS as
services that provide some combination of high security, high
reliability, low packet drop rate, low delay, and low jitter; in
general ATM is an example of a network technology that
provides good QoS.
Although ATM can be used to transmit both IP packets
and ATM data, it is less suitable for best effort services IP
packets mainly because ATM supports only a small part of IP
services. The most common and major QoS problem in the
backbone network is unevenly distributed traffic. MPLS-TE
can distribute traffic evenly and optimize network utilization
TE ensures that all available network resources are optimally
used during times of failure or traffic routing, which is needed
when congestion happens. Network congestion is not easily
solved by IP because of its characteristics: connectionless and
best effort service. As results, bursts of traffic appear
unexpectedly, routers are easily congested, and packets are
dropped. Therefore, the current Internet has poor reliability,
unbounded delay and jitter, and varied throughput.
B. Traditional IP Routing
The IP was created as a connectionless network layer
protocol that makes no attempt to discriminate between
various application types. IP uses routing protocols as
traditional Interior Gateway Routing Protocol (IGRP) [10],
Intermediate System-to-Intermediate System (IS-IS) [14],
Open Shortest Path First (OSPF) [18] to build routing tables
for active number the equations consecutively. Equation
numbers, within parentheses, are to position flush right, as in
(1), using a right tab stop.
Links in an area of network, and therefore transferring data
between the source and the destination [16], the operation of
these protocols depends on how to promote and distribute
information on the state network that are broadcast regularly
and depends also on how to update the routing tables of all
routers located in the same autonomous system (AS).
Each router uses the information on the overall state of
network to maintain an independently its own routing tables so
that it can transfer data successfully using the shortest path or
the link state as metric maintain before deciding to send data.
The major problem of some of these protocols is that they
transfer the data on the paths with minimal hops, and since
they do not use the paths with many hops that can lead the data
to the destination, then this strategy produces quite congested
links.
So, the traditional routing IP traffic is routed by the same
types of paths (short), and therefore a fairly large amount of
packets is lost.
To tackle the problem of low delay and packet loss during
the delivery of multimedia applications, it is necessary to think
of improvement methods to use more effectively the available
network resources. MPLS and MPLS-TE (MPLS Traffic
Engineering) are some process that provides this functionality.
C. MPLS
Multiprotocol Label Switching (MPLS) can speed up the
flow of network traffic and make it easier to manage. MPLS is
flexible, fast, cost-efficient and allows for network
segmentation and quality of service (QoS). MPLS also offers a
better way of transporting latency-sensitive applications like
voice and video. While MPLS technology has been around for
several years, businesses are now taking advantage of service
provider offerings and beginning their own corporate
implementations.
MPLS can be considered a technology that has brought an
oriented connection for IP protocol. Therefore, network
services and applications can exploit all of the advantages of
MPLS. In other words, MPLS is a connection oriented
technology that uses a label swapping technique with IP
network routing [12]. A label is a small, fixed index, which
identifies a Forward Equivalence Class; a group of IP packets
that are forwarded over the same path with the same packet
treatments. With MPLS, the packet is faster than with use IP
address because MPLS uses labels to quickly check the next
hop that leads to the destination without going to the network
layer to analyze the packets along the path.
MPLS consists of routers: Label Switching Routers (LSR)
and Label Edge Routers (LER). These routers use labels to
quickly send packets to the destination.
An LSR is a router that forwards both conventional IP
packets and MPLS labelled packets. An LER is an LSR at the
edge of the MPLS network to add and remove labels. An LER
connects between the MPLS domain and the non-MPLS
domain such as IP network.
A flow of packets coming from a non-MPLS domain is
first assigned a label at an incoming LER and its forward
along the path as an old label is replaced with a new label at
LSRs on the path. Therefore, a label is used to reach the next
node.
Although the exchange of label is required on the path, and
the search of the network layer is not required at LSRs routers
due to transmission of the link layer with labels. In routers
LERS the labels are completely removed and the packets are
transmitted directly to other networks. MPLS label switched
paths are an essential element in delivering end-to-end QoS.
Without them, it is not possible to control the path of packet
flows from requested packet treatments.
The assignment of labels to packets is based on the concept
of forwarding equivalence class (FEC). According to this
concept, packets which belong to the same FEC are assigned
the same label at an ingress node to an MPLS domain. A FEC
consists of packets entering a network through the same
ingress node and exiting the network through the same ingress
node. A FEC also consists of packets requiring similar QoS or
packet treatment across the MPLS domain. The path traversed
by a FEC is called a Label Switched Path (LSP). A signal
protocol such as LDP (Load Distribution Protocol) or RSVP
(Resource reservation Protocol) [17] is used to establish and
release LSPs [13].
Vol. 3, No.8, 2012

10 | P a g e

Figure 1. MPLS Network
D. Traffic Engineering within MPLS
MPLS Traffic Engineering [11] is an obligation for
network operators to provide a fairly reliable infrastructure
and provides quality performance. Traffic Engineering
provides efficient routing of traffic in the network to the use of
network resources. This allows operators the ability to better
exploit bandwidth resources on the network [8].
As a result of the unprecedented growth in demand for
network resources and the competitiveness amongst providers,
Traffic Engineering has become the primary application for
MPLS.
MPLS Traffic Engineering responds to the ineffectiveness
of some routing protocols in terms of datagram processing in
the case of congestion. It allows a wider distribution flow of
traffic across all available resources. Load balancing for TE in
IP network requires an ability to control traffic flow precisely.
In the traditional metric-based control, an administrator can
change only link metrics, and the changes of some link metrics
may affect the overall traffic flow. To manage the
performance of a network, it is necessary to have explicit
control over the paths that traffic flows traverse so that traffic
flows can be arranged to maximize resource commitments and
utilization of the networks [13].
The connection-oriented nature of MPLS allows ISPs to
implement TE in their networks and achieve a variety of goals,
including bandwidth assurance, different routing, load
balancing, path redundancy, and other services that lead to
QoS [9].
MPLS networks can use native TE mechanisms to
minimize network congestion and improve network
performance. TE modifies routing patterns to provide efficient
mapping of traffic streams to network resources. This efficient
mapping can reduce the occurrence of congestion and
improves service quality in terms of the latency, jitter, and loss
that packets experience. Historically, IP networks relied on the
optimization of underlying network infrastructure or Interior
Gateway Protocol (IGP) tuning for TE. Instead, MPLS extends
existing IP protocols and makes use of MPLS forwarding
capabilities to provide native TE. In addition, MPLS TE can
reduce the impact of network failures and increase service
availability. RFC 2702 discusses the requirements for TE in
MPLS networks.
MPLS TE brings explicit routing capabilities to MPLS
networks. An originating label switching route (LSR) can set
up a TE label switched path (LSP) to a terminating LSR
through an explicitly defined path containing a list of
intermediate LSRs. IP uses destination-based routing and does
not provide a general and scalable method for explicitly
routing traffic. In contrast, MPLS networks can support
destination-based and explicit routing simultaneously. MPLS
TE uses extensions to RSVP and the MPLS forwarding
paradigm to provide explicit routing. These enhancements
provide a level of routing control that makes MPLS suitable
for TE.
E. Problem Context And Enhacem Ent Of Mpls-Te
We propose an improvement for the FEC group treatment
in order to consider the throughput as an important parameter
for multimedia applications that allows it to select the best
paths in its routing.
MPLS-TE determines LSP as a sequence of labels in the
packet to construct a path and to convey through these paths
established by the protocol for distributing labels. The problem
of MPLS-TE is how to select the FEC groups that satisfy some
parameters of quality of service and in particular the
throughput which can be considered as important parameter
for some types of applications. The choice of FEC group is
also according to several parameters (source address,
destination address, QoS parameters). To solve this problem,
we group all packets for multimedia applications in a specific
FEC, with a high throughput, and LSP that consists of a
sequence of labels for multimedia packets is associated with
this FEC by the LDP protocol which provides this information
to routers LSR on the throughput which we have chosen for
packets multimedia in a specified FEC. Hence these
multimedia packets take the paths that correspond to
throughput as an important quality of service parameter. After
the improvement in the FEC group associated with multimedia
packets which are labeled, these packets with the principle of
MPLS-TE are switched toward the MPLS-TE network by
using number of label and the LSP paths. The LSR routers of
MPLS-TE network switches the FEC labels that we improved
to LER routers, taking into account the throughput that we
have set for these multimedia packets.
F. Simulation And Analysis Of The Solution
To analyze the proposed solution and the effectiveness of
our suggested enhancement in MPLS-TE, we use an event-
driven network simulator targeted at networking research. The
software version used in this paper is ns-2.34 with MPLS
Network Simulator (MNS 2.0).
MPLS-TE and EMPL-TE as discussed in the previous
sections have several desirable capabilities. However in this
paper, the simulation was chosen to demonstrate the ability of
EMPLS-TE in providing Traffic Engineering. To demonstrate
this capability, the simulations were setup using a normal
MPLS, and a normal MPLS with Traffic engineering
implemented (MPLS-TE). The results from these simulations
are used for the comparison between the three approaches and
evaluate our proposed scheme. Both simulations are based on
the common topology.
Vol. 3, No.8, 2012

11 | P a g e
G. Simulation environment
The network consists of 90 nodes (in backbone, sources
and destination). All links were set up as duplex with 15 ms
delay and using Drop Tail Queuing, which serve packets on a
First Come First Serve (FCFS) basis. The simulation time is
200s and the links have a capacity of 1.5 Mbps and the
transmitted flux in the network is multimedia.
TABLE I. SIMULATION PARAMETRS
Simulation Parameter

Value

Simulator NS-2.34
Simulation Time 200s
Node Max. IFQ Length 50
Data Packet Size 512 bytes
Traffic type CBR(UDP)
Packet rate 4pkt/sec
H. Performance Metrics
The following metrics are used in varying scenarios to
evaluate different protocols:
Packet delivery ratio - This is defined as the ratio of the
number of data packets received by the destinations to those
sent by the CBR sources.
End-to-end delay of data packets - This is defined as the
delay between the time at which the data packet was
originated at the source and the time it reaches the destination.
Data packets that get lost en route are not considered. Delays
due to route discovery, queuing and retransmissions are
included in the delay metric.
The metrics are measured against various mobility
scenarios and with varying number of data connections.
I. Comparison between MPLS, MPLS-TE and EMPLS-TE
In this subsection, we present a comparative analysis
of the performance metrics of the MPLS, MPLS-TE and our
approach EMPLS-TE.

Figure 2. Average delay Vs number of sessions CBR
Packet Delivery Ratio:
Figure 3 gives the packet delivery ratio when the number
of sessions (CBR) varies. With number of sessions from 60 to
80 both EMPLS-TE and MPLS-TE has almost same packet
delivery ratio but as with number of sessions from 10 to 60 the
packet delivery fraction of EMPLS-TE is better. The ratio
decreases rapidly in case of MPLS whereas MPLS-TE
maintains the same ratio. Thus with the increase in number of
sessions EMPLS-TE gives more packet delivery fraction
thereby outperforming MPLS and MPLS-TE.

Figure 3. Packet delivery fraction Vs number of sessions CBR
Packets loss :
EMPLS-TE has less packets loss than both MPLS and
MPLS-TE under almost all possible values of number of
sessions. The difference is magnified under high number of
sessions (40 and 60). The primary reason is that in MPLS-TE
architecture, we group all packets for multimedia applications
in a specific FEC, with a high throughput as compared to that
in MPLS. MPLS-TE performs considerably better than both
MPLS and MPLS-TE, because MPLS and MPLS-TE focus on
LSP routes with the fewest hops, while MPLS-TE tends to
choose the least congested route with a specific FEC. Also,
when utilizing promiscuous listening MPLS-TE has to spend
time processing any control packet it receives, even if it is not
the intended recipient. For the time of simulation, the packets
loss increases with an increase in the number of sessions.

Figure 4. Packets Vs Number of sessions
Throughput with UDP :
From Figure 5, it is clear that at 10 second, MPLS-TE
gives better throughput than EMPLS-TE. As the simulation
times increases to 50, both MPLS and MPLS-TE have almost
the same throughput but as the simulation times increases
beyond 200 EMPLS-TE outperforms MPLS-TE and MPLS (as
the throughput of all MPLS, MPLS-TE and EMPLS-TE
increase with simulation times).
The throughput of EMLPS-TE is similar to MPLS and
MPLS-TE between 10s and 50s. The architecture suffers a
Vol. 3, No.8, 2012

12 | P a g e
little at fewer simulation times. At low simulation time, the
throughput does not exceed 0, 4 Mbps in MPLS, MPLS-TE
and EMPLS-TE due to packet collisions. This is because the
number of collisions increases in EMPLS-TE due to the
additional pending data packets sent by the intermediate routes
during route discovery. The throughput increases quickly with
increase in simulation times from 100s. Our EMPL-TE
solution is very efficient at 200s. The obtained results show
that EMPL-TE is an architecture designed for long periods.

Figure 5. Throughput Vs Simulation times
Throughput with TCP :
In the first two source nodes send the multimedia stream to
their destination through the MPLS network, MPLS-TE
network and EMPLS-TE network. And we calculate the TCP
throughput and UDP throughput at two destinations (see
Figure. 5 and Figure. 6), we note that between 0 second and 10
seconds, our approach EMPLS-TE is more efficient, we obtain
an important value of packet delivery ratio, but with author
approach (MPLS-TE), the result is not efficient between 0s
and 10s.
This improvement of the packet delivery ratio is due to
enhanced throughput with FEC that we changed in MPLS-TE,
and as result it performs the transmission of multimedia
stream.

Figure 6. Throughput Vs Simulation times
III. Conclusion
Through simulation results and analysis, it was clear that
MPLS-TE does not provide a reliable service and improved
packet delivery ratio as an important performance metric to
ensure the arrival of received packets for sensitive applications
as multimedia packets.
ACKNOWLEDGMENT
This work was supported in part by the networks
laboratory at Institute of the Post and Telecommunications-
Rabat. We would like to thank for his help and the anonymous
referees for their valuable comments.
REFERENCES
[1] A. Jamali, N. Naja, D. El Oudghiri, and R. Benaini, Improving quality
of service (QoS) in multi-protocol label switching module, in
MMS2009 IEEE Xplore.
[2] R. Callon, E. Rosen, and A. Viswanathan, Multiprotocol label
switching architecture, In IETF RFC3031, Janvier, 2001.
[3] B. Jamoussi, and al. Constraint based isp setup using idp, In IETF
RFC3212, Junuary 2002 ggdd
[4] D. Awduche and al. Rsvp-te : extensions to rsvp for isp tunnels, In
IETF RFC3209, December 2001.
[5] W. Lai, Requirements for support of differentiated services-aware mpls
traffic engineering, In IETF RFC3564, July 2003.
[6] A. Lund, C. Reingold, N. Rexford, J. Feldmann, and A. Greenberg,
Traffic engineering for ip networks, IEEE Network Magazine, vol. 14,
pp. 1119, April 2000.
[7] C. L. Hedrick, An introduction to igrp, In cisco white paper, August
1991.
[8] H. Wang, J. Liebeherr, J. Wang, and S. Patek, Traffic engineering with
aimd in mpls networks, In LNCS, May 2002.
[9] A. Toguyni, A. Rahmani, K. Lee, and A. Noce, Comparison of
multipath algorithms for load balancing in a mpls network, ICOIN05,
Published by Springer as Lecture Notes in Computer Science (LNCS),
pp. 463-470, February 2005.
[10] L. Kyeongja, Global model for quality of service in fai networks :
integration of diffserv and traffic engineering based on mpls, In Thesis,
order number 0025. university of science and technology , Central
School .
[11] A. Martey, Integrated is-is routing protocol concepts, In Cisco Press,
May 2002.
[12] K. A. M. Lutfullah, M. Z. Hassan, M. R. Amin, M. Arifur Rahman, and
A. H. Kabir, Performance analysis and the study of the behavior of
mpls protocols, In International Conference on Computer and
Communication Engineering, pp. 1315, Kuala Lumpur, Malaysia, May
2008.
[13] P. Lewis James, Cisco tcp/ip routing professional reference,revised
and expanded. 1998.
[14] S. Berson, S. Herzog, S. Jamin, R. Braden, and L. Zhang, Resource
reservation protocol (rsvp), In RFC2205, Septembre 1997.
[15] D. Williams R. Guerin, and A. Orda, Qos routing mechanisms and ospf
extensions, In of the Global Internet Miniconference, 1997.
AUTHORS PROFILE
Jamali Abdellah is a researsh and professor at Hassan
1
er
University, Settat, Morroco, since Ocyober 2011. He
was born in Ouarzazate, Morocco. He received a thesis
in Computer Science from the University of Hassan II,
Mohammedia, Morocco. He is a founding member, in
2007, of a research group e-NGN (e-Next Generation
Networks) for Africa and Middle East. His research
interests include the computer networks, IPv6
Networks, Quality of Service in MPLS and QoS in Ad Hoc and
Networks performance analysis.
Najib NAJA was born in Maohammadia, Morroco. In
1989, He received his engineering degree from
TELECOM Bretagne, France, option : Computer
Science and Networks. In 1994, he received the PhD
thesis in Computer Science from the University of
Rennes I, France. In 1997, he received a thesis in
telecommunications from the University of
Mohammed V, Morocco. His research interests include the Information
Vol. 3, No.8, 2012

13 | P a g e
system, Ad Hoc Network, signal processing, information technology, QoS in
Ad Hoc, Networks performance analysis. He is Professor in The National
Institute of Posts and Telecommunications (INPT) in Rabat, Morocco since
1994.
Driss El Ouadghiri is a research and an associate professor at Science
Faculty, Moulay Ismail University, Meknes, Morocco, since September 1994.
He was born in Ouarzazate, Morocco. He got his License in applied
mathematics and his Doctorat de Spcialit de Troisime Cycle in computer
networks, respectively, in 1992 and 1997 from Mohamed V
University, Rabat, Morocco. In 2000 he got his PhD in
performance evaluation in wide area networks
from Moulay Ismail University, Meknes, Morocco. He is
a founding member, in 2007, of a research group e-NGN
(e-Next Generation Networks) for Africa and Middle
East. His research interests focus on performance
evaluation in networks (modelling and simulation),
DiffServ architecture (mechanisms based active queue
management) and IPv6 networks.

Vol. 3, No.8, 2012
14 | P a g e
A New Algorithm for Data Compression
Optimization
I Made Agus Dwi Suarjaya
Information Technology Department
Udayana University
Bali, Indonesia

Abstract People tend to store a lot of files inside theirs storage.
When the storage nears it limit, they then try to reduce those files
size to minimum by using data compression software. In this
paper we propose a new algorithm for data compression, called j-
bit encoding (JBE). This algorithm will manipulates each bit of
data inside file to minimize the size without losing any data after
decoding which is classified to lossless compression. This basic
algorithm is intended to be combining with other data
compression algorithms to optimize the compression ratio. The
performance of this algorithm is measured by comparing
combination of different data compression algorithms.
Keywords- algorithms; data compression; j-bit encoding; JBE;
lossless.
I. INTRODUCTION
Data compression is a way to reduce storage cost by
eliminating redundancies that happen in most files. There are
two types of compression, lossy and lossless. Lossy
compression reduced file size by eliminating some unneeded
data that wont be recognize by human after decoding, this
often used by video and audio compression. Lossless
compression on the other hand, manipulates each bit of data
inside file to minimize the size without losing any data after
decoding. This is important because if file lost even a single bit
after decoding, that mean the file is corrupted.
Data compression can also be used for in-network
processing technique in order to save energy because it reduces
the amount of data in order to reduce data transmitted and/or
decreases transfer time because the size of data is reduced [1].
There are some well-known data compression algorithms.
In this paper we will take a look on various data compression
algorithms that can be use in combination with our proposed
algorithms. Those algorithms can be classified into
transformation and compression algorithms. Transformation
algorithm does not compress data but rearrange or change data
to optimize input for the next sequence of transformation or
compression algorithm.
Most compression methods are physical and logical. They
are physical because look only at the bits in the input stream
and ignore the meaning of the contents in the input. Such a
method translates one bit stream into another, shorter, one. The
only way to understand and decode of the output stream is by
knowing how it was encoded. They are logical because look
only at individual contents in the source stream and replace
common contents with short codes. Logical compression
method is useful and effective (achieve best compression ratio)
on certain types of data [2].
II. RELATED ALGORITHMS
A. Run-length encoding
Run-length encoding (RLE) is one of basic technique for
data compression. The idea behind this approach is this: If a
data item d occurs n consecutive times in the input stream,
replace the n occurrences with the single pair nd [2].
RLE is mainly used to compress runs of the same byte [3].
This approach is useful when repetition often occurs inside
data. That is why RLE is one good choice to compress a bitmap
image especially the low bit one, example 8 bit bitmap image.
B. Burrows-wheeler transform
Burrows-wheeler transform (BWT) works in block mode
while others mostly work in streaming mode. This algorithm
classified into transformation algorithm because the main idea
is to rearrange (by adding and sorting) and concentrate
symbols. These concentrated symbols then can be used as input
for another algorithm to achieve good compression ratios.
Since the BWT operates on data in memory, you may
encounter files too big to process in one fell swoop. In these
cases, the file must be split up and processed a block at a time
[3]. To speed up the sorting process, it is possible to do parallel
sorting or using larger block of input if more memory
available.
C. Move to front transform
Move to front transform (MTF) is another basic technique
for data compression. MTF is a transformation algorithm which
does not compress data but can help to reduce redundancy
sometimes [5]. The main idea is to move to front the symbols
that mostly occur, so those symbols will have smaller output
number.
This technique is intended to be used as optimization for
other algorithm likes Burrows-wheeler transform.
D. Arithmetic coding
Arithmetic coding (ARI) is using statistical method to
compress data. The method starts with a certain interval, it
reads the input file symbol by symbol, and uses the probability
of each symbol to narrow the interval. Specifying a narrower
interval requires more bits, so the number constructed by the
algorithm grows continuously. To achieve compression, the
algorithm is designed such that a high-probability symbol
Vol. 3, No.8, 2012
15 | P a g e
narrows the interval less than a low-probability symbol, with
the result that high-probability symbols contribute fewer bits to
the output [2].
Arithmetic coding, is entropy coder widely used, the only
problem is its speed, but compression tends to be better than
Huffman (other statistical method algorithm) can achieve [6].
This technique is useful for final sequence of data compression
combination algorithm and gives the most for compression
ratio.
III. PROPOSED ALGORITHM
J-bit encoding (JBE) works by manipulate bits of data to
reduce the size and optimize input for other algorithm. The
main idea of this algorithm is to split the input data into two
data where the first data will contain original nonzero byte and
the second data will contain bit value explaining position of
nonzero and zero bytes. Both data then can be compress
separately with other data compression algorithm to achieve
maximum compression ratio. Step-by-step of the compression
process can be describe as below:
1. Read input per byte, can be all types of file.
2. Determine read byte as nonzero or zero byte.
3. Write nonzero byte into data I and write bit 1 into
temporary byte data, or only write bit 0 into
temporary byte data for zero input byte.
4. Repeat step 1-3 until temporary byte data filled with 8
bits of data.
5. If temporary byte data filled with 8 bit then write the
byte value of temporary byte data into data II.
6. Clear temporary byte data.
7. Repeat step 1-6 until end of file is reach.
8. Write combined output data
a) Write original input length.
b) Write data I.
c) Write data II.
9. If followed by another compression algorithm, data I
and data II can be compress separately before
combined (optional).
Figure 1 shows visual step-by-step compression process for
J-bit encoding. Inserted original input length into the beginning
of the output will be used as information for data I and data II
size. As for step-by-step of the decompression process can be
describe below:
1. Read original input length.
2. If was compressed separately, decompress data I and
data II (optional).
3. Read data II per bit.
4. Determine whether read bit is '0' or '1'.
5. Write to output, if read bit is '1' then read and write
data I to output, if read bit is '0' then write zero byte to
output.
6. Repeat step 2-5 until original input length is reach.

Figure 1. J-bit Encoding process
IV. COMBINATION COMPARISON
Five combinations of data compression algorithm are used
to find out which combination with the best compression ratio.
The combinations are:
1. RLE+ARI.
2. BWT+MTF+ARI.
3. BWT+RLE+ARI.
4. RLE+BWT+MTF+RLE+ARI (as used in [3]).
5. RLE+BWT+MTF+JBE+ARI.
Those combinations are tested with 5 types of files. Each
type consists of 50 samples. Each sample has different size to
show real file system condition. All samples are uncompressed,
this include raw bitmap images and raw audio without lossy
65 01000001
0 00000000
65 01000001
0 00000000
0 00000000
0 00000000
66 01000010
65 01000001

65 01000001
65 01000001
66 01000010
65 01000001

1 00000001
0 00000000
1 00000001
0 00000000
0 00000000
0 00000000
1 00000001
1 00000001

163 10100011

Original

Data I

Data II

Temporary byte
data

8 bytes

4 bytes

1 byte

Vol. 3, No.8, 2012
16 | P a g e
compression. Average compression ratio for each type of file is
used. Samples for the experiment are show in table 1.
TABLE I. SAMPLES FOR COMBINATION INPUT
No Name Qty Type Spec.
1 Image 50 Bitmap Image Raw 8 bit
2 Image 50 Bitmap Image Raw 24 bit
3 Text 50 Text Document
4 Binary 50 Executable, library
5 Audio 50 Wave Audio Raw
V. RESULT
Figure 2 shows that 8-bit bitmap images are compressed
with good compression ratio by algorithms that combined with
J-bit encoding.

Figure 2. Ratio comparison for 8-bit bitmap image
Figure 3 shows that 24-bit bitmap images are compressed
with better compression ratio by algorithms that combined with
J-bit encoding. A 24 bit bitmap image has more complex data
than 8 bit since it is store more color. Lossy compression for
image would be more appropriate for 24 bit bitmap image to
achieve best compression ratio, even thought that will decrease
quality of the original image.

Figure 3. Ratio comparison for 24-bit bitmap image
Figure 4 shows that text files are compressed with better
compression ratio by algorithms that combined with J-bit
encoding.

Figure 4. Ratio comparison for text
Figure 5 show that binary files are compressed with better
compression ratio by algorithms that combined with J-bit
encoding.
0
25
50
75
100
Combination
33.32
71.77
30.69
22.75 22.56
Based on average ratio of 50 samples
RLE+ARI BWT+MTF++RLE BWT+RLE+ARI
RLE+BWT+MTF+RLE+ARI RLE+BWT+MTF+JBE+ARI
0
25
50
75
100
Combination
85.65
85.73
66.76
56.73
55.42
0
25
50
75
100
Combination
65.19
66.87
44.8
34.08
33.58
Vol. 3, No.8, 2012
17 | P a g e

Figure 5. Ratio comparison for binary
Figure 6 shows that wave audio files are compressed with
better compression ratio by algorithms that combined with J-bit
encoding.

Figure 6. Ratio comparison for wave
VI. CONCLUSION
This paper proposes and confirms a data compression
algorithm that can be used to optimize other algorithm. An
experiment by using 5 types of files with 50 different sizes for
each type was conducted, 5 combination algorithms has been
tested and compared. This algorithm gives better compression
ratio when inserted between move to front transform (MTF)
and arithmetic coding (ARI).
Because some files consist of hybrid contents (text, audio,
video, binary in one file just like document file), the ability to
recognize contents regardless the file type, split it then
compresses it separately with appropriate algorithm to the
contents is potential for further research in the future to achieve
better compression ratio.
REFERENCES
[1] Capo-chichi, E. P., Guyennet, H. and Friedt, J. K-RLE a New Data
Compression Algorithm for Wireless Sensor Network. In Proceedings of
the 2009 Third International Conference on Sensor Technologies and
Applications.
[2] Salomon, D. 2004. Data Compression the Complete References Third
Edition. Springer-Verlag New York, Inc.
[3] Nelson, M. 1996. Data compression with Burrows-Wheeler Transform.
Dr. Dobb's Journal.
[4] Campos, A. S. E. Run Length Encoding. Available:
http://www.arturocampos.com/ac_rle.html (last accessed July 2012).
[5] Campos, A. S. E. Move to Front. Available:
http://www.arturocampos.com/ac_mtf.html (last accessed July 2012).
[6] Campos, A. S. E. Basic arithmetic coding. Available:
http://www.arturocampos.com/ac_arithmetic.html (last accessed July
2012).
AUTHORS PROFILE
I Made Agus Dwi Suarjaya received his Bachelors
degree in Computer System and Information Science
in 2007 from Udayana University and Masters degree
in Information Technology in 2009 from Gadjah Mada
University. He served as a full-time lecturer at Faculty
of Engineering, Information Technology Department in Udayana University.
His research interest include software engineering, networking, security,
computing, artificial intelligent, operating system and multimedia.

0
25
50
75
100
Combination
63.86 63.36
48.45
40.99
40.71
0
25
50
75
100
Combination
80.88
95.43
78.89 78.07
77.06
Vol. 3, No.8, 2012
18 | P a g e
Monte Carlo Based Non-Linear Mixture Model of
Earth Observation Satellite Imagery Pixel Data
Kohei Arai
1

Graduate School of Science and Engineering
Saga University
Saga City, Japan

Abstract Monte Carlo based non-linear mixel (mixed pixel)
model of visible to near infrared radiometer of earth observation
satellite imagery is proposed. Through comparative studies with
actual real earth observation satellite imagery data between
conventional linear mixel model and the proposed non-linear
mixel model, it is found that the proposed mixel model represents
the pixels in concern much precisely rather than the conventional
linear mixel model.
Keywords- remote sensing satellite; visible to near infrared
radiometer; mixed pixel: mixel; Monte Carlo simulation model.
I. INTRODUCTION
The pixels in earth observed images which are acquired
with Visible to Near Infrared: VNIR sensors onboard remote
sensing satellites are, essentially mixed pixels (mixels) which
consists of several ground cover materials [1]. Some mixel
model is required for analysis such as un-mixing of the mixel
in concern [2],[3]. Typical mixel is linear mixing model which
is represented by linear combination of several ground cover
materials with mixing ratio for each material [4]. It is not
always true that the linear mixel model is appropriate [5]. Due
to the influences from multiple reflections between the
atmosphere and ground, multiple scattering in the atmosphere
on the observed radiance from the ground surface, pixel
mixture model is essentially non-linear rather than linear.
These influence is interpreted as adjacency effect [6].[7].
Method for representation of non-linear mixel model is not
so easy. In particular, there is not sophisticated multi reflection
model between ground materials. The representation method
for non-linear mixel model is based on Monte Carlo Ray
Tracing: MCRT model [8]. It is rather easy to designate
surface slopes on the ground and multi reflection among trees
for MCRT model. The proposed MCRT based non-linear
mixel model is applied to real earth observation satellite
imagery data of Advanced Spaceborn Thermal Emission and
Reflection Radiometer / Visible and Near Infrared
Radiometer: ASTER/VNIR onboard on Terra satellite. A
comparison of radiance between the conventional linear mixel
model and the proposed non-linear mixel model is conducted.
As a result, validity of the proposed model is confirmed.
The following section describes the proposed non-linear
mixel model based on MCRT followed by some experiments
for validation of the proposed model. Then, finally,
conclusions with some discussions are described.
II. PROPOSED NON-LINEAR MIXEL MODEL
A. Monte CarloRay Tracing Simulation
In order to show a validity of the proposed non-linear
mixel model, MCRT simulation study and field experimental
study is conducted. MCRT allows simulation of polarization
characteristics of sea surface with designated parameters of the
atmospheric conditions and sea surface and sea water
conditions. Illustrative view of MCRT is shown in Fig.1.

Figure 1 Illustrative view of MCRT for the atmosphere and sea water

Photon from the sun is input from the top of the
atmosphere (the top of the simulation cell). Travel length of
the photon is calculated with optical depth of the atmospheric
molecule and that of aerosol. There are two components in the
atmosphere; molecule and aerosol particles while three are
also two components, water and particles; suspended solid and
phytoplankton in the ocean. When the photon meets molecule
or aerosol (the meeting probability with molecule and aerosol
depends on their optical depth), then the photon scattered in
accordance with scattering properties of molecule and aerosol.
The scattering property is called as phase function
1
. In the
visible to near infrared wavelength region, the scattering by
molecule is followed by Rayleigh scattering law [10] while
that by aerosol is followed by Mie scattering law [10].
Example of phase function of Mie scattering is shown in Fig.2
(a) while that of Rayleigh scattering is shown in Fig.2 (b).

1
http://ejje.weblio.jp/content/phase+function
Vol. 3, No.8, 2012
19 | P a g e

(a)Mie scattering (aerosol)

(b)Rayleigh scattering (Molecule)
Figure 2 Phase functions for Mie and Rayleigh scattering
In the atmosphere, there are absorption due to water vapor,
ozone and aerosols together with scattering due to the
atmospheric molecules, aerosols. Atmospheric Optical Depth:
AOD (optical thickness) in total, Optical Depth: OD due to
water vapor (H
2
O), ozone (O
3
), molecules (MOL), aerosols
(AER), and real observed OD (OBS) are plotted in Fig.3 as an
example.
For simplifying the calculations of the atmospheric
influences, it is assumed that the atmosphere containing only
molecules and aerosols. As shown in Fig.3, this assumption is
not so bad. Thus the travel length of the photon at once, L is
expressed with equation (1).
L=L
0
RND(i) (1)
L
0
=Z
max
/ (2)
where Z
max,
, RND(i) are maximum length, altitude of the
atmosphere, optical depth, and i-th random number,
respectively. In this equation, is optical depth of molecule or
aerosol. The photon meets molecule when the random number
is greater than . Meanwhile, if the random number is less than
, then the photon meats aerosol. The photon is scattered at the
molecule or aerosol to the direction which is determined with
the aforementioned phase function and with the rest of the
travel length of the photon.

Figure 3 Example of observed atmospheric optical depth in total and the best fit
curves of optical depth due to water vapor, ozone, molecules, and aerosols
calculated with MODTRAN of atmospheric radiative transfer software code..
B. Ground Surface with Slopes
When the photon reaches on the ground, the photon
reflects at the ground surface to the direction which is
determined by random number. Lambertian surface [11] is
assumed. Therefore, reflectance is constant for all the
directions. The reflected photon travels with the rest of travel
length. Two adjacent slopes of Lambertian surfaces are
assumed on the ground as shown in Fig.4. Slope angles for
both are while their reflectance are
A
and
B

Figure 4 Two adjacent slopes of Lambertian surfaces which are assumed on the
ground
C. Top of the Atmosphere: TOA Radiance Calculation
If the photon reaches on the wall of the simulation cell, the
photon disappears at the wall and it appears from the
corresponding position on the opposite side wall. Then it
travels with the rest of travel length. Eventually, the photons
which are reached at the top of the atmosphere are gathered
with the Instantaneous Field of View: IFOV of the Visible to
Near Infrared Radiometer: VNIR onboard satellite. At sensor
radiance, I
+
with direction and IFOV of ,
0
can be calculated
with equation (3)
I
+
(,
0
)=I N
+
(,
0
)/N
total
(3)
where N
+
is the number of photons which are gathered by
VNIR, N
total
denotes the number of photons input to the
simulation cell. Also I denote extraterrestrial irradiance at the
top of the atmosphere.
Vol. 3, No.8, 2012
20 | P a g e
III. EXPERIMENTS
A. Validity of the Monte CarloRay Tracing Simulation
In order to confirm that the developed MCRT is valid, a
comparative study is conducted between radiative transfer
code of Gauss Seidel method and the MCRT derived TOA
radiance. Because the Gauss Seidel method allows calculation
of TOA radiance with flat surface of ground, 0.2 of reflectance
of flat surface is assumed in the comparison. Also, 0.02 and
0.03 of optical depths are assumed for aerosol and molecule.
The size of simulation cell is determined as 50 km by 50
km by 50 km. Solar zenith angle is set at 30 degree while solar
azimuth is set at 120 degree. 700,000 of photons are input to
the simulation cell. TOA radiance derived from the Gauss
Seidel method is 0.565 (mW/m
2
/sr/m) while that from the
MCRT is 0.579 (mW/m
2
/sr/m) at the 500nm of wavelength.
For both cases, IFOV of the VNIR radiometer is assumed to
be 2 , all of the photons output from the top of the
atmosphere are counted. Therefore, the developed MCRT
seems valid enough.
B. TOA Radiance for the Different Combination of Optical
Depths of Aerosol and Molecule and for the Ground with
the Different Slopes
TOA radiance at 500 nm of wavelength for the different
combination of optical depths of aerosol and molecule which
ranges from 0.01 to 0.04 and for the ground with the different
slopes, 0 and 20 degree are calculated.
Again, IFOV of the VNIR radiometer is assumed to be 2,
all of the photons output from the top of the atmosphere are
counted. The reflectance for both slopes are same as 0.5. The
results are shown in Table 1. In the table,
aer
,
mol
are optical
depths of aerosol and molecule, respectively.
TABLE I. TOP OF THE ATMOSPHERE: TOA RADIANCE FOR THE
COMBINATIONS OF ATMOSPHERIC CONDITIONS

TOA radiance (mW/m
2
/sr/m)
aer
\
mol

0.01 0.02 0.03 0.04
0.01 3.51 0.317 3.99 6.51
0.02 5.65 1.91 1.09 3.04
0.03 5.7 3.08 0.622 10.7
0.04 3.29 3.85 3.97 7.45

C. Validity of the Proposed Non-Linear Mixel Model with
Real VNIR Data
The proposed non-linear mixel model based on MCRT is
validated with real earth observation satellite imagery data of
ASTER/VNIR onboard Terra satellite which is acquired at
11:09 Japanese Standard Time: JST on December 15 2004.
IFOV of ASTER/VNIR is 15m with 60km of swath width.
Whole scene of ASTER/VNIR is shown in Fig.5 (a) while
Fig.5 (b) shows a portion of the scene.

(a)Whole scene of ASTER/VNIR image

(b)A portion of the scene
Figure 5 ASTER/VNIR image used for experiment
Three test sites, Area #1, 2, 3 are extracted from the scene.
Attribute information of these sites are listed in Table 2.
Fig.6 (a) shows three test sites on ASTER/VNIR image
while Fig.6 (b) shows three test sites on Google map. Other
than these, topographic map of three test sites which is
corresponding to the Google map is shown in Fig.6 (c) while
the extracted portion of each test site on ASTER/VNIR image
is shown in Fig.6 (d), (e) and (f), respectively. These digital
elevation models for three test sites are taken into account in
the MCRT simulations.

Vol. 3, No.8, 2012
21 | P a g e
Also, solar zenith angle of 58 degree and solar azimuth
angle of 17 degree are taken into account in the simulations.
From the atmospheric optical depth measurement data with
sun photometer, optical depth of total atmosphere is calculated.
TABLE II. ATTRIBUTIONS FOR THE TEST SITE WITH SLOPES
Area #1 Area #2 Area #3
Area Name
Korai-cho,
Ochiai-gawa
Korai-cho,
Ochiai-gawa
Konagai Golf
Club
Latitude 3257'30" 3256'33" 3256'13"
Longitude 1307'19" 1307'25" 13010'21"
Slope () 24 30 20
Slope () 28 26 0
A 0.14 0.2 0.14
Material Deciduous Bare Soil Deciduous
B 0.08 0.08 0.12
Material Coniferous Coniferous Paddy
OD-Aerosol 0.35 0.35 0.35
OD-Molecule 0.14 0.14 0.14

(a)Three test sites on ASTER/VNIR image

(b)Three test sites on Google Map

(c) Topographic map of corresponding area of three test sites on Google map

Coniferous (above), Deciduous (bottom)
(d)Area#1 (Korai-cho, Ochiai-gawa, Nagasaki, Japan)

Coniferous (above), Bare Soil (bottom)
(e)Area #2 (Korai-cho, Ochiai-gawa, Nagasaki, Japan)

Deciduous (above), Paddy field (bottom)
(f)Area #3 (Konagai Country Club, Nagasaki, Japan)
Figure 6 Three test site, Area #1, 2, 3.
Furthermore, molecule optical depth
R
is calculated with
equation (4) as a function of atmospheric pressure P which is
measured on the ground.
(4)
where P
0
denotes standard atmospheric pressure on the
ground (1013 hPa) while denotes wavelength. Then aerosol
optical depth is calculated from total atmospheric optical depth
by subtracting molecule optical depth.
Vol. 3, No.8, 2012
22 | P a g e
Comparative study is conducted between ASTER/VNIR
derived radiance of Band 2 (Green band) and the radiance
which derived from the conventional linear mixel model and
the proposed non-linear mixel model.
Table 3 shows the calculated radiance in unit of
W/m
2
/sr/m and the radiance difference between
ASTER/VNIR and the estimated with the conventional and the
proposed mixel models.
TABLE III. COMPARISON OF RADIANCE BETWEEN REAL ASTER/VNIR
AND THE CONVENTIONAL LINEAR MIXEL MODEL AS WELL AS THE PROPOSED
NON-LINEAR MIXEL MODEL DERIVED RADIANCE
Area #1 Area #2 Area #3
ASTER/VNIR 14.1 15.5 16
Linear 12.9 13.7 14.6
Non-Linear 13.7 14.3 15
VNIR-Linear 1.2 1.8 1.4
VNIR-Non-Linear 0.4 1.2 1

It is found that the estimated radiance with the proposed
non-linear mixel model is much closer rather than that with the
conventional linear mixel model.
IV. CONCLUSION
Monte Carlo based non-linear mixel (mixed pixel) model
of visible to near infrared radiometer of earth observation
satellite imagery is proposed. Through comparative studies
between ASTER/VNIR derived radiance and the conventional
linear mixel model derived radiance as well as the proposed
non-linear mixel model derived radiance, it is found that the
estimated radiance with the proposed non-linear mixel model
is much closer to ASTER/VNIR derived radiance (around 6%)
rather than that with the conventional linear mixel model.
One of the disadvantages of the proposed non-linear mixel
model based on MCRT is time consumable computations.
Acceleration is highly required.
ACKNOWLEDGMENT
The author would like to thank Dr. Yasunori Terayama and
Mr. Kohei Imaoka of Saga University for their effort to
simulation study and experiments.
REFERENCES
[1] Masao Matsumoto, Hiroki Fujiku, Kiyoshi Tsuchiya, Kohei Arai,
Category decomposition in the maximum likelihood classification,
Journal of Japan Society of Phtogrammetro and Remote Sensing, 30, 2,
25-34, 1991.
[2] Masao Moriyama, Yasunori Terayama, Kohei Arai, Clafficication
method based on the mixing ratio by means of category decomposition,
Journal of Remote Sensing Society of Japan, 13, 3, 23-32, 1993.
[3] Kohei Arai and H.Chen, Unmixing method for hyperspectral data based
on subspace method with learning process, Techninical Notes of the
Science and Engineering Faculty of Saga University,, 35, 1, 41-46, 2006.
[4] Kohei Arai and Y.Terayama, Label Relaxation Using a Linear Mixture
Model, International Journal of Remote Sensing, 13, 16, 3217-3227,
1992.
[5] Kohei Arai, Yasunori Terayama, Yoko Ueda, Masao Moriyama, Cloud
coverage ratio estimations within a pixel by means of category
decomposition, Journal of Japan Society of Phtogrammetro and Remote
Sensing, 31, 5, 4-10, 1992.
[6] Kohei Arai, Non-linear mixture model of mixed pixels in remote sensing
satellite images based on Monte Carlo simulation, Advances in Space
Research, 41, 11, 1715-1723, 2008.
[7] Kohei Arai, Kakei Chen, Category decomposition of hyper spectral data
analysis based on sub-space method with learning processes, Journal of
Japan Society of Phtogrammetro and Remote Sensing, 45, 5, 23-31,
2006.
[8] Kohei Arai, Adjacency effect of layered clouds estimated with Monte-
Carlo simulation, Advances in Space Research, Vol.29, No.19, 1807-
1812, 2002.
[9] Ramachandran, Justice, Abrams(Edt.),Kohei Arai et al., Land Remote
Sensing and Global Environmental Changes, Part-II, Sec.5: ASTER
VNIR and SWIR Radiometric Calibration and Atmospheric Correction,
83-116, Springer 2010.
[10] Kohei Arai, Lecture Note for Remote Sensing, Morikita Publishing Inc.,
(Scattering), 2004.
[11] Kohei Arai, Fundamental Theory for Remote Sensing, Gakujutsu-Tosho
Publishing Co., Ltd.,(Lambertian), 2001.
AUTHORS PROFILE
Kohei Arai, He received BS, MS and PhD degrees in 1972, 1974 and 1982,
respectively. He was with The Institute for Industrial Science, and Technology
of the University of Tokyo from 1974 to 1978 also was with National Space
Development Agency of Japan (current JAXA) from 1979 to 1990. During
from 1985 to 1987, he was with Canada Centre for Remote Sensing as a Post-
Doctoral Fellow of National Science and Engineering Research Council of
Canada. He was appointed professor at Department of Information Science,
Saga University in 1990. He was appointed councilor for the Aeronautics and
Space related to the Technology Committee of the Ministry of Science and
Technology during from 1998 to 2000. He was also appointed councilor of
Saga University from 2002 and 2003 followed by an executive councilor of
the Remote Sensing Society of Japan for 2003 to 2005. He is an adjunct
professor of University of Arizona, USA since 1998. He also was appointed
vice chairman of the Commission A of ICSU/COSPAR in 2008. He wrote
30 books and published 332 journal papers.

Vol. 3, No.8, 2012

23 | P a g e
A Modified Feistel Cipher Involving Substitution,
shifting of rows, mixing of columns, XOR operation
with a Key and Shuffling

V.U.K Sastry
Dean R&D, Department of Computer Science and
Engineering, Sreenidhi Institute of Science & Tech.
Hyderabad, India.
K. Anup Kumar
Associate Professor, Department of Computer
Science and Engineering, SNIST,
Hyderabad, India

Abstract In this paper, we have developed a modification to the
Feistel cipher by taking the plaintext in the form of a pair of
matrices and introducing a set of functions namely, substitute,
shifting of rows, mixing of columns and XOR operation with a
key. Further we have supplemented this process by using another
function called shuffling at the end of each round of the iteration
process. In this analysis, the cryptanalysis clearly indicates that
the strength of the cipher is quite significant and this is achieved
by the introduction of the aforementioned functions.
Keywords- encryption; decryption; cryptanalysis; avalanche effect;
XOR operation.
I. INTRODUCTION
The study of the Feistel cipher has been a fascinating
fundamental area in the development of block ciphers in
cryptography. In the recent years, we have offered several
modifications [1-4] to the classical Feistel cipher by taking the
plaintext in the form of a pair of matrices. In all these
investigations, we have made use of the multiplication with a
single key matrix or the multiplication with a pair of key
matrices as a fundamental tool in the development of the
cipher. This is associated with the mod operation. Further, we
have introduced some operations such as mixing, permutation,
blending or shuffling in order to achieve confusion and
diffusion, so that, the strength of the cipher becomes
significant.
In the present investigation, our objective is to study a
modification of the Feistel cipher, wherein we use the
fundamental operations such as substitution, shifting of rows,
mixing of columns, XOR operation and Shuffling. It may be
noted here that the operations, substitution, shifting of rows
and mixing of columns are very well utilized in Advanced
Encryption Standard (AES) [5]. Our interest here is to develop
a strong block cipher which exceeds, in strength, almost all the
other ciphers available in the literature.
In what follows we present the plan of the paper. In
section 2, we deal with the development of the cipher and
present the flowcharts and algorithms required in this analysis.
In section 3, we mention an illustration of the cipher and
describe the avalanche effect. We study the cryptanalysis in
section 4. Finally, in section 5, we discuss the computations
and draw conclusions.
II. DEVELOPMENT OF THE CIPHER
Consider a plaintext P containing 2m
2
characters. On using
the EBCIDIC code, the characters occurring in the plaintext
can be represented in terms of decimal numbers wherein each
number lies in [0 - 255]. Then, these numbers can be written in
the form of a pair of square matrices P
0
and Q
0
, wherein each
one is of size m.
Let us consider a key matrix K, where K is a square matrix
whose size is m.
The flowcharts depicting the encryption and the decryption
are given below.

Fig 1. The process of Encryption

Q
i-1
= Mix (Q
i-1
)
Q
i-1
= Shift (Q
i-1
) P
i-1
= Shift ( P
i-1
)
P
i-1
= Mix ( P
i-1
)
P
i-1
= P
i-1
K
C = P
r
Q
r

for i = 1 to r
Pi , Qi
P
i-1
= Sub ( P
i-1
) Q
i-1
= Sub ( Q
i-1
)
Pi-1 Qi-1
Read P
0
, Q
0
, K
Q
i-1
= Q
i-1
K
( P
i
, Q
i
) = Shuffle ( P
i-1
, Q
i-1
)
Vol. 3, No.8, 2012

24 | P a g e

Fig 2. The process of Decryption
Now we write the algorithms for the processes encryption
and decryption as given below.
A. Algorithm for Encryption

1. Read P, K
2. P
0
= Left half of P.
3. Q
0
= Right half of P.
4. for i = 1 to r
begin

P
i-1
= Sub (P
i-1
)

P
i-1
= Shift (P
i-1
)
P
i-1
= Mix (P
i-1
)
P
i-1
= P
i-1
K
Q
i-1
= Sub (Q
i-1
)
Q
i-1
= Shift (Q
i-1
)
Q
i-1
= Mix (Q
i-1
)
Q
i-1
= Q
i-1
K

(P
i
, Q
i
) = Shuffle ( P
i-1
, Q
i-1
)

end
5. C = P
r
Q
r
/* represents concatenation */

6. Write(C)
B. Algorithm for Decryption
1. Read C, K
2. P
r
= Left half of C.
3. Q
r
= Right half of C.
4. for i = r to 1
begin
(P
i-1
, Q
i-1
) = IShuffle ( P
i
, Q
i
)
P
i-1
= P
i-1
K
P
i-1
= IMix (P
i-1
)
P
i-1
= IShift (P
i-1
)
P
i-1
= ISub (P
i-1
)
Q
i-1
= Q
i-1
K

Q
i-1
= IMix (Q
i-1
)
Q
i-1
= IShift (Q
i-1
)

Q
i-1
= ISub (Q
i-1
)

end
5. P = P
0
Q
0
/* represents concatenation */

6. Write (P)
Let us now explain the basic ideas underlying in the
functions Sub ( ), Shift ( ), Mix ( ), used for substitution,
shifting of rows, mixing of columns respectively.
Firstly, Let us focus our attention on the substitution
process involved in the function Sub ( ).
Consider the EBCIDIC code which can be written in the
form a matrix given by
E (i, j) = 16*(i-1) + (j1), i = 1 to 16 and j = 1 to 16
(2.1)
All these numbers can be placed in the form of a table.
Let us arrange these numbers, which are lying in the
interval [0-255], in a random manner.
We represent these numbers in the hexadecimal notation.
All these numbers can be written in the form of a table given
below (table 2).
In the encryption process, when we come across a number
lying in [0-255], we will replace it by the corresponding
number in the substitution table. For example, if we come
across the number 70, in the process of encryption, this will be
converted into hexadecimal number as 46. Then, 70 will be
replaced by the number which is occurring in the 4
th
row, 6
th

column of the substitution table, i.e by 5A ( = 90 in decimal
notation). This is the process of substitution. Keeping the
EBCIDIC code matrix and the substitution table in view, we
form the inverse substitution table which is given in Table 2.
The inverse substitution table will be utilized while
carrying out the decryption process and it is denoted by
function ISub ( ).

Read Pr, Qr, , K
P
i-1
= IShift ( P
i-1
) Q
i-1
= IShift (Q
i-1
)
P
i-1
= IMix ( P
i-1
)
P = P
0
Q
0

for i = r to 1
Pi-1 , Qi-1
Q
i-1
= IMix ( Q
i-1
)
P
i-1
= P
i-1
K Q
i-1
= Q
i-1
K
( P
i-1
, Q
i-1
) = IShuffle ( P
i
, Q
i
)
Pi -1 Qi -1
Q
i-1
= ISub ( Q
i-1
) P
i-1
= ISub ( P
i-1
)
Vol. 3, No.8, 2012

25 | P a g e

0

1

2

3

4

5

6

7

8

9

A

B

C

D

E

F

0

63

7C

77

7B

F2

6B

6F

C5

30

01

67

2B

FE

D7

AB

76

1

CA

82

C9

7D

FA

59

47

F0

AD

D4

A2

AF

9C

A4

72

C0

2

B7

FD

93

26

36

3F

F7

CC

34

A5

E5

F1

71

D8

31

15

3

04

C7

23

C3

18

96

05

9A

07

12

80

E2

EB

27

B2

75

4

09

83

2C

1A

1B

6E

5A

A0

52

3B

D6

B3

29

E3

2F

84

5

53

D1

00

ED

20

FC

B1

5B

6A

CB

BE

39

4A

4C

58

CF

6

D0

EF

AA

FB

43

4D

33

85

45

F9

02

7F

50

3C

9F

A8

7

51

A3

40

AF

92

9D

38

F5

BC

B6

DA

21

10

FF

F3

D2

8

CD

0C

13

EC

5F

97

44

17

C4

A7

7E

3D

64

5D

19

73

9

60

81

4F

DC

22

2A

90

88

46

EE

B8

14

DE

5E

0B

DB

A

E0

32

3A

0A

49

06

24

5C

C2

D3

AC

62

91

95

E4

79

B

E7

C8

37

6D

8D

D5

4E

A9

6C

56

F4

6A

65

7A

AE

08

C

BA

78

25

2E

1C

A6

B4

C6

E8

DD

74

1F

4B

BD

8B

8A

D

70

3E

B5

66

48

03

F6

0E

61

35

57

B9

86

C1

1D

9E

E

E1

F8

98

11

69

D9

8E

94

9B

1E

87

E9

CE

55

28

DF

F

8C

A1

89

0D

BF

E6

42

68

41

99

2D

0F

B0

54

BB

16

Table 1. Substitution Table

0

1

2

3

4

5

6

7

8

9

A

B

C

D

E

F

0

52

09

6A

D5

30

36

A5

38

BF

40

A3

9E

81

F3

D7

FB

1

7C

E3

39

82

9B

2F

FF

87

34

8E

43

44

C4

DE

F9

CB

2

54

7B

94

32

A6

C2

23

3D

EE

4C

95

0B

42

FA

C3

4E

3

08

2E

A1

66

28

D9

24

B2

76

5B

A2

49

6D

8B

D1

25

4

72

F8

F6

64

86

68

98

16

D4

A4

5C

CC

5D

65

B6

92

5

6C

70

48

50

FD

ED

B9

DA

5E

15

46

57

A7

8D

9D

84

6

90

D8

AB

00

8C

BC

D3

0A

F7

E4

58

05

B8

B3

45

06

7

D0

2C

1E

8F

CA

3F

0F

02

C1

AF

BD

03

01

13

8A

6B

8

3A

91

11

41

4F

67

DC

EA

97

F2

CF

CE

F0

B4

E6

73

9

96

AC

74

22

E7

AD

35

85

E2

F9

37

E8

1C

75

DF

6E

A

47

F1

1A

71

1D

29

E5

89

6F

B7

62

0E

AA

18

BE

1B

B

FC

56

3E

4B

C6

D2

79

20

9A

DB

C0

FE

78

CD

5A

F4

C

1F

DD

A8

33

88

07

C7

31

B1

12

10

59

27

80

EC

5F

D

60

51

F7

A9

19

B5

4A

0D

2D

E5

7A

9F

93

C9

9C

EF

E

A0

E0

3B

4D

AE

2A

F5

B0

C8

EB

BB

3C

83

53

99

61

F

17

2B

04

7E

BA

77

D6

26

E1

69

14

63

55

21

0C

7D

Table 2. Inverse Substitution Table

Vol. 3, No.8, 2012

26 | P a g e
Now let us see the process of shifting involved in the
function Shift ( ), during the encryption process we come
across plaintext P
i-1
and Q
i-1
in the process of iteration. As P
i-1

is a square matrix of size m, it can be written in the form

p
11
p
12
p
13
p
1m

p
21
p
22
p
23
p
2m

p
31
p
32
p
33
p
3m

p
m1
p
m2
p
m3
p
mm

On converting each decimal number in (2.4) into its binary
form, we get

p111p112...p118 p121p122.p128 p1m1p1m2p1m8

p111p112...p118 p121p122.p128 p1m1p1m2p1m8

p111p112...p118 p121p122.p128 p1m1p1m2p1m8

p111p112...p118 p121p122.p128 p1m1p1m2p1m8

Here each row contains 8m binary bits. In the process of
shifting, we offer a right shift of 4 bits in the first row, 12 bits
in the second row, 20 bits in the third row and in general,
4 + 8 * (i-1) bits right shift in the i
th
row.
This process is carried out till we exhaust all the rows. It
may be noted here that IShift ( ) denotes the reverse process of
Shift ( ).
In this, the binary bits are obviously given a left shift in an
appropriate manner.
To have a clear insight into the mixing process denoted by
the function Mix ( ), let us consider again the matrix P
i-1
,
which is represented in the form (2.5).
Let us restrict our attention only to a plaintext matrix,
wherein, m=4. This can be written in the form given below

p
111
p
112
p
118
p
121
p
122
p
128
p
141
p
142.
p
148

p
211
p
212
p
218
p
221
p
222
p
228
p
241
p
242.
p
248

p
311
p
312
p
318
p
321
p
322
p
328
p
341
p
342.
p
348

p
411
p
412...
p
418
p
421
p
422...
p
428
p
441
p
442.
p
448

This has 4 rows and 32 columns. On concatenating the
binary bits of the 1
st
column and the 17
th
column we get a
string of binary bits, which can be converted into a decimal
number. This can be considered as new p
11
.
On considering the binary bits of the 2
nd
column and the
18
th
column and concatenating them, we get another decimal
number which will be called as p
12
.
On adopting the same process till we exhaust all the
columns taken in pairs, we get the decimal numbers which
correspond to the other elements of the matrix written in the
row wise order. Thus we have, the new plaintext matrix,
which is obtained after the completion of mixing. Imix ( ) is
the reverse process of Mix ( ).
For a detailed discussion of the function shuffle ( ),
wherein we are shuffling the columns of two matrices, we
refer to [4].
III. ILLUSTRATION OF THE CIPHER
Consider the plaintext given below
My dear young lady! We both are well qualified. You have
done your B.Tech and I have completed my M.S, where is the
problem! We can fly anywhere. Why your father and mother
are not accepting our marriage. We both belong to the same
cast, we both are farmers. What is the objection of your father
and your mother, are they having any thinking regarding my
financial status? We are having as much landed property as
your father is having. My father and your father both are well
trained seasonal politicians. I wonder why your father is not
accepting and why your mother is not accepting. Our marriage
must happen soon. Yours loving Mr.X (3.1)
Let us focus our attention on the first 32 characters of the
plaintext. This is given by
My dear young lady! We both are
On using EBCIDIC code, we get the plaintext matrix P in
the form

77 121 32 100 101 97 114 32

121 111 117 110 103 32 108 97

100 121 33 32 87 101 32 98

111 116 104 32 97 114 101 32

This can be written in the form of a pair of matrices given
by

77 121 32 100

121 111 117 110

100 121 33 32

111 116 104 32

(2.6)
(3.2)
P =
(3.3)
P
0
=
(2.4)
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
(2.5)
Vol. 3, No.8, 2012

27 | P a g e
and

101 97 114 32

103 32 108 97

87 101 32 98

97 114 101 32

Let us take the key matrix K in the form

45 128 192 53

133 200 150 16

100 150 33 120

13 189 164 55

On applying the encryption algorithm, given in section 2,
we get the ciphertext C in the form

51 145 164 146 108 237 147 173

155 18 82 72 85 155 19 71

182 102 90 237 150 142 218 60

11 150 219 226 237 177 36 100

On using the decryption algorithm on (3.6), we get back
the original plaintext P given by (3.2).
Let us now study the avalanche effect which throws some
light on the strength of the cipher.
On changing the first row, first column element of P
0
, from
77 to 76, we get a 1 bit change in the plaintext. On applying
the encryption algorithm on the modified plaintext, keeping up
the key as it is, we get the ciphertext C in the form

218 88 129 219 201 58 54 101

157 209 7 186 109 153 44 75

219 120 243 158 95 55 38 117

43 233 147 229 81 38 133 187

On comparing (3.6) and (3.7), after converting them into
their binary form, we notice that they differ by 128 bits out of
256 bits. This indicates that the cipher is quite good from the
view point of its strength.
Let us now consider a one bit change in the key. This is
achieved by changing first row, first column element of the
key K, given by (3.5), from 45 to 44.
Now on using the modified key and applying the
encryption algorithm, keeping the plaintext as it is, we get the
cipher text C in the form

79 149 68 154 22 239 105 98

232 131 221 63 57 229 243 114

103 82 190 152 14 222 73 209

179 44 237 153 44 75 219 120

Now on comparing (3.6) and (3.8), after converting both
into their binary form, we find that these two ciphertexts differ
by 134 bits out of 256 bits.
This also shows that, the strength of the cipher is expected
to be significant.
IV. CRYPTANALYSIS
In cryptography, determination of the strength of the
cipher is a very important aspect. In the literature of
cryptography, it is well known that the cryptanalysis can be
carried out by the following approaches.
1. Ciphertext only attack ( Brute force attack )
2. Known plaintext attack
3. Chosen plaintext attack
4. Chosen ciphertext attack
As William Stallings [6] has pointed out that every cipher
must be designed so that it withstands the first two attacks at
least.
Let us now consider the brute force attack.
Here the key is containing m
2
decimal numbers. Thus the
size of the key space
8m
2

= 2 .
Let us suppose that, the time required for the computation
of the cipher with one value of the key is 10
-7
seconds. Then
the time required for processing the cipher with all the
possible values of the key in the key space is
8m
2
(2.4) m
2
-7 (2.4)m
2
-15

(2)

x 10
-7
= 10 = 3.12 x 10 years
365x24x60x60 365x24x60x60

This time is very large when m is greater than or equal to
3.
In our example as we have taken m=4, the attack on this
cipher, by the brute force approach, is totally ruled out.
Let us now investigate the known plaintext attack. In this
case, we know as many plaintext and ciphertext pairs as we
require, making an attempt for breaking the cipher. In the light
of the above information, we have as many pairs of P and C as
we require.
(3.4)
Q
0
=
(3.5) K

=
(3.6)
C =
.(3.7)
C =
(3.8)
C =
Vol. 3, No.8, 2012

28 | P a g e
If we take r=1, that is, if we confine our attention to a
single round of the iteration process, then we have the
relations connecting C and P as follows:
P
0
= Sub (P
0
) (4.1)

P
0
= Shift (P
0
) (4.2)

P
0
= Mix (P
0
) (4.3)

P
0
= P
0
K (4.4)

Q
0
= Sub (Q
0
) (4.5)
Q
0
= Shift (Q
0
) (4.6)
Q
0
= Mix (Q
0
) (4.7)

Q
0
= Q
0
K (4.8)

(P
1
, Q
1
) = Shuffle ( P
0
, Q
0
) (4.9)

C = P
1
|| Q
1
(4.10)

In the known plaintext attack, we know P
0
and Q
0

corresponding to the initial stage. We also know the C
obtained at the end.
As C is known to us, we can determine P
1
and Q
1
from
(4.10)
On using the IShuffle ( ), on (4.9), we get the current P
0

and Q
0
which are occurring on the left hand side of (4.4) and
(4.8). On using initial the P
0
and the Sub ( ), we get P
0
on the
left hand side of (4.1). After that, on using shift ( ) on the
available P
0
, we get P
0
occurring on the Left hand side of
(4.2). Then on using the function Mix( ) on the current P
0
, we
have the P
0
occurring on the left side of (4.3). Thus, we can
readily determine the key K from (4.4). Hence this cipher can
be broken by the known plaintext attack if we confine only to
one step in the iteration process.
Let us now study the cipher when r = 2. Then the equations
governing the cipher are (4.1) to (4.10) and the following
P
1
= Sub (P
1
) (4.11)

P
1
= Shift (P
1
) (4.12)

P
1
= Mix (P
1
) (4.13)

P
1
= P
1
K (4.14)

Q
1
= Sub (Q
1
) (4.15)
Q
1
= Shift (Q
1
) (4.16)
Q
1
= Mix (Q
1
) (4.17)

Q
1
= Q
1
K (4.18)

(P
2
, Q
2
) = Shuffle ( P
1
, Q
1
) (4.19)

C = P
2
|| Q
2
(4.20)

In the known plaintext attack, we know C, obtained at the
end of the iteration process, and the corresponding P
0
and Q
0
,
which are available at the very beginning of the iteration
process.
As we know C, we can determine P
2
and Q
2
from (4.20).
On using IShuffle on (4.19), we get P
1
and Q
1
which are
occurring on the left side of (4.14) and (4.18). We cannot
determine K as we do not know the P
1
and Q
1
occurring in the
right hand side of (4.14) and (4.18). Here, we notice that,
though P
0
and Q
0
are known to us, we cannot determine the P
1

and Q
1
which are occurring on the right hand side of (4.14)
and (4.18), by starting at the beginning as the key K is
occurring in (4.4) and (4.8). In the light of these facts, this
cipher cannot be broken by the known plaintext attack, when
we have confined to r=2. This shows that it is impossible to
break the cipher by the known plaintext attack when we carry
out all the sixteen rounds in the iteration.

Intuitively choosing a plaintext or ciphertext and
determining the key or a function of the key is a formidable
task in the case of this cipher.
From the above discussion we conclude that this cipher is
not breakable by all the possible attacks that are available in
cryptography.
V. COMPUTATIONS AND CONCLUSIONS

In this investigation, we have offered a through
modification in the Feistel cipher by taking the plaintext in the
form of a pair of matrices, and by applying several procedures,
namely, substitution, shifting, mixing, XORing with the key
and shuffle operation. Each one of these procedures modifies
the plaintext in a through manner and creates confusion and
diffusion in the development of the cipher. The iteration
process, which is the basic one in this cipher, supports all the
above procedures in a strong way.
Here it may be noted that the substitution table generated
in a random manner by using the numbers [0-255] is to be sent
to the receiver by the sender.
The programs for encryption and decryption are written in
C language.
The plaintext given in (3.1) is divided into 20 blocks,
wherein each block is containing 32 characters. We have
appended in the last block by adding 13 blank characters, so
that it becomes a complete block. On applying the encryption
algorithm given in section 2 we get the cipher text
corresponding to the entire plaintext (excluding the first block
for which the cipher text is already given in (3.6) ), in the form

212 111 166 213 179 183 219 102 51 84 223 38 165 45 198 253
244 153 37 69 150 119 82 206 223 122 100 147 82 145 190 142
122 45 157 190 115 140 161 154 229 63 77 179 44 237 243 158
140 154 148 153 53 41 110 76 146 115 202 111 223 77 50 100
147 158 94 147 126 250 105 153 103 121 34 63 71 62 155 102
51 93 211 211 35 125 54 173 157 186 100 149 22 94 115 140
161 154 229 63 77 179 44 237 243 158 140 154 148 153 53 41
110 55 38 73 81 237 201 146 84 89 103 121 34 63 71 62
155 102 51 93 211 211 14 113 148 51 92 228 201 42 61 185
79 211 108 203 59 124 231 142 242 68 126 142 140 154 148 153
53 41 110 76 146 115 218 100 201 39 60 189 38 253 244 211
50 206 242 68 126 142 125 54 204 102 187 167 166 70 250 109
91 59 124 231 25 67 53 202 126 155 102 89 219 231 61 25
53 41 14 113 148 51 92 228 201 42 61 185 79 211 108 203
59 124 231 142 242 68 126 142 140 154 148 153 53 41 110 76
146 115 218 100 201 39 60 189 38 253 244 211 50 206 242 68
126 142 125 54 204 102 187 167 166 91 81 190 155 86 206 223
217 140 219 103 172 102 143 209 207 108 198 109 70 250 109 91
59 125 182 99 53 77 242 106 82 220 111 223 73 146 84 89
103 117 44 237 247 166 73 53 41 27 232 231 162 217 219 231
56 202 25 174 83 244 219 50 206 223 57 232 201 169 73 147
82 150 228 201 39 60 166 253 244 211 38 73 57 229 233 55
239 166 153 150 119 146 35 244 115 233 182 99 53 221 61 50
55 211 106 217 219 166 73 81 101 231 56 202 25 174 83 244
219 50 206 223 57 232 201 169 73 147 82 150 227 114 100 149
30 220 153 37 69 150 119 146 37 221 61 48 231 25 67 53
206 76 146 163 219 148 253 54 204 179 183 206 120 239 36 71
232 232 201 169 73 147 82 150 228 201 39 61 166 76 146 115
203 210 111 223 77 51 44 239 36 71 232 231 211 108 198 107
186 122 100 111 166 213 179 183 206 113 148 51 92 167 233 182
Vol. 3, No.8, 2012

29 | P a g e
101 157 190 115 209 147 82 144 231 25 67 53 206 76 146 163
219 148 253 54 204 179 183 206 120 239 36 71 232 232 201 169
73 147 82 150 228 201 39 61 166 76 146 115 203 210 111 223
77 51 44 239 36 86 233 233 150 212 111 166 213 179 183 206
203 206 113 148 51 92 167 233 182 101 157 190 115 209 147 82
147 38 165 45 198 228 201 42 61 185 50 74 139 44 239 36
218 182 118 250 198 104 253 26 93 211 211 14 113 148 51 92
228 201 42 61 185 79 211 108 203 59 124 231 142 242 68 126

The cryptanalysis, carried out in this investigation, clearly
shows that this cipher is a strong one. This has become a very
good cipher as we have taken the length of the plaintext as
large as possible (2048 bits), and supported the encryption
process with a good number of functions so that the plaintext
undergoes a through transformation ( in each round of the
iteration process) before it becomes the ciphertext. In this
analysis, the substitution table generated in the random
manner plays a very important role.
REFERENCES

[1] V.U.K Sastry and K. Anup Kumar, A Modified Feistel Cipher
involving a key as a multiplicant on both the sides of the Plaintext matrix
and supplemented with Mixing Permutation and XOR Operation,
International Journal of Computer Technology and Applications ISSN:
2229-6093. Vol. 3, No.1, pp. 23-31, 2012.
Involving a Key as a Multiplicant on Both the Sides of the Plaintext
Matrix and Supplemented with Mixing, Permutation, and Modular
Arithmetic Addition, International Journal of Computer Technology
and Applications ISSN: 2229-6093. Vol. 3, No.1, pp. 32-39, 2012.
Involving a Pair of Key Matrices, Supplemented with XOR Operation,
and Blending of the Plaintext in each Round of the Iteration Process,
International Journal of Computer Science and Information
Technologies ISSN: 0975-9646. Vol. 3, No.1, pp. 3133-3141, 2012.
involving a pair of key matrices, Supplemented with Modular Arithmetic
Addition and Shuffling of the plaintext in each round of the iteration
process, International Journal of Computer Science and Information
Technologies ISSN: 0975-9646. Vol. 3, No.1, pp. 3119-3128, 2012.
[5] Daemen J, and Rijmen V, Rijndael, the Advanced Encryption Standard
(AES), Dr. Dobbs Journal, Vol. 26(3), pp. 137 -139, Mar 2001.
[6] William Stallings, Cryptography and Network Security, Principles and
Practice, Third Edition, Pearson, 2003.
AUTHORS PROFILE
Dr. V. U. K. Sastry is presently working as Professor
in the Dept. of Computer Science and Engineering
(CSE), Director (SCSI), Dean (R & D), SreeNidhi
Institute of Science and Technology (SNIST),
Hyderabad, India. He was Formerly Professor in IIT,
Kharagpur, India and
Worked in IIT, Kharagpur during 963 1998. He
guided 12 PhDs, and published more than 40 research
papers in various international journals. His research interests are Network
Security & Cryptography, Image Processing, Data Mining and Genetic
Algorithms.

Mr. K. Anup Kumar is presently working as an
Associate Professor in the Department of Computer
Science and Engineering, SNIST, Hyderabad India. He
obtained his B.Tech (CSE) degree from JNTU
Hyderabad and his M.Tech (CSE) from Osmania
University, Hyderabad. He is now pursuing his PhD
from JNTU, Hyderabad, India, under the supervision
of Dr. V.U.K. Sastry in the area of Information Security and
Cryptography. He has 10 years of teaching experience and his interest in
research area includes Cryptography,Steganography and Parallel Processing
Systems.

Vol. 3, No.8, 2012

30 | P a g e
Automatic Association of Strahlers Order and
Attributes with the Drainage System

Mohan P. Pradhan
Department of CSE,
SMIT
Sikkim India
M. K. Ghose
Department of CSE,
SMIT
Sikkim India
Yash R. Kharka
Department of CSE,
SMIT
Sikkim India

Abstract A typical drainage pattern is an arrangement of river
segment in a drainage basin and has several contributing
identifiable features such as leaf segments, intermediate segments
and bifurcations. In studies related to morphological assessment
of drainage pattern for estimating channel capacity, length,
bifurcation ratio and contribution of segments to the main
stream, association of order with the identified segment and
creation of attribute repository plays a pivotal role. Strahlers
(1952) proposed an ordering technique that categories the
identified stream segments into different classes based on their
significance and contribution to the drainage pattern. This work
aims at implementation of procedures that efficiently associates
order with the identified segments and creates a repository that
stores the attributes and estimates of different segments
automatically. Implementation of such techniques not only
reduces both time and effort as compared to that of manual
procedures, it also improves the confidence and reliability of the
results.
Keywords- Stream; digitization; Strahlers order.
I. INTRODUCTION
A drainage pattern pertaining to a terrain is a mesh of
interconnected streams. This mesh may be of different types
such as Dendritic, Trellied or Lattice, Radial or Concentric,
Parallel, Rectangular, Deranged, Centripetal and Violent. The
formation of mesh depends upon the morphological aspect of
the terrain the drainage system is subjected to such as slope,
varied resistance of rocks and its geological and
geomorphological past[1]. In studies related to drainage system
and its effects on the terrain demands classification of the
system into identifiable classes or identifiable orders.
There are different systems for ordering drainage pattern
designed by Horton (1945), Strahler (1952), Scheideggar
(1965) and Shreve (1967) for associating order with stream
segment in drainage segment.[1]
The order of a stream segment depends upon the order of its
tributaries. The point where tributaries meet is called as a
junction. In studies related to drainage system streams are also
referred to as links.
Links are typically of two types namely internal and
external. Links are classified based on whether they do or do
not have tributaries. Link that stretches from source to a
junction is referred to as external links whereas link that
stretches from on junction to another is referred to as internal
links. Each of these identified streams have their own order,
length, channel capacity and bifurcation ratio.
In order to associate order with the stream segment in a
drainage pattern either traditional tedious manual technique can
be used or a process can be designed based on certain criteria
or knowledge of any ordering techniques.
Traditional techniques for associating order with segment
involve manual digitization, manual association of order and
attributes etc. This demands greater effort, time and cost
investment. Design of automated procedure for the same would
greatly reduce investments; in addition it enhances quality and
reliability of the results.
This proposed work automatically extracts segments from a
drainage pattern, associates order and also estimates various
attributes related to the same. This work relies on the concept
of Strahlers ordering technique for performing qualitative and
quantitative assessment of the drainage system. In order to
minimize the time and space complexity of identification and
storing attributes, two important procedures are used.
On analyzing the drainage pattern it was observed that the
contributing tributaries always converges inwards to form a
segment of higher order and these converging streams are
always located towards the bound. So, in order to identify these
streams traditional row column traversal technique proves
ineffective and would consume more time.
To minimize the time required for determining segments an
efficient and effective spiral navigation technique is used [2].
In order to store the indentified segments of variable length an
efficient 2D data structure (jazzed array) is used to store the
detail that optimally utilizes memory space. The proposed
schema is shown in figure 1 below.
Vol. 3, No.8, 2012

31 | P a g e

Figure 1. Schema for stream order & attribute association.
II. STREAM ORDERING
In Strahlers (1952) system [3]
if a stream has no contributing tributaries, then order of the
stream is 1
else if the stream has more than on tributaries, and their
orders are i and j then
o if i==j then the order of the resulting stream will be i+1
or j+1
o else if i<j then the order of the resulting stream will be j
o else if i>j then the order of the resulting stream will be i

Figure 2. Strahlers Ordering Scheme
Two streams with same order i unite to give a stream of
order i+1 and if the streams of different order unite the new
stream retains the order of the highest order stream.
III. RELATED WORK
Stream ordering has wide range of application in
hydrological studies for e.g. run-off modeling, fisheries etc.
Association of stream ordering is tedious and time consuming
task, thus method for determining stream order automatically
for various drainage network topologies plays a crucial role in
various studies related to drainage pattern.
Andy Ward et al [4] in their study related to stream
classification has expressed that Stream order plays significant
role in determining the expected ecological function of
drainage system. Stream order is extensively used in River
Continuum Concept RCC which is used for classifying,
describing flowing water and classification of sections of
waters based on presence of indicator organisms and Flood
Pulse Concept (FPC), a concept that describes the interaction
between water and land or hydrological conditions takes into
basis the orientation of the stream in a drainage pattern for
analysis.
Dawson et al [5] have developed an automated computer
based extraction procedure on GIS system which helps in
classification of rivers in Britain.
Many researchers have used terrain analysis of a DEM for
extracting drainage pattern and associating attributes with each
stream using Strahlers ordering technique. P. Venkatachalam
et al. [5] have suggested a procedure for delineating drainage
and watersheds from DEM of a terrain. David et al. [7] have
used DEM as source for automatic delineation of flow path,
sub-watershed and flow networks for hydrographic modeling.
Storey and Wadhwa [8] have also estimated the length of each
stream using terrain analysis of DEM based on LiDAR survey
to produce stream channel maps. Ejstrud [9] have also
constructed digital dataset of streams, lakes and wetlands using
DEMS.
Alper Sen et al. [10] in their study related to drainage
pattern used k-means clustering method for grouping rivers into
different categories for creating reduced scale map for a
drainage system taking into consideration various river
attributes.
In spite of wide hydrological application, very few works
has been carried for automatic digitization and association of
order.
IV. METHODOLOGY USED
A typical stream segment is represented by (Start, End,
Length, Order and Bifurcation Ratio). Start represents starting
coordinates of a stream segment. End represents ending
coordinates of a stream segment. Length represents Length of
the stream. Order represents the order of the segment.
Bifurcation ratio is the number of streams of order u divided by
number of streams of order (u-1).
A. Data Traversal Procedure
In order to decrease the total amount of time required for
determining the peripheral stream this work uses an efficient
traversal scheme based on spiral navigation [1] rather than the
traditional row column approach for navigation.
The peripheral streams that converge to form a main stream
are often oriented around the main stream, so to identify these
contributing stream spiral traversal proves efficient. The
traversal process in case of spiral traversal may end at any step
between 0 to n
2
against n
2
in case of traditional row column
approach.
This traversal scheme exhaust the values in the data set in
spiral manner either in a clockwise direction or an anti-clock
wise direction, in order to ensure that the procedure terminates
at a single point or coordinate the dataset has to be odd order
so, in case if the dimension is not odd then row or column of
non-significant values are added.
Vol. 3, No.8, 2012

32 | P a g e
B. Procedure for Determination of segment initiation point
The implemented process takes in skeletonized binary
image (0 (non-significant) or 1(significant)) of the drainage
system for performing the classification process.
While traversing the data set in a spiral manner two
possible types of values might be encountered either 0 or 1.
On encountering 0 (insignificant value) the spiral traversal
continues.
On encountering 1 (significant value) the spiral traversal
temporarily stop and either of the two actions are
performed.
o If the encountered 1 is in previously visited
coordinate, then spiral traversal continues.
o If the encountered 1 is an unvisited
coordinate, then the spiral navigation is
temporarily put to halt and segment
navigation procedure is initiated.
On completing segment navigation procedure the spiral
traversal process resumes.
The process is repeated until all external links are exhausted.

C. Segment Navigation Procedure
This procedure aims at determining various segments, their
starting coordinate, intermediate coordinates and terminating
coordinate. It also determines length of the identified segment
and associates order for an identified stream.
Identification of leaf segments
On encountering a significant unvisited value during spiral
navigation, the significant value is traversed until a junction is
not encountered. On encountering a junction the segment
navigation process stops and the navigation status is saved in a
data structure.
The traversal status includes information such as starting
coordinate, intermediate coordinates and terminating
coordinates along with the length of the segment.
In order to prevent misinterpretation of traversed
coordinates a status variable is maintained of equal dimensions
as the actual data. As and when a coordinate value is visited the
status of that coordinate is changed to visited (1 for visited and
0 for not visited).
The dataset is traversed until and unless all the leaf
segments are determined. The traversed leaf segments have
variable length, so the process demanded an efficient data
management scheme that utilizes the memory efficiently, so in
this work a jazzed data structure is implemented in order to
store the segment information.
The individual rows in the jazzed data structure represent a
segment and store the coordinate points for the segment.
Additional information such as segment coordinate count is
also maintained for determining length. Every leaf stream
encountered is assigned a Strahlers order 1.

Figure 3. Spiral Navigation Scheme
Identification of non leaf segments
Upon completing identification of leaf segments, the
process for identifying non leaf segment is initiated. For
determining the non leaf segment, spiral traversal is not used
rather in order to reduce the complexity, the initially created
jazzed structure is taken as input. The entries in the jazzed
structure are scanned for determining junction. Junction is
determined by determining entries that terminate at same
coordinate. All such junctions are identified. For each
identified junction segment navigation is done in order to
identify second order streams and their status is saved in the
data structure.
In order to determine i
th
order streams, all entries related to
(i-1)
th
order streams in the jazzed structure are taken into
consideration.
This process terminates when there are no junctions left to
be extended.
Procedure for Strahlers ordering
If in case
i. the terminating coordinates in the jazzed structure are
same and their orders are same (say i) then the order of
the segment extending from their intersection will be
(i+1)
ii. the terminating coordinates in the jazzed structure are
same and their orders are different then the order of the
segment will be the highest of the orders
This process is repeated for all orders until and unless all
the streams are not exhausted.

Figure 4. Identification of 1st Order Streams using spiral traversal
Vol. 3, No.8, 2012

33 | P a g e

Figure 5. Identification of 2nd Order Stream using junction extension

Figure 6. Identification of 3nd Order Stream using junction extension
V. RESULTS AND DISCUSSION
The skeletonized images of the drainage pattern were taken
into consideration for evaluation the performance of the
techniques.
The skeletonized image was processed in order to
determine the number of segments, coordinate points, length,
and order and bifurcation ratio. The input and the output image
along with the various attributes are represented below in table
1.
VI. CONCLUSION
This works aim at developing an automatic procedure for
digitizing and vectoring drainage pattern. Implementation of
such process tremendously reduces the amount of effort and
time required in order to digitize and classify segments in a
drainage pattern. In addition to classification it also associates
attributes with the identified segments of drainage pattern.
REFERENCES
[1] Garde R J, River Morphology, 2nd edition, New Age International
Publishers Ltd.-New Delhi, pp 14-15.
[2] Mohan P. Pradhan, M.K. Ghose, Automatic Association of Stream
Order for Vector Hydrograph Using Spiral Traversal Technique, IOSR
Journal of Computer Enginering, Volume 1, Issue 5, pp 09-12, 2012.
[3] Strahlers A. N., Quantitative Analysis of Watershed Geomorphology,
American Geophysical Union Transactions, Volume 38, Pages 912-920,
1957.
[4] Andy Ward, Jessica L. DAmbrosio, Dan Mecklenburg, Stream
Classification, Environmental hydrology, CRC Press.
[5] F.H. Dawson, D.D. Hornby, J. Hilton, A method for the automated
extraction of environmental variables to help the classification of rivers
in Britain, Special Issue: Sustainable River Basin Management in the
UK: Needs and Opportunities, Volume 12, Issue 4, pages 391403,
July/August 2002.
[6] P. Venkatachalam et al., Automatic delineation of watersheds for
Hydrological Applications, Proc. of 22
nd
Asian Conference on Remote
Sensing, 2011.
[7] David Tarboton, Terrain Analysis Using Digital Elevation Models in
Hydrology, 23rd ESRI International Users Conference, 2003, San
Diego, California.
[8] Storey, R., Wadhwa, S. , An Assessment of the Lengths of Permanent,
Intermittent and Ephemeral Streams in the Auckland Region, Prepared
by NIWA for Auckland Regional Council. Auckland Regional Council
Technical Report 2009/028, 2009.
[9] Bo Ejstrud, Reconstructing drainage networks, [available online:
web.sdu.dk/ejstrud/forskning/GIS/Arbejdspapir_Streams_Lakes.pdf ]
[10] Alper Sen, Turkay Gokgoz, Clustering Approaches for Hydrographic
Generalization, GIS Ostrava 2012 - Surface Models for Geosciences,
2012.
AUTHORS PROFILE
Mohan P Pradhan, He received B. Tech, M. Tech degrees in
2006 and 2009 respectively from Sikkim Manipal University,
Sikkim, India. He is currently persuing his Ph. D for Sikkim
Manipal University in the field of automatic digitization for
GIS applications. He is currently working as Assistant
Professor in department of Computer Science and Engineering,
Sikkim Manipal Institute of Technology, Sikkim Manipal University since
August 2006. His research interest includes RS and GIS, Algorithms, Image
Processing and Software Engineering.
Professor (Dr.) M. K. Ghose is the Dean (R & D), SMIT and
Professor & Head of the Department of Computer Science &
Engineering at Sikkim Manipal Institute of Technology,
Mazitar, Sikkim, India since June, 2006. Prior to this, Dr.
Ghose worked in the internationally reputed R & D
organization ISRO during 1981 to 1994 at Vikram Sarabhai
Space Centre, ISRO, Trivandrum in the areas of Mission
Simulation and Quality & Reliability Analysis of ISRO Launch vehicles and
Satellite systems and during 1995 to 2006 at Regional Remote Sensing
Service Centre, ISRO, IIT Campus, Kharagpur in the areas of RS & GIS
techniques for the natural resources management. He was also associated with
Regional Engg. College (NIT), Silchar (1979 1981) as Teaching Asst. and
Assam Central University, Silchar as COE and HOD of Computer Science
Department (1997-2000).
His areas of research interest are Data Mining, Simulation & Modelling,
Network, Sensor Network, Information Security, Optimization & Genetic
Algorithm, Digital Image processing, Remote Sensing & GIS and Software
Engineering.
Yash R Kharka, he received his B. Tech in Computer
Science from Sikkim Manipal University, Sikkim, India, 2011.
He is currently working as Assistant Systems Engineer at Tata
Consultancy Services.
His research interest includes Algorithms, Networking and
Image Processing.

Vol. 3, No.8, 2012

34 | P a g e
TABLE I. DIGITIZATION OF DRAINAGE PATTERN USING STRAHLERS ORDERING TECHNIQUE AND ATTRIBUTE ASSOCIATION
Input Output Segment Points Count Order Bi. R

0 (0,0)(1,1)(2,2)(3,3)(4,4) 5 1 2.000
1 (0,7) (1,7) (2,6)(3,5) (4,4) 5 1 2.000
2 (7,14)(8,13)(9,12)(9,11) (10,10) 5 1 2.000
3 (14,12)(13,11)(12,10) (11,10)(10,10) 5 3 -
4 (2,13)(2,12)(2,11)(2,10) (3,9)
(4,8)(5,7)(6,6)
8 1 2.000
5 (9,2) (9,3) (8,4) 3 1 2.000
6 (6,2) (7,3) (8,4) 3 1 2.000
7 (4,4) (5,5) (6,6) 3 2 1.500
8 (10,10) (9,9) (8,8) (7,7) 4 3 -
9 (8,4) (8,5) (8,6) (7,7) 4 2 1.500
10 (6,6) (7,7) 2 2 1.500

0 (0,0)(1,1)(2,1)(3,2)(4,3) 5 1 2.500
1 (14,10)(13,9)(12,9)(11,9) 4 3 -
2 (1,5)(2,5)(3,4)(4,3) 4 1 2.500
3 (1,13)(2,12)(3,11)(3,10)(4,9) 5 1 2.500
4 (8,13)(9,12)(10,11)(10,10)(11,9) 5 1 2.500
5 (2,8)(3,8)(4,9) 3 1 2.500
6 (11,9)(10,8)(9,8)(8,7)(7,6) 5 3 -
7 (4,3)(5,4)(6,5)(7,6) 4 2 1.000
8 (4,9)(5,8)(6,7)(7,6) 4 2 1.000

Vol. 3, No.8, 2012
35 | P a g e
Performance model to predict overall defect density

Dr. J. Venkatesh
Associate Professor, School of
Management Studies
Anna University: Chennai
600025, Regional Office
Coimbatore
Jothipuram Post,
Coimbatore 641 047,
Tamilnadu, India.

Mr. Priyesh Cherurveettil
Ph.D Part Time Research
Scholar, School of
Management Studies, Anna
University: Chennai 600025
Regional Office: Coimbatore,
Jothipuram Post, Coimbatore
641 047, Tamilnadu, India.

Mrs. Thenmozhi. S
Assistant Professor
Department of Computer
Applications
Gnanamani College of
Technology,
AK Sumuthiram, Namakkal
District. Tamilnadu, India.

Dr. Balasubramanie. P
Professor
Department of Computer
Science & Engineering
Kongu Engineering College
Perundurai, Erode - 638 052.
Tamilnadu, India

Abstract Management by metrics is the expectation from the IT
service providers to stay as a differentiator. Given a project, the
associated parameters and dynamics, the behaviour and outcome
need to be predicted. There is lot of focus on the end state and in
minimizing defect leakage as much as possible. In most of the
cases, the actions taken are re-active. It is too late in the life cycle.
Root cause analysis and corrective actions can be implemented
only to the benefit of the next project. The focus has to shift left,
towards the execution phase than waiting for lessons to be learnt
post the implementation. How do we pro-actively predict defect
metrics and have a preventive action plan in place. This paper
illustrates the process performance model to predict overall
defect density based on data from projects in an organization.
Keywords- process; performance; defect density; metrics.
I. INTRODUCTION TO OVERALL DEFECT DENSITY
Number of defects leaked into production is a key metric
that IT service providers will track month on month and intend
to show a downward trend to their clients. Number of defects
as a measure alone might not make sense, its relationship with
size or effort is important.
For example, if we have 40 defects leaked in January, 30 in
February and 5 in March, it doesnt mean that there is a
downward trend. 30 defects could be for a project effort of
1000 hrs whereas 5 defects could be for a project size of 50
hrs. Hence, the metric that need to be closely tracked is overall
defect density, number of defects leaked against the project
effort.
II. PROCESS PERFORMANCE MODELS IN ORGANIZATIONS
Process Performance Models (PPM) is probabilistic,
statistical and simulative in nature. It can predict interim and
final outcome, it is a proactive measure of tracking the end
goal instead of a reactive one. It can model the variation of
factor and help us understand the predicted range or the
variation of its outcomes. Mid-course correction can be made
to achieve desired outcome. Interestingly, PPMs enable
What-if analysis for project planning, dynamic re-planning
and problem resolution during project execution. We can run
what if exercises holding one or more values constant. We
can see the effect of tradeoffs between schedule, effort,
defects, staff and functionality.

CMMI Dev 1.2 predominantly focuses on development
and enhancement type of projects. CMMI for Services focuses
on production support or maintenance type of projects. At an
organization level, the respective Software Engineering
Process Group (SEPG) develops few standard process
performance models. These models are developed based on
the data gathered from different types of projects within the
organization. The organization focuses on three to four key
models; typically they are around defect density, productivity,
and schedule variance. The important step is to agree to
standard definitions for the entities and their measured
attributes. When we use terms like defect, productivity, size,
and even project, different teams will tend to interpret in their
own context. Hence it is important to have a common
definition.
Managers should have a good understanding of process
performance models and should use it to pro-actively manage
the customer needs. Project Managers will check the
availability of organizational developed process performance
models with respect to the project objectives and quality and
process performance objectives. During planning phase,
managers are expected to implement process models to
manage the objectives. Project manager has to define the
project objective and consider the organization objective if
there is no client objective defined. The specification limit of
the objective is derived from process performance baseline
arrived at the organization level. Project Manager is expected
to provide the values for the controllable factors in process
model prediction section. The prediction intervals are
automatically generated based on the values provided
III. VARIABLES ASSOCIATED IN PREDICTION MODEL
Based on brain storming session with the project team in
the organization the different parameters that influence overall
defect density were looked at. The team shortlisted following
factors to start with, domain experience, technical experience,
defects identified during design and coding phase, overall
review efficiency and usage of tools. Operational definitions
for these parameters were baseline and data was collected
from projects in a particular account against these parameters.
Linear regression was performed against the data to find out
which are the key variables that influences the overall defect
density.
Vol. 3, No.8, 2012
36 | P a g e
After many trial and error methods the below three
variables were established as the x factors.
1. Y - Overall defect density No of defects identified
in the entire life cycle of the project against total
effort for the project
2. X1 - Technical experience Average technical
experience of the team, in person months
3. X2 DDD - Design Defect Density - Defects
attributed to design identified during design review
against effort spent for design
4. X3 CDD Coding Defect Density - Defects
attributed to coding, identified during code review
against effort spent for coding.
IV. DEFECT DENSITY REGRESSION EQUATION
The project data collated for the x and y factors are as
shown in the Table 3.1. Data points from 25 projects in an
organization were collected and considered for analysis.
Projects factored in were similar in nature.

Y X1 X2 X3
Overall
Defect
Density
Technical
Experience
(in months)
Design
Defect
Density
DDD
Coding Defect
Density
CDD
0.092 36 1.813 0.231
0.093 31 0.158 0.052
0.095 35 0.258 0.140
0.114 35 0.044 0.083
0.120 47 0.425 0.192
0.119 42 0.458 0.069
0.128 33 0.510 0.250
0.134 40 0.520 0.125
0.126 34 1.525 0.310
0.139 22 0.650 0.055
0.093 30 0.058 0.057
0.064 38 0.143 0.022
0.066 44 0.035 0.213
0.074 54 0.079 0.051
0.080 41 0.390 0.310
0.083 44 0.090 0.290
0.131 33 0.540 0.150
0.134 39 0.510 0.125
0.136 36 1.625 0.300
0.322 12 0.026 0.004
0.565 20 0.558 2.125
0.350 50 5.000 0.500
0.089 59 0.014 0.071
0.083 58 0.435 0.424
0.320 38 0.500 0.500
Table 3.1 Project data values

Figure 3.1 Residual Plot
Mirror pattern is not found in Figure 3.1, Residual Plot and
hence no heteroscedasticity is found. The normal probability
plot is approximately linear. This would indicate that the
normality assumption for the errors has not been violated.
Looking at the p value, since it is 0.0003 which is < 0.05,
null hypothesis is not valid, which means the variables
selected have an impact to overall defect density

Intercept
Technical
Experience
(in
months)
Design
Defect
Density
DDD
Coding
Defect
Density
CDD
0.21375 -0.0035 0.0295 0.1892
Table 3.2 Regression Equation
As shown in Table 3.2, technical experience has a negative
influence on overall defect density. As the teams technical
experience increases the overall defect density is reduced. The
influence of Design Defect Density and Coding Defect
Density is positive.
This means that when the values of Design Defect Density
and Coding Defect Density are low the overall defect density
will be low and vice versa.

Vol. 3, No.8, 2012
37 | P a g e
V. DEFECT DENSITY COMPONENTS OF PREDICTION MODEL
Based on project data analyzed it is evident that overall
defect density is critically influenced by design sub process,
code sub process and technical experience. Organization
Metrics group would help with the baseline data for these
metrics. For code sub process and CDD metric, based on the
technology (Java, .Net) and review type (manual, tool, FxCop)
the baseline values can be tabled. Organization Metrics group
will share the baseline values for these combinations. Baseline
values will include lower specification limit (LSL), goal and
upper specification limit (USL).
The same can be gathered for Design Defect Density as
well. Project team needs to choose the process that they would
be following for coding or design sub process. Based on the
composition of sub process, project goal for DDD and CDD
would be calculated. It is also important for the project team to
justify why they have gone with a particular sub process and
the rationale. Table 4.1 gives the sub process performance
baseline for Coding Defect Density and Design Defect
Density. The values are represented by A1, A2, A3 and so on.
Organization Metrics team would have the actual baseline
values for LSL, Goal and USL for these identified metrics.
Based on the current project context, the type of technology
and review type will selected as shown in Table 4.2, Selected
Sub process performance baseline.
Sub
process
Metric Technology LSL Goal USL
Code
Review
CDD .Net A1 A2 A3
Code
Review
CDD Java C1 C2 C3
Code
Review
CDD J2ee D1 D2 D3
Design
Review
DDD .Net E1 E2 E3
Design
Review
DDD Java F1 F2 F3
Table 4.1 - Sub process performance baseline
Sub
process
Metric Technology Goal Comments
Code
Review
CDD .Net A2
Design
Review
DDD .Net E2
Table 4.2 Selected Sub process performance baseline
VI. DEFECT DENSITY MODEL PRACTICAL USAGE
One of the current releases in design phase was considered for
the practical usage of this model. The below steps will
illustrate the prediction model.

1. X factors baseline data was used as input. Technical
experience goal is 28 months with LSL of 14 months
and USL of 48 months
2. Sub process performance baseline data was reviewed
and based on the current project context the below
selection was made. As shown in Table 5.1 the sub
process code and design review were selected. Based
on the project usage, .net technology was selected.
The goal, upper specification limit and lower
specification limit are chosen from organization
baseline report.

Sub
process
Metric Technology LSL Goal USL
Code
Review
Coding
Defect
Density
(defects/
person
day)
.Net 0.10 0.38 1.5
Design
Review
Design
Defect
Density
(defects/
person
day)
.Net 0.15 0.22 0.45
Table 5.1 Selected Sub process

3. Update the actual technical experience in the team
and predict the overall defect density. The predicted
value is based on Monte Carlo simulation.

T
a
b
l
e
Table 5.2 Predicted Overall Defect Density

4. The data was compared against goal. The client goal
for overall defect density is 0.15 whereas the
predicted value is 0.21.
5. Perform what-if analysis and look at various
combinations of the x factors and analyze the
predicted overall defect density based on these
factors. Based on the project experience choose the
one which is close to reality. In this case the option of
24months technical experience, DDD 0.22 and CDD
0.38 was considered as the closest option.
6. List down the assumptions considered when the final
decision is made on the values of x factors. Ensure all
the relevant assumptions are documented. As need
be, the assumptions need to be validated with the
relevant stakeholders before the baseline process.
7. Understand the deviation and prepare preventive
action plan
Expected
client
overall
defect
density
Predicted
overall defect
density
Preventive Action Responsibility
0.15 0.21 List down the top
three preventive
action items
Project
Manager
Table 5.3 Deviation Analysis
Average
Technical
Experience
(in months)
DRDDE
(Defects/
Personday)
CRDDE
(Defects/
Personday)
Predicted
Overall
Defect
Density
24 0.22 0.38 0.21
Vol. 3, No.8, 2012
38 | P a g e
8. Estimated effort in person days for the project is 1000
person days. Based on the predicted defect density
and organizations standard effort distribution across
phases, the defects that could be injected at each
phase are predicted as show in Table 5.4

Phases Expected
Injection
Actual Defects
Captured
Remarks
Requirements 12 10
Design 38 28
Coding 70
Unit Testing 62
System
Testing
108
Table 5.4 Predicted-Actual Defects phase wise
9. Based on the actual data collated, keep updating
Table 5.4 to compare the expected and actual defects
captured. Based on the actual value in each phases,
the predicted value for next phases are accordingly
impacted. If there any specific inputs or
considerations on the actual values, those are
highlighted in the remarks column.
10. Prepare the detailed defect prevention plan. Against
each phase, list down the defect type, defect cause,
root cause, preventive action planned, responsible
person, target date and the status. Defect types could
be incorrect functionality or missing functionality or
incorrect user interface or missing user interface.
Defect causes could be lack of knowledge, missing
information or incidental. Root cause should be as
detailed as possible to plan for preventive and
corrective action. 5-Why analysis can be used to
identify the root causes.
Defect prevention plan is an ongoing document that
need to be tracked very closely. It is meant both for
planning and tracking defect prevention activities.
This plan has to be revisited after completion of each
stage. If defects detected during the completed stage
fall under different defect types and defect causes not
identified for preventing at that stage, then these new
types need to be included in the on-going phases.
VII. CONCLUSION
IT organizations focus on customer satisfaction is the key
for survival. Unfortunately the element of predictive behavior
during planning phase is very minimal and subjective. While
Capability Maturity Model recommends pro-active
management using quantitative models, the practical
implementation is very low. The context of the organization is
important in building these models. It is also important to
understand that the project managers need to be equipped with
the right information, metrics baseline and subject matter
expertise. Defect leakage is a standard concern in the industry.
The practical case study demonstrated the influence of Coding
Defect Density and Design Defect Density. The case study
also helped us understand how the values need to be
determined, the steps around what-if analysis, the defect
prevention plan and the tracking mechanism. This illustration
gives us the confidence that the predictive mechanism can be
planned and executed well in an organization.
REFERENCES
[1] CMMI Product Development Team, CMMI for Development, Version
1.2, CMMI-DEV, V1.2. Carnegie Mellon University, Pittsburgh, PA.,
2006.
[2] Stephen H. Kan, Metrics and Models in Software Quality Engineering
2
nd
ed., 2002
[3] Richard D. Stutzke, Estimating Software-Intensive Systems: Projects,
Products, and Processes, Addison-Wesley, 2005
[4] Robert W. Stoddard, CMMI Process Performance Models and
Reliability, Carnegie Mellon University, IEEE Reliability Society
Annual Technology Report, 2007
[5] Umesh Kumar Mishra, K. Harihara sudhan, Shalu Gupta. Establishing
Process Performance Baselines and Models for Statistical Control of
Software Projects, Proceedings of ASCNT-2011, CDAC, Noida, India
[6] Tim Kasse, Practical Insight into CMMI, 2
nd
ed., Artech House, 2008
[7] Bill Self, Greg Roche and Nigel Hill, Customer Satisfaction
Measurement for ISO 9000: 2000, 2002
[8] Michael Johnson and Anders Gustafsson, Improving Customer
Satisfaction, Loyalty and Profit, 2000
[9] Margaret K. Kulpa and Kent A. Johnson, Interpreting the CMMI: A
Process Improvement Approach, 2
nd
ed., 2008
[10] Tim Kasse, Practical Insight into CMMI, 2
nd
ed., 2008
[11] CMMI Product Development Team, CMMI for Development, Version
1.2, CMMI-DEV, V1.2. by Carnegie Mellon University, Pittsburgh, PA,
2006
[12] Don Peppers and Martha Rogers, Managing Customer Relationships: A
Strategic Framework, John Wiley & Sons, 2004
[13] Forrest W. Breyfogle III, Implementing Six Sigma: Smarter Solutions
Using Statistical Methods, John Wiley & Sons, 2
nd
ed., 2003
Vol. 3, No.8, 2012

39 | P a g e
Spontaneous-braking and lane-changing effect on
traffic congestion using cellular automata model
applied to the two-lane traffic

Kohei Arai
1

Saga University
Saga City, Japan
Steven Ray Sentinuwo
2

Saga University
Saga City, Japan

Abstract In the real traffic situations, vehicle would make a
braking as the response to avoid collision with another vehicle or
avoid some obstacle like potholes, snow, or pedestrian that
crosses the road unexpectedly. However, in some cases the
spontaneous-braking may occur even though there are no
obstacles in front of the vehicle. In some country, the reckless
driving behaviors such as sudden-stop by public-buses,
motorcycle which changing lane too quickly, or tailgating make
the probability of braking getting increase. The new aspect of this
paper is the simulation of braking behavior of the driver and
presents the new Cellular Automata model for describing this
characteristic. Moreover, this paper also examines the impact of
lane-changing maneuvers to reduce the number of traffic
congestion that caused by spontaneous-braking behavior of the
vehicles.
Keywords- spontaneous-braking; traffic congestion; cellular
automata; two-lane trafficcomponent.
I. INTRODUCTION
The study of traffic flow has received a lot of attention for
the past couple of decades. The simulations of traffic
congestion become the most important aspect in the field of
traffic analysis and modeling. Traffic congestion can be
defined as the saturation condition of road network that occurs
as increased traffic volume or interruption on the road, and is
characterized by slower speed, longer trip times, and increased
vehicular queuing. The investigated situations in the real
traffic condition are those of traffic congestion caused by some
main reason, such as insufficient road capacity, incidents,
work zones (e.g., road maintenance or constructions near the
road that requires space), weather events (e.g., in the case of
rain or snow) which can hampers visibility therefore a driver
have to slowdown its vehicle to compensate, or emergencies
situations (e.g., hurricanes or severe snowstorms). However, in
this paper, we concern to investigate the effect of individual
braking behavior of the driver towards traffic congestion.
In more detail, this paper interests to describe and
reproduce the characteristic of spontaneous-braking
probability and its effects to the traffic behavior. In the real
traffic situations, vehicle would make a braking as the
response to avoid collision with another vehicle or avoid some
obstacle like potholes, snow, or pedestrian that crosses the
road unexpectedly. However, in some cases the spontaneous-
braking may occur even if there are no obstacles in front of the
vehicle. In some country, the reckless driving behaviors such
as sudden-stop by public-buses, motorcycle which changing
lane too quickly, or tailgating make the probability of braking
getting increase.
One of the famous microscopic models for the simulation
of road traffic flow is Cellular Automata (CA) model. In
comparison with another microscopic model, the CA model
proposes an efficient and fast performance when used in
computer simulation [18]. CA is a dynamic model developed
to model and simulates complex dynamical system. The set of
CA rules may illustrate complex evolution patterns, such as
time and space evolution in a system. Those evolutions can be
shown just by use simple rules of CA. Furthermore, the
utilization of CA successfully explains the phenomenon of
transportation. These so-called traffic cellular automata (TCA)
are dynamical systems that are discrete in nature and powerful
to capture all previously mentioned basic phenomena that
occur in traffic flows [18]. The one dimensional cellular
automata model for single lane freeway traffic introduced by
Nagel and Schreckenberg (NaSch) [1] is simple and elegant
that captures the transition from laminar flow to start-stop
waves with increasing vehicle density. The space of CA is
discrete and consists of a regular grid of cells, each one of
which can be in one of finite number of possible states. The
number and array of cells in the grid depends on the specific
transportation behavior that is described. The simplicity of the
NaSch model has prompted the use of it for studying many
traffic situations.
This paper presents a new Cellular Automata model for
describing the phenomena of spontaneous-braking behavior
and lane-changing character in traffic flow. In this model, we
investigate the effect of spontaneous-braking probability and
lane-changing maneuver in two-lane highway with one-way
traffic character. This proposed model extends the NaSch
model that first introduced CA for traffic simulation. The set
of rules in NaSch model are modified to better capture and
describe the behavior of the driver while making spontaneous-
braking and lane-changing maneuver in traffic flow. The base
deceleration rule of NaSch model is applicable only to
stationary vehicles, which is vehicles that are blocked by the
Vol. 3, No.8, 2012

40 | P a g e
leading vehicle in the previous time step. This rule is not
applicable to two conditions, in the condition of those vehicles
which are stopped due to spontaneous-braking behavior, and
in the two-lane highway that allows vehicle to make lane-
changing maneuvers. Compared with the original NaSch
model, this proposed model exhibits spontaneous-braking
probabilities effect combined with acceleration, deceleration,
and lane-changing maneuvers effects. Though it is well known
that spontaneous-braking is extremely reducing the local speed
of vehicles, the impact on the global system has not been
studied.
This paper uses a two-lane highway character with a
periodic boundary condition. The periodic boundary approach
has been used to conserve the number of vehicles and the
stability of the model. The goal of this paper is to analyze the
phenomena of spontaneous-braking behavior in traffic flow
then propose a new cellular automata model to describing this
phenomena. Moreover, this paper also investigates the impact
of lane-changing maneuvers towards traffic congestion that is
caused by spontaneous-braking behavior.
This paper is organized as follows. Some studies relating
with CA based traffic flow is quick reviewed in Section 2.
Section 3 presents a short description of the theoretical aspect
of traffic CA model. Section 4 explains about the proposed
model. Section 5 contains simulation process and the results in
the form of fundamental diagrams and space-time diagrams.
Finally, Section 6 contains conclusion and a summary of
findings.
II. RELATED RESEARCH WORKS
The one dimensional cellular automata model for single
lane freeway traffic introduced by Nagel and Schreckenberg
(NaSch) [1] is a probabilistic CA model that captures the
transition from laminar flow to start-stop waves with
increasing vehicle density. NaSch model update the state of
cells synchronously in discrete time steps. There is a finite set
of local interaction rules. This set of rules manages the new
state of a cell by taking into account the actual state of the cell
and its neighbor cells. This local interaction allows capture
micro-level dynamics and propagates it to macro-level
behavior. This single-lane system consists of a one-
dimensional grid of L sites with periodic boundary conditions.
A site can either be occupied, or empty by one vehicle with
integer velocity between zero and v
max
. The velocity of each
vehicle is equivalent to the number of sites that a vehicle
advances in one update, if there is no obstacle ahead. Each of
vehicles moves only in one direction. Refer to the Ricket et. al
[6], they outlined the rules of single-lane model. The index i
denotes the number of vehicle, x(i) is the position of vehicle i,
v(i) is the vehicles current velocity, v
d
(i) is the maximum
speed, pred(i) is the number of preceding vehicle, gap(i) =
x(pred(i)) x(i) 1 indicates the width of the gap to the
predecessor. The rules are applied to all vehicles at the
beginning of each time step by simultaneously, which mean
using parallel update. Then the vehicles are advanced
according to their new velocities [6].
The parallel update rules are the following:
v(i) v
d
(i) v(i) := v(i) +1 (1)
v(i) > gap(i) v(i) := gap(i) (2)
v(i) > 0 rand < p
d
(i)v(i) := v(i) 1 (3)

The first rule represents the linear acceleration of each
vehicle which is not at the maximum speed to accelerate its
speed by one site (cell) until the vehicle has reached its
maximum velocity v
d
. Second rule ensures that vehicles
having predecessors in their way slowdown in order not to run
into them. In this rule, all vehicles are checked for their
distance between the vehicle and its predecessor. If the
distance is smaller than its speed then the speed is reduced to
the number of empty cells between them to avoid the collision.
Third rule consider the stochastic noise parameter.
The probability p
d
is the probability number of each car to
reduce its speed by one unit (cell) per time step. This NaSch
model encouraged another study toward traffic flow conditions
[2]-[7]. Ricket, et al. [8] investigated a simple model for two-
lane traffic. Their model introduced the lane changing
behavior for two lanes traffic. It was found that the
fundamental diagram for each lane is asymmetric but the
maximum is shifted towards large values of vehicular density
(
max
> 1/2 ). They proposed a symmetric rule set where the
vehicle changes lanes if the following criteria are fulfilled:
v
move
> gap
same
v
move
= min (v
n
+ 1, v
max
)
gap
target
> gap
same

gap
back
v
max

The variable gap
same
, gap
target,
and gap
back
denote the
number of unoccupied cells between the vehicle and its
predecessor on its current lane, and between the same vehicle
and its two neighbor vehicles on the desired lane, respectively.
The advance analysis about lane-changing behavior has
been done, which includes symmetric and asymmetric rules of
lane-changing [9-14]. Symmetric rule can be considered as
rules that threat both lanes equally, while asymmetric rule can
be applied in special characters highway, like German
highways simulation [15], where lane changes are dominated
by right lane rather than left lane. Another studies focus on the
effect of lane-changing behavior on a two-lane road in
presence of slow vehicle and fast vehicle [13], [16-18]. While
the NaSch model could reproduce some of basic phenomenon
observed in real traffic situations such as the start-stop waves
in congested traffic, but it has been observed that the base
NaSch model lacks the ability to produce other more realistic
traffic patterns [19].
In this paper, we consider two parameters in traffic
behavior; those are the spontaneous-braking behavior and
lane-changing maneuver that occurs in the real traffic
situation. This proposed model using two-lanes traffic and also
adopts the symmetric lane-changing rules.
III. TRAFFIC CELLAR AUTOMATA MODEL
Cellular automaton (CA), at the basis of the model
presented in this paper, is a discrete model studied in
computability theory, mathematics, physics, complexity
science, theoretical biology and microstructure modeling.
Vol. 3, No.8, 2012

41 | P a g e
Currently, various fields have been using CA models to
model the phenomena of their system, such as vehicular traffic
flow, pedestrian behavior, escape and panic dynamic,
collective behavior, and self-organization. CA model uses a
simple approach for modeling and simulation of complex
dynamical systems. The behavior of complex systems can be
described by considering at the local interactions between their
elementary parts. CA decomposes a complex phenomenon
into a finite number of elementary processes.
The CA model consists of two components, a cellular
space and a set of state. The state of a cell is completely
determined by its nearest neighborhood cells. All
neighborhood cells have the same size in the lattice. Each cell
can either be empty, or is occupied by exactly one node. There
is a set of local transition rule that is applied to each cell from
one discrete time step to another (i.e., iteration of the system).
This parallel updating from local simple interaction leads to
the emergence of global complex behavior.
The Nagel-Schreckenberg (NaSch) model is one of the
theoretical CA models for the simulation of freeway traffic
[1]. This NaSch model known as the simple CA model for
illustrate road traffic flow that can reproduce traffic
congestion, like slow down car behavior in a high-density road
condition. This model shows how traffic congestion can be
thought of as an emergent or collective phenomenon due to
interactions between cars on the road, when the density of cars
is high and so cars are close to each on average. The NaSch
model also known as stochastic traffic cellular automaton
(STCA) because it included a stochastic term in one of its
rules. Like in deterministic traffic CA models (e.g., CA-184 or
DFI-TCA), this NaSch model contains a rule that reflect
vehicle increasing speed and braking to avoid collision.
However, the stochasticity term also introduced in the system
by its additional rule. In one of its rules, at each time-step t, a
random number (t) [0,1] is generated from a uniform
distribution. This random number is then compared with a
stochastic noise parameter p [0,1]. For it is based on this
probability p then a vehicle will slow down to v(i) 1
cells/time-step. According to Nagel and Schreckenberg, the
randomization rule captures natural speed fluctuations due to
human behavior or varying external conditions [20].
IV. PROPOSED METHOD
This paper extends a probabilistic CA model that
introduced by Nagel-Schrekenberg [1] for the description of
single-lane highway traffic. While the original NaSch model
uses a single lane that is represented by a one-dimensional
array of L sites (cells), this paper considers two-lane highway
with unidirectional traffic character in periodic boundaries
condition. The two-lane model is needed to describe the more
realistic traffic condition which has several types of vehicles
with multiple desired velocities. In single-lane model, the
vehicles with multiple desired velocities just resulting in the
platooning effect with slow vehicle being followed by faster
ones and the average velocity reduced to the free-flow velocity
of the slowest vehicle [8].
The simulation model in this paper presents two additional
elements. The first additional element is spontaneous-braking
parameter. This element is needed to illustrate the probability
of spontaneous-braking behavior of the vehicle that occur in
the real traffic situation. The concept of spontaneous-braking
probability is introduced for the description of the spontaneous
reaction of the drivers while making a spontaneous-braking
behavior. This reaction can be caused by several things e.g., as
the response to avoid collision with another vehicles, the
reckless driving behaviors such as sudden-stop by public-
buses, motorcycle which changing lane too quickly, or
tailgating. Those behaviors make the probability of braking
getting increase.
In original NaSch model [1], there is no rule accommodate
the spontaneous-braking behavior. NaSch model introduced a
stochastic noise parameter p [0,1] that can make a
slowdown vehicle to v(i) 1 cells/time-step. However, in real
traffic situations this rule is difficult to describe the nature of
the braking, especially on spontaneous-braking behavior of the
vehicle. In our opinion, the value of braking is a variable
number and the spontaneous-braking represent the extreme
value of a braking behavior. Thus, the slow-down rule of
vehicle v(i) 1 cells/time-step cannot describe the
characteristic of spontaneous-braking. This paper introduces a
new additional rule to represent the behavior of spontaneous-
braking by using a spontaneous-braking probability P
b
: v(i)
v(i) b
x
. Here bx denotes the characteristic of driver while
make a braking. The value of b
x
is equal or less than the
current speed v(i). This rule takes into account the dynamic
characteristic of the driver while make a braking of its car.
Already mentioned before, a two-lane unidirectional highway
model with periodic boundary system is used in this
computational model. Refer to the discrete NaSch model, a
one-dimensional chain of L cells of length 7.5 m represents
each lane. There are just two possibility states of each cell.
Each cell can only be empty or containing by just one vehicle.
The speed of each vehicle is integer value between v = 0, 1, . .
., v
max
. In this model, all vehicles are considered as
homogeneous then have the same maximum speed v
max
. In
order to investigate the effect of spontaneous-braking behavior
then the state of a road cell at the next time-step, from t to t +
1 is dependent on the states of the direct frontal neighborhood
cell of the vehicle and the core cell itself of the vehicle. The
state of the road cells can be obtained by applying the
following rules to all cells (vehicles) by parallel updated:
Acceleration: v(i) min(v(i) +1, v
max
) (4)
Deceleration: v(i)min(v(i), gap(i)) (5)
Spontaneous braking probability p
b
: v(i) v(i) b
x
(6)
Driving: x(i)x(i) + v(i) (7)

As this simulation model try to investigate the effect of
spontaneous-braking behavior on traffic flow then this model
deliberately eliminates the randomization rule of original
NaSch (v(i) 1 cells/time-step). Here for the reason to avoid
the speed reduction of vehicles caused by this rule that could
influence our simulation results. The variable gap(i) indicates
the distance between a vehicle x(i) and its predecessor
x((i)+1). vmax represents the maximum speed of the vehicle.
The second additional element is lane-changing parameter.
By using two-lane highway model and applying multiple
Vol. 3, No.8, 2012

42 | P a g e
desired velocity types, then this paper also accommodates the
lane-changing maneuvers of vehicles. In the real traffic
situation, driver tends to make a lane-changing maneuver
while encounter traffic congestion along its lane. This paper
also intends to evaluate the impact of lane-changing
maneuvers towards the traffic congestion that caused by
spontaneous-braking behavior of the driver. In this model, the
lane-changing maneuver is analogous as the movement of
liquid. There is a different from the lane-changing model of
Ricket et al. In this model, a vehicle would consider changing
its lane only if the vehicles see another vehicle on its cell
ahead and do so if possible. It means, as long as there is a cell
free ahead on their lane then the vehicles would still remain on
their lane. This lane-changing model will preserve the
deceleration rule in our model that is showed in equation (5).

Figure 1. Schematic diagram of a lane-changing operation
The lane-changing rule is applied to vehicles to change
from right lane to left lane and conversely. Vehicles are only
move sideways and they do not advance. Fig. 1 shows the
schematic diagram of lane-changing operation. A vehicle
changes to the next lane if all of the following conditions are
fulfilled:
Cell
next
> 0 (8)
Cell
target
= 0 (9)
x
(cellsback )
+ v
(cellsback )
t+1
cell
t arget
(10)

Cell
next
, Cell
target
, and Cell
back
are the parameters that
inform the state of one cell ahead, state of next cell, and state
of cells behind on the other lane, respectively. If one cell is
unoccupied or free-cell then its state is 0. In the real traffic
situation, a driver also has to look back on the other lane and
estimate the velocity of another cars-behind to avoid a
collision. Equation (10) accommodates the driver behavior to
estimate the velocity of vehicles before change the lane.
V. SIMULATIONA DN RESULTS
The simulation starts with an initial configuration of N
vehicles, with random distributions of positions on both lanes.
This simulation use the same initial velocity for all vehicle v
min

= 0 and the maximum vehicle speed has been set to v
max
= 5
cell/time-step. Many simulations performed with different
density . The density can be defined as number of cars N
along the highway over number of cells on the highway L.
During one simulation, the total number of cars on the
highway cannot change. Vehicles go from left to right. If a
vehicle arrives on the right boundary then it moves to the left
boundary. Fig. 2 illustrates an environment, which exhibits a
certain configuration.

Figure 2. An environment with a certain configuration

This paper divides the analysis into two stages. The first
stage investigates the effect of spontaneous-braking on the
traffic flow. In this simulation stage, we analyze the traffic
flow for the spontaneous-braking probability b
p
= 0; 0.3; and
0.7. The simulation was running 1000 time steps to let the
system reaches its stable condition. The system automatically
increase the vehicles density from minimum density = 0
until maximum density = 100 percent. Once the transient
dies out, then the data extraction was started. The data was
analyzed using fundamental diagrams, which plot the velocity
of vehicle vs vehicle flow vs global density.
To show the system dynamics then the graph had written
the last ten steps for each density before the end of simulation.
Fig. 3 and Fig. 4 present the fundamental diagrams of this
model. Fig. 3 shows the measurement of the average velocity v
(t ) over all vehicles at each density. The red color, black
color, and blue color of scatter graph present the average
velocity in the condition with spontaneous-braking probability
P
b
= 0, P
b
= 0.3, and P
b
= 0.7, respectively.
One can be observed that in the traffic without spontaneous
braking probability, the maximum velocity 5 unit of distance
per unit of time could be achieved in the density 0.12.
When the probability of spontaneous-braking increased then
the critical density point that maximum velocity can be
achieved became lower than normal condition.
For the spontaneous-braking probability P
b
= 0.3, the
critical point of maximum velocity v
max
= 5 is around = 0.04.
While in the situation that spontaneous-braking probability P
b

= 0.7, the vehicles were very difficult to reach their maximum
speed v
max
= 5.

Figure 3. Average velocity (cell/time-step) vs density (cars/highway site)
In the phase after the critical density point of maximum
velocity was reached, the vehicles reduced their velocity to
synchronize with the gap between them and the vehicle ahead.
However, in the transition phase after the critical density
point of maximum velocity, the vehicles still maintained their
velocity. Regarding this average velocity graph, the traffic jam
obviously appeared when the average velocity v < 1 cell/time.

Vol. 3, No.8, 2012

43 | P a g e

Figure 4. Traffic flow (cars/time step) vs density (cars/highway site)

Fig. 4 illustrates the traffic flow over vehicles density for
the spontaneous-braking probability P
b
= 0, P
b
= 0.3, and P
b
=
0.7, respectively. The traffic flow indicates the number of
moving vehicles per unit of time. While the density parameter
means the number of vehicles per unit area of the highway. As
can be seen from the graph, there is a reduction in traffic flow
in the presence of spontaneous-braking parameter. We also
consider the critical density k
c
that appeared in each traffic
flow. Here, the critical density means a maximum density
achievable under free flow. In the traffic flow with P
b
= 0, the
critical density k
c
situated at the density = 0.18.
The critical density k
c
was getting lower when the
spontaneous-braking parameter increased. Below the critical
density k
c
, all vehicles can make a movement. However, in the
density after the critical density point, not every vehicle can
move at each time step. This critical density point also
indicates when the traffic congestion started to happen. To get
an intuitive feel for the dynamics, we provide a set of space-
time diagrams in Fig. 5, Fig. 6, and Fig. 7 for various density
values.
The horizontal axis represents space and vertical axis
represents the time. In order to get data to analyze, we
simulate this model for density = 0.25; 0.50; and 0.75 that
represent light traffic, moderate traffic, and heavy traffic
situations.
For density = 0.25, it can be seen that the spontaneous-
braking behavior has given a significant impact to produce
traffic congestion (Fig. 5). The single vertical line which is
shown in these time-space diagrams represents a stationary
vehicle that is making a spontaneous-braking behavior. In the
traffic with density value = 0.50, there is a moderate impact
of the spontaneous-braking behavior on the traffic congestion.
It can be seen that before the spontaneous-braking
parameter was applied, the congestion already occurred on the
traffic (Fig. 6). While in Fig. 7, the effect of spontaneous-
braking on traffic congestion just a slightly impact is shown.
That because in density value = 0.75, the traffic congestion
already appeared although in the condition without
spontaneous-braking behavior.

(a)
(b)
(c)
Figure 5. Space-time diagram for density = 0.25 and Pb = 0 (a), Pb = 0.3
(b), and Pb = 0.7 (c); without lane-changing maneuvers
Vol. 3, No.8, 2012

44 | P a g e
(a)
(b)
(c)
(b), and Pb = 0.7 (c); without lane-changing maneuvers

The lane-changing effect on traffic congestion is discussed
from here. As shown before that the spontaneous-braking
behavior can contribute to the traffic congestion.
(a)
(b)
(c)
Figure 7. Space-time diagram for density = 0.75 and Pb = 0 (a), Pb = 0.3 (b),
and Pb = 0.7 (c); without lane-changing maneuvers

Therefore, in this section we evaluate the effect of lane-
changing to reduce the congestion level. This lane-changing
model was applying the equations (8), (9), and (10).
Vol. 3, No.8, 2012

45 | P a g e
In this simulation, the vehicles can look back and estimate
the situation along 5 cells behind on the other lane before
make a lane-changing. We provide a set of space-time
diagrams in Fig. 8, Fig. 9, and Fig. 10 for the density values
= 0.25; 0.50; and 0.75.
(a)
(b)
(c)
and Pb = 0.7 (c); with lane-changing maneuvers
(a)
(b)
(c)
and Pb = 0.7 (c); with lane-changing maneuvers

Vol. 3, No.8, 2012

46 | P a g e
(a)
(b)
(c)
(b), and Pb = 0.7 (c); with lane-changing maneuvers.

The comparative graph shows that for the traffic density
< 0.75, the lane-changing maneuvers have given a good
impact to reduce the congestion level. However, in all
spontaneous-braking parameter value condition, the result
shows that there is no significant impact that is contributed by
lane-changing maneuver.
VI. CONCLUSION
In this work, we simulate the braking behavior of the
driver and present the new Cellular Automata model for
describing this characteristic. The original NaSch model has
been modified to accommodate the parameter of spontaneous-
braking probability. This spontaneous-braking probability rule
captures the natural of braking behavior due to human
behavior. This simulation shows that the traffic congestion can
be caused not only by the road capacity condition but also by
driver behavior. Moreover, we also evaluate the effect of lane-
changing to reduce the congestion that is caused by the
parameter of spontaneous-braking probability.
REFERENCES
[1] K. Nagel and M. Schreckenberg, A cellular automaton model for
freeway traffic, Journal of Physics I France, vol. 2, no. 12, pp.2221-
2229, 1992.
[2] A. Schadschneider and M. Schreckenberg. Cellular automaton models
and traffic flow, Physics A, 1993.
[3] L. Villar and A. de Souza, Cellular automata models for general traffic
conditions on a line, Physica A, 1994.
[4] M. E. Lrraga, J. a. D. Ro, and L. Alvarez-lcaza, Cellular automata for
one-lane traffic flow modeling, Transportation Research Part C:
Emerging Technologies, vol. 13, no. 1, pp. 63-74, Feb. 2005.
[5] K. Nagel, Particle hopping models and traffic flow theory, Physical
review. E, vol. 53, no. 5, pp. 4655-4672, May 1996.
[6] K. Arai and Tri Harsono Agent and diligent driver behavior on the car-
following part of the micro traffic flow in a situation of vehicles
evacuation from Sidoarjo Prong roadway, International Journal of
Computer Science and Network Security, 11, 1, 137-144, 2011.
[7] K. Arai, Tri Harsono, Ahmad Basuki, Car-Following Parameters by
Means of Cellular Automata in the Case of Evacuation, International
Journal of Computer Science and Security (IJCSS), Vol (5), 2011.
[8] M. Rickert, K. Nagel, M. Schreckenberg, and A. Latour, Two Lane
Traffic Simulations using Cellular Automata, vol. 4367, no. 95, 1995.
[9] W. Knospe, L. Santen, A. Schadschneider, and M. Schrekenberg,
Disorder effects in cellular automata for two lane traffic, Physica A,
vol. 265, no. 3-4, pp. 614633, 1998.
[10] A. Awazu, Dynamics of two equivalent lanes traffic flow model:
selforganization of the slow lane and fast lane, Journal of Physical
Society of Japan, vol. 64, no. 4, pp. 1071 1074, 1998.
[11] E. G. Campri and G. Levi, A cellular automata model for highway
traffic, The European Physica Journal B, vol. 17, no. 1, pp. 159166,
2000.
[12] L. Wang, B. H. Wang, and B. Hu, Cellular automaton traffic flow
model between the Fukui-Ishibashi and Nagel- Schreckenberg models,
Physical Review E, vol. 63, no. 5, Article ID 056117, 5 pages, 2001.
[13] B. Jia, R. Jiang, Q. S. Wu, and M. B. Hu, Honk effect in the two-lane
cellular automaton model for traffic flow, Physica A, vol. 348, pp. 544
552, 2005.
[14] D. Chowdhury, L. Santen, and A. Schadschneider, Statistical physics of
vehicular traffic and some related systems, Physics Report, vol. 329,
no. 4-6, pp. 199329, 2000.
[15] W. Knospe, L. Santen, A. Schadschneider, and M. Schreckenberg, A
realistic two-lane traffic model for highway traffic, Journal of Physics
A, vol. 35, no. 15, pp. 33693388, 2002.
[16] D. Chowdhury, L. Santen, and A. Schadschneider, Statistical physics of
vehicular traffic and some related systems, Physics Report, vol. 329,
no. 4-6, pp. 199329, 2000.
[17] R. J.Harris and R. B. Stinchcombe, Ideal and disordered two- lane
traffic models, Physica A, vol. 354, no. 14, pp. 582596, 2005.
Vol. 3, No.8, 2012

47 | P a g e
[18] X. G. Li, B. Jia, Z. Y. Gao, and R. Jiang, A realistic two-lane cellular
automata traffic model considering aggressive lane- changing behavior
of fast vehicle, PhysicaA, vol. 367, pp. 479 486, 2006.
[19] W. Knospe, L. Santen, A. Schadschneider, and M. Schreckenberg,
Empirical test for cellular automaton models of traffic flow, Phys.
Rev. E, vol. 70, 2004.
[20] S. Maerivoet and B. D. Moor, Transportation Planning and Traffic
Flow Models, 05-155, Katholieke Universiteit Leuven, Department of
Electrical Engineering ESAT-SCD (SISTA), July 2005.
AUTHORS PROFILE
Kohei Arai, He received BS, MS and PhD degrees in 1972, 1974 and 1982,
respectively. He was with The Institute for Industrial Science and Technology
of the University of Tokyo from April 1974 to December 1978 and also was
with National Space Development Agency of Japan from January, 1979 to
March, 1990. During from 1985 to 1987, he was with Canada Centre for
Remote Sensing as a Post Doctoral Fellow of National Science and
Engineering Research Council of Canada. He moved to Saga University as a
Professor in Department of Information Science on April 1990. He was a
councilor for the Aeronautics and Space related to the Technology Committee
of the Ministry of Science and Technology during from 1998 to 2000. He was
a councilor of Saga University for 2002 and 2003. He also was an executive
councilor for the Remote Sensing Society of Japan for 2003 to 2005. He is an
Adjunct Professor of University of Arizona, USA since 1998. He also is Vice
Chairman of the Commission A of ICSU/COSPAR since 2008. He wrote 30
books and published 307 journal papers

Vol. 3, No.8, 2012

48 | P a g e
Enhancing eHealth Information Systems for chronic
diseases remote monitoring systems
Amir HAJJAM
Laboratoire Systmes et Transports
Universit de Technologie de Belfort Montbliard, UTBM
Belfort, France

Abstract Statistics and demographics for the aging population
in Europe are compelling. The stakes are then in terms of
disability and chronic diseases whose proportions will increase
because of increased life expectancy. Heart failure (HF), a serious
chronic disease, induces frequent re-hospitalizations, some of
which can be prevented by up-stream actions. Managing HF is
quite a complex process: long, often difficult and expensive. In
France, nearly one million people suffer from HF and 120,000
new cases are diagnosed every year. Managing such patients, a
telemedicine system tools associated with motivation and
education can significantly reduce the number of hospital days
that believes therefore that the patient is hospitalized for acute
HF. The current development projects are fully in prevention,
human security, and remote monitoring of people in their living
day-to-day spaces, from the perspective of health and wellness.
These projects encompass gathering, organizing, structuring and
sharing medical information. They also have to take into account
the main aspects of interoperability. A different approach has
been used to capitalize on such information: data warehouse
approach, mediation approach (or integration by views) or
integration approach by link (or so-called mashup).
In this paper, we will focus on ontologies that take a central place
in the Semantic Web: on one hand, they rely on modeling from
conceptual representations of the areas concerned and, on the
other hand, they allow programs to make inferences over them.
Keywords- Ontologies; Web Semantic; Remote Monitoring;
Chronic Diseases.
I. INTRODUCTION
The pervasiveness of chronic diseases is highly growing
with increased life expectancy. In most developed countries,
those diseases are responsible for increasingly growing health
spending. Today, there are more than 15 million patients
suffering from such diseases in France as we do expect this
number to grow over 20 million by 2020 [1]. Having those
patients in specialized institutions (hospitals, nursing homes
...) is not only really desired but even not possible. A
European study Catalan Remote Management Evaluation
(CARME) [2] has shown that there was a 68% decrease in
heart failure related hospitalization and a 73% reduction of
days spent in hospital from 646 days to 168 days. The move is
towards solutions known as "home care", where patients are to
be cared for, medically and paramedically, by remaining in
their own homes. These remote monitoring solutions provide
unquestionably higher quality of care and greater security than
conventional practices and better quality of life for patients.
They incorporate the most innovative technological aspects
(monitoring and remote transmission of vital signs, detect
falls, alarms, etc.) and organizational aspects necessary for the
coordination of the different players contributing this home
care. These solutions are still widely at an experimental
stage, especially to assess their economic viability.
Pilot projects, with various concepts and objectives, were
born throughout the world: Gator Tech [3] for the USA,
Prosafe [4] for France, the work of Ogawa [5] [6] in Japan or
yet CarerNet [7] for England. Most recently, we have the
systems based on ontologies proposed by [8-12]. These
projects vary both in scale deployment and diseases monitored
(daily activities, asthma, Alzheimer's, cardiovascular disease,
falls, etc.). However, they all put back up relevant information
on the evolution of the patient's health including information
on daily activities.
Most of these projects include various sensors to monitor
the person's home (medical sensors, motion sensors, infrared
sensors etc..). Some, like the Gator Tech project and the work
of Tamura, focus on the instrumentation of domicile to study
the lifestyle of the occupant (electronic bathroom scales, ECG
in the bathtub, intelligent floors for fall detection, etc.) and
make his life easier. Other projects such as TelePat [13], Ailisa
[14] [15], CarerNet adjoin the sensors and home automation
physiological sensors to be placed directly on the person to
bring up more detailed medical data and allow a finer tracking
of changes in his condition.
All this information is daily backed up to monitor patients
to early detection of any abnormalities, behavioral changes or
vital signs, to raise an alert. The objective of such platforms is
to monitor a large number of patients. If we take the single
case of heart failure patients, actually they account for France
about 1 million patients with more than 120,000 new cases per
year. The amount of information stored and processed is
designed logically to an explosion of their volume. This has
prompted the community to build integrated systems where
semantics and data are coupled. The challenge in these
systems is to achieve semantic interoperability.
II. KNOWLEDGE MODELING
A. The main approaches
Today, databases cover most of biomedical information:
patients administrative data, clinical chemistry, clinical
diagnostics, images, or even genetic data. The use of this mass
of information to improve care and patient safety is still very
limited. Literature offers different approaches to address some
Vol. 3, No.8, 2012

49 | P a g e
of the issues raised above, including: data warehouse approach
as in BioWarehouse [16] and BioDWH [17] projects, the
integration by views approach in Hemsys [18] and Tambis
[19] projects or the so-called mashup in SRS [20] and Integr8
[21] projects. These different approaches offer methods and
techniques to solve problems related to access to information
regardless of its informative content, ie their semantics.
The increasing use of terminology, in the field of health, or
ontologies in health information systems encourages the use of
methodologies [22] and technologies from the Semantic Web
community.
B. Ontologies
Artificial intelligence has allowed knowledge to be
represented in the form of a domain knowledge base and to
automate their use in problem solving. These knowledge bases
are generally not reusable which by the way limit their
interest. To overcome this problem, the notion of ontology has
been introduced [23]. An ontology is seen as a set of concepts
for modeling knowledge in a given field. A concept may have
several thematic senses. The concepts are linked by semantic
relations, composition relations and inheritance. Many
researchers have proposed definitions including:
- Guarino introduced the formal ontology notion, defined
as a conceptual modeling: "An ontology is an agreement on a
shared and possibly partial conceptualization" [24].
- The ontology is defined by Uschold [25] as a formal
description of entities and their properties, relationships,
constraints and behaviors. Furthermore, the authors introduced
the notion of ontology explicit "An explicit ontology may take
a variety of forms, but necessarily it will include a vocabulary
of terms and some specification of their meaning".
- Thomas R. Gruber [26] which describes ontology as an
explicit specification of a conceptualization of modeling
concepts and relationships between concepts: "An ontology is
a specification of a conceptualization. That is, an ontology is a
description (like a formal specification of a program) of the
concepts and relationships that can exist for an agent or a
community of agents. This definition is consistent with the
usage of ontology as set-of-concept-definitions, but more
general".
- John F. Sowa [27] clarified this concept and defined
ontology as a type catalog from the study of categories of
abstract and concrete entities that exist or may exist in a
domain: "The subject of ontology is the study of the categories
of things that exist or may exist in some domain. The product
of such a study, called ontology, is a catalogue of the types of
things that are assumed to exist in a domain of interest D from
the perspective of a person who uses a language L for the
purpose of talking about D. The types in the ontology
represent the predicates, word senses, or concept and relation
types of the language L when used to discuss topics in the
domain D".
Christophe Roche [28] gave a simple and generic
definition that encompasses and summarizes the above
definitions "An ontology is a conceptualization of a domain to
which are associated one or more vocabularies of terms. The
concepts are structured into a system and participate in the
meaning of terms. Ontology is defined for a particular purpose
and expresses a view shared by a community. An ontology is
expressed in language (representation) based on a theory
(semantics) that guarantees the properties of the ontology in
terms of consensus, consistency, reuse and sharing"
Ontologies are widely accepted as an appropriate form for
the conceptualization of knowledge. They represent a basic
step in the knowledge representation process which integrates
terminology, taxonomy (organization of concepts) and
description of relations among concepts and/or classes of
concepts.
Using ontology enables appropriate organization of
procedural knowledge and that can be beneficial for the
implementation and maintenance of any complex system.
Ontologies are reusable and facilitate interoperability among
the application. They enable easier verification and
comparison and ensure comparability of results coming from
applications using the same ontology.
Ontologies can be described by meta-languages such as the
Unified Modeling Language (UML), expressing the concepts
in classes with attributes and operations as well as the
interrelations in associations. HL7 Version 3 Normative
Edition shows how to map the HL7 data types to the Object
Management Groups (OMG) Unified Modeling Language
(UML) [29].
C. What does one represent in ontology?
Ontologies allow representing knowledge and the way to
automatically handle it, while preserving their semantics.
Knowledge is defined through concepts linked together by
relationships. The ontology is then presented, usually in the
form of a hierarchical organization of concepts.
Concepts are represented by a set of properties and could
be equivalent, not connected or dependent. They can be linked
by relations defined as a connection concept between entities,
often expressed by a term or a literal symbol or other. We
have two types of links: hierarchical and semantic. The
hierarchical relationship resumes Hyperonymy / hyponymy
structuring, while the semantic relationship links the concepts
through a link, said part-whole, which corresponds to the
Holonymie / meronymy structuring. A hierarchical
relationship links a higher member, said the hypernym
element, and a lower member, said hyponym element, having
the same properties as the first element with at least one
additional one.
As the concepts, relationships can have algebraic
properties (symmetry, reflexivity, transitivity). To describe the
concepts and relationships of ontology, it is expressed in a
language and is based on formalism.
1) The representation formalisms
Ontology, as described above, needs to be formally
represented. Moreover, it must represent the semantic relations
linking concepts. To this end, much formalism has been
developed:
- The diagrams represent complex data structures. They are
considered as a prototype describing a situation or standard
Vol. 3, No.8, 2012

50 | P a g e
object. They provide a benchmark for comparing objects that
we wishes to recognize, analyze or classify. The prototypes
must consider all possible forms of expression of knowledge.
A scheme is characterized by attributes (data structure), facets
(the attributes semantics) and relationships (the inheritance
semantics).
- Semantic networks represent a graph structure that
encodes the knowledge and their properties. The nodes of the
graph represent objects (concepts, situations, events, etc..) and
edges express relations between these objects. These relations
can be links "kind of" expressing the inclusion relation or links
"is a" showing the relationship of belonging. It includes a set
of concepts describing an area completely. The interest of
these graphs is their non-ambiguity and ease of use. This
prompted the designers of multiple applications to use them,
whether in knowledge acquisition, information retrieval and
reasoning about conceptual knowledge.
- A script is a data structure that contains knowledge about
a situation and which combines representations. It can be seen
as a set of elementary actions or references to other scenarios,
ordered according to their sequence in time.
2) Building ontologies
The method chosen to build ontology should be strongly
guided by the desired type of ontologies and objectives of its
use.There are three types of methods for the construction of
ontology: manual, automatic and semi-automatic. For the first,
experts create a new ontology of a domain or extend an
existing ontology. In the automatic method, the ontology is
built by knowledge extraction techniques: concepts and
relations are extracted and then verified by the inferences.
Finally, the semi-automatic, ontologies are automatically built
and used to extend ontologies that was built manually.

Figure 1. Steps for creating a medical ontology: intervention at different
phases of the ontology development
For the medical field and chronic disease management, the
creation of an ontology should go through the following steps
(Figure 1):
- Establishment of a corpus of work from a thesaurus using
a morpho-syntactic analysis for a list of candidate terms. A
tool such as Syntex software, including working on verb
phrases is particularly interesting. Furthermore, a study of the
context of each candidate term would highlight additional
concepts and / or to specify other relationships between
concepts.
- Semantic analysis for validating candidate terms as a
term of the domain by a medical expert. It would facilitate the
grouping of terms validated in concepts, defining relations
between concepts and between symptoms and function.
- Structuring by semantic groups.
- Finalization of the process in a language, based on
description logics.
Process of designing the ontology begins after a language
and a tool have been selected. There are two standard
approaches to the ontology design: bottom-up approach
(smaller parts of the ontology are constructed first and then
later intagretaed) and top-down approach (design upper
classes and the developp small pieces of the hierarchy).
Though, probably the best way of creating an ontology is to
combine both approaches in an iterative way.
III. REQUIREMENTS FOR MONITORING PATIENTS WITH
CHRONIC DISEASES
The home care solutions (Figure 2) support usually
innovative technological aspects (monitoring and remote
transmission of vital signs, detect falls, alarms ...) and
organizational aspects necessary for the coordination of
different factors contributing to remain at home [30]. These
solutions are still largely at an experimental stage, in order to
assess the relationship between cost, reliability, the medical
service and economies of scale they are likely to make to the
health system overall.

Figure 2. An evolving architecture, encompassing a full set of domestic and
medical devices along with the analysis and interpretation of data
Vol. 3, No.8, 2012

51 | P a g e
Most of these systems are designed and intended to follow
a particular chronic disease (heart failure, respiratory failure,
diabetes ...) and require fairly large technological equipment
(sensors, computers located at home, set-top boxes for the
transmission information ...). However, older people are multi-
disease, with both of several chronic disorders related to age.
Technically and economically it cant be considered to
increase the monitoring systems number. It is therefore
necessary to focus on the interoperability of these systems, so
as to factor out the common elements, thereby reducing costs
of deployment and operation.
Elderly patients are often multi-pathological so currently it
is necessary to multiply the patient's home systems. This
profusion of systems has little interest because most of them
uses similar equipment to perform their measurements. Motion
sensors for example are found in virtually all existing
solutions.
To avoid this multiplication of equipment we must adopt
architecture to pool them. Thus, even a motion sensor can be
used by different applications. This pooling of equipment
meets both an economic need but also a demand of patients
seen in general who wish to limit the proliferation of such
equipment in their homes.
In practice this mutualization and this consideration multi-
phatologiques will translate by monitoring platforms
necessarily evolving which can therefore integrate knowledge
about various diseases. Ontologies represent then formalism
well adapted to enable the integration of new knowledge and /
or to make available the knowledge.
A. The heterogeneity of medical knowledge
The information and resources used to consider and treat
various diseases are necessarily heterogeneous and make their
understanding and analysis very difficult. Meaning
preservation of information shared is then an important
problem. This is what is called semantic interoperability. A
commonly accepted definition for semantic interoperability,
"it gives meaning to the information shared and ensures that
this is common sense in all systems between which exchanges
must be implemented" [31-33]. Consideration of this semantics
enables distributed systems to combine received information
with local information and treat all consistently.

Figure 3. Ontologies are an effective way for the representation and the
sharing of knowledge
To ensure semantic interoperability, information shared
between systems (Figure 3) must first be described in a formal
structure for preserving its semantics. This is a recurring
problem in the field of knowledge engineering, where
methodologies and techniques are proposed to collect,
identify, analyze, organize and share knowledge between
different entities. Among these techniques, ontologies are
experiencing a rapid development over the ten years past and
appear as an effective way for knowledge representation.
B. Techniques for semantic interoperability.
A number of techniques have been proposed in the
literature to achieve interoperability [34]. They are often used
to allow data sharing between heterogeneous knowledge bases
and for the re-use of these bases.
We can distinguish three main categories which are:
- The alignment of ontologies [35], for whom the goal is to
find correspondences between ontologies. It is usually
described as an application of the MATCH operator [65],
whose input consists of a set of ontologies and output, formed
correspondences between these ontologies
- The mapping of ontologies which allows, for example, to
query heterogeneous knowledge bases using a common
interface or transforming data between different
representations.
- The merging of ontologies, which creates a new
ontology, called the merged ontology with the knowledge of
the original ontologies. The challenge then is to ensure that all
correspondences and differences between ontologies are
properly reflected in the resulting ontology.
Generally speaking, providing semantic interoperability
among heterogeneous ontologies is still primarily a semi-
automated process.
IV. DISCUSSION
Ontologies are necessary to both facilitate goals semantic
structuring with their relations and take into account the
heterogeneity of knowledge in a growing field such as
monitoring patients at home and especially patients with
chronic diseases. Their increasing use in this area, leads to in
significant availability of ontologies that drives us to think
about when to re-use them.
It is then important to take into account both their
popularity and also the simplicity of their implementation. The
ratio of these two parameters allows us to gauge the interest to
investigate their interoperability.
Ontology engineering and management have to encompass
the entire ontology lifecycle: creation, coordination and
merging [37]. Merging tools or alignment of ontologies allow
the integration of information from a distributed environment
or heterogeneous systems. It is essential to establish semantic
correspondences between ontologies that describe this shared
information.
The role of alignment tools [1] is to search for matches
between the concepts of distinct ontologies, to allow the joint
consideration of the resources they describe. This is to
Vol. 3, No.8, 2012

52 | P a g e
combine the techniques and methods of matching linguistic,
syntactic, semantic or structural. Reference [38] enriched
ontologies based on thin semantic analysis of concept of labels
and in the fact that regularities exist in the way of naming
them. These naming conventions are used to establish
mappings between these labels and axioms of the ontology,
which makes semantic information explicit and then use it to
automatically reason above.
Currently there exist a variety of heuristics and other
techniques that can be utilised for semantic interoperability,
but there is still plenty of scope for refinement and for
providing fully automated frameworks.
V. CONCLUSION
In order to provide a consistent solution in the field of
medical telemonitoring, monitoring systems must take into
account different pathologies in order to avoid duplication of
equipment. They must therefore be open and scalable to allow
the sharing and management of heterogeneous knowledge.
Ontologies are particularly suited for understanding,
sharing and integrating information. However, various
problems are still open, others appear: design method
ontologies, representation and reasoning on ontologies,
automatic generation of ontologies, ontology alignment and
development, representation and data persistence based
ontological systems integration based ontological design
databases accessible from ontologies, integration of blur in
ontologies, etc..
Beyond the issues raised by the heterogeneity of available
data, the sequence of algorithmic processes that can exploit
this data represents a scientific and technical challenge.
REFERENCES
[1] Simon, P.; Acker, D. La place de la tlmdicine dans lorganisation des
soins. Rapport du Ministre de la sant et des sports Direction de
lHospitalisation et de lOrganisation des Soins. 2008.
[2] Lupon, J.; http://www.acmcb.es/files/425-933-DOCUMENT/Lupon-
41010-7.pdf
[3] Helal, S. ; Mann, W. ; El-Zabadani, H. ; King, J. ; Kaddourra, Y.;
Jansen, E. ; The Gator Tech Smart House: A Programmable Pervasive
Space , In Computer, 2005, 38, 50-60.
[4] Chan, M.; Campo, E.; Estve, D. PROSAFE, a multisensory remote
monitoring system for the elderly or the handicapped in Independent
Living for Persons with Disabilities and Elderly People In Proceedings
of the ICOST2003, Paris, France, September 2003, pp. 89-95.
[5] Ogawa, R.; Togawa, T.; Attempts at monitoring health status in the
home , In the Proceedings of the 1st Annual International Conference
On Microtechnologies in Medicine and Biology, Lyon, France, October
2000, 52-556.
[6] Tamura, T.; Togawa, T.; Ogawa, M.; Yoda, M. Fully automated health
monitoring system in the home, In Medical Engineering & Physics,
1998, 20, 8, 573-579.
[7] Williams, G.; Doughty, K.; Bradley, D. A. A systems approach to
achieving CarerNet - an integrated and intelligent telecare system, In
IEEE Transactions on Information Technology in Biomedicine, 1998, 2,
1, p. 1-9.
[8] Aniello Minutolo, Giovanna Sannino, Massimo Esposito, Giuseppe De
Pietro. 2010. A rule-based mHealth system for cardiac monitoring. 2010
IEEE EMBS Conference on Biomedical Engineering & Sciences.
[9] Federica Paganelli, Dino Giuli. 2011. An Ontology-Based System for
Context-Aware and Configurable Services to Support Home-Based
Continuous Care. IEEE Transactions on Information Technology in
Biomedicine, 2011, 15, 2.
[10] Miguel A. Valero, Laura Vadillo, Ivn Pau, and Ana Pealver. 2009. An
Intelligent Agents Reasoning Platform to Support Smart Home Telecare.
IWANN 2009, Part II, LNCS 5518. 679686.
[11] F. Latfi, B. Lefebvre and C. Descheneaux. 2007. Le rle de lontologie
de la tche dans un Habitat Intelligent en Tl-Sant. 1res Journes
Francophones sur les Ontologies JFO, Sousse, Tunisie, 18 - 20 Octobre
2007.
[12] F. Latfi, C. Descheneaux, B. Lefebvre. 2007. Habitat intelligent en tl
Sant : ontologie de lquipement. FICCDAT, 16-19 Juin. Toronto,
Canada, 2007.
[13] Boudy, J. ; Baldinger, J. ; Delavaut, F. ; I., B.; Dorizzi, B. ; Farin, I.
Tlvigilance mdicale au service du patient domicile In Proceedings
of the ASSISTH2007, Toulouse, France, November 2007, 1, 397-40.
[14] Noury N. AILISA : Plateformes dvaluations pour des technologies de
tlsurveillance mdicale et dassistance en grontologie, In
Grontologie et Socit, 2005, 113, 97-119.
[15] Rialle, V.; Lamy, J.B.; Noury, N.; Bajolle, L. Telemonitoring of patients
at home : a software agent approach , In Computer Methods and
Programs in Biomedicine, 2004, 72, 3, 257-26.
[16] Lee, T.; Pouliot, Y.; Wagner, V.; Gupta, P.; Stringer-Calvert D. W.J.;
Tenenbaum, J. D.; Karp, P. D. BioWarehouse: a bioinformatics database
warehouse toolkit In BMC Bioinformatics, 7, 170, 2006.
[17] Tpel, T.; Kormeier, B.; Klassen, A.; Hofestdt, R. BioDWH: A Data
Warehouse Kit for Life Science Data Integration, Journal of Integrative
Bioinformatics, 2008, 5, 2.
[18] Pillai, S.; Gudipati, R.; Lilen, L. Design issues and an architecture for a
heterogenous multidatabase system In Proceedings of the 15th ACM
Computer Science Conference, St. Louis, Missouri, USA, February
1987.
[19] Goble, C. A.; Stevens, R.; Ng, G.; Beshhofer, S.; Paton, N. W.; Baker, P.
G.; Peim, M.; Brass, A. Transparent access to multiple bioinformatics
information sources In IBM Systems Journal, 2001, 40, 2, 532-551.
[20] Etzold, T.; Argos, P. SRS an indexing and retrieval tool for flat file
data libraries In Computer Applications in the Biosciences, 1993, 9, 1,
49-57.
[21] Kersey, P.; Bower, L.; Morris, L.; Horne, A.; Petryszak, R.; Kanz, C.;
Kanapin, A.; Das, U.; Michoud, K.; Phan, I.; Gattiker, A.; Kulikova, T.;
Faruque, N.; Duggan, K.; Mclaren, P.; Reimholz, B.; Duret, L.; Penel,
S.; Reuter, I.; Apweiler, R. Integr8 and genome reviews: integrated
views of complete genomes and proteomes In Nucleic Acids Res, 2005,.
33, 297-302.
[22] Blobel, B. & Oemig, F. Ontology-driven health information systems
architectures. Stud Health Technol Inform 150, 195-199 (2009),
doi:10.3233/978-1-60750-044-5-195.
[23] Charlet, J. ; Bachimont, B.; Troncy, R. Ontologies pour le web
smantique In Revue I3, 2004, Hors srie Le Web smantique, 43-63.
[24] Guarino, N.; Poli, R. Formal ontology in conceptual analysis and
knowledge representation. In Special issue of the International Journal of
Human and Computer Studies, 1995, 43, 5/6, 625640.
[25] Uschold, M.; Gruninger, M. Ontologies: Principles, Methods and
Applications. In Knowledge Engineering Review, 1996, 11, 2, 93 - 155.
[26] Gruber, T. R. A Translation Approach to Portable Ontology
Specifications In Knowledge Acquisition, 1993, 5, 2, 199-221.
[27] Sowa, J. F. Building, Sharing, and Merging Ontologies. In
http://users.bestweb.net/~sowa/ontology/ontoshar.htm, 2001.
[28] Roche, C. Terminologie et ontologie. In Revue Langages, 157, 2005.
[29] Health Level 7, Inc, http://www.hl7.org.
[30] Ahmed Benyahia, A. ; Hajjam, A. ; Hilaire, V.; Hajjam, M. Ontological
architecture for management of telemonitoring system and alerts
detection. In A. Hajjam (eds), eHealth and remote monitoring, ISBN
980-953-307-305-2, 2012.
[31] Charlet, J. LIngnierie des connaissances : dveloppements, rsultats et
perspectives pour la gestion des connaissances mdicales. Habilitation
diriger des recherches, Universit Paris 6, 2002.
[32] Jouanot, F. Un modle smantique pour linteroprabilit de systmes
dinformation. In INFORSID, 2000. 347364.
Vol. 3, No.8, 2012

53 | P a g e
[33] Vernadat, F. Interoperable enterprise systems: Architectures, methods
and metrics. In Rapport technique, LGIPM, Universit de Metz, France,
2007.
[34] Choquet, R.; Teodoro, D.; Mels, G.; Assele, A.; Pasche, E.; Ruch, P.;
Lovis, C.; Jaulent, M.C. Partage de donnes biomdicales sur le web
smantique, Proceedings of the IC2010, Nmes, France, june 2010.
[35] Scharffe, F. ;Fensel, D. Correspondence patterns for ontology alignment.
In Proceedings of the 16th EKAW 08, Berlin, Heidelberg, 2008, 8392.
[36] Shvaiko, P.; Euzenat, J. Ontology matching: state of the art and future
challenges. In IEEE Transactions on Knowledge and Data Engineering,
2012.
[37] Rebstock, M., Fengel, J., Paulheim, H. (2008) Ontologies-Based
Business Integration. Springer-Verlag, Berlin.
[38] Fernandez-Breis, J. ; Iannone, L. ; Palmisano, I. ; Rector, A. ; Stevens, R.
Enriching the gene ontology via the dissection of labels using the
ontology pre-processor language. In Knowledge Engineering and
Management by the Masses, 2010, 6317, 5973.
AUTHORS PROFILE
Amir HAJJAM received his Ph.D. in Computer
Science from University of Haute-Alsace (UHA) in
1990. He was with the UHA from 1990 to 2002. He
joined the University of Technology of Belfort-
Montbeliard (UTBM) in 2002 and performs research
activity at the Systems and Transportation Laboratory
(SeT). His current research interest includes heuristic
mechanisms and Artificial Intelligence applied in a
distributed dynamic framework to networks, telecommunications, intelligent
transportation services and e-Health. He is responsible for the research area of
e-Health. He is the author of 3 books and many international publications in
refereed journals and conferences. He is Editorial Board Member of many
international journals and serves on the steering committee of many
workshops and conferences. He is an expert for various organizations such as
the Association Nationale de la Recherche et de la Technologie (ANRT
France), Agence Rgionale de l'Innovation d'Alsace (ARI Alsace) and Centre
National pour la Recherche Scientifique et Technique du Maroc (CNRST
Morocco).
Vol. 3, No.8, 2012

54 | P a g e
E-commerce Smartphone Application

Abdullah Saleh Alqahtani
School of Computer Science, Engineering and
Mathematics, Faculty of Science and Engineering,
Flinders University,
Adelaide SA 5001, Australia
Robert Goodwin
School of Computer Science, Engineering and
Mathematics, Faculty of Science and Engineering,
Flinders University,
Adelaide SA 5001, Australia

Abstract Mobile and e-commerce applications are tools for
accessing the Internet and for buying products and services.
These applications are constantly evolving due to the high rate of
technological advances being made. This paper provides a new
perspective on the types of applications that can be used. It
describes and analyses device requirements, provides a literature
review of important aspects of mobile devices that can use such
applications and the requirements of websites designed for m-
commerce. The design and security aspects of mobile devices are
also investigated. As an alternative to existing m-commerce
applications, this paper also investigates the characteristics and
potential of the PhoneGap cross-mobile platform application. The
results suggest that effective mobile applications do exist for
various Smartphones, and web applications on mobile devices
should be effective. PhoneGap and Spree applications can
communicate using JSON instead of the XML language. Android
simulators can be used for ensuring proper functionality and for
compiling the applications.
Keywords- E-commerce ; PhoneGap ; M-commerce ; Smartphones
; Spree commerce ; Ruby on Rails.
I. INTRODUCTION
The internet has changed many aspects of society, from
business to recreation, from culture to communication and
technology, as well as shopping and travelling. This new form
of communication has provided new ways of doing business
with the help of technological development. E-commerce is
the new way of shopping and doing business. Technology has
allowed companies to promote and sell their products on new
markets, overcoming geographical borders as never before.
Consumers have access to a wider market of products when
they use wireless and internet technologies. Mobile devices
with wide access to the Internet have allowed companies to
reach consumers in more diverse ways, thus ensuring deep
market penetration.
This study investigates the opportunities generated through
mobile telephone access to the Internet [13]. Faster wireless
networking standards allow wireless devices to use more e-
commerce applications, and consequently, permit wider access
to mobile commerce (m-commerce) [14]. M-commerce has
been defined as a special branch of e-commerce, in which
mobile devices and their network connection medium are used
to buy, sell, and promote products, services, and information
[20]. According to Koukia, Rigou and Sirmakessis (2006),
wireless technologies have improved traditional e-commerce
by providing the additional aspects of mobility (of
participation) and portability (of technology). On this theme,
mobile and e-commerce application developments are an
important factor for the expansion of m-commerce among
consumers. The technical characteristics of devices and
corresponding applications, as well as Internet access
facilities, are determining the level of acceptance of m-
commerce and its development. Aspects like processing
power, display and device size, mobile internet coverage,
standardization and quality of devices, are only some of the
important factors that decide the level of use of m-commerce,
and consequently, the level of its development [15].
The purpose of designing interfaces for mobile
applications should be to increase consumers interest in using
and dedication to m-commerce. Among the inhibiting factors
is that m-commerce applications were developed based on e-
commerce applications. The most important thing when
designing such applications is to design the application in such
a way that it does not distract the user from the main purpose
of the application [15]. However, aspects concerning security
and accessibility should not be neglected. Even though storing
sensitive data such as medical, financial, or personal
information on mobile devices can help people, the risks of
losing such information or of unauthorized access are higher
and should be considered when an m-commerce transaction
begins[16].
This paper will review the latest trends in mobile and e-
commerce applications and will develop an application
architecture that describes the internal architecture of both web
and mobile components. Moreover, the focus will be on
developing a more sophisticated demonstration mobile
application regime that will employ web-services to
communicate with web servers. Furthermore this paper will
discuss the main characteristics of devices used for m-
commerce, the available survey design guidelines, and the
important role of these characteristics for increasing the
potential of m-commerce will be articulated.
This paper will also investigate the requirements of e-
commerce applications and why normal websites are not
suitable for mobile devices. More specifically this paper will
analyse the characteristics required for websites so that they
function properly on mobile devices. These characteristics are
screen size, input device, task-based interfaces for mobile
devices, m-loyalty, design aesthetics and website design.
Other research questions relate to the reason why mobile
native applications are preferred over mobile websites and
Vol. 3, No.8, 2012

55 | P a g e
what can be achieved using PhoneGap cross-mobile platform
applications. Finally, the paper will attempt to provide
solutions for mobile application development and make
recommendations for future directions.
II. RESEARCH METHODOLOGY
A literature review on the two major issues of e-commerce
mobile applications has been conducted, these being: firstly,
the interface usability of mobile applications; and secondly,
design and security considerations. The literature review
concluded that mobile applications must effectively operate on
different Smartphones and have the ability to use different
ecommerce web applications through web services.
A. Research Procedure
The main question answered here is: How can we exploit
the usability and security of e-commerce application(s) for
mobile devices (m-commerce) with maximized mobile
platform independence?
The sub-questions indicated below should be answered in
order to fully explain the main themes in this paper :
1) Why is mobility required for e-commerce application?
2) Why are normal websites not useful for mobile devices?
3) Why are mobile native applications preferred over mobile
websites?
4) What can we achieve using the PhoneGap (cross-mobile
platform application development framework)?
5) How will the initial application architecture use Spree-
commerce, PhoneGap and web applications?
6) What are the different solutions for developing mobile
applications?
In this paper the focus was also on developing a
prototype mobile application. The following steps were taken
in order to complete a demonstration application:
1) Understanding how the Spree commerce system was
developed and how it works.
2) Improved the application architecture based on more
research and actual development experience gained while
developing the application.
3) Developed a more sophisticated demonstration mobile
application that will communicate with web servers
through web services.
a) Developed of the PhoneGap application and tested it on
iOS and Android
Page constraints have limited the papers scope to
actual functionality of the application. This paper will
therefore discuss basic functionality.
Step 1: In order to achieve the first goal based on the web
server was set up a Spree-commerce Ruby on Rails
application. Spree is basically an open source e-commerce
system and it has the ability to deal with web services in
obtaining a list of products, product details and cart system.
Step 2: A mobile application gap via PhoneGap (mobile
application development framework). Then product listing and
view page was initiated through managing data from web
services
Step 3: The application was tested on different
Smartphones as follow:
1) Test on iPhone Simulator
2) Test on iPad Simulator
3) Test on Android Simulator
4) Test on actual iPhone device by paying 100$ apple
developer account
B. Software development methodology
Based on the questions developed for this study a general
search was conducted in the first phase of the research.
Software architecture was developed based on this research.
The results of the research enabled the initial scope of the
research to be more precise and achievable. Various software
development methodologies were utilized since software
engineering is a diverse field and encompasses many diverse
factors and contexts. Experimental software engineering was
used in the initial development phase so that risk and
uncertainty were reduced. An iterative software model was
used as it is the best choice for prototype development.
C. Application Architecture
The application architecture is explained in terms of how:
the Spree web application is hosted on the Ruby on Rails
equipped webserver. Web services have been built inside the
Spree application; and the mobile phone native applications
built using PhoneGap which communicates with the webserver
through web services to obtain data and information.
(Figure1). This diagram shows how the application
architecture used the Spree web server and a native mobile
application using PhoneGap which allows the same
application to run in different mobile operating systems. This
corresponds with the web server which uses web services to
obtain data.
Figure 1: Application Architecture
Vol. 3, No.8, 2012

56 | P a g e
III. SCOPE
A. Scope of application
The basic functionality delivered by the application
features two main factors, product listing and a product detail
page.
Two changes were made,
Spree web services were modified in order to show
complete data in our mobile application.
The functionality needed so that the mobile application
can list products by requesting web services provided by
the Spree-commerce application was added.
The functionality of the mobile application was improved
to show complete information on a product by requesting
a second web-service from the Spree-commerce
application.
Best practices were employed for making the user
interface of the mobile application functional.
IV. RESULT
A. User/System scenario
The process starts when a user starts the mobile
application.
The mobile application requests the product listing web
service.
The web application (Spree-commerce application),
which operates on remote web servers receives product
listing web service requests.
The web application finds the published products from
the database and prepares JSON response.
The mobile application receives JSON response and will
convert it into HTML and render it. The mobile
application also makes sure that all links in the product
list should only work as AJAX.
When the user selects a product the mobile application
sends on AJAX request to the server for the product
details web service.
The server finds full details of the product and sends
JSON response.
The mobile application prepares HTML from JSON
response for the display on the screen.
B. iOS and Android Demo Application
The application was tested on Android and iOS
(iPhone/iPad/iPod operating system). The application provided
two web services. Figures 2 and 4 depict the web service
which enables the user to see a list of products available.
Figure 3 shows that when you tap (select) any of these
products, a new detailed page relating to the product selected
will open and display information such as image, colour and
price.
1) iPhone simulator tested

Figure 3: Product detail screen on iPhone

2) Android simulator tested

Figure 4: List of products on Android
Figure 2: List of products on iPhone

Vol. 3, No.8, 2012

57 | P a g e
V. SET UP AND SETTING UP
A. Setting up Spree web application
1) Before starting the Spree application the following
applications must be installed:
a) Unix bases OS
b) Mysql5
c) Ruby 1.8.7 or 1.9.2
d) Ruby on Rails 3.0.9
e) ImageMagick (library for image processing; used for
generating different sizes of images)
2) Once these applications are installed then the following
steps are needed:
a) Copy the ecommerce application code.
b) On UNIX terminal go to ecommerce application
folder
c) Run bundle install [Make sure machine is connected
to internet because this command will install all dependencies
for the application]
d) rake db: create [ It will create MYSQL database]
e) rake db: migrate [ It will create all tables]
f) rake db: bootstrap [ It will load sample data in
database]
g) Rails S [ run web application server]
h) Use frontend and admin panel of Spree application
for exploration
B. Mobile applications
1) Deploy to iOS:
a) Run Xcode and open iOS version of PhoneGap
project
b) Deploy to actual iOS device
Add your device to apple provisioning portal
(remember paid developer account is needed)
Select device to be used in execution list and Click
run button
c) Or run on iPhone simulator
Select iPhone simulator in execution list and click run
button
2) Deploy to Android:
a) Download and install Eclipse Classic.
b) Download and install Android SDK..
c) Download and install ADT Plugin for Eclipse.
d) Download the latest copy of PhoneGap and extract
its contents.
e) Setup Project.
f) Merge our HTML code to be I used.
VI. DISCUSSION RESULT
E-commerce functionality was provided by installing a
Spree application on a web server. The research suggested that
mobile phones such as Android-based Smartphones, iPad and
iPhones were appropriate for installing the PhoneGap
application on them. The research also illustrated that using
JSON-based web services promoted data communication
between PhoneGap and the Spree application. The native
support of JSON in JavaScript and it being comparatively
lightweight was the significant reason why this report
suggested JSON should be selected instead of the XML data
format.
The actual Spree application provided several web services
but the research forewarned that these web services would not
be feasible with every application. For instance, often there
was too much unneeded information on a products listing web
service. For this reason and that of network latency, not only
did an application slow down but in response to the JSON
object. The processing of JSON in JavaScript was increased.
The research also suggested that the required data such as the
ID, title, the short description and slug were obtained by
overriding that particular web service.
As the research elaborated, depending on the screen size of
mobile phones, it was often the case that image sizes for the
Spree application were either too large or too small.
Fortunately, the paperclip library that is available in the Spree
application was able to regenerate all the images after their
smaller sizes were introduced. The research also suggested
that for the UI components of a Smartphone, jQuery and
jQueryMobile should be employed. The research also
illustrated the fact that no image was used for UI components
but rather CSS3 features were used, making this library the
preferred option. The research indicated that jQuery guidelines
were used to create a robust code and jQunit, which is a
jQuery-based unit testing library dedicated to unit testing.
According to the research a PhoneGap template was used
to make an Xcode project once this functionality became
functional on WebKit based browsers such as Chrome and
Safari. CSS, HTML and JavaScript were embedded into this
project template. The research suggested that iPhone and iPad
simulators had to be used for testing until it was definite that
they worked appropriately. Once they indeed worked
appropriately, the application was then tested on the actual
device by purchasing a $100 Apple developer account.
The research highlighted the fact that Android
Smartphones followed a similar process. Android development
and the Eclipse IDE plug-ins had to be installed in this case.
The research suggested that an Android template was then
used to make an eclipse project, merging CSS, HTML and
JavaScript into the template as well. Ultimately, the research
indicated that the Android simulator had to be utilized to
compile and test the application. This was the process used to
execute the application and ensure its success.
VII. CONCLUSION AND FUTURE DIRECTIONS
Mobility of people and technologies are key factors in
todays economy. Mobile applications are of utmost
importance when companies market their products or services.
Mobile phones have generated an incredible opportunity for
accessing the Internet, while m-commerce has increased the
level of using a mobile phone for business. Applications for
electronic and mobile commerce are, however, developed
Vol. 3, No.8, 2012

58 | P a g e
sufficiently to cover all aspects of the market. The ideal
applications do not distract the user from his/her intent and
they provide on appropriately level of security, accessibility
and speed. Screen size, input device, urgency, task based
interfaces, and design are important aspects that developers
have to consider when designing mobile phone applications.
Such applications should be implemented on different
operating systems, whether using JavaScript, CSS3 or
HTML5, or combining these three together. The installation of
a Spree application allows partial web services to function
because not all applications may work with these web
services. The device screen size proved to be of great
importance but the Spree application solved these issues.
Simulators for iPhone and iPad represent an excellent way to
test applications.
A. Future directions
In order to bring this research from its current analysis on
the basic model to an advanced level, the points listed under
the scope of excluded and discussed below must also
considered.
Payment transactions need to be made through a more
secure and safe channel like Pay-Pal, which will ensure a safe
and sound transaction system. Transaction payments should be
password protected or should have a PIN code that confirms
its reliability. Even if the gadget used - for example a cell
phone - is lost, stolen or otherwise incapable of being used,
consumers would not have to worry about theft or
mismanagement of money. Furthermore, they can obtain a
password to improve protection and security.
Products listed on the product information pages should
have paged display pictures, which will be more appealing to
the viewer. The product should be listed with information such
as the product specifications: price, part number, technical
details, features and packaging details. Yet some product
information pages do not provide this information which is
very inconvenient for the consumer and does not present a
professional image.
The applications should allow users to change the web
services URLs. Settings should be made flexible so that users
can improvise and access the web services they require by
changing settings. The application should allow the user to add
multiple sources for the same web server by cell phone so that
the user can code flex, which means that a user can develop
and deploy cross-platform Internet applications being run on
cell phones. This implies that a mobile application should be
able to locate a list of products from different remote
applications, which provide the same web services being
supported by a mobile application.
REFERENCES
[1] Ngai, E.W.T. & A. Gunasekaran (2007). A review for mobile commerce
research and applications. Decision Support Systems, vol. 43, no. 1, pp.
3-15.
[2] Cyr, D., M. Head & A. Ivanov (2006). Design aesthetics leading to m-
loyalty in mobile commerce. Information & Management, vol. 43, no. 8,
pp. 950-963.
[3] [Sumita, Ushio & Jun Yoshii (2010). Enhancement of e-commerce via
mobile accesses to the Internet. Electronic Commerce Research and
Applications, vol. 9, pp. 217-227.
[4] Lee, Ching-Chang, Hsing Kenneth Cheng & Hui-Hsin Cheng (2007). An
empirical study of mobile commerce in insurance industry: Task
technology fit and individual differences. Science Direct, vol. 43, 2007,
pp.95-110.
[5] Wu, Jen-Her & Yu-Min Wang (2006). Development of a tool for
selecting mobile shopping site: A customer perspective. Science Direct,
vol. 5, pp. 192-200.
[6] Wu, Jen-Her & Shu-Ching Wanga (2005). What drives mobile
commerce? An empirical evaluation of the revised technology
acceptance model. Information & Management, vol. 42, pp. 719-729.
[7] Yung-Ming, Li & Yung-Shao Yeh (2010). Increasing trust in mobile
commerce through design aesthetics. Computers in Human Behavior,
vol. 26, pp. 673-684.
[8] Chang, Yung-Fu & C.S. Chen (2005). Smart phonethe choice of client
platform for mobile commerce. Computer Standards & Interfaces, vol.
27, pp. 329-336.
[9] Barnes, Stuart & Eusebio Scornavacca (2007). Chapter 7: The
emergence of mobile commerce. In Stuart Barnes, E-Commerce and V-
Business: Digital Enterprise in the Twenty-First Century, 2
nd
ed. Oxford,
U.K.: Butterworth-Heinemann, pp. 157-178.
[10] Lehtinen, Toni (2011). Native versus Web - which approach is best for
mobile apps?, mobilesolutions, White Paper, 2011, pp. 2-6.
[11] Power, Mark (2011). Mobile Web Apps: A Briefing Paper. Centre for
Educational Technology & Interoperability Standards, March. Mark
Power@CETIS.
[12] Descartes (2010). Why PhoneGap, 4p., Descartes website. Retrieved
July 28, 2011 from http://www.phonegapmobileappdev.com/wp-
content/uploads/Why-PhoneGap.pdf.
[13] Soriano, M. & D. Ponce (2002). A security and usability proposal for
mobile electronic commerce. Communications Magazine, vol. 40, no. 8,
pp. 62-67.
[14] Buranatrived, J. & P. Vickers, P (2002). An investigation of the impact
of mobile phone and PDA interfaces on the usability of mobile-
commerce applications. Proceedings of the IEEE 5th International
Workshop on Networked Appliances. Liverpool: pp. 90-95.
[15] Koukia, Spiridoula, Maria Rigou & Spiros Sirmakessis (2006). The Role
of Context in m-Commerce and the Personalization Dimension.
In International Conference on Web Intelligence and Intelligent Agent
Technology Workshops, 2006. WI-IAT 2006 Workshops. Hong Kong: pp.
267-276.
[16] Leavitt, N. (2010). Payment Applications Make E-Commerce Mobile.
Computer, vol. 43, no. 12, pp. 19-22.
[17] Github Inc. (2011). Github Social Coding: spree, Github Social Coding
website. Retrieved Jun 28, 2011 from https://github.com/spree/spree.
[18] Spree (n.d.). The Worlds Most Flexible E-Commerce Platform, Spree
website. Retrieved August 3, 2011 from http://spreecommerce.com/.
[19] Nitobi (2011). Home page, PhoneGap website. Retrieved August 20,
2011 from http://www.phonegap.com/.
[20] Kurvosky, Stan, Vladimir Zanev & Anatoly Kurkovsky (2005).
SMMART: using context-awareness in m-commerce. In MobileHCI '05:
Proceedings of the 7th International Conference on Human Computer
Interaction with Mobile Services & Devices. New York: ACM, pp. 383-
384.
[21] PocketHacks.com (2011). iOS page, All about Windows Mobile,
Windows Phone 7 and Android devices, latest news, freeware apps and
hacks.. Retrieved August 20, 2011 from http://pockethacks.com/ios/.
[22] ImageMagick Studio LLC (2011). About Imagemagick.
Imagemagick.org website. Retrieved October 5, 2011 from
http://www.imagemagick.org/script/index.php.
[23] Nitobi (2011). How PhoneGap Works. PhoneGap website. Retrieved
August 20, 2011 from http://www.phonegap.com/about.
[24] TechTerms.com (2010). Android. TechTerms website. Retrieved
September 17, 2011 from http://www.techterms.com/definition/android.
[25] PC Magazine (2011). Definition of AJaX. PC Magazine website.
Retrieved August 11, 2011 from
http://www.pcmag.com/encyclopedia_term/0,2542,t=AJaX&i=55346,00
.asp#fbid=cGwEB1ODoJ6.
[26] The jQuery Project (2011). Homepage, jQuery Mobile Framework
website. Retrieved September 5, 2011 from http://jquerymobile.com/.
[27] Internet.com. (2011). jQuery, in Webopedia website. Retrieved
September 7, 2011 from
http://www.webopedia.com/TERM/J/jQuery.html.
Vol. 3, No.8, 2012

59 | P a g e
[28] About.com (2011). JSON: JavaScript Object Notation, About.com
website. Retrieved August 3, 2011 from
http://javascript.about.com/library/bljson.htm.
[29] Motive Ltd. (2010). The Motive Internet Glossary: MySQL. Motive
Glossary website. Retrieved August 3, 2011 by
http://www.motive.co.nz/glossary/mysql.php.
[30] Howard, Alexander B. (2006). Definition: Ruby on Rails (RoR or Rails),
SearchSOA.com website. Retrieved July 3, 2011 from
http://searchsoa.techtarget.com/definition/Ruby-on-Rails.
[31] Linux Information Project (2005). Unix-like Definition [updated June
18, 2006], Linfo website. Retrieved July 1, 2011 from
http://www.linfo.org/unix-like.html.
Vol. 3, No.8, 2012

60 | P a g e
SW-SDF Based Personal Privacy with QIDB-
Anonymization Method

Kiran P
Research Scholar
VTU,Belgaum
Karnataka, India

Dr Kavya N P
Prof & Head
Dept of MCA,RNSIT
Bangalore,Karnataka,India

Abstract Personalized anonymization is a method in which a
guarding node is used to indicate whether the record owner is
ready to reveal its sensitivity based on which anonymization will
be performed. Most of the sensitive values that are present in the
private data base do not require privacy preservation since the
record owner sensitivity is a general one. So there are only few
records in the entire distribution that require privacy. For
example a record owner having disease flu doesnt mind
revealing his identity as compared to record owner having
disease cancer. Even in this some of the record owners who have
cancer are ready to reveal their identity, this is the motivation for
SW-SDF based Personal Privacy. In this paper we propose a
novel personalized privacy preserving technique that over comes
the disadvantages of previous personalized privacy and other
anonymization techniques. The core of this method can be
divided in to two major components. The first component deals
with additional attribute used in the table which is in the form of
flags which can be used to divide sensitive attribute. Sensitive
Disclosure Flag (SDF) determines whether record owner sensitive
information is to be disclosed or whether privacy should be
maintained. The second flag that we are using is Sensitive Weigh
(SW) which indicates how much sensitive the attribute value is as
compared with the rest. Second section deals with a novel
representation called Frequency Distribution Block (FDB) and
QuasiIdentifier Distribution Block(QIDB) which is used in
anonymization. Experimental result show that it has lesser
information loss and faster execution time as compared with
existing methods.
Keywords- Privacy Peserving Data Mining(PPDM);Privacy
Preserving Data Publishing(PPDP); Personal Anonymization.
I. INTRODUCTION
Personal information present in different organizations can
be used by research for understanding patterns there by
achieving betterment of the community. For example a
personal detail of the patient is present in different hospitals,
this information can be used by researchers to understand the
patterns for a particular disease and hence improve the
identification of the diagnosis. The raw data present in
hospitals contain detailed information regarding the patient
like name, address, DOB, zip code, symptoms & disease.
From this raw data, details regarding name and address which
are considered personal are removed before it is given to Data
Recipient and this information is also called Microdata. This
microdata however contains details like zip, DOB that can be
linked with other external publicly available data bases for re-
identification of sensitive value.
This re-identification of the record by linking public data
to Published data is called as linking attack. For example
consider the details of the patient Published by the hospital in
table 1, which does not contain details regarding name,
address and other personal information. The attacker can use
the publicly available external data base shown in table 2 and
join these details with table 1 thereby personal details can be
revealed. The query may look like
SELECT NAME, DISEASE
FROM VOTERS_TABLE AS V, PAIENT_TABLE AS P
WHERE V.ZIP=P.ZIPAND V.AGE=P.AGE;

The result of this query gives me entire details regarding
sensitive information i.e. disease and the identity of the
individual which is of great concern because the individuals
are not ready to share their sensitive information. The join
may give me a value <RAMA, Gastric ulcer > for zipcode
48677 & age 26 and is called Record Level Disclosure. The
approaches used by researchers to mask sensitive data from
Data Recipients come under a category called Privacy
Preserving Data Publishing (PPDP). Attributes present in
Published Patient Data that can be linked to external publicly
available data bases like ZIP, DOB, are called Quasi-
Identifier (Q) attributes.
TABLE 1. PATIENT PUBLISHED DATA
ZIP Code Age Disease
48677 26 Gastric ulcer
48602 28 Stomach cancer
48678 32 Flu
48685 36 Flu
48905 42 Flu
48906 46 Flu
48909 43 Flu
48673 48 Heart Disease
48655 58 Stomach_cancer

Modification of data is done in such a way that the
resultant table has duplicated records there by restricting the
disclosure. Indirectly there must be more than one link to the
Vol. 3, No.8, 2012

61 | P a g e
external data base and is done by using generalization [1, 2, 3,
4]. Once the table is generalized various methods were used to
check the property of duplication and distribution. To measure
this Samarati and Sweeney [6,7] introduced k-anonymity. A
table satisfies k-anonymity if every record in the table is
indistinguishable from at least k 1 other records with respect
to every set of quasi-identifier attributes; such a table is called
a k-anonymous table. In other words each group of quasi
identifier values must have at least k-1 records and can be
cheeked by linking a record in released data to multiple
records publicly available data base. Table 3 shows a 2-
anonymus generalization for table 1. Let us assume that the
attacker uses the publicly available data base and finds that
Ramas zip code is 48677 and his age is 26 and wants to
know the disease of Rama, the attacker observes the
anonymized table 3 from which attacker understands that
48677 & 26 has been generalized to 486** & [20-30] which
can be linked to two records of published table and hence the
disease cannot be inferred. In this table <486**,[40-50],Heart
Disease> has been suppressed and is not considered for
publication. Similarly if the attacker tries to infer Sitas
disease who is related to group 3 but since the entire group
contains the same sensitive attribute the attacker infers that his
disease is Flu. This leakage of sensitive value leads to
Attribute Level Disclosure. This happens if all the diseases
indicated in a group are related to the same disease. To
overcome this l-diversity [8] was defied. An equivalence class
is said to have l-diversity if there are at least l well-
represented values for the sensitive attribute. A table is said
to have l-diversity if every equivalence class of the table has l-
diversity. l-diversity also has the disadvantage that it suffers
from skewness and similarity attack. To overcome this t-
closeness was defined [9]. In this technique distribution of
sensitive attribute must be equal to the anonymized block.
This suffers from information loss
A. Motivation
Major disclosures that take place are record level and
attribute level to avoid this various anonymity techniques have
been proposed in literature. Among them most important are
k-anonymity, l-diversity, t-closeness, but each of them have
several drawbacks as indicated above it includes data utility
loss and sensitivity disclosure. To overcome this author in [5]
had indicated a method personalized privacy preservation
which takes in to account record owners privacy requirement.
In [5] the record owner can indicate his privacy by indicating
in terms of a guarding node. The values of it are based on a
sensitive hierarchy which is framed by Data Publisher. The
core of this technique is to divide the sensitive value based on
importance so that more privacy is given to those values and
data utility is improved. The drawback of this method is that it
may require several iterations based on the guarding node,
sensitive attribute is also generalized which has larger
information loss. The most important drawback is that
distribution of sensitive attribute has not been taken in to
account while anonymization.
B. Contribution and paper outline
In this paper we propose a novel privacy preserving
technique that over comes the disadvantages of [5] and other
anonymization techniques. The core of this method can be
divided in to two major components. The first component
deals with additional attribute used in the table which is in the
form of flags. Sensitive Disclosure Flag (SDF) determines
whether record owner sensitive information is to be disclosed
or whether privacy should be maintained. The second flag that
we are using is Sensitive Weigh (SW) which indicates how
much sensitive the attribute is. SDF is dependent on SW.
TABLE 2. EXTERNAL VOTERS DATA BASE
Name ZIP Code Age
Rama 48677 26
Laxman 48677 35
Suresh 48602 28
Nagesh 48602 22
Anuma 48678 32
Sita 48905 42
Kushal 48909 43
Vihan 48906 46
.
.
TABLE 3. 2-ANONYMUS TABLE
486** [20-30] Gastric ulcer
486** [20-30] Stomach cancer
486** [30-40] Flu
486** [30-40] Flu
489** [40-50] Flu
489** [40-50] Flu
489** [40-50] Flu
486** [40-50] Heart Disease
486** [50-60] Stomach_cancer

SDF can be easily obtained from the individual when
he/she is providing her data. SW can be based on the prior
knowledge of sensitive attribute. General privacy methods
provide the same level of security for all sensitive attributes
which has been overcome in this method by the use of SDF
and SW. The flag SDF=0 means that the record owner is not
ready to disclose his sensitive attribute whereas SDF=1
doesnt mind revealing his sensitivity. SW is indicated by the
publisher for those Sensitive attribute where privacy is at most
important. For example record owner who has Flu or Gastritis
doesnt mind revealing his identity as compared to a record
owner who has Cancer. The value of SW=0 is used when the
sensitive attribute is a common disease like Flu or Gastritis
and SW=1 for sensitive attribute like Cancer which is not
common. For SW=0 default value of SDF=1 & if SW=1 SDF
values are accepted from record owner.
Second section deals with a novel representation called
Frequency Distribution Block (FDB) and QuasiIdentifier
Distribution Block (QIDB) used for measuring the
distribution. FDB contains distribution of every disease with
respect to original private data. For every record with SW=1
and SDF=0 QIDB is created. There will be multiple QIDB
Vol. 3, No.8, 2012

62 | P a g e
blocks. These blocks are used to ensure that distribution of
FDB is matched with individual QIDB.
In section II we have indicated Model and Notations used
in our Personalized Privacy. Personalized Privacy Breach has
been discussed in section III. Section IV gives the QIDB-
Anonymization Algorithm. Experiment in section V has been
analyzed. Related work has been discussed in section VI. Last
section deals with conclusion and future work.
TABLE 4. SW FOR DISEASES
Disease SW
Gastric ulcer
0
Stomach cancer
1
Flu
0
Heart Disease 1
TABLE 5. PATIENT PUBLISHED DATA WITH SW & SDF
ZIP Code Age Disease SW SDF
48677 26 Gastric ulcer 0 1
48602 28 Stomach cancer 1 0
48678 32 Flu 0 1
48685 36 Flu 0 1
48905 42 Flu 0 1
48906 46 Flu 0 1
48909 43 Flu 0 1
48673 48 Heart Disease 1 1
1 0
48655 58 Stomach_cancer
1 1
TABLE 6. FREQUENCY DISTRIBUTION BLOCK
Disease Probability
Gastric ulcer 0.1
Stomach cancer 0.2
Flu 0.5
Heart Disease 0.2
II. MODEL AND NOTATION FOR PERSONALIZED PRIVACY
Let T be a relation containing private data about a set of
individuals. there are four categories of attributes in T i)
unique Identifiers UI
i
which can be used for identification of a
person and is removed from T ii) quasi identifiers Q
i
whose
values can be used for revealing the identity of a person by
joining Q
i
with publicly available data iii) sensitive attributes
S
i
which is confidential or sensitive to the record owner.
iv)Non quasi identifiers NQ
i
which do not belong to the
previous three categories.
Objective of our approach is to find a generalized table T*
such that distribution of each QIDB is approximately equal to
the diversity of the overall distribution which is there in FDB.
For simplicity the entire quasi identifiers are represented as Q
and their values as q. similarly we assume there is a single
sensitive attribute S and its value is s. Relation T is made of n
number of tuples T={t
1
,t
2
,,t
n
}. Record owner information
can be retrieved by referring as t
i
.s to indicate sensitive value
and t
i
.q for quasi identifier value 1 .
A. Requirement for personal privacy
DEFINITION 1 (SENSITIVE WEIGHT) For each tuple t
T, its sensitive weight is added. This value is taken from
Relation W(d,sw) where d disease and sw sensitive weight. W
contains k records.
t
i
.sw={ w
j
.sw if w
j
.d=t
i
.s 1 }
For example table 4 shows the sw value for each disease.
This distribution is taken from Table 1.
DEFINTION 2 (SENSITIVE DISCLOSURE FLAG) for
each tuple t T, its sensitive Disclosure Flag is indicated as
t.sdf.
t
i
.sdf={1 if t
i
.sw=0
ud t
i
.sw=1 }

ud represents user defined and the value is either 0 or 1.
t
i
.sdf=0 then user is not ready to disclose his information and
t
i
.sdf=1 then user is ready to disclose his information. In table
5 value of sw and sdf are indicated assuming that sdf value is
accepted from record owner for SW=1. We can also observe
that if sw=0 its correspondent sdf is initialized to 1 indicating
that the sensitivity of this record is not of much relevance.
B. Thresholds for Personalized Privacy
Threshold values are defined for various dimensions of
personalized privacy to improve the overall performance of
generalization, suppression and disclosure.
i)
n
minimum number of records in T.
ii)
iter
maximum number of iterations that must be
performed .it indicates the amount of generalization &
Height(VDH)
iii)
suppr
minimum number of sensitive values for
suppression.
iv)
disc
minimum number of sensitive values for
disclosure.
v)
acct
minimum threshold that can be added or
subtracted.
Since we are considering the distribution aspect we can
indicate different threshold values. The first value indicates
the minimum number of tuples that must be present for
applying anonymization which was never defined in the
previous representations. T
iter
based on the knowledge of
the height of Value domain hierarchy. The larger the value of
T
iter
higher the generalization and consequently information
loss is more.
suppr
indicates the minimum number of
sensitive distribution that may be there in QIDB for removal
of that block after T
iter
.
disc
indicates the threshold value
that can be added or subtracted to each frequency distribution
for each disease such that it is equivalent to the distribution
FDB. The frequency of QIDB block and FDB will not be
exactly same so while checking the distribution of each
disease is checked whether the frequency in that
qidb.v.sT
acct
always
disc
T
acct
.
Vol. 3, No.8, 2012

63 | P a g e
C. Additional Block Creations for personal privacy
DEFINITION 3 (FREQUENCY DISTRIBUTION
BLOCK) Distribution of each w
j
.d with respect to the original
distribution t
i
.s is saved in relation FDB(d,p) where d indicates
disease and p indicates probability distribution of it. Each p for
d is calculated by mapping each d in T (values of t
i
.s=fdb
u
.d)
to the total number of tuples in T i.e. n, . let us
assume there are m records in the relation.
DEFINITION 4 (QuasiIdentifier Distribution Block ) for
each t
i
.s where t
i
.sw=1 & t
i
.sdf=0 a new QIDB is created
containing t
i
.s . The relation QIDB.V(q,s) where
qidb.v
l
.q=t
i
.q & qidb.v
l
.s=t
i
.s. Let us assume there are dn
QIDB blocks.
For example Table 6 shows the frequency distribution of
each disease. This distribution shows that the disease flu is a
common disease so its frequency is more, around 50% in the
published data. The same distribution is maintained in each of
the QIDB. In the first iteration two blocks of QIDB will be
created for the qasi value <48602,28> and <48607,55> since
its SW=1 & SDF=0 which is shown in table 7 & 8.
TABLE 7. QIDB.1 CONTENTS
48602 28 Stomach cancer
TABLE 8. QIDB.2 CONTENTS
D. Functions For Personal Privacy
DEFINITION 5 (GENERALIZATION) A general domain
of an attribute T.Q is given by a generalization function.
Given a value t.q in the original domain, function returns a
generalized value within the domain.
For each t we use t* to represent its generalized tuple
in T*.we denote it as G(T)
This is similar to earlier representations let us assume that
Domain Generalization Hierarchy and Value Generalization
Hierarchy are defined for each Quasi Identifiers. The distance
vector of quasi attributes has also been generated. In figure 1
Value and Domain Generalization Hierarchy of zipcode has
been indicated. Age is also generalized similarly. Distance
vector is calculated which is shown in figure 2.

DEFINITION 6 (CHECK FREQUENCY) for any QIDB,
we check CF(QIDB.V

) wither QIDB.V frequency of
distribution is equal to the frequency distribution in FDB. It is
done as follows
Let c be the no of records in QIDB.V. for each
UNIQ(qidb.v
l
.s) find total no of mappings which match
qidb.v
l
.s to the no of records i.e. c in QIDB.V, thus CF will
return true if
such that fdb
u
.d=qidb.v
l
.s
fdb
u
.p
acct

this is checked in every iteration if a QIDB satisfies the
frequency distribution then this block will not be considered
for the next iteration.

Figure 1. An example of Value and Domain generalization hierarchy for
zipcode and Age

Figure 2. Hierarchy DGH<Z0,A0> and corresponding Hierarchy of distance
vectors
DEFINITION 7 (SUPPRESSION) After
iter
iterations,
SUPP(QIDB.v) suppress the block if it satisfies the following
condition
such that for every fdb
u
.d=qidb.v
l
.s ^
fdb
u
.d=w
j
.d ^ w
j
.sw=1 j 1

suppr
DEFINITION 8 (DISCLOSURE) After
iter
iterations,
DIS(QIDB.v) adds additional records if it satisfies the
following condition
such that for every fdb
u
.d=qidb.v
l
.s ^
fdb
u
.d=w
j
.d ^ w
j
.sw=1 for some j 1

disc
fdb
u
.p
III. PERSONALIZED PRIVACY BREACH
Consider an attacker who attempts to infer the sensitive
data of a record owner x. the worst case scenario assumes that
the adversary knows Q of X, therefore the attacker observes
only those tuples t*T
*
whose Q value t
i
*.q covers x.q for all i
such that 1 . These tuples form a Q-group. That is, if
t
i
* and t
ip
* are two such tuples then t
i
*.q=t
ip
*.q for all i such
that 1 .if this group is not formed the attacker cannot
infer sensitive attribute of x.
<Z0,A0>
[2,2]

[2,1] [1,2]

[2,0] [1,1] [0,2]

[1,0] [0,1]
[ 0,0]

<Z2,A0> <Z1,A1> <Z0,A2>
<Z1,A0> <Z0,A1>
<Z2,A1> <Z1,A2>
<Z2,A2>
<Z0,A0>
<Z2,A2>
<Z0,A0>
486**
4860* 4867*
48602 48607 48677 48678

[20-30] [30-40] [40-50] [50-60]
Z0={48602,48607,48677,48678}
Z1={4860*,4867*}
Z2={486**}
[20-40] [40-60]
[20-40]

A0={[20-30],[30-40],[40-50],[50-
60}]
A1={[20-40],[40-60]}
A2={[20-60]}
Vol. 3, No.8, 2012

64 | P a g e
DEFINITION 9 (REQUIRED Q-GROUP/ ACT(X)).
Given an individual x, the Required Q-group RG(X) is the only
Q-group in t* covers x.q. let us assume ACT(X) refers to those
records which are generalized to RG(X).
ACT(X) is unknown to the attacker. To obtain ACT(X), the
attacker must find some external data base EXT(X) that must
be covered in RG(X).
DEFINITION 10(EXTERNAL DATA BASE EXT(X))
EXT(X) are set of individuals whose value is covered by
RG(X)
In general ACT(X) EXTX
The attacker adopts a combinational approach to infer
sensitive attribute of x. let us assume that x.s is present in
one of ti
*
and the repetition of x is not present. The possible
reconstruction of the RG(X) includes
r distinct record owners x1,x2,x3,,xr who belong to
EXT(X) are taken but there can be only y in RG(X). this can
be understood by the probabilistic nature and can be
indicated as perm(r,y). perm(r,y) is Possible
Reconstruction(PR) that can be formed by using r owners
and y mappings. Breach Probability (BP) indicates the
probability of inferred knowledge. Let us assume ACTN
indicates actual number of records with sensitive attribute
that can be inferred to x.
BP=
,

BP will decide the privacy parameter, BP is 100% then x
can be inferred if it is very low than the inference will be very
much difficult for the attacker.
IV. QIDB-ANONYMIZATION ALGORITHM
In this algorithm we are using a sequential processing of
quasi values since the assumption is that in each region usually
the distribution of sensitivity is approximately same. The
algorithm is as follows
Algorithm QIDB-Anonymization
Input: private data T with SW-SDF, threshold values
n
,
iter
,
suppr
,
disc
, TH
acct

and initialized
FDB(d,p)
Output: publishable table T
*
1. if (n<
n
) then return with 1
2. for every t
i
.s where t
i
.sw=1 & t
i
.sdf=0 a new QIDB is
created containing t
i
.s and t
i
.q .
3.ini_itr=0, accept_flag=0 and gen=first G(T)
4.while (ini_itr<
iter
and accept_flag=0)
4.1.QIDB blocks are removed if CF() returns true
then check the number of QIDB if it is equal to zero then
accept_flag=1
4.2.itr=itr+1 and gen=next G(T)
5. if accept_flag=0 then invoke supp() & dis()
6. check number of QIDB if it is equal to zero
accept_flag=1
7. publish T* if accept_flag=1
The resultant anonymization after applying Personal
Anonymization of one of the QIDB with
acct
=0.1 block is
shown in Table 9.
TABLE 9. RESULTANT SW-SDF BASED QIDB-
ANONYMIZATION WITH
ACCT
=0.1
486** [20-40] Stomach cancer
486** [20-40] Gastric ulcer
486** [20-40] Flu
486** [20-40] Flu
V. EXPERIMENTS
In this section we try to evaluate the effectiveness of our
technique as compared to k-anonymity and l-diversity. We
have used a standard dataset used in the literature[7,8,9] for
our experiment. We have considered Americal adult dataset of
400 records, with the following quasi attributes Age,
Education , Maritial status & Occupation. The attribute age is
numerical and the rest of the attributes are categorical. The
sensitive attribute income has been converted to disease.
Probability is used to find SDF value for SW=1.
We have defined and used generalization hierarchy for
each qasi identifier and distance vector is generated which has
been used in our algorithm. The maximum height of our
generalization hierarchy is 10. Information loss parameter is
shown in figure 3. Less the information loss better is the data
quality. Minimal distortion (MD)is based on charging penalty
for each value which is generalized or suppressed. Each
hierarchy is assigned a penalty when it is generalized to the
next level with in the domain generalization hierarchy. MD is
shown in figure 4. In our experiment we have used a penalty
of 10 for every generalization. This Discernibility Metric
(DM) calculates the cost by charging a penalty to each tuple
for being indistinguishable from other tuples which is shown
in figure 5. Execution time is shown in figure 6. For our
experiment the threshold values
n
=400,
iter
=10,
suppr
=1,
disc
=0.01 and
T
acct
=0.1 was used. Experiment was conducted using
Matlab 7 in which our algorithm out performs k-anonymity
and l-diversity.
VI. RELATED WORK
Different methods of PPDM exist, among them the most
important are Randomization Method [13], Data Swapping
[14], Cryptographic Approach [15] and Data Anonymization.
Data Anonymization is considered as one of the most
important anonymization technique since it has lesser
information loss and higher data utility. There are different
anonymization algorithms has been proposed in literature [1,
3, 4, 6, 10, 11, 12]. Initial anonymization algorithm was called
k-anonymity [6] but the drawback of this approach is that it is
prone to record level disclosure. To overcome this
Vol. 3, No.8, 2012

65 | P a g e
disadvantage l-diversity[8] was proposed. Disadvantage is that
it is prone to Skewness and Back ground Knowledge Attack. t-
closeness[9] is used to overcome the disadvantages of l-
diversity but it has larger information loss. Personalized
Privacy[5] was added on to anonymization which gave lesser
information loss. This is the motivation of our approach.

Figure 3. Information Loss Of SW-SDF Personal Anonymization As
Compared With K-Anonymity & L-Diversiy

Figure 4. Minimal Distortion Parameter Of SW-SDF Personal Anonymization
As Compared With K-Anonymity & L-Diversiy
VII. CONCLUSIONS AND FUTUREWORK
Personalized privacy is an important research direction in
PPDP since its data quality and execution time is less. Usage
of SW not only improves the indication of sensitivity as the
entire records do not require privacy but also improves the
data utility. SDF is an additional flag which once again
improves data utility with in SW record since some of the
record owners are ready to reveal their identity. Thus the
combination of SW-SDF is a better option for personalized
privacy as compared to just using a guarding node.
QIDB based anonymization allows different quasi group to
be generalized independently. In this approach each quidb
block is checked for the frequency distribution of sensitive
value approximately equal to the frequency distribution of the
sensitive value in original contents thereby improving privacy.
It also overcomes record linkage, attribute linkage and
even probabilistic attack. This approach works well when the
frequency distribution of a particular sensitivity is
concentrated within a region of individual pattern.

Figure 5. Discernibility Metric Parameter of SW-SDF personal anonymization
as compared with k-anonymity & l-diversiy

Figure 6. Execution Time of SW-SDF personal anonymization as compared
with k-anonymity & l-diversiy
There are several future research directions along the way
of analyzing SW-SDF personal privacy with QIDB
anonymization. First we havent considered the effect of
Sequential Release and Multiple Release of published data.
Research on giving different Weight on sensitivity can be
considered. In this approach we have used sequential
processing of records to check the generalized record matches
with QIDB generalized value if they are same then it would be
included in the block. Instead of sequential processing
alternative methods can be looked in to. This method can be
extended to unstructured schema and multi-dimensional data.
REFERENCES
[1] Lefevre K., Dewitt D. J. and Ramakrishnan R., Incognito: Efficient
full-domain k-anonymity. In Proceedings of ACM SIGMOD. ACM,
New York, pp. 4960, 2005.
[2] Fung B. C. M., Wang K. and Yu P. S., Anonymizing classification data
for privacy preservation. IEEE Trans. Knowl. Data Engin, pp. 711
725,2007.
[3] Fung B. C. M.,Wang K. and Yu P. S., Top-down specialization for
information and privacy preservation. In Proceedings of the 21st IEEE
International Conference on Data Engineering (ICDE). 205216, 2005.
[4] Iyengar V. S., Transforming data to satisfy privacy constraints". In
Proceedings of the 8th ACMSIGKDD. ACM, New York, pp. 279
288,2002.
[5] Xiao X. and Tao Y., Personalized privacy preservation. In Proceedings
of the ACM SIGMOD Conference.ACM, New York, 2006.
[6] P. Samarati, Protecting Respondents Privacy in Microdata Release.
IEEE Trans. on Knowledge and Data Engineering (TKDE), vol. 13,
no.6, pp. 1010-1027, 2001.
100 150 200 250 300 350 400
200
400
600
800
1000
1200
1400
1600
No Of Records
I
n
f
o
r
m
a
t
i
o
n

l
o
s
s
Information Loss Parameter
k-anonymity
l-diversity
Sw-Sdf
100 150 200 250 300 350 400
0
0.5
1
1.5
2
2.5
3
3.5
x 10
4
No Of Records
M
i
n
i
m
a
l

D
i
s
t
o
r
t
i
o
n
Minimal Distortion Parameter
k-anonymity
l-diversity
Sw-Sdf
100 150 200 250 300 350 400
0
0.5
1
1.5
2
2.5
3
x 10
4
No Of Records
D
i
s
c
e
r
n
i
b
i
l
i
t
y

M
e
t
r
i
c
Discernibility Metric Parameter
k-anonymity
l-diversity
Sw-Sdf
100 150 200 250 300 350 400
0
2
4
6
8
10
12
14
No Of Records
T
i
m
e
Execution Time
k-anonymity
l-diversity
Sw-Sdf
Vol. 3, No.8, 2012

66 | P a g e
[7] L. Sweeney, k-Anonymity: A Model for Protecting Privacy. Intl
J.Uncertain. Fuzz., vol. 10, no. 5, pp. 557-570, 2002.
[8] Machanavajjhala A, Gehrke J, Kifer D and Venkitasubramaniam M, l-
diversity: Privacy beyond k-anonymity. In Proceedings of the 22nd
IEEE International Conference on Data Engineering(ICDE), 2006.
[9] Ninghui Li , Tiancheng Li , Suresh Venkatasubramanian, t-Closeness:
Privacy Beyond k-Anonymity and lDiversity. ICDE Conference,
2007.
[10] R. Bayardo and R. Agrawal. Data privacy through optimal k-
anonymization. In ICDE, pp. 217228, 2005.
[11] K. LeFevre, D. J. DeWitt and R. Ramakrishnan, Mondrian
multidimensional k-anonymity. In ICDE, 2006.
[12] K. Wang, P. S. Yu, and S. Chakraborty, Bottom-up generalization: A
data mining solution to privacy protection. In ICDM, pp. 249
256,2004.
[13] Agrawal D. Aggarwal C. C., On the Design and Quantification of
Privacy- Preserving Data Mining Algorithms. ACM PODS Conference,
2002.
[14] Fienberg S.,McIntyre J., Data Swapping: Variations on a Theme by
Dalenius and Reiss. Technical Report, National Institute of Statistical
Sciences, 2003.
[15] Pinkas B.,Cryptographic Techniques for Privacy-Preserving Data
Mining. ACM SIGKDD Explorations, 2002.

Vol. 3, No.8, 2012

67 | P a g e
Integration of data mining within a Strategic
Knowledge Management framework:
A platform for competitive advantage in the Australian mining sector

Sanaz Moayer
Murdoch University
Perth, Australia
Scott Gardner
Murdoch University
Perth, Australia

Abstract In todays globally interconnected economy,
knowledge is recognised as a valuable intangible asset and source
of competitive advantage for firms operating in both established
and emerging industries. Within these contexts Knowledge
Management (KM) manifests as set of organising principles and
heuristics which shape management routines, structures,
technologies and cultures within organisations.When employed as
an integral part of business strategy KM can blend and develop
the expertise and capacity embedded in human and technological
networks. This may improve processes or add value to products,
services, brands and reputation. We argue that if located within a
suitable strategic framework, KM can enable sustainable
competitive advantage by mobilising the intangible value in
networks to create products, processes or services with unique
characteristics that are hard to substitute or replicate. Despite
the promise of integrated knowledge strategies within high
technology and professional service industries, there has been
limited discussion of business strategies linked to Knowledge
Management in traditional capital intensive industries such as
mining and petroleum. Within these industries IT-centric
Knowledge Management Systems (KMS) have dominated, with
varying degrees of success as business analysis, process
improvement and cost reduction tools.
This paper aims to explore the opportunities and benefits arising
from the application of a strategic KM and Data Mining
framework within the local operations of large domestic or
multinational mining companies, located in Western Australia
(WA). The paper presents a high level conceptual framework for
integrating so called hard, ICT and soft, human systems
representing the explicit and tacit knowledge embedded within
broader networks of mining activity. This Strategic Knowledge
Management (SKM) framework is presented as a novel first step
towards improving organisational performance and realisation of
the human and technological capability captured in
organisational networks. The SKM framework represents a
unique combination of concepts and constructs from the
Strategy, Knowledge Management, Information Systems, and
Data Mining literatures. It was generated from the Stage 1-
Literature and industry documentation review of a two stage
exploratory study. Stage 2 will comprise a quantitative case based
research approach employing clearly defined metrics to describe
and compare SKM activity in designated mining companies.
Keywords- Knowledge Management (KM); data mining, sustainable
competitive advantage; Strategic Knowledge
Management (SKM) framework; integration; hard and soft
systems; Australian mining organisation.
I. INTRODUCTION:LINKING BUSINESS STRATEGY AND
KNOWLEDGE MANAGEMENT
In a complex and challenging world, organizational
success depends on the ability to configure corporate assets to
reflect rapidly shifting markets and environmental conditions
(Hafizi & Nor Hayati, 2006). In recent years the competitive
significance of tangible assets such as financial capital,
technology or inventories has declined in the disrupted,
globally interconnected markets of early 21
st
century. This has
led to an increased focus on strategic deployment of unique,
hard to imitate, intangible assets such as expert knowledge
which as a basis for sustainable competitive advantage (Hafizi
& Nor Hayati, 2006). (Shih, Chang, & Lin, 2010).
Knowledge is actionable information which helps people
to make better decisions and to be more creative in their
approach to a range of problem solving activities. Many
organizations seek to identify, stocks of knowledge embedded
in their human, information and communication networks and
link them to value adding flows, using Knowledge
Management. Knowledge Management(KM) is defined by
Jashapara (2011, p14) as- The effective knowledge processes
associated with exploration, exploitation and sharing of human
knowledge (tacit and explicit) that use appropriate technology
and cultural environments to enhance an organizations
intellectual capital and performance. Knowledge
Management (KM) processes help organizations to define,
select, organize, distribute, and transfer information,
knowledge and expertise retained in the organizations
memory in an unstructured manner (Turban & Leidner,
2008).Effective KM improves operational efficiency, enhances
products and services and creates customer satisfaction (Lee,
2009). Knowledge Management has many potential benefits
such as retaining expertise, capturing and sharing best
practice, corporate support, improved customer service, better
decision making, increased profitability and competitive
advantage (Duvall, 2002). Over the past fifteen years KM has
been increasingly recognised in the strategy, management and
information systems literature, with senior managers treating it
as a focal point for improving organizational performance,
adding value to goods and services, building brands and
reputation.
Vol. 3, No.8, 2012

68 | P a g e
While KM is often associated with professional services,
biotechnology, IT and internet businesses, it also has
significant potential for adding value and reducing costs in
mining and allied industries. Mining organizations employ
many skilled knowledge professionals in science, engineering,
and technology, including geologists and other earth scientists.
Computer scientists also have a key role in mining
organizations working with managers from different
functional areas to integrate and exploit the knowledge
capacity of these companies (Fernandez, 2010). Employing
high value knowledge workers and retaining knowledge in-
house has a strong economic justification. The migration
towards Strategic Knowledge Management systems and
practices is inevitable part of this process (Fernandez, 2010).
II. CONVERTING DATA INTO ACTIONABLE KNOWLEDGE
In the IT literature Data Mining (DM) is often represented
uncritically as a powerful tool for knowledge discovery. As
noted by (Lee, 2009) it is helpful for discovering patterns of
data and creating new information. Arguably Data Mining
gives organisations the ability to exceed their goals and
enables decision makers to deploy the results (Noonna, 2000),
for improving business performance in highly competitive
environments. These statements point to the practical benefits
and applications of data mining as a knowledge generation and
decision support tool. However there has been little discussion
in the literature of how DM processes can be integrated into
dynamic business strategies supported by KM organizing
principles, management routines, structures, information
systems and organisational culture. Malhotra highlights the
dynamic connectivity between strategy, change, human and
information systems stating that- Knowledge Management
caters to the critical issues of organisational adaptation,
survival, and competence in the face of increasingly
discontinuous environmental change(It) seeks a synergistic
combination of the data and information processing
capabilities of information technology and the creative and
innovative capacity of human beings (Malhotra 2000&2001
in Haslinda and Sarinah 2009).
Just how to achieve these synergies between hard
(technological) and soft (human) systems remains one of the
central questions yet to be addressed in organisational studies.
Managing the interface between people and technology within
complex organisations often manifests as a black art for even
the most adept computer scientist or organisational
psychologist. The ability to frame or ask the right questions
and arrive at satisfactory answers is deeply rooted in our own
worldviews. It is subject to our interpretation of KM concepts
drawn from philosophy and the information, cognitive, social
sciences. As observed by Liebowitz and Megbolugbe (2003)
converting data into actionable information or useful
knowledge in organisations is not a simple, mechanistic
process. It is dependent on heuristic rules and the broader
ontologies or worldviews of key actors within the system
(Liebowitz & Megbolugbe, 2003, p. 189).The Strategic
Knowledge Management (SKM) framework presented in this
paper (Figure 1) acknowledges this complexity while outlining
key elements and broad interrelationships, which subject to
further empirical investigation may advance KM and DM
practice.
III. OBJECTIVES, SCOPE AND METHOD OF THE PAPER
The aim of this paper is to present a novel Strategic
Knowledge Management (SKM) framework (Figure1), as a
platform for building competitive advantage within a mining
industry context. The defining characteristics and claimed
benefits of Knowledge Management and Data Mining are
discussed in relation to the SKM framework incorporating
four related perspectives on strategic management. These
characteristics have been derived from a limited review of the
Strategic Management, Knowledge Management and Data
Mining literatures, with relevant academic sources and reports
profiling the Australian mining industry. This review of
concepts and industry data has been used to produce a high
level representation of strategic management and Data Mining
processes applicable to the Australian based operations of
large mining companies. It represents Stage 1 of two stage
study combining a a high level SKM framework generated
from a review of relevant strategy, KM and DM concepts with
a quantitative case based research method. The SKM
framework in this paper is presented for peer and industry
stakeholder feedback prior to developing a detailed Stage 2
model employing hard metrics. These will be used to describe
measure and compare Strategic Knowledge Management
(SKM) activities displayed in three large mining organisations.
IV. FOUR VIEWS OF STRATEGY AND COMPETITIVE
ADVANTAGE
Strategy is the scope for an organisation to gain benefits
and advantages, with available resources in a challenging
environment in the long-term (Johnson, Scholes, &
Whittington, 2005). Strategic management is the art and
science of formulating, implementing and evaluating
functional decisions that empower the organisation to gain its
goals and objectives (David, 2011). The long term
performance of organisation is defined by a set of managerial
decisions or strategic choices (Hunger & Wheelen, 2003).
There a number of interrelated views of strategy identified
as the basis of competitive advantage in the strategic
management literature. These include: variations on Porters
(1980) economic perspective or Market-Based View (MBV);
Freeman and McVeas (2001) political perspective or
Stakeholder-Based View, (SBV); Barneys (1993) internal
human, structural and capital asset capability perspective, or
Resource-Based View (RBV); and more recently the portfolio
of expertise or Knowledge-Based View (KBV) popularised by
Spender (1996),Grant (1996). Both RBV and KBV have
subsequently been elaborated and linked the theory of
dynamic capabilities (Mouritsen, Larsen, & Bukh,
2005).Dynamic capabilities theory focuses on the firms
ability anticipate and adapt to dynamic, discontinuous or
disruptive market conditions. This is achieved when managers
synergistically combine portfolios of knowledge assets with
organizational learning routines, sense making and strategy to
process market signals and anticipate emerging conditions
(Choo, 1998). These actions serve to renew organizational
structures and systems whilst generating unique assets as a
basis for competitive advantage. This follows the
Schumpeterian logic of competition based on creative
destruction of existing resources and novel combinations of
Vol. 3, No.8, 2012

69 | P a g e
new functional competences (Pavlou and El Sawy, 2004 in
Easterby-Smith and Prieto, 2008, p 236).
All four strategic perspectives and elements of dynamic
capabilities theory incorporated into the SKM (Figure 1), are
particularly applicable to multinational mining companies,
which can exercise control over local, regional, and global
markets and supply chains. The local operations of these larger
companies can also develop a strong internal human resource
profile and workforce capability through attractive salary
packages or significant investments in workforce planning,
training, information technology expertise and infrastructure.
The Market-Based View focuses on achieving an attractive
position within a designated industry. (Poser, 2003). It can
help mining companies to exercise strategic choice and
identify which factors of production should be prioritised to
gain competitive advantage in specific industry structures or
market segments (Porter, 1980). The Stakeholder-Based View
acknowledges the political dimensions of strategy. It
highlights the importance of working with constituents or
stakeholders to facilitate the achievement of business goals
and competitive advantage through informed decision making.
It proposes that managers formulate and implement political
processes that identify, classify and build productive
relationships with people who have stake in the business
(Freeman & McVea, 2001); (Gardner, 2001). The Resource-
Based View emphasises that organisational performance
depends on internal resource configurations and capabilities
including physical resources, human resources, and
organisational resources (David, 2011). By extension the
Knowledge-Based View is built on the logic of the Resource-
Based View. It revisits many tenets of knowledge conversion
and creation. It also identifies organizational learning and
management routines as potential sources of competitive
advantage (Jashapara, 2011). Taking the Knowledge-Based
View (KBV) of strategy, knowledge is a valuable resource and
basis for competitive advantage in organizations. We argue
that this view is particularly applicable to the emerging
knowledge and high technology sectors and the traditionally
capital intensive industry sectors, such as mining and
petroleum.
V. THE AUSTRALIAN MINING INDUSTRY
Mining is a major industry in Australia. One third of the
words mineral resources are produced in Australia
(Nimmagadda & Dreher, 2009). The Australian mining sector
generated revenue of about $138.8 billion in 2006-7 growing
to $203.9 billion in 2011-12. Mining is expected to generate
about 8.0% of Australias GDP in 2012 with a profit forecast
of $58.3 billion. (IBIS World, 2012). These statistics indicate
that the mining industry plays important role in maintaining
revenue growth in Australia within a global context of
economic slowdown or contraction. No other industry in
Australia has gained a superior significance in economic
development terms (Fernandez, 2010).
The scope of the mining industry includes all operations
for extracting minerals or hydrocarbons. Coal, oil and gas,
metal ore, and non-metallic mineral commodities are products
of this industry (IBIS World, 2012). Exploration, drilling,
production, and marketing are significant business functions in
the resource industry. In recent years the major resource
companies and primary contractors have increasingly
recognised relationships with suppliers, customers, regulators
and other stakeholders as critical determinant of firm and
industry performance. According to Richards (2009), suppliers
in particular can drive organisations to produce new services
in different ways (Richards, 2009). The knowledge-base of
suppliers is an important element to increase performance and
maintain the competitive advantage of firms in the mining
industry (Urza, 2011).
VI. DATA MINING AND ITS APPLICATION IN MINING
INDUSTRIES
Data Mining (DM) is a technique for identifying patterns
and relationships between data in large databases (Lee, 2009).
It also informs strategic and operational decisions in
organisations through dashboards, and interrogation or
scanning of relational databases. Data mining aids
organisational problem solving by employing programs that
can search for patterns and relationships without human
intervention (Paddock & Lemoine, 2012, p. 4). Giudici (2003,
p.2) offers a more complete definition of data mining as:
...the process of selection, exploration, and modelling of large
quantities of data to discover regularities or relations, that are
at first unknown with the aim of obtaining clear and useful
results for the owner of the database.
Data Mining encompasses major tasks such as data
exploration, data archaeology, data pattern processing, data
dredging, information harvesting, and knowledge extraction
(Lee, 2009). Data mining technology is becoming a significant
aspect of strategy for many organisations. It has become major
component of (often complex multi-interface) enterprise
decision support systems (Brusilovsky & Brusilovskiy, 2008).
In general, business problems can be categorised as
structured or unstructured. Statistical analysis is useful for
overcoming structured problems and DM is often employed to
deal with unstructured problems. This capability to interpret
problem characteristics and dimensions makes DM potentially
compatible with the human cognitive processes required to
generate useful context specific, knowledge and address
complex problems. This is consistent with the logic of gaining
competitive advantage through unique processes, products and
services that are hard to replicate. As noted by (Brusilovsky &
Brusilovskiy, 2008) the strategic strength of DM resides in the
ability to deal with unstructured problems because competitors
are not familiar with the characteristics of, or solutions to,
these kinds of problems (Brusilovsky & Brusilovskiy, 2008, p.
131).Data Mining clearly has the potential to deliver
significant benefits if framed within broader KM enabling
architecture and aligned through heuristic questioning and
iteration to business goals and an unfolding strategy process.
The different characteristics of KM enabling architecture are
outlined in the models below. These key elements can most
usefully be incorporated into the SKM framework are then
briefly discussed.
VII. RELEVANT MODELS OF KNOWLEDGE MANAGEMENT
Senior managers, KM and IT specialists within the mining
industry must choose an appropriate model which fits the
Vol. 3, No.8, 2012

70 | P a g e
strategic goals, processes and changing environment of their
organisations. Table 1 below illustrates the different
dimensions of knowledge, creation; individual cognition and
shared learning captured in some of the more widely cited
models from the KM literature: (Dalkir, 2005, pp. 49-72);
(Haslinda & Sarinah, 2009, pp. 189-196); (McAdam &
McCreedy, 1999, pp. 95-98).
The decision to incorporate any of these KM elements into
an organisational Knowledge Management systems and
practices is context or domain dependent. According to
Sanchez and Heene (1997) organisational knowledge and
learning cannot be understood from a purely (top down)
strategic perspective, so organisations should also generate
(bottom up) KM activities based on analysis of the context in
which organisations knowledge is applied (Sanchez & Heene,
1997, p. 12). All the models outlined above contribute a
perspective or position on the nature of knowledge, knowledge
as an asset, and knowledge as a capability, knowledge
enabling structures, cultures, or leadership practices germane
to the SKM framework in Figure 1. However Nonakas (1995)
knowledge spiral model and Hedland and Nonakas (1993)
KM framework are the most pertinent to our discussion of
integrating strategy, KM structures and processes, with data
mining activities. These models focus on surfacing, combining
and actioning tacit knowledge (based on human cognition) and
explicit knowledge (repositories of data and information) to
add value in organisations. This is achieved through the SECI
(Socialisation, Externalisation, Combination and
Internalisation), knowledge conversion process. The SECI
process is in turn enabled by Ba - Nonaka and Tackeuchis
concept of a safe space (or cyberspace). This supports
conversion of knowledge assets into value added products,
processes or services, enabled by Information and
Communication Technology (ICT) infrastructure,
management and teamwork practices (Nonaka & Takeuchi,
1995). The converted knowledge assets are simultaneously
carried up through the organisational structure in a dynamic
spiral to inform senior management decision making and
support the strategy process. This model is most applicable
within project based industries like mining which typically
rely on matrixes superimposed on functional structures to
align staff expertise and capacity with business requirements.
We propose that SKM can overcome these limitations by
using clearly articulated organising principles to drive KM and
Organisational Learning (OL) activities. These are embedded
in management routines, which continuously align human, and
technology interactions, structures, cultural norms and values
with dynamic changes in the competitive environment.
VIII. TOWARDS A STRATEGIC KNOWLEDGE MANAGEMENT
(SKM) FRAMEWORK FOR THE AUSTRALIAN MINING INDUSTRY
Figure 1, below, illustrates the integration of data mining
practices into a SKM framework applicable to global and
Australian based mining operations. It presents a high level
guide to exploiting the knowledge embedded in human and
ICT networks to create process efficiencies, improve decision
making and by extension, productivity for large multinational
miners domiciled in Australia. SKM is premised on the idea
that data mining should not be conducted in isolation from a
broader KM strategy that incorporates the following elements:
1)Simultaneous application of interrelated strategic
perspectives notably: The Market Based View paying
attention to product, price or supply chain concerns; The
Resource-Based View focusing on how to build human
capability and physical asset capacity; The Stakeholder Based
View concerned with building relational capital and the salient
stakeholders who can affect or are affected by the goals and
activities of the firm; and finally the Knowledge-Based View
which emphasises the importance of managing human
networks, knowledge portfolios, stocks and flows as a key
determinant of organisational performance and sustainable
competitive advantage. Adoption of these strategic
perspectives and organisational learning processes links the
firms dynamic capabilities to KM and day to day
management practices. As such SKM supports a dynamic
strategy process which calibrates internal capability with
changes in the external environment.
2) A knowledge enabling architecture based on reciprocal
hard and soft system organising principles. This living
architecture is generated from different ontological positions,
heuristics, and taxonomies. Its design elements are comprised
of transparent organising principles, shared goals and priorities
negotiated between key internal actors representing the tension
between commercial, humanistic and technologically
orientated worldviews. These in turn drive management
routines and practices, information and communications
infrastructure design (including DM tools), organisational
structure, culture and reward systems. Using iterative action
learning loops these first and second order design elements are
continuously re-configured to support knowledge creation and
adaptability to dynamic competitive conditions.
3) Alignment mechanisms: The proposed SKM approach
to action learning is consistent with Nonakas knowledge
spiral and tacit to explicit continuous knowledge conversion
model. Both are generative bottom up approaches. They are
aim to align individual and team behaviours with
organisational structures, rewards and management routines
and broader KM based strategy. This requires ongoing
dialogue between DM and information systems specialists,
vendors and senior managers. This is something akin to the Ba
or a safe physical or virtual space for knowledge sharing. High
trust, open protocol environments of this type is essential for
effective knowledge sharing, problem solving, and
identification of common ground between senior managers
and IT specialists. This safe space and common ground allows
for the surfacing and testing of different ontological
viewpoints and creation of shared heuristics. This precedes the
dialogue on shared organising principles which once agreed
can lead practical discussion of how best to manage data,
information and unique knowledge assets to achieve
competitive advantage for the organisation.This preliminary
framework will be further refined based on peer review, tested
and empirically validated through application to KM and data
mining systems and practices in three West Australian mining
organisations.
A broad inventory of data mining tasks and Strategic
Knowledge Management processes will be created for each
firm as part of the process for testing the model.
Vol. 3, No.8, 2012

71 | P a g e

Model Features
The von Krogh and Roos Model of
organisational Epistemology (Von Krogh &
Roos, 1995)
Individual knowledge
Social knowledge
The Nonaka and Takeuchi Knowledge
Spiral Model (Nonaka & Takeuchi, 1995)
Knowledge creation
Knowledge conversion
(Socialisation, externalisation,
combination, internalisation),
Ba safe space,
Knowledge assets.
Hedlund and Nonakas Knowledge
Management Model (Hedlund & Nonaka,
1993)
Articulated knowledge-
Individual
Tacit knowledge-
Individual
Group
Tacit knowledge- Group
Organisation
Tacit knowledge-
Organisation
Inter- Organisational
Domain
Tacit knowledge- Inter-
Organisational Domain
The Choo Sense-making KM Model (Choo,
1998)
Sense making
Knowledge creation
Decision making
The Wiig Model for Building and Using
Knowledge (Wiig, 1993)
Public Knowledge
Shared experience
Personal knowledge
The Boisot knowledge category Model
(Boisot, 1987)
Propriety knowledge
Personal knowledge
Public knowledge
Common sense
The Boisot I-Space KM Model (Boisot,
1998)
Codified- Uncodified
Abstract- Concrete
Model Features
Diffused- Undiffused
Skandia Intellectual Capital Model of
Knowledge Management (Chase, 1997);
(Roos & Roos, 1997)
Equity
Human Capital
Customer
Capital(Customer Base,
Relationsips, Potential)
Innovation Capital
Process Capital
Demerests Knowledge Management Model
(Demerest, 1997)

Knowledge construction
Knowledge embodiment
Knowledge
dissemination
Use
Frids Knowledge Management Model
(Frid, 2003)
Knowledge Chaotic
Knowledge Aware
Knowledge Focused
Knowledge Managed
Knowledge Centric
Stankosky and Baldanzas Knowledge
Management Framework (Stankosky &
Baldanza, 2001)
Learning
Leadership
Organisation, structure
& culture
Technology
Kogut and Zanders Knowledge
Management Model (Kogut & Zander,
1992)
Knowledge Creation
Knowledge Transfer
Process &
Transformation Of
Knowledge
Knowledge capabilities
Individual Unsocial
sociality
Complex Adaptive System Model of KM
(Bennet & Bennet, 2004)
Creating new ideas
Solving problems
Making decisions
Taking actions to
achieve desired results

Table 1: Overview of Widely Cited Knowledge Management Models
IX. CONCLUSION
Australia is the one of the largest producer of mineral
resources and the mining industry plays a significant role in
national revenue growth. Finding new ways to identify and
activate the knowledge capabilities embedded in human and
technological networks is a critical concern for mining
organisations seeking increased efficiencies, productivity and
competitive advantage in Australian and global markets. The
SKM framework and discussion presented in this paper
represents a useful point of departure for companies pursuing
this goal.
REFERENCES
[1] IBIS World. (2012, May). Retrieved from Mining in Australia: Market
Research Report:
http://www.ibisworld.com.au/industry/default.aspx?indid=55
[2] Barney, J. B. (1991). Firm resources and sustained competitive
advantage. Journal of Management, 17(1), 99-120.
[3] Bennet, A., & Bennet, D. (2004). Organizational survival in the new
world: the inteligent complex adaptive system. A new theoary of the
firm. Burlington: MA:Elsevier Butterworth-Heinemann.
[4] Boisot, M. (1987). Information and Organizations. The Manager as
Anthropologist, , Fontana/Collins, London.
[5] Boisot, M. (1998). Knowledge assets: Securing competitive advantage
in the information economy. Oxford University Press, USA.
[6] Brusilovsky, D., & Brusilovskiy, E. (2008). White Paper: Data mining:
the means to competitive advantage. retreived from
http://www.bisolutions.us/web/graphic/BISolutions-DATA-MINING-
THE-MEANS-TO-COMPETITIVE-ADVANTAGE.pdf.
[7] Chase, R. (1997). The Knowledge based Organization: An International
Survey. Journal of Knowledge Management, 1(1).
[8] Choo, C. (1998). The knowing organization. New York: Oxford
University Press.
[9] Dalkir, K. (2005). Title Knowledge management in theory and practice
. Amsterdam ; Boston : Elsevier/Butterworth Heinemann.
[10] David, F. R. (2011). Strategic Management concepts and cases.
Prentice Hall.
[11] Demerest, M. (1997). Understanding knowledge management. Journal
of long range Planing, 30(3), 374-84.
[12] Duvall, M. (2002). Knowledge Management Vendors Go Vertical.
Intranet journal.
[13] Easterby-Smith, M., & Prieto, I. M. (2008). Dynamic Capabilities and
Knowledge Management: An integrative role for learning. British
Journal of Management, 19, 235-249.
[14] Fernandez, M. C. (2010). Knowledge-intensive service activities in the
success of the Australian mining industry. The Service Industries
Journal, 30(1), 55-70.
Vol. 3, No.8, 2012

72 | P a g e
[15] Freeman, R., & McVea, J. (2001). A stakeholder approach to strategic
management. .
[16] Frid, R. (2003). A Common KM Framework For The Government Of
Canada: Frid Framework For Enterprise Knowledge management.
Canadian Institute of Knowledge Management, Ontario.
[17] Gardner, R. S. (2001, May). Stakeholder and Reputation Management
as a tool for effective governance in corporatised government agencies.
Journal of Contemporary Issues in Business and Government, 7(1), 29-
40.
[18] Grant, R. M. (1996). Toward a knowledge-based theory of the firm.
Strategic Management Jurnal, 17, 109-22.
[19] Hafizi, M. A., & Nor Hayati, A. (2006, September). KNOWLEDGE
MANAGEMENT IN MALAYSIAN BANKS: A NEW PARADIGM.
Journal of Knowledge Management Practice, Vol. 7, No. 3, September
2006, 7(3).
[20] Haslinda, A., & Sarinah, A. (2009). A Review of Knowledge
Management Models. The Journal of International Social Research,
2(9), 187-198.
[21] Hedlund, G., & Nonaka, I. (1993). Models of Knowledge Management
in the West and Japan. Implementing Strategic Process, Change,
Learning and, 117-44.
[22] Hunger, D., & Wheelen, T. L. (2003). Essentials of Strategic
Management. Prentice Hall.
[23] Jashapara, A. (2011). Knowledge management an integrated approach
(second ed.). Sydney: Prentice Hall.
[24] Johnson, G., Scholes, K., & Whittington, R. (2005). Exploring
corporate strategies (7 ed.). Financial Times Prentice Hall.
[25] Kogut, B., & Zander, U. (1992). Knowledge of the Firm, Combinative
Capabilities, and the Replication of Technology. Organization Science,
3(3), 383-97., 3(3), 383-397.
[26] Lee, M.-C. (2009, September). The Combination of Knowledge
Management and Data mining with Knowledge Warehouse.
International Journal of Advancements in Computing Technology, 1(1),
39-45.
[27] Liebowitz, J., & Megbolugbe, I. (2003). A set of framework to aid the
project manager in conceptualizing and implementing knowledge
management initiatives. International Journal of Project Management,
21, 189-98.
[28] McAdam, R., & McCreedy, S. (1999). A critical review of knowledge
management models. The learning organization, 6(3), 91-100.
[29] Mouritsen, J., Larsen, H. T., & Bukh, P. N. (2005). Dealing with the
knowledge economy: intellectual capital versus the balanced scorecard.
Journal of Intellectual Capital, 16(1).
[30] Nimmagadda, S. L., & Dreher, H. (2009). On issues of data warehouse
architectures-managing australian resources data. IEEE International
Conference on Digital Ecosystems and Technologies.
[31] Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company:
How Japanese companies create the dynamics of innovation. New
York: Oxford University Press.
[32] Noonna, J. (2000, Jul). Data Mining strategies. Retrieved from
information management: http://www.information-
management.com/issues/20000701/2367-1.html?zkPrintable=true
[33] Paddock, J., & Lemoine, K. (2012). Business Intelligence and Data
Mining Review.
[34] Porter, M. E. (1980). Competitve Strategy: thechniques for Analyzing
Industries and Competitors. New York: Free Press.
[35] Poser, T. B. (2003). Poser,T. B.(2003). The Impact of Corporate
Venture Capital: Potentials of Competitive Advantages for the
Investing Company. Germany: DUV.
[36] Richards, J. (2009). Mining, society and sustaiable world. Springer.
[37] Roos, G., & Roos, J. (1997). Measuring your Companys Intellectual
Performance. Journal of Long Range Planning,, 30(3), 413-26.
[38] Sanchez, R., & Heene, A. (1997). Strategic learning and knowledge
management. John Wiley & Sons, Inc.
[39] Shih, K.-H., Chang, C.-J., & Lin, B. (2010). Assessing knowledge
creation and intellectual capital in banking industry. Journal of
Intellectual Capital, 11(1), 74-89.
[40] Spender, J. C. (1996). Making knowledge the basis of a dynamic
theory of the firm. Strategic Management Journal, 17, 45-62.
[41] Stankosky, M., & Baldanza, C. (2001). A systems approach to
engineering a KM system. Unpublished manuscript.
[42] Turban, E. D., & Leidner, e. a. (2008). Information Technology For
Management.
[43] Urza, O. (2011, June 29). World-class suppliers to the Mining
Industry Supporting the development of a knowledge-based. Retrieved
from
http://www.corfo.cl/transferenciatecnologica/PDF/Technology_Transfe
r_Presentation_BHP_Billiton_Santiago_2011.pdf.
[44] Von Krogh, G., & Roos, J. (1995). Organizational Epistemology. New
York:St.Martin press.
[45] Wiig, K. M. (1993). Knowledge management foundations: thinking
about thinking: how people and organizations create, represent, and use
knowledge. Arlington, TX: Schema Press.

Figure 1: Creating competitive advantage through integration of Data Mining and Strategic Knowledge Management
Strategic
knowledge
management
Market based
view
Resource
based view
Data mining
Knowledge
based view
Stakeholder
based view
I
n
t
e
g
r
a
t
i
o
n

Competitive Advantage
Iterative learning
processes
Shared organising principles
Vol. 3, No.8, 2012

73 | P a g e
Managing Changes in Citizen-Centric Healthcare
Service Platform using High Level Petri Net

Sabri MTIBAA
LI3 Laboratory / University of Manouba
National School of Computer Sciences
2010 Manouba, Tunisia
Moncef TAGINA
LI3 Laboratory / University of Manouba
National School of Computer Sciences
2010 Manouba, Tunisia

Abstract The healthcare organizations are facing a number of
daunting challenges pushing systems to deal with requirements
changes and benefit from modern technologies and telecom
capabilities. Systems evolution through extension of the existing
information technology infrastructure becomes one of the most
challenging aspects of healthcare and the adaptation to changes is
a must. The paper presents a change management framework for
a citizen-centric healthcare service platform. A combination
between Petri nets model to handle changes and reconfigurable
Petri nets model to react to these changes are introduced to fulfill
healthcare goals. Thanks to this management framework model,
consistency and correctness of a healthcare processes in the
presence of frequent changes can be checked and guaranteed.
Keywords- Healthcare; requirements changes; evolution;
information technology; healthcare service platform; handle
changes; reconfigurable Petri nets; consistency.
I. INTRODUCTION
The traditional method of receiving healthcare required a
patient to visit their doctors office, or a hospital; now doctors,
hospitals, and healthcare ecosystems are increasingly brought
to the patient. The baby boomer generation is entering a new
stage of their life; they are adopting and demanding better
delivery models for quality and access to care [1]. This has
caused a paradigm shift in the healthcare system.
The new culture is patient centered utilizing care
coordination that is focused on successful outcomes that
depend on new innovations, technology and meaningful data,
including collection, delivery, ease of use, intelligence, and
reporting [2]. As human life expectancy continues to increase
and aging populations lead to higher health treatment costs,
telehealth is on the top of the regulators agenda.
The focus on the climbing cost to deliver and maintain
quality healthcare is no longer in the peripheral view, but a
clear line of sight for the patients, healthcare
providers, regulators, and payers. This brings new
requirements for healthcare professionals to share information,
communicate and collaborate in real time from multiple
locations, because medicine is a collaborative science.
Communications become a strategic asset for a strongly
needed healthcare transformation technology. It must be
deployed to this field in order to ensure better context for
medical decisions, reduce administrative costs and improve
patient safety by reducing errors. The healthcare community
has recognized the need to transform from the current hospital
centralized treatment-based mode to prevention-oriented
comprehensive healthcare mode in which hospitals,
communities, families and individuals are closely involved.
The new mode needs to provide individuals with intelligent
health information management and healthcare services. It
allows them to enjoy medical prevention and healthcare
services in their daily life.
The advancement of information technology (IT) brings
more opportunities for innovations in the healthcare area. The
use of service oriented technologies such as SOA, Web
Services (WS) allows service providers to reduce and simplify
integration process, to abstract network capabilities (e.g., call
control, presence, location, etc.), and create personalized and
blended services (both internally and with 3rd party partners)
[3]. These technologies facilitate the construction of service
systems with higher reusability, flexibility, extensibility, and
robustness.
Cloud computing is evolving as an important IT service
platform with its benefits of cost effectiveness and global
access. Built upon Enterprise Service Bus (ESB) as an
integration backbone [4], this paper presents a novel citizen-
centric healthcare service platform. One of the much-touted
potentials of this platform is the ability to construct healthcare
composite Web services on demand, relieving telecom
operators from the intricate details of how technologies work
so they can focus on the business aspects.
As healthcare services aggregated in the proposed
healthcare service platform increases, the complexity of
managing changes will grow and the manual management of
changes becomes not practical [5]. One of the greatest
promises of the healthcare platform is ability to self-adapting
to guarantee goals achieving. Abstracting changes in business
relationships claim a framework to manage changes without
any impact. For instance, the proposed healthcare platform
needs to achieve the plug-in/plug-out Web services with little
overhead while guaranteeing properties. These properties can
be classified either functional or non-functional. Functional
properties address the functionalities that healthcare platform
have to fulfill. The non-functional properties refer to events
surrounding the functional properties.
Vol. 3, No.8, 2012

74 | P a g e
Change management is a critical component in the
deployment of healthcare platform. We identify two main
approaches dealing with changes: top-down and button-up [6].
A top-down approach focuses on changes that are usually
business mandated. These changes are motivated by the
business goal, and do not consider the uncertainty of the
underlying member services. The second type of changes is
referred to as bottom-up changes because Web service
providers are the initiators of changes. Bottom-up changes are
initiated by the member services. These changes are initiated
in the Web service environment, and eventually translate into
top-down changes. A service operation may become
unavailable and trigger dependencies services and users in
order to replace this service. In this paper, we will concentrate
on this aspect.
In this paper, we present a conceptual module for
management of bottom-up changes using Petri nets. In our
work, we use Petri nets to model handling and adaptive
changes in healthcare platform. We model changes using Petri
nets because of their applicability to a Web services
composition modelling.
The behaviour of a composite Web services is described by
the evolution of its Petri net model [7]. As the Petri net
evolves, the system goes through different safe and unsafe
states that can be completely defined by the marking of a Petri
net model. Furthermore, Petri nets map directly to our change
specification. They also preserve all the details of our change
specification while modelling the changes accurately. For
instance, Petri nets can easily represent the safe an unsafe
states of web services composition. They represent changes
between these states as transitions. In addition, the use of
reconfigurable Petri nets allows us to incorporate our mapping
rules into the Petri net model. This allows us to completely
model our change specification.
The remainder of this paper is organized as follows. In
Section 2, we use the healthcare platform architecture of and a
scenario from this domain to motivate our work. It will also be
used as a running example. Section 3 presents a bottom-up
specification of changes. In Section 4, we describe our change
management model which is based on Petri nets. Finally, we
conclude in section 5.
II. HEATHCARE SERVICE PLATFORM ARCHITECTURE
In this section, we present the global context of our work
and an overview about service oriented architecture, cloud
computing and enterprise service bus. Then, the healthcare
services platform architecture is exposed and some basic
concepts and definitions are explained.
A. Context and Background
Below we have summarized a few key notions and
technologies that should be of significant value to the design
of healthcare architecture.
Service Oriented Architecture (SOA): In this IT
architecture, applications and more discrete software functions
are network-based, loosely coupled and available on demand
to authorized users or to other applications or services.
Although SOA is not a new concept, the emergence of Web
services as a standard way to expose, describe, access and
combine services has given new life to this approach to
computing.
The key idea of SOA is the following: a service provider
publishes services in a service registry [7]. The service
requester searches for a service in the registry. He finds one or
more by browsing or querying the registry. The service
requester uses the service description to bind service. These
ideas are shown in Fig. 1.

Fig. 1. Reference Architecture Of Web Services- SOA
1) Cloud Computing: Cloud computing called also utility
computing refers to an IT service model and platform that
provides on-demand based IT services over the internet (see
Fig. 2). The five essential characteristics are: on-demand self-
service, broad network access, resource pooling, rapid
elasticity, and measured service [8]. The three services models
include:
- SaaS (Software as a Service) which delivers software
service on demand, such as, salesforce.com Customer
Relationship Management (CRM) service and Google
Gmail;
- PaaS (Platform as a Service) which provides the
computing platform for companies to deploy and
customize business applications on demand, such as,
Google App Engine and Microsofts Azure;
- IaaS (Infrastructure as a Service) which offers data
center, infrastructure hardware and software resources
on demand, such as, Amazon Elastic Compute Cloud
(EC2) and VMware vCloud Datacenter. Both of these
resources provide virtual computers for renters to run
their business applications.
The four major deployment models include: private cloud,
public cloud, community cloud, and hybrid cloud. Companies
normally adopt different service models and deployment
models depending on their unique business processes and
demands on IT services.
Cloud computing today is an evolution and application of
modern ICT including server virtualization, autonomic
computing, grid computing, server farm, network storage, and
web service.
Vol. 3, No.8, 2012

75 | P a g e

2) Enterprise Service Bus:
ESB is one piece of an infrastructure that might help
facilitating the implementation of a SOA, but it is not a
perquisite. There are many aspects of an ESB that fit well with
the SOA model, and denying its possible usefulness would be
counterproductive, but the two are not completely inter-
dependent [4].

Fig. 2. Cloud Computing Architecture
Fig. 3 Depicts The Base Functional Elements Within An
ESB. It Includes:

- Data transformation.
- Application adapters.
- Automation of processes.
- Transformation.
- Routing.
- Messaging.
- Event triggering.
If we consider some of these functional elements it can be
seen that items such as application adapters fall neatly into the
product category, while routing and messaging are more of an
architectural consideration.
B. System Architecture
First, we present our Healthcare Service Platform (HSP). It
intends to provide personalized healthcare services for the
public. The healthcare value chain is complex. It consists not
only of healthcare providers, but also of payers (government,
employers and patients), fiscal intermediaries, distributors and
producers of pharmaceuticals and devices [9].
The HSP does not attempt to address this complete value
chain. It focuses on the delivery of healthcare services. It is an
end-to-end reference architecture that focuses on meeting the
needs of citizens, patients and professionals. Its architectural
diagram is given in Fig. 4.

Fig. 3. ESB architecture
We distinguish three main components, i.e. body sensor
networks (BSN), IaaS cloud, healthcare delivery environment.
- BSN: according to circumstances and personalized
needs, appropriate health information collection
terminals (i.e. sensors) are configured for different
individuals. BSN is used to provide long term and
continuous monitoring of patients under their natural
physiological states. It performs the multi-mode
acquisition, integration and real-time transmission of
personal health information anywhere [10].
- IaaS cloud: modern healthcare is information driven.
Healthcare providers are making progress in building
an integrated profile of patients. This data sits in
systems throughout the enterprise including the HER
and many other electronic systems throughout the
enterprise and community [11]. This component
achieves the rapid storage, management, retrieval, and
analysis of massive heath data. It mainly includes
Electronic Medical Record (EMR) repository. It
considers also personal health data acquired from BSN.
- Healthcare delivery environment: it includes a personal
health information management system. It replaces
expensive in-patient acute care with preventative,
chronic care, offers disease management and remote
patient monitoring and ensures health
education/wellness programs.

Fig. 4. HSP Architecture
Vol. 3, No.8, 2012

76 | P a g e
C. Healthcare Web Services Provided by HSP
In PHISP, we adopt the design idea of SOA and Web
service technology for its design and implementation. The
majority of its functional modules are developed and packaged
in the form of services [8]. Here, we overview some of them
as follows.
- PhysInfoWS: this service can acquire some general
physiological signals such as body temperature, blood
pressure, and saturation of blood oxygen,
electrocardiogram, and some special physiological
signals according to different sensor deployment for
different users. Users ID number is required.
- EnvInfoWS: for a unique ID number, this service can
acquire temperature, humidity, air pressure and other
environmental information for this user.
- SubjFeelWS: it can acquire the user subjective feelings,
food intake, etc., and the information is often provided
by the user from the terminal.
- CoronaryDiagWS: it can analyze the information
according to a series of analysis models, which are
built for coronary heart disease, and then produce
preliminary diagnostic results.
- AssessmentWS: this service can assess the status of the
patients health risk based on the diagnostic results and
the EMR information of the patient.
- EmrWS: this service can output the users medical
history information.
- GeoWS: it can return the users location.
- EmerWS: it can raise an alarm to the user in case of
illness.
- GuideWS: it can provide the patient with preventive
measures especially items that need attention.
D. Healthcare Service Scenario
A way to motivate and illustrate this work, we presents an
example of healthcare service scenario. We distinguish three
main layers: service, business and HSP. The service layer
consists of available web services, and the business layer
represents the Web service like operations typically ordered in
a particular application domain. We refer to the selected
services as member services (see Fig. 5).
Key healthcare environment objectives include:
- Allowing people to stay in their homes to an older
age. By doing this, we can reduce the economic
burden of dedicated care facilities and improve quality
of life for a substantial proportion of the aging
population.
- Using televisions to keep in touch. Another use of
camera technology is in conjunction with an IPTV set-
top box and connection back to a Contact Center.
- Using wireless toys for always-on monitoring and
communications. The wireless home network itself
enables a new class of device that has significant
healthcare implications.

Fig. 5. Healthcare Service Scenario Based On HSP
Let us assume that a citizen establishes a need for a
business objective (healthcare service). Typically, he starts
with formulating the business strategy (or goal). During the
planning, some services can be identified: HealthService,
AccountingService, SpecialistService, FinancialServcie and
InsuranceService. Second, the senior citizen develops a
specification listing the services to be composed through a
graphical interface. We assume that HS, AS, SS, FS and IS are
selected and orchestrated. The third step is the orchestration
where member services that match the specified high level
configuration are selected and invoked.
We describe here the ideal scenario: the senior citizen
subscribe to HealthService. Then all information regarding
who contacts it and when are forwarded to AccountingService.
HealthService forwards also the received data to
SpecialistService in charge of checking the received values.
After analyzing the received values, the team sends a
confirmation or an adjustment of the medication doses. The
FinancialService and InsuranceService are executed to
finalize the process.
E. Modeling Healthcare process using High Level Petri
Nets
The modeling of healthcare process is as crucial as the
implement of healthcare service platform. The formal
representation of real healthcare process with Hierarchical
Petri-Nets is easy to understand. Fig. 6 shows a Hierarchical
Petri Net that describes our healthcare service scenario offered
by HSP.
From the above, the first part of hierarchical healthcare
process net N with refinable transition named Health Service
is shown below. It is refined with the attachment of web
services net N. The planned web service is the assembly of
the set of web services presented previously (PhysInfoWS,
EnvInfoWS, SubjFeelWS, CoronaryDiagWS, AssessmentWS,
EmrWS, GeoWS, EmerWS, GuideWS).
However some changes to member services may handle
some inconsistency in the composition and orchestration. Each
Vol. 3, No.8, 2012

77 | P a g e
service layer change presents a functional and non-functional
change that may happen in a member service.

Fig. 6. Healthcare Process Modeled By Hierarchical Petri Nets
For example, in case of non-availability of a
SpecialistService, a change management is required to ensure
that the healthcare system is remaining profitable.
III. CHANGE SPECIFICATION
Managing bottom-up changes is highly dependent on the
services that compose the system. Therefore, it is quite
important to define the changes that may occur to web services
and then map them to into the system level.
In this section, we present bottom-up changes. We define
the set of handling changes. Handling change ( u ) is defined
for changes that occurred at the service level (for example
Web service availability) while adaptive changes ( O ) are
related to changes at the business level (for instance the
selection of alternative service).
A. Changes Overview
For each change, a transition will be associated between
two states: precondition and postcondition. In our scenario
presented, a precondition for SS unavailability is that it was
available and the postcondition is that it has switched to
unavailable state.
Handling changes will be modeled using Petri nets. Our
classification of triggering changes is based on the traditional
approaches from the fields of software engineering and
workflow systems [7]. A handling change is initiated at the
service level such as the operations, the availability, etc.
Therefore, we can distinguish several handling changes based
on Web service properties.
The Web service properties can be sorted into two
categories: functional and non-functional. Fig. 7 shows the
handling changes: functional and non-functional.
- Non-functional changes: we assume that the non-
functional parameters represent the dependability and response
aspects associated with a member service. Service
dependability can be set to two possible values (i.e., available
or unavailable). Alternatively, service cost values may take
more than two possible values.

Fig. 7. Handling Changes Categories
- Functional changes: this category of changes is related
to a service WSDL description [12]. We consider functional
changes as a combined execution of elementary operations:
remove and then add. We distinguish two different functional
changes: structural and behavioral (see Fig. 6). Structural
changes are related to the operational aspects of a Web
service. For example, a structural change in a healthcare
service can be a consequence of change in the operations
offered to a citizen. Functional changes to a member Web
service occur when its WSDL description is modified.
- Adaptive changes: Adaptive changes may occur at the
composition and orchestration levels. Fig. 8 shows the adaptive
changes considered in our model. In our scenario, when the
healthcare system is interrupted by a change in SS, it reacts to
the change after suspending execution. This may be
accomplished by raising a fault, compensating for the change
at the composition layer, and calling of an alternate service.
For example, if SS becomes unavailable, business layer
will be search for equivalent service to continue execution to
ensure that there is no high level impact on user demands.

Fig. 8. Handling Changes Categories
Now, we will discuss the impact of u changes to the
healthcare business layer. A mapping details how change
instances in one layer correspond to changes in another layer.
These mapping must remain consistent in the presence of
frequent changes.
Handling changes have a reactive impact on the business
layer. For instance, a u change in availability maps to O
change of change instance.
Vol. 3, No.8, 2012

78 | P a g e
IV. CHANGE MODEL
In this section, we introduce a change model to accurately
identify eventual types of changes in a composite Web
services.
A. Handling Changes Model using Petri Nets
Petri nets or PN are a well-founded process modeling
techniques that have formal semantics. They have been used to
model and analyze several types of processes including
protocols, and business processes.
Visual representations provide a high-level, yet precise
language, which allows reasoning about concepts at their
natural level of abstraction. Services are basically a partially
ordered set of changes. Therefore, it is a natural choice to map
it into a Petri net. Moreover, the semantics delivered by Petri
nets can be used to model the standard behavior of composite
Web services described by BPEL [7].
We formalize the change model for triggering changes by
introducing Petri-Net-Handle (PNH) which is defined as
follows.
The algebraic structure of PNH = (P, T, F, P
0
, P
n
) if the
following conditions hold:
- F _ (P x T) (T x P)
- P T = C
- P T = C
- P
i
e P
- P
0
e P
where:
- P is a finite set of places representing the states of
Web service.
- T is a finite set of transitions representing changes to
Web service.
- F is called the web services action flow.
- P
0
is the input place, or starting state of the Web
service
- P
n
is the output place, or the ending state of the Web
service

Fig. 9 represents the model of non-functional changes to
Web services. It is composed from five places and four
transitions. PS is the initial place of PNH
N
. It represents the
initial state of the Web Service. PS consists of four tokens,
each representing one of the four non-functional changes. The
token corresponding to change is fired each to represent
dynamic evolution. If more than one change occurs, the
corresponding token for each change type is fired.
For instance, if a member services (i.e Web service)
becomes unavailable, the transition will be enabled and the
corresponding token will be fired.
The subnet representing dependability changes in PNH
d
=
(P
d
, T
d
, F
d
, P
0d
, P
nd
), where P
d
={PS,PSRe,PSA} and
T
d
={TRe,TA},. The place PS is corresponding to the state of
available and reliable service. PSA represents a service that
becomes unavailable.
Table I. gives summary about non-functional changes.
TABLE I. NON FUNCTIONAL CHANGES
Change
u
Pre Post
alterAvailibility
A
u PSA PSA
alterReliability
R
u PSR PSR
alterCost
C
u PSC PSC
alterResponsivenss
Re
u PSRe PSRe

Fig. 9. Handling non-functional changes categories
When a service becomes unavailable, the token (representing
the availability property) is moved from PS to PSA. A similar
behavior is observed when the service becomes unreliable. TA
represents alterAvailibility, and TRe represents
alterReliability.
The same logical structure can be applied for functional
changes.
B. Modeling Adaptive Changes with Reconfigurable Petri
Nets
We have surveyed extensions of Petri nets for modeling
reactive changes. Reconfigurable Petri-nets provide formalism
for modeling these changes. It is a class of high level Petri
nets.
They support internal and incremental description of
changes over a uniform description. Reconfigurable Petri nets
are an extension of Petri nets with local structural modifying
rules performing the replacement of one of its subnets by other
subnets [13]. The tokens in a deleted place are transferred to
the created one.
We formalize the change model for adaptive changes by
introducing PNAC (PNAC) which is defined as follows. The
algebraic structure of PNAC= (P, T, F, R, I) where:
- P= {P
1
,P
n
} is a non-empty and finite set of places
- T={T
1
,...,T
n
} is a non-empty and finite set of
transitions disjoint from ) ( u = T P P I
Vol. 3, No.8, 2012

79 | P a g e
- F: ) ( ) ( P T T P Y IN is a weighted flow
relation. A rewriting rule is a map
2 1
: P P r
whose domain and co domain are disjoint subsets of
places u = _ P P P, P ,
1 1
I P
- R={r
1
,..,r
n
} is a finite set of structure modifying
structure rules.
- I represent the initial state: the first configuration of
composition in business layer. The domain of I is
HCE
0
.

Table I. gives summary about adaptive changes.
TABLE II. ADAPTIVE CHANGES
Change
O
Pre Post
alterState
ST
O VEST VEST
alterServiceInstance
S
O VES VES
alterCost
C
O VEC VEC

We consider the scenario containing five places
corresponding to adaptive changes:
- HCE
S
is the set of places {HCE
0
, HCE
1
, HCE
2
,
HCE
3
, HCE
4
} where S represents alterState.
- HCE
v
5
, HCE
6
, HCE
7
,
HCE
8
, HCE
9
} where V denotes
alterServiceInstance.
- HCE
w
10
, HCE
11
, HCE
12
,
HCE
13
, HCE
14
} where W denotes alterOrder.

Fig. 10 shows a PNAC representing initial statechange in
service orchestration. The adaptive changes using
Reconfigurable Petri net representing modification on service
state (see Fig. 11), removal of service (see Fig. 12), and
addition of service (see Fig. 13) are presented.
C. Change Management Framework
We use our Petri net change specification as the basis for
handling changes in our healthcare environment. The
framework of change management is divided into two
modules: detection and reaction.
After the change specification is defined, we begin the
management. Detecting the respective changes is the first step
of change management.
All changes identified in the handling changes models are
subject to detection. Detection involves an agent that monitors
the Web service. Each change type has an associated set of
rules for detection. For example, a SpecialistService may
change the input parameters (i.e required information provided
by patient), when this change occurs, the healthcare system
must detect this change using some predefined detection rules.

Fig. 10. Reconfigurable Petri Nets For Reactive Changes- Initial State

Fig. 11. Reconfigurable Petri Nets For Reactive Changes- After Change

Fig. 12. Reconfigurable Petri Nets For Reactive Changes- After Removal Of
Service

Vol. 3, No.8, 2012

80 | P a g e
Each detected change must be forwarded to monitoring
service; and then the composition strategy must be updated.
The notification and polling mechanism are mainly the
techniques to awareness that a change has occurred. These
techniques require that a monitoring service periodically send
Refresh and Alive messages to detect unavailable services
and also renew membership.

Fig. 13. Reconfigurable Petri Nets For Reactive Changes- After Addition Of
New Service
The changes are detected at the service layer and
represented as an incidence matrix. Some rules are identified
for detecting functional and non functional changes.
We define a rule for mapping the change into the defined
Petri-net: first of all, the current service state is corresponded
to a set of precondition places in the triggering Petri net and
the updated service state as the set of postcondition places in
the triggering Petri net. Then, a comparison between the
values precondition and postcondition places of the Petri net is
done. Depending on result retuned, a token is placed in the
respective precondition place. This token will enable the
change transition only if a difference is found.
Let us consider the example where the service WS service
change availability (due to maintenance reasons) and the other
attributes remain constant. In this case, we map the change
into the non-functional Petri net. The service agent responsible
of monitoring of WS will generate the incidence matrix
corresponding to the Petri net model. This agent is in
interaction with healthcare service platform to detect effective
changes in the execution environment and map it into a Petri
net model defined in this section ( for example, unavailability
of a WS). The Centralized agent is the module that reacts to
these changes and purpose a reconfiguration in the execution
environment to guarantee consistency and correctness of the
healthcare processes. Fig. 14 shows the different modules
designed for management framework to detect and react to
different changes in healthcare services.
Based on the information sent by service agent we define
how to execute adaptive change. After receiving the matrix
indicating the change that occurred, the handling change is
mapped to the appropriate reactive change. We can list some
considered reaction techniques in our system:
- In case of add, the newly service member will be
considered. It can be taken into account in load
balancing context or as back-up alternative.
- In case of unavailability of a service, if it is critical
then the orchestration will be paused. Since a
heartbeat is activated to check the status of the
service; the orchestration will wait the service
availability otherwise orchestration will exit
(depending on configurable heartbeat number).

Fig. 14. Adaptive Petri Net Generator Module In Change Management
Framework
V. CONCLUSION AND FUTURE WORK
In this paper, we first presented a novel architecture of
healthcare services platform. Second, we exposed the bottom-
up approach focusing on handling changes that may occur in
this system and then mapped to adaptive changes. We used a
formal change model based on High Level Petri Nets to
accurately represent these changes.
Future work includes an extension of change management
framework. We plan to include a top-down approach to
specifying changes. A full simulation prototype taking into
account priority in changes and the estimation of their
frequency based on measures represent actions planned to
enhance actual healthcare platform.
REFERENCES
[1] M. d. Reuver, and T. Haaker, Designing viable business models for
context-aware mobile services, Telematics and Informatics, Vol. 26 ,
pp. 24024, 2009.
[2] V. .s. .Kondapall, J. Raju, K. R. Rao, and K. tanuja , Patient-Centric-
Integrated EHA OF Physician Pratice Portal Through Cloud Computing
Technology, International Journal of Computer Science &
Communication Networks, ,Vol 2, pp. 172-18, 2012.
[3] N. Kryvinska, C. Strauss, and P. Zinterhof, Next Generation Service
Delivery Network as Enabler of Applicable Intelligence in Decision and
Management Support Systems Migration Strategies, Planning
Methodologies, Architectural Design Principles, Studies in
Computational Intelligence, Vol. 352, pp. 473-502, 2011.
Vol. 3, No.8, 2012

81 | P a g e
[4] L. Chen, Integrating Cloud Computing Services Using Enterprise
Service Bus (ESB), Business and Management Research,Vol.1, No. 1,
pp. 26-31, 2012.
[5] L. Baresi, and L. Pasquale., Adaptive Goals for Self-Adaptive Service
Compositions, Proceedings of Web Services (ICWS), pp. 353 - 360,
2010.
[6] S. Akram, A. Bouguettaya, X. Liu, A. Haller, and F. Rosenberg, A
Change Management Framework for Service Oriented Enterprises,
International Journal of Next-Generation Computing (IJNGC), Vol. 1.
No. 1, 2010.
[7] S. Mtibaa, and M. Tagina An Automated Petri-Net Based Approach for
Change Management in Distributed Telemedicine Environment,
Journal of Telecommunications,Vol.15, No. 1, pp. 1-9, 2012.
[8] P. Wang, Z. Ding, C. Jiang, and M. Zhou, Web Service Composition
Techniques in a Health Care Service Platform, IEEE International
Conference on Web Services, pp. 355-362, 2011.
[9] A. Azees, Challenges and Performance Enhancement in Could
Computing, IRACST - International Journal of Computer Science and
Information Technology & Security (IJCSITS),Vol.1, No. 1, pp. 49-54,
2011.
[10] M. Brenner, M. Grech, and M. Unmehopa, An enabler gateway for
service composition using SIP 13th International Conference on
Intelligence in Next Generation Networks, , pp. 1-6, 2009.
[11] V. .s. .Kondapall, J. Raju, K. R. Rao, and K. tanuja , Patient-Centric-
Integrated EHA Of Physician Pratice Portal Through Cloud Computing
Technology, International Journal of Computer Science &
Communication Networks, ,Vol 2, pp. 172-18, 2012.
[12] A. Chehri, and H. T. Moutah, Survivable and Scalable Wireless
Solution for E-health and Eemergency Applications, EICS4Med, pp.
25-29, 2011.
[13] P. Shanchen, L. Yin, H. Hua and L. Chuang, A Model for Dynamic
Business Processes and. Process Changes, Chinese Journal of
Electronics, Vol. 20, No. 4, 2011.
AUTHORS PROFILE
Sabri Mtibaa, He is currently a Ph.D. student in the National School for
Computer Sciences of Tunis, Tunisia (ENSI). He received the master degree
from High School of Communication of Tunis, University of Carthage,
Tunisia (Sup'Com) in 2008. His current research interest includes web service
composition using Petri nets as well as system verification and QoS aware.
Moncef Tagina, He is a professor of Computer Science at the National
School for Computer Sciences of Tunis, Tunisia (ENSI). He received the
Ph.D. in Industrial Computer Science from Central School of Lille, France, in
1995. He heads research activities at LI3 Laboratory in Tunisia (Laboratoire
d'Ingnierie Informatique Intelligente) on Metaheuristics, Diagnostic,
Production, Scheduling and Robotics.

Vol. 3, No.8, 2012

82 | P a g e
Software Architecture- Evolution and Evaluation
S.Roselin Mary
Department of Information Technology,
Hindustan University,
Chennai, India.
Dr.Paul Rodrigues
CTO,
WisdomTree Software Solutions,
Chennai, India.

Abstract The growth of various software architectural
frameworks and models provides a standard governing structure
for different types of organizations. Selection of a suitable
framework for a particular environment needs much more
detailed information in various aspects and a reference guide of
features should be provided. This paper brings out the history of
software architecture with a new evolution tree. It also
technically analyses well known frameworks used in industries
and other governmental organizations and lists out the
supportive tools for them. This paper presents the comparative
chart that can be used as a reference guide to understand top
level frameworks and to further research to enable and promote
the utilization of these frameworks in various environments.
Keywords- Framework; Software Architecture; Views.
I. INTRODUCTION
Architecture is playing a vital role to reveal the complexity
of a given system. Number of steps will be increased when the
system becomes complex. Planning should be done in a
detailed manner when the system becomes complex.
Architecture comprises the combination of process and
product of planning, designing and constructing space to
reflect functional, social and aesthetic considerations [21].
Planning for buildings and complexity behind this will be
interrelated in civil engineering. As customers and
constructors have their own views in a particular subject, the
architecture should solve it in a unique manner by covering all
of them [13].
Likewise the same concept in software is called software
architecture. The term and concept of Software architecture
was brought out by the research work of Dijikstra in 1968 and
David parnas in 1970s. The interconnected basic building
components and the views of end user, designer, developer
and tester are needed to build a complicated, critical system.
The design and implementation of the high-level structure of
the software are the backbone of software architecture. The
architectural elements will be interconnected in well-known
manner to get the major functionality and performance
requirements of the system and to obtain non-functional
requirements such as reliability, scalability, portability, and
availability [12]. Software frameworks points out the suitable
places in the architecture where specific functionality can be
adapted by application programmers [17]. A software
framework provides an abstraction where generic functionality
can be selectively overridden or specialized by user code. The
overall development time will be cut into minimum as it
concentrates on the low level details of a working system. So,
the designers and programmers can concentrate only on the
software requirements. [7].
The rest of the paper is organized as follows. Section II
briefly describes the history of Software architecture and the
figure Fig.1 given below clearly portrays the evolution.
Section III classifies the frameworks. Section IV and V
summarizes and compares the different frameworks.
II. HISTORY OF SOFTWARE ARCHITECTURE
The basic principles of software architecture have been
applied since the mid 1980s and it crossed various stages
from algorithms era by borrowing the concepts from others to
get a shaped form. In 1928, An Algorithm was formulated to
solve the problem by the finite sequence of instructions. Von
Neumann developed Flow Chart that has a visual
representation of the instruction flow, to plan computer
programs in 1947 by inheriting the idea from the flow process
chart(1921) and multi flow chart(1944)which were used
mostly in the area of electrical engineering. But, there is a gap
to point out the flow of control. So, Control Flow Diagram
(CFD) was developed in the late 1950s to describe the control
flow of a business process and program. This was not enough
to view the complex systems. The high level view of the work
and immediate access of particular points cant be represented
using this diagram. So, to reveal the entire system by dividing
into blocks, Block Diagram was developed in late 1950s. A
specific function for each block and the connection between
blocks will be shown in a diagram.
The introduction of abstraction concept became a booster
in the field of software architecture. It made a revolution and
tremendous growth to that area. By that way, data structures
that have similar behaviour, data structures that have similar
behaviour, certain data types and modules of one or more
programming languages that have similar semantics are
grouped in the late 1960s.This was happened by the
introduction of Abstract data types. It again leads to Modular
Programming that introduces the concept of separate parts
called modules in software in 1968. Separation of concerns
with the logical boundaries between components is called as
modules.
In 1977, Three Schema Approach that adopts layered
architecture based on the modular programming was
developed. It is used to build information systems using three
different views in systems development. Here an application
will be broken into tiers and developers have to modify a
Vol. 3, No.8, 2012

83 | P a g e
Figure 1. Evolution of software Architecture
specific layer not to rewrite the entire application over.
Flexible and reusable applications can be developed using this
scheme.
Later, based on this three tier approach, a layer of six
perspectives was introduced in 1987 by John Zachman .That is
called as The Zachman Framework which still plays an
important role in the era of Enterprise Architecture and
influenced frameworks DODAF, TOGAF, TEAF and FEAF.
In 1993Zachman released the modified version of Zachman
Framework with more number of views. In 1995, 4+1 view
model was developed by Kruchten.
Views are used to analyze the complex systems, and to list
out the problem elements and the solution. A view of a system
suppresses details. It focuses on specific concerns of the
system. It provides a simplified model [13] [12].
U.S Government encouraged the researchers to develop the
frameworks for defense side applications and it leads to the
C4ISR Architecture Framework in 1996. The Department of
Defense Architecture Framework (DODAF) was released n
2003, which restructured the C4ISR framework ver2.0 [19]
[6].
The restructured C4ISR framework ver2.0 was released as,
The Department of Defense Architecture Framework
(DODAF) in 2003[19] [6]. The Open Group Architecture
Framework (TOGAF) was developed by the members of
open architecture forums in 1995. Recently in 2009, TOGAF
Version 9 was released [15].
To integrate its myriad agencies and functions under single
common and enterprise architecture, the Federal enterprise
Architecture Framework (FEAF) was developed in 1999 by
the Federal Government [18].
Treasury Enterprise Architecture Framework (TEAF)
was developed to support the Treasurys business processes in
terms of products of the US Department of Treasury and
published in July 2000 [20].
A reference model RM-ODP was developed by Andrew
Herbert in 1984. It combines the concepts of abstraction,
composition and emergence on the distributed processing
developments. By including the set of UML profiles in the
ODP and UML4ODP was released in 2004[10].
In 2001, Aspect oriented programming boom out by
inheriting the principles of OOPS. And, it leads to the Aspect
oriented software development in later 2002.
IBM announced Service Oriented Modeling Architecture
(SOMA) in 2004 opposing the distributed processing and
Modular programming. It is the first publicly announced SOA
related methodology. In addition to this, to provide tactical and
strategic solutions to enterprise problems, the SOMF ver 1.1
was released by Michael Bell [4][5].
Vol. 3, No.8, 2012

84 | P a g e
This section clearly portrays that Zachman framework
paves a way to build so many frameworks on it. The
application of UML on RM-ODP derives a new framework.
This analysis invokes why not to develop new frameworks by
combining some existing technology to yield a better
framework. The frameworks dealt in the next sections are
most widely used for the commercial and Government
departments. So, it is necessary to classify and compare them.
III. CLASSIFICATION OF FRAMEWORKS
Classification is the problem of identifying which of a set
of categories a new observation belongs to. As the frameworks
were developed under the interests of different field masters,
they were influenced by various perspectives. So, it is
necessary to classify them as whether they are developed by
standard bodies or individual interests or by private agencies.
The frameworks developed by standard bodies fall under
the standard category and others fall under nonstandard
category. And also they are subcategorized based on their
usage in commercial or Government purpose.
Commercial ( ISO RM-ODP)
Standard
Government ( DODAF, FEAF, TEAF)
Frameworks
Commercial ( TOGAF, Zachman)
Non-Standard
Government(TOGAF,Zachman)
Figure 2. Classification
Frameworks developed and used for the Government
departments and for Defense side applications are classified
under the Government frameworks. Frameworks used for
commercial purpose are classified under the commercial
frameworks.
The Open Distributed model ISO RM-ODP falls under the
standard and commercial frameworks. DODAF, FEAF and
TEAF which were developed for the U.S Government
agencies are coming under the standard and government
frameworks.The well accepted and most widely used
frameworks, TOGAF and Zachman frameworks are used by
both the commercial and government agencies.
Even though TOGAF and Zachman frameworks are
falling under non-standard category, mapping of these
frameworks to DODAF, FEAF and other standard frameworks
yielded good products in the industry. The classification
described in this section will be very much useful for the
customer to choose the suitable framework quickly for his
organization based on the job nature also. The next subsection
deals the comparison parameters that can be used by the
customer to choose the appropriate tool. The following section
analyses the well-known frameworks and lists out their
criteria.
IV. EVALUATION OF VARIOUS FRAMEWORKS
In this paper, we have taken the survey of few frameworks
which are most widely used. The parameters used for
comparison in existing surveys are not suitable for a customer
to choose the tool. So, the methodologies, techniques and tools
used in these frameworks are considered for the comparison.
A. Zachman Framework
The Zachman Framework describes the complex thing in
different ways using different types of descriptions. It provides
thirty-six categories to describe anything completely.
1) Views / Viewpoints: It has six different views to
facilitate each player to view the system in their own particular
way.
Planner's View (Scope)
Owner's View (Enterprise or Business Model)
Designer's View (Information Systems Model)
Builder's View (Technology Model)
Subcontractor View (Detailed Specifications)
Actual System View
2) Domain: It mainly focuses on Categorizing
Deliverables [8].
3) Origin: This framework is well suited for
Manufacturing Industries [8].
4) Focus: It focuses mainly on Business process.
5) Phase of SDLC: In the Design stage or planning stage,
it can be used [8].
6) System development methodology: Organizations own
methodology can be followed.
7) System modeling Technique: OMG-Model driven
Architecture, Organizations own technique
8) Business Modeling Technique: BPML is used for this
framework.
9) Advantages :
Provides improved professional communications
within community [22].
Understanding the reasons for and risks of not
developing any one architectural representation
[22].
Provides variety of tools and/or methodologies [22].
Developing improved approaches [22].
10) Weakness:
It may lead to more documentation depending on
the cases [2]
It may guide to a process-heavy approach to
development [2].
It isnt well accepted by all the developers [2].
It seems in its first appearance as a top-down
approach to developers. [2].
It is to be biased towards traditional and data-
centric techniques. [2].
B. NATO Architecture Framework / C4ISR / DoDAF
The Department of Defense Architecture Framework
(DoDAF) provides the organization of an enterprise
architecture (EA) into consistent views. It is well suited for
large complicated systems and interoperability challenges.
"Operational views" used here are to deal with the external
customer's operating domain.
Vol. 3, No.8, 2012

85 | P a g e
1) Views / Viewpoints: DoDAF provides multiple views,
each of which describes various aspects of the architecture.
DoDAF defines the following views:
Overarching All View (AV).
Operational View (OV).
Systems View (SV).
Technical Standards View (TV).
2) Domain: It mainly focuses on operating domain [8].
3) Origin: This framework is developed for Defence [8].
4) Focus: It focuses mainly on Architecture data and
Business process.
5) Phase of SDLC: It is used in a Process or Planning
stage [8].
6) System development methodology: The Framework
does not advice the use of any one methodology. It depends on
the organizations decision.
7) System modeling Technique: If the system to be
developed is larger, then UML tools are likely the best choice.
8) Business Modeling Technique: IDEF Family
9) Advantages:
Defines a common approach for describing,
presenting, and comparing DoD enterprise
architectures [19].
Common principles, assumptions and terminology
are used [19].
Across the organizational boundaries architecture
descriptions can be compared [19].
Deployment costs and reinvention of same system
can be reduced. [9].
10) Weakness:
No common ontology of architecture elements [1].
Baseline (current) and objective (target)
architectures are not addressed [1].
How the architectures can be used to measure
effectiveness is not dealt [1].
Business-financial plans are not addressed. [1].
C. TOGAF
The Open Group Architecture Framework (TOGAF)
provides a comprehensive approach to the design, planning,
implementation, and governance of enterprise information
architecture.
1) Views / viewpoints : TOGAF identifies many views to
be modeled in an architecture development process. The
architecture views, and corresponding viewpoints come under
the following categories:
Business Architecture Views
Information Systems Architecture views
Technology Architecture views
Composite views
2) Domain: It mainly focuses on Business, data and
applications [8].
3) Origin: This framework is developed due to the
motivation in Defence side framework.
4) Focus: It focuses mainly on Business process, Data,
applications and Technology.
5) Phase of SDLC: It is used in a Process or Planning
stage [8].
6) System development methodology: Rational Unified
process (RUP) is used as a system development Methodology.
7) System modeling Technique: UML, BPMN are widely
used in TOGAF system modeling.
8) Business Modeling Technique: IDEF is used for
business modeling in TOGAF
9) Advantages :
Increased transparency of accountability [24].
Controlled risk [24].
Protection of assets [24].
Proactive control [24].
Value creation [24].
2) Weakness:
Lots of Detail [16].
Planning methods and governance framework [15].
Weak on Information Architecture [15].
Can lead startup efforts into too much too soon
[16].
D. TEAF
Treasury Enterprise Architecture Framework (TEAF) was
developed by the US Department of the Treasury and
published in July 2000. It is based on the Zachman
Framework.
The Treasury Enterprise Architecture Framework (TEAF)
supports Treasurys business processes in terms of products.
This framework guides the development and redesign of the
business processes for various bureaus.
1) Views / Viewpoints: It provides four different views.
Functional Views
Information View
Organizational View
Infrastructure View
2) Domain: It has a domain on Business processes [20][8].
3) Origin: This framework is developed for Treasury
department [20].
4) Focus: It focuses mainly on Business process.
5) Phase of SDLC: It is used in a communication or
Planning stage [8].
6) System development methodology: It does not refer any
specific methodology. It depends on the organizations
decision.[23].
7) System modeling Technique: Flow chart, UML can be
used in TEAF.
8) Business Modeling Technique: IDEF, ERD can be used
as business modeling techniques.
9) Advantages :
Provides the guidance to the treasury bureaus and
offices in satisfying OMB and other federal
requirements [20].
Vol. 3, No.8, 2012

86 | P a g e
Support Treasury bureaus and offices based on their
individual priorities and strategic plans [20].
Leads to Treasury-wide interoperability and
reusability [20].
10) Weakness:
The TEAF does not contain a detailed description of how
to generate the specification documents (work products) that
are suggested for each cell of the TEAF Matrix [14].
E. FEAF
Federal Enterprise Architecture (FEA) was developed for
the Federal Government to provide a common methodology
for information technology (IT) acquisition, use, and disposal
in that Federal government. It was built to develop a common
taxonomy and ontology to describe IT resources. The FEAF
provides documenting architecture descriptions of high-
priority areas. It guides to describe architectures for functional
segments in multi-organization manner of the Federal
Government.
1) Views / Viewpoints: Like zachman framework, FEAF is
also having five different views in its framework.
Planners View (Scope)
Owners View (Enterprise or Business Model)
Designer's View (Information Systems Model)
Builder's View (Technology Model)
Subcontractors View (Detailed Specifications)
2) Domain: It has a domain on provision of services [8].
3) Origin: This framework is well suited for Enterprise
Architecture planning.
4) Focus: It focuses mainly on Business process, Data,
Application and Technology.
5) Phase of SDLC: It is used in a Communication or
Planning stage [8].
6) System development methodology: RUP (Rational
Unified process) is followed in FEAF.
7) System modeling Technique: UML is used as a system
modeling tool in FEAF.
8) Business Modeling Technique: BPML is the technique
used in FEAF.
9) Advantages :
Serve customer needs better, faster, and cost
effectively [18].
Promote Federal interoperability [18].
Promote Agency resource sharing [18].
Reduced costs for Federal and Agency [18].
Improve ability to share information [18].
Supports capital IT investment planning in Federal
and Agency [18].
10) Weakness:
The Federal Government could risk allocating too
much time and resources to an enterprise
architecture description effort yielding potentially
little return at significant cost [18].
The Federal Enterprise Architecture program
requires technical and acquisition expertise [18].
The Federal IT community must keep its eyes on
the basic principles rather than near-term objectives
and achievements [18].
The Federal Government has to pay up-front for the
right to exercise options in the future [18].
Concern over territoriality and loss of autonomy
may impede the Federal Enterprise Architecture
effort due to long-term, realignment of Agency
functions and responsibilities [18].
It is hard to have common, cross-Agency models
and standards to ensure interoperability [18].
F. ISO RM-ODP
The ISO Reference Model for Open Distributed Processing
provides a framework standard to support the distributed
processing in heterogeneous platforms. Object modeling
approach is used to describe the systems in distributed
environment.
1) Views / Viewpoint: The five viewpoints described by
RM-ODP are:
The enterprise viewpoint
The information viewpoint.
The computational viewpoint
The engineering viewpoint
The technology viewpoint
2) Domain: It has a domain on information sharing in
distributed environment.
3) Origin: This framework is well suited for major
computing and telecommunication companies.
4) Focus: It focuses mainly on Business process,
Technical Functionality and Solution.
5) Phase of SDLC: It is used in a Processing and
communication stage.
6) System development methodology: Object oriented
method and IAD can be followed here [3].
7) System modeling Technique: UML, OMG (Model
driven Architecture) are used as system modeling techniques
[3].
8) Business Modeling Technique: BPMN is used as
business modeling technique in RM-ODP.
9) Advantages :
It provides lot of details for the analysis phases of
the development of applications [3].
It provides the platform to integrate the
requirements from different languages consistently.
[3].
It provides a set of established reasoning patterns to
identify the fundamental entities of the system and the
relations among them. It provides the appropriate degrees of
abstraction and precision for building useful system
specifications [3].
TABLE 1 COMPARATIVE CHART FOR FRAMEWORKS
Vol. 3, No.8, 2012

87 | P a g e

S/
N
Frame
work

TERMS
ZACHMAN
FRAMEWORK
http://zachmaninternati
onal.com
DoDAF
Cio-nii.defense.gov/docs/
DoDAF_Volume_II.pdf
TOGAF
http://www.opengro
up.
org/architecture/
TEAF
www.treas.gov/
cio
FEAF
www.cio.gov/documents/
fedarch1.pdf
ISO RM-ODP
http://www.rm-odp.net/
1 No of Views Six Four Four Four Five Five
2 Domain
Categorizing
deliverables
Operating domain
Business, Data and
Applications
Business
processes
Provision of services
information sharing in
distributed environment
3
Origin

In- Manufacturing Defence Defence
Treasury
Department
Enterprise Architecture
planning
major computing and
telecommunication
companies
4 Focus Business process
Architecture Data,
Business process
Business process,
Data, Applications,
Technology
Business
processes
Business process, Data,
Applications&
Technology
Business process,
Technical functionality
& Solution
5 Phase of SDLC Planning (Design) Process/Planning Process/Planning
planning /
communication
Planning &
communication
Processing &
communication
6
System
development
methodology
Organization own
methodology

Organization own
methodology

RUP
Organization
own
methodology

RUP
Object oriented method,
IAD

7
System
modeling
technique
OMG-Model driven
Architecture,
Organizations own
technique
UML UML, BPMN
Flow chart,
UML

UML
UML, OMG(Model
driven Architecture)

8
Business model
technique
IDEF IDEF Family IDEF Family IDEF, ERD BPML BPMN
9 Advantages
Improving
professional
Communications
wide variety of
tools
improved
approaches For
Architectural
representations
common Approach
common principles,
assumptions and
terminology
comparable
architecture
descriptions across
organizational
boundaries
reduction of
deployment costs
increased
transparency of
accountability
controlled risk
protection of
Assets
proactive
Control
value creation
satisfying
OMB
support
individual
Priorities
and
strategic
Plans
interoperabi
lity and
reusability
serve customer needs
better, faster and cost
effectively
promote federal
Interoperability
provide agency
resource sharing
reduced costs
improve ability to
share information
support Federal and
agency capital IT
investment planning
improved
requirement
collection and
analysis phases
consistently
integrated
requirements
expressed in
separate languages
set of already
established
reasoning patterns
used for building
robust,
efficient and
competitive
applications
backed by industrial
products
with enough
acceptance
10 Weakness
documentation
heavy approach
process heavy
approach to
development
seems like Top
down
Approach
biased towards
traditional,
data centric
techniques
No common ontology
Of architecture
elements
not addressing baseline
and objective
architectures
not addressing
capabilities and
measures of
effectiveness
lack of business
financial artifacts
lots of detail
planning
methods
and governance
framework
weak on
information
Architecture
can lead startup
efforts into too
much too soon
No detailed
description
of
Specificatio
n
document
for each
cell
Missing the
techniques
for creating
specificatio
n
document
little return at
significant cost
need technical and
acquisition expertise
need a watch on
future
less future
Maneuverability
loss of autonomy
may impede effort
difficult to ensure
interoperability
problem of inter-
view
Consistency
Not a truly
guaranteed
cross-view checks
No precise notion of
Consistency
11 Tools
Adaptive EA
Manager
Mega V6.1
SystemArchitect
V10
Simon Tool
EA Webmodeler
Corporate Modeler
Enterprise Edition
10
SystemArchitect V10
Metis product family
System Architect
10
Metastorm
ProVision EA
V6.0
IDS Scheer
EA Webmodeler
EA
Webmode
ler
Corporate
Modeler
Enterprise
Editionv1
0
FEAMS
V0.2
Metis
product
family
Adaptive EA
Manager
Flashline4
FEAMS V0.2
SystemArchitect
V10
ConsVISor
TINA
Simon Tool
MagicDraw

Vol. 3, No.8, 2012

88 | P a g e
It provides a set of mechanisms and common
services to build robust, efficient and competitive
applications, interoperable with other systems [3].
10) Weakness:
RM-ODP has the problem of inter-view and inter-view
consistency. A number of cross-view checks to be done to
maintain the consistency. These checks dont guarantee the
consistency [11].
V. COMPARATIVE CHART OF FRAMEWORKS
Table 1 given above describes the comparison between all
the discussed frameworks. It has precise data for the user with
the additional information of available supportive tools.
VI. CONCLUSION
This paper presents an overview of software architecture
and reviewed the evolution of software architecture. By seeing
the evolution tree, one can easily understand the evolution.
Well known frameworks have been studied and discussed in
detail in this paper. It summarizes the frameworks based on
the industry side criteria and it discusses the benefits and
drawbacks of each framework. The comparative chart
included here clearly figures out the frameworks and it can be
used as the reference guide also. It will invoke the user to
choose the right framework for their industry, organization and
business based on their requirement. Users can easily identify
the supporting tools available for their frameworks. All the
frameworks analyzed here are mainly focusing on business
and IT solutions. In future ancient Indian architecture styles
can be mapped to the familiar Frameworks to yield new
frameworks to focus on quality.
REFERENCES
[1] Alessio Mosto,: DoD Architecture Framework Overview [Online].
Available : www.enterprise-architecture.info/Images/.../DODAF.ppt
(2004, May.)
[2] Ambler, S. Extending the RUP with the Zachman
Framework.,http://www.enterpriseunifiedprocess.com/essays/ZachmanF
ramework.html (2007).
[3] Antonio Vallecillo,: RM-ODP: The ISO Reference Model for Open
Distributed Processing. ETSI Informtica, Universidad de Mlaga ,
www.enterprise-architecture.info/Images/Documents/RM-ODP.pdf
[4] Bell, Michael,: "Introduction to Service-Oriented Modeling", in Service-
Oriented Modeling: Service Analysis, Design, and Architecture. Wiley
& Sons, (2008)
[5] Buckalew P M.: Service Oriented Architecture.,http://www.pmbuckalew
.com/soa.htm (2009).
[6] Cris Kobryn and Chris Sibbald.,: Modeling DODAF Complaint
Architectures., www.uml-forum.com//White_Paper_Modeling_
DoDAF_ UML2.pdf (2004, Oct. 25).
[7] HighBeam Research,:Software Framework., http://www.reference.com
/browse/SoftwaRRre_framework (2008).
[8] Jaap schekkerman.: A comparative survey of Enterprise Architecture
Frameworks. Institute for Enterprise Architecture Developments,
Capgemini.,, www.enterprise-architecture.info (2003).
[9] Jim.,: Applicability of DODAF in Documenting Business Enterprise
Architectures, http://www.thario.net/2008/08/applicability-of-dodaf-in-
documenting.html (2008, Aug. 9).
[10] Juan Ignacio.; UML4ODP PLUGIN User guide Version 0.9, Atenea
Research Group, Spain., http://issuu.com/i72jamaj/docs/uml4odp_plugin
(2009).
[11] Mark Maier and Eberhardt Rechtin, Architecture Frameworks in The
art of systems architecting,2nd ed.CRC Press, Florida, pp. 229-
250.(2000).
[12] Philippe Kruchten: Architectural Blueprints The 4+1 View model
of software Architecture IEEE Softw. Vol. 12, pp. 42-50, (1995)
[13] Roger Session: A Comparison of Top Four Enterprise Architecture
Methodologies, ObjectWatch, Inc., www.objectwatch.com/white_
papers.htm(2007, May)
[14] Susanne Leist, Gregor Zellner.,: Evaluation of Current Architecture
Frameworks . University of Regensburg, Germany,
www.dcc.uchile.cl/~vramiro/d/p1546-leist.pdf (2006).
[15] The Open Group.: Module 2 TOGAF9 Components. www.opengroup.
org/togaf/ (2009).
[16] Tim Westbrock.,: Do Frameworks Really Matter? , EADirections,
ww.eadirections.com/.../ EAdirections%20Frameworks%20Breakout%
20updated.pdf (2007, Oct. 24).
[17] Tony C Shan ,: Taxonomy of Java Web Application Frameworks in
Conf. Rec. 2006 IEEE Int. Conf. e-Business Engg., pp. 378-385.
[18] U.S. Chief Information officers ( CIO) Council. :Federal Enterprise
Architecture Framework Version 1.1 .,www.cio.gov/documents/
fedarch1.pdf (1999, Sep).
[19] U.S. Dept. of Defense.: DoD Architecture Framework Version 1.5. cio-
nii.defense.gov/docs/DoDAF_Volume_II.pdf (2007, Apr. 23).
[20] U.S. Treasury Chief Information officer Council.: Treasury Enterprise
Architecture Framework Version 1, www.treas.gov/cio (2000, Jul.).
[21] Will Conely: About Architecture . http://www.ehow.com/about_
4565949_architecture.html (2009).
[22] Zachman, J. A.,: A Framework for Information Systems Architecture.
IBM Syst. J. Vol. 26, No. 3.pp. 276 292, Apr. (1987.)
[23] Treasury Enterprise Architecture Framework , en.wikipedia.org/.../
Treasury_Enterprise_Architecture_Framework
[24] What is TOGAF? ,http://www.articlebase.com/ information-
technology-articles/what-is-togaf-626259.html
AUTHORS PROFILE
Mrs. S. Roselin Mary is presently working as an Assistant professor in the
department of Computer science and engineering, Anand Institute of Higher
technology at Chennai. She obtained her B.E(CSE) degree from Madurai
Kamaraj University and M.Tech(CSE) from Sastra University, Thanjavur. She
is now pursuing her research in Hindustan University, chennai; India.She has
8 years of teaching experience.
Dr.Paul Rodrigues is a Chief Technology Officer, WSS at chennai. He
has many years of experience in industries and teaching. He guided many
Ph.ds and published more than 25 papers in various journals. He was
formerly professor in Kalasalingam college of engineering, Tamilnadu and
worked as Dean (IT & MCA )in Hindustan University.
Vol. 3, No.8, 2012

89 | P a g e
A hybrid Evolutionary Functional Link Artificial
Neural Network for Data mining and Classification

Faissal MILI
Applied Economics and Simulation,
Faculty of Management and Economic Sciences,
5100 Mahdia, Tunisia
Manel HAMDI
International Finance Group Tunisia , Faculty of
Management and Economic Sciences of Tunis, Tunisia
2092, El Manar Tunisia.

Abstract This paper presents a specific structure of neural
network as the functional link artificial neural network
(FLANN). This technique has been employed for classification
tasks of data mining. In fact, there are a few studies that used this
tool for solving classification problems. In this present research,
we propose a hybrid FLANN (HFLANN) model, where the
optimization process is performed using 3 known population
based techniques such as genetic algorithms, particle swarm and
differential evolution. This model will be empirically compared to
FLANN based back-propagation algorithm and to others
classifiers as decision tree, multilayer perceptron based back-
propagation algorithm, radical basic function, support vector
machine, and K-nearest Neighbor. Our results proved that the
proposed model outperforms the other single model.
Keywords- component Data mining; Classification; Functional link
artificial neural network; genetic algorithms; Particle swarm;
Differential evolution.
I. INTRODUCTION
Classification task is a very important topic in data mining.
A lot of research ([1], [2], [3]) has focused on the field over the
last two decades. The Data mining is a knowledge discovery
process from large databases. The extracted knowledge will be
used by a human user for supporting a decision that is the
ultimate goal of data mining. Therefore, classification decision
is our aim in this study. A various classification models have
been used in this regard. M. James [4] has employed a
linear/quadratic discriminates techniques for solving
classification problems. Another procedure has been applied
using decision trees ([5], [6]). In the same context, Duda et al.
[7] have proposed a discriminant analysis based on the
Bayesian decision theory. Nevertheless, these traditional
statistical models are built mainly on various linear
assumptions that will be necessary satisfied. Otherwise, we
cannot apply these techniques for classification tasks. To
overcome the disadvantage, artificial intelligent tools have been
emerged to solve data mining classification problems. For this
purpose, genetic algorithms models were used [8]. In a recent
research, Zhang ([9], [10]) have introduced the neural networks
technique as a powerful classification tool. In these studies, he
showed that neural network is a promising alternative tool
compared to various conventional classification techniques. In
a more recent literature, a specific structure of neural network
has been employed for classification task of data mining as the
functional link artificial neural network (FLANN). In fact,
there are a few studies ([11], [12], [13]) used this tool for
solving classification problems.
In this present research, we propose a hybrid FLANN
(HFLANN) model based on three metaheuristics population
based optimization tools such: genetic algorithms (GAs),
particle swarm optimization (PSO) and differential evolution.
This model will be compared to the trained FLANN based
backpropagation and multilayer perceptron (MLP) as the most
famous model in the area.
II. CONCEPTS AND DEFINITION
A. Population based algorithms
Population based algorithms are classed as a computational
intelligence techniques representing a class of robust
optimization ones. These population based ones make use of a
population of solution in the same time based on natural
evolution.
Many population based algorithms are presented in the
literature such evolutionary programming [14], evolution
strategy [15], genetic algorithms [16], genetic programming
[17], Ant Colony [18], particle swarm [19] and differential
evolution [20]. These algorithms differ in selection, offspring
generation and replacement mechanisms. Genetic algorithms,
particle swarm and differential evolutions represent the most
popular ones.
1) Genetic algorithms
Genetic algorithms (GAs) are defined as a search technique
that was inspired from Darwinian Theory. The idea is based on
the theory of natural selection. We assume that there is a
population composed with different characteristics. The
stronger will be able to survive and they pass their
characteristics to their offsprings.
The total process is described as follows:
1- Generate randomly an initial population;
2- Evaluate this population using the fitness function;
3- Apply genetic operators such selection, crossover, and
mutation;
4- Turn the process Evaluation Crossover mutation until
reaching the stopped criteria fixed in prior.
Vol. 3, No.8, 2012

90 | P a g e
2) Particle swarm
Presented in 1995 by L. Kennedy and R. Eberhart [19],
particle swarm optimization (PSO) represents one of the most
known population-based approaches, where particles change
their positions with time. These particles fly around in a
multidimensional search space, and each particle adjusts its
position according to its own experience and the experience of
their neighboring, making use of the best position encountered
by itself and its neighbors. The direction of a particle is
defined by the set of neighboring and its correspondent
history of experience.
An individual particle i is composed of three vectors:
- Its position in the V-dimensional search space
X
= (X
, X
, , X
)
- The best position that it has individually found
= (
, ,
)
- Its velocity
= (
, ,
)
Particles were originally initialized in a uniform random
manner throughout the search space; velocity is also randomly
initialized.
These particles then move throughout the search space by a
fairly simple set of update equations. The algorithm updates the
entire swarm at each time step by updating the velocity and
position of each particle in every dimension by the following
rules:
.
= (
)) (1)
X
= X
(2)
Where in the original equations:
C is a constant with the value of 2.0

and
are independent random numbers uniquely

generated at every update for each individual dimension (n = 1
to V).

is the best position found by the global population of
particle.

is the best position found by any neighbor of the
particle.
: the weight
: the constriction factor.
3) Differential evolution

Proposed by Storn and Price in 1995 [20], differential
evolution represents a new floating evolutionary algorithm
using a special kind of differential operator. Easy
implementation and negligible parameter tuning makes this
algorithm quite popular.
Like any evolutionary algorithm, differential evolution
starts with a population. Differential evolution is a small and
simple mathematical model of a big and naturally complex
process of evolution. So, it is easy and efficient.
Firstly, there are five DE strategies (or schemes) that were
proposed by R. Storn and K. Price [20]:
Scheme DE/rand/1 :
= x
1
+ F* (x
2
x
3
) (3)
Scheme DE/rand/2 :
= x
5
+ F * (x
1
+ x
2
x
3
x
4
) (4)
Scheme DE/best/1:
= x
best
+ F * (x
1
x
2
) (5)
Scheme DE/best/2:
= x
best
+ F * (x
1
+ x
2
x
3
x
4
) (6)
Scheme DE/rand-to best/1:
= x+* (x
best
x
1
)+F * (x
2
x
3
) (7)
Later, two more strategies were introduced [21].
We present the trigonometric scheme defined by:
= (x
1
+ x
2
+ x
3
)/3 + (p
2
p
1
) * (x
1
x
2
)
+ (p
3
p
2
) * (x
2
x
3
) + (p
1
p
3
) * (x
3
x
1
) (8)
p
i
=|f(x
i
)/ (f(x
1
) + f(x
2
) + f(x
3
)) |, i= 1, 2, 3 ; (9)
F define the constriction factor generally taken equal to 0.5
x define the selected element
x
1
, x
2
, x
3
, x
4
and x
5
represent random generated elements
from the population.
Many others schemes can be found in the literature [20].
B. Functional Link Artificial Neural Networks
The FLANN architecture was originally proposed by Pao et
al. [22]. The basic idea of this model is to apply an expansion
function which increases the input vector dimensionality. We
say that the hyper-planes generated provide greater
discrimination capability in the input pattern space. By
applying this expansion, we neednt the use of the hidden layer,
making the learning algorithm simpler. Thus, compared to the
MLP structure, this model has the advantage to have faster
convergence rate and lesser computational cost.
The conventional nonlinear functional expansions which
can be employed are trigonometric, power series or Chebyshev
type. R. Majhi et al. [23], shows that use of trigonometric
expansion provides better prediction capability of the model.
Hence, in the present case, trigonometric expansion is
employed.
Let each element of the input pattern before expansion be
represented as X(i), 1 < i < I where each element x(i) is
functionally expanded as Zn(i) , 1 < n < N , where N = number
of expanded points for each input element. In this study, we
take N=5.
I= the total number of features
As presented in figure 1, the expansion of each input
pattern is done as follows.
Vol. 3, No.8, 2012

91 | P a g e
Z
1
(i) = X(i), Z
2
(i) = f
1
(X(i)),...., Z
5
(i) = f
5
(X(i)) (10)
These expanded inputs are then fed to the single layer
neural network and the network is trained to obtain the desired
output.
III. HYBRID FLANN DESCRIPTION
The proposed hybrid FLANN is based on evolutionary
algorithms as genetic algorithms, particle swarm and
differential evolution.
A. Resampling technique:
In order to avoid overfitting, we use the (2*5) K fold cross-
validation resampling technique. We proceed as follows:

Figure 1. Functional expansion of the first element
We divide initial database into 5 folds (K=5) where each
one contain the same repartition of classes. For example, if
initial population contains 60% of class 1 and 40% of class 2,
then all the resulted K folds must have the same repartition.
B. Generation
We begin the process by generating randomly initial
solution. We execute partial training using differential
evolution in order to improve initial state.
C. Fitness fuction and evaluation
In order to evaluate each solution, two criterions are used
such the mean square error (MSE) and the misclassification
error (MCE) rate. If we have to compare solutions A and B, we
apply the following rules: A is preferred to B If and only if
MCE(A)< MCE(B) Or MCE(A)= MCE(B) and MSE(A)<
MSE(B).
D. Selection
Many selections are defined in the literature such the
Roulette wheel method, the N/2 elitist method and the
tournament selection method. The last method will be used
here. The principle is to compare two solutions, and the best
one will be selected.
N/2 elitist is used at the beginning of the process in order to
select 50% of generated solution.
E. Crossover
Two parents are selected randomly in order to exchange
their information. Two crossovers are applied and described as
follows:
1) Crossover 1 (over input feature):
An input feature is chosen randomly to exchange his
correspondent weight between the selected two parents.
2) Crossover 2 (over output nodes):
An output is chosen randomly to exchange his
correspondent weight.
3) Crossover 3 (Crossover over connection):
A connection position is chosen randomly and his
correspondent weight is exchanged between the two parents.
F. Mutation
1) Mutation 1(over connection)
A connection position is chosen randomly and his
correspondent weight has been controlled. If this connection is
connected, his correspondent weight is disconnected by setting
his value equal to zero. Else, this connection is connected.
2) Mutation 2 (over one input feature)
An input feature is chosen randomly and his correspondent
weights have been controlled. If this input feature is connected
(there is at least one weights of his correspondent ones is
different from zero), it will be disconnected by putting all his
entire weight equal to zero. Else if this input feature is totally
disconnected, it will be connected there by generating weights
different from zero.
3) Mutation 3 (over two input feature)
We do the same like mutation 2 but here simultaneously for
the two selected features.
4) Mutation 4 ( over three input feature)
In this mutation, the same principle is used for three input
features.
We note that many input features connection and
disconnection can be executed in the same time when having a
large number of features. This crossover helps to remove
undesirable features from our classification process and can
improve the final performance process.
G. Particle swarm optimization (PSO)
In the presented paper, we define three PSO model based
on the notion of neighbor.
1) PSO based on resulted genetic offsprings
First, we apply genetic operators. Each offspring that
improve our fitness function define a neighbor, and used in
equation (1).
2) PSO based on Euclidian distance:
() =
()
( ()) =
()
( ()) =
()
( ()) =
()
( ()) =
()
Vol. 3, No.8, 2012

92 | P a g e
For each particle, we compute the Euclidian distance
between this particle and the rest of the population. Next we
choose the five nearest particles based on this distance.
From the selected subset of neighbors, we choose the best
one which has the best fitness value. This selected one defines
our neighbor to be replaced in equation (1).
3) PSO based on the last best visited solution:
In this case, each particle flies and memorizes his best
reached solution. This memory defines the neighbor to be used
in equation (1).
H. Differential evolution
In this work, we proceed as follows:
- First, for each candidate x, we generate five random
solution x
1
, x
2
, x
3
, x
4
and x
5
.
- Next we apply seven chosen schemes as follows:
DE1: Scheme DE/direct/1 :
= x + F* (x
2
x
1
) (11)
DE2: Scheme DE/best/1 :
= x
best
+ F* (x
2
x
1
) (12)
= x
best
+ F* (x
3
x
2
) (13)
= x
best
+ F* (x
3
x
1
) (14)
= x
best
+ F * (x1 + x2 x3 x4) (15)
DE6: Scheme DE/rand/2 :
= x5 + F * (x1 + x2 x3 x4) (16)
DE7: with Trigonometric Mutation:
= (x1 + x2 + x3)/3 + (p2 p1) * (x1 x2)
+ (p3 p2) * (x2 x3) + (p1 p3) * (x3 x1) (17)
p
i
= |f(x
i
)/ (f(x1) + f(x2) + f(x3)) |, i= 1, 2, 3 ; (18)
I. Stopping criterion:
The process turns in a cycle until reaching a maximum
number of epochs without any improvement. We fix the
maximum number of epochs equal to 30 epochs.
IV. EXPERIMENTAL STUDIES:
11 real-world databases were selected there to be used in
simulation works. They are chosen from the UCI repository
machine learning, which is commonly used to benchmark
learning algorithms [24].
We compare the results of the proposed hybrid FLANN
(HFLANN) with FLANN based on the gradient descent
algorithm. Next, Comparison with other classifiers will be
done.
A. Description of the databases
A brief description of used databases for experimental setup
is presented in table I. Num. is the numeric features, Bin. is the
binary ones, and Nom. is the nominal inputs that mean discrete
with three or more distinct labels.
TABLE I. SUMMARY OF THE DATASET USED IN SIMULATION STUDIES
Dataset Inputs Ex. Cls
Num. Bin. Nom. Total
IRIS
VOTING
BREAST
PRIMA
CREDIT
BALANCE
WINE
BUPA
ECOLI
GLASS
ZOO
4
0
0
8
6
4
13
6
7
10
1
0
16
0
0
4
0
0
0
0
0
15
0
0
9
0
4
0
0
0
0
0
0
4
16
9
8
14
4
13
6
7
10
16
150
435
699
768
690
625
178
345
336
214
101
3
2
2
2
2
3
3
2
8
6
7
B. Convergence test:
In order to test the convergence of the proposed hybrid
FLANN, a comparison will be done with trained FLANN using
the back-propagation algorithm. Results are presented in figure
2 and figure 3. Comparison is done based on the required time
and number of epochs for convergence.
From figure 2, we find that our process needs less than 200
seconds 20 epochs to converge. Figure 3 present results for
FLANN based on back-propagation. This model requires less
than 150 seconds and 15 epochs to converge.
The proposed hybrid FLANN has a strong ability to
converge fast and requires approximately the same time and
epochs than FLANN based back-propagation.
C. Comparative results:
The classification accuracy of the proposed hybrid FLANN
are compared with the results obtained from FLANN trained
using the back-propagation algorithm. Results are presented in
table II. Bold typeface is used to highlight the results that are
significantly better.
We find that the proposed model gives better results for all
used databases. So, our proposed evolutionary process trains
better than the back-propagation algorithm.
Next, a basic comparison between the HFLANN and five
others classifiers is performed using fourth databases. These
classifiers are:
- The decision tree based C.45,
- The multilayer perceptron (MLP) based back-propagation
algorithms,
- The radical basic function (RBF),
Vol. 3, No.8, 2012

93 | P a g e
- The support vector machine (SVM),
- The K-nearest Neighbor (KNN),
Results are presented in table II. From this table, we find
that HFLANN is better for 3 databases and SVM is better for
the forth. We say that HFLANN is a good classifier giving
better results in the majority of used databases.
Table IV presents the number of different local and global
improvement of used population based algorithms. We find
that PSO represents the best local population based technique
with 104476 improvements, and differential evolution is the
best global one with 1854 of improvement.

a.MSE vs Time

b.MSE vs epochs

Figure 2. The MSE Hybrid FLANN results vs. time and epochs applied to
the iris database

a. MSE vs Time

b. MSE vs epochs
Figure 3. The MSE FLANN based back-propagation results vs. time and
epochs applied to the iris database
TABLE II. AVERAGE COMPARATIVE PERFORMANCE OF HFLANN AND
FLANN WITH A CONFIDENCE INTERVAL LEVEL OF 5%
Data bases Folds HFLANN FLANN BASED BP
IRIS training 0,9833 0,0054 0,8958 0,0351
validation 0,9600 0,0303 0,9000 0,0228
test 0,9600 0,0288 0,8933 0,0522
Time 184,3925 142,6253
VOTING training 0,9787 0,0057 0,7796 0,0695
validation 0,9654 0,0227 0,7990 0,0765
test 0,9469 0,0190 0,7829 0,0781
Time 292,5714 78,4182
BREAST training 0,9787 0,0022 0,9277 0,0192
validation 0,9699 0,0101 0,9342 0,0192
test 0,9527 0,0139 0,9298 0,0179
Time 211,6272 156,4693

PRIMA training 0,7906 0,0061 0,6865 0,0130
validation 0,7773 0,0170 0,7161 0,0153
test 0,7501 0,0255 0,6536 0,0213
Time 161,3451 133,8419
CREDIT training 0,8926 0,0056 0,5830 0,0540
validation 0,8771 0,0264 0,6021 0,0444
test 0,8615 0,0284 0,5935 0,0817
Time 355,2951 100,7891
BALANCE training 0,9212 0,0036 0,6123 0,0157
validation 90,7400 0,0137 0,6454 0,0383
test 0,9101 0,0104 0,6036 0,0367
Time 314,7218 168,9399
WINE training 0,9972 0,0036 0,9244 0,0549
validation 0,9833 0,0174 0,9379 0,0719

test 0,9508 0,0329 0,9035 0,0399
Time 173,4038

61,0904

BUPA training 0,7666 0,0175 0,5630 0,0249
validation 0,7147 0,0395 0,6202 0,0276
test 0,7027 0,0199 0,5392 0,0506
Time 232,4010 42,2137
ECOLI training 0,8077 0,0157 0,6197 0,0559
Vol. 3, No.8, 2012

94 | P a g e
validation 0,7703 0,0342 0,6282 0,0630
test 0,7889 0,0221 0,6279 0,0811
Time 315,8144 255,8217
GLASS training 0,7103 0,0181 0,3792 0,0419
validation 0,6901 0,0342 0,4328 0,0450
Test 0,6054 0,0464 0,3463 0,0577
Time 508,9287 174,2889
ZOO training 0,8935 0,0310 0,4683 0,0789
validation 0,8977 0,0406 0,5005 0,1055
test 0,8322 0,0422 0,4163 0,0757
Time 193,7053 53,7702
TABLE III. SUMMARY OF THE RESULTS COMPARING THE PERFORMANCE
THE HFLANN AGAINST EXISTING WORKS. THE TABLE SHOWS THE REPORTED
MEAN CLASSIFICATION ACCURACY OF THE VARIOUS WORKS AND THE BEST
RESULTS ARE HIGHLIGHTED IN BOLD
TABLE IV. NUMBER OF LOCAL IMPROVEMENT AND GLOBAL
IMPROVEMENT OF GENETIC ALGORITHMS, DIFFERENTIAL EVOLUTION AND
PARTICLE SWARM
TABLE V. BEST LOCAL IMPROVEMENT AND BEST GLOBAL
IMPROVEMENT OF GENETIC ALGORITHMS, DIFFERENTIAL EVOLUTION AND PSO
Table V presents best local improvement and best global
improvement of genetic algorithms, differential evolution and
particle swarm. For differential evolution, trigonometric
mutation scheme represent the best local search strategy and
DE/best/1 scheme is the best global one. For PSO, basic model
based on genetic algorithms represent the best local search
strategy, and the PSO based Euclidian distance is the best
global one. Comparing genetic operators, we find that
crossovers improve results more than mutation.
V. CONCLUSION
In this paper, a HFLANN was proposed based on three
populations based algorithms such genetic algorithms,
differential evolution and particle swarm. This classifier shows
his ability to converge faster and gives better performance than
FLANN based on back-propagation.
When comparing different population based algorithms, we
find that PSO is the best local technique for improvement and
differential evolution is the best global one. For differential
evolution, trigonometric mutation scheme represent the best
local search strategy and DE/best/1 scheme is the best global.
For PSO, basic model based on genetic algorithms
represent the best local search strategy, and the PSO based
Euclidian distance is the best global. Comparing genetic
operators, we find that crossovers improve results more than
mutation. Following this comparison, we are able to identify
best local strategy and best global strategy. Compared to the
MLP, FLANN has the advantage to optimize the process
without using hidden nodes.
Future works can be addressed to compare other classifiers
and others evolutionary algorithms. Others comparison criteria
can be used such the needed speed and the robustness of the
algorithm. A wrapper approach can be included in the proposed
process in order to delete simultaneously irrelevant features
over the optimization process.
REFERENCES
[1] R .Agrawal, T. Imielinski and A. Swami, Database mining: A
performance perspective. IEEE Trans. Knowledge Data Eng., 5, pp.
914-925 (1993).
[2] U. M. Fayyad, Piatetsky-Shapiro, G., and Smyth, P. ,From data mining
to knowledge discovery: An overview. In U. M. Fayyad, G. Piatetsky-
Shapiro, & P. Smyth (Eds.), Advances in knowledge discovery and data
mining ,pp. 134,. Menlo Park, CA: AAAI Press ,1996.
[3] H.P. Kriegel, et al., Future trends in data mining. Data Mining and
Knowledge Discovery, 15(1), 8797. Netherlands: Springer (2007).
[4] M., James, Classification Algorithms. Wiley,1985.
[5] L. Breiman, , J.H. Friedman, R.A. Olshen and C.J. Stone, Classification
and Regression Learning. Morgan Kaufman, 1984.
[6] H.P. Kriegel, et al., Future trends in data mining. Data Mining and
Knowledge Discovery, 15(1), 8797. Netherlands: Springer (2007).
[7] R.O. Duda, Hart PE and Stork D.G., Pattern classification. Wiley,
New York ,2001.
[8] D.E. Goldberg, Genetic algorithms in search, optimization and machine
learning. Morgan Kaufmann ,1989.
[9] G. P., Zhang, Neural networks for classification, A survey. IEEE
Transactions on Systems, Man, Cybernetics-Part C: Application and
Reviews, 30(4), pp. 451461, 2000.
[10] G. P Zhang, Avoiding pitfalls in neural network research. IEEE
Transactions on Systems, Man and Cybernetics-Part C: Applications and
Reviews, 37(1), pp. 316 , 2007. M. Young, The Technical Writer's
Handbook. Mill Valley, CA: University Science, 1989.
[11] R. Majhi, G. Panda and G. Sahoo, Development and performance
evaluation of FLANN based model for forecasting of stock markets,
Expert Systems with Applications 36, pp. 68006808,2009.
[12] S. Dehuri and Cho S.B. ,A hybrid genetic based functional link
artificial neural network with a statistical comparison of classifiers over
multiple datasets. Neural Comput & Applic (19), pp.317-328. ,2010a.
Evolutionary
algorithm
Best Local
improvement Best Global improvement
GA Crossover Crossover
DE

trigonometric
mutation scheme DE/best/1
PSO

PSO based
genetic
algorithms
PSO based Euclidian
distance
Methods/
references
Data set
Cancer Pima Iris Balance
HFLANN
C4.5
MLP
RBF
SVM
K-NN
0,9527
0,947
0,92
-
95,43
-
0,7501
0,7313
0,7330
076
0,6780
0,6760
0,9600
0,9400
0,9453
0,3850
-
-
0,9101
0,783
0,853
-
0,876
0,876
Evolutionary
algorithm
Local
improvement
Global
improvement
GA 28423 662
DE 26148 1854
PSO1 104476 1246
Vol. 3, No.8, 2012

95 | P a g e
[13] S. Dehuri and Cho S.B.,Evolutionarily optimized features in functional
link neural network for classification. Expert Systems with
Applications (37), pp.43794391, 2010b .
[14] L., Fogel, J. Owens and J. Walsh, Artificial Intelligence through
Simulated Evolution. John Wiley, Chichester,1966.
[15] D.E. Goldberg, Genetic algorithms in search, optimization and machine
learning. Morgan Kaufmann ,1989.
[16] J., Holland, Adaptation in natural and artificial systems. Univ. of
Michigan Press, Ann Arbor ,1975.
[17] J. Koza, , Genetic programming on the programming of computers by
means of natural selection, Cambridge MA: MIT Press, 1992.
[18] M. Dorigo, , Stutzle, T.,Ant Colony Optimization. MIT Press,
Cambridge. ,2004.
[19] J., Kennedy Eberhart R., Particle Swarm Optimization, In Proceedings
of IEEE International Conference on Neural Networks, pp. 1942
1948,1995.
[20] R. Storn Kenneth P. Differential evolution A simple and efficient
adaptive scheme for global optimization over continuous spaces.
Technical Report TR, pp. 95-012, International Computer Science
Institute, Berkeley, CA, 1995.
[21] F. Hui-Y. and J. Lampinen,A trigonometric mutation approach to
differential evolution. In K. C. Giannakoglou, D. T. Tsahalis, J.
Periaux, K. D. Papailiou, & T. Fogarty, editors, Evolutionary Methods
for Design Optimization and Control with Applications to Industrial
Problems, pp. 6570, Athens, Greece. International Center for
Numerical Methods in Engineering (Cmine) , 2001.
[22] Y.-H. Pao, S.M. Phillips and D.J. Sobajic, Neural-net computing and
intelligent control systems. Int. J. Contr., 56, pp. 263-289 , 1992.
[23] R. Majhi, G. Panda and G. Sahoo, Development and performance
evaluation of FLANN based model for forecasting of stock markets,
Expert Systems with Applications 36, pp. 68006808,2009.
[24] C.L. Blake,. and Merz, C.J., UCI Repository of machine learning
databases, em Irvine, CA: University of California, department of
information and Computer Science. Available on-line at :
http//www.ics.uci.edu/~mlearn/ML.Repository.html. , 1998.
AUTHORS PROFILE
Faissal MILI , Ph. Doctor in quantitative methods, A member of Applied
Economics and Simulation laboratory, Faculty of Management and Economic
Sciences, 5100 Mahdia, Tunisia. E-Mail: [email protected]. Phone:
0021620545728
Manel HAMDI, Doctoral student in finance, Member of International Finance
Group Tunisia, Faculty of Management and Economic Sciences of Tunis, El
Manar University, Tunis cedex, C. P. 2092, El Manar Tunisia. E-Mail:
[email protected]. Phone: 0021627220884

Vol. 3, No.8, 2012

96 | P a g e
Automatic Aircraft Target Recognition
by ISAR Image Processing based on Neural Classifier

F. Benedetto, IEEE, Member, F. Riganti Fulginei, A. Laudani, IEEE, Member, G. Albanese
Dept. of Applied Electronics
University of ROMA TRE, via della Vasca Navale 84, 00146 Rome, Italy
Corresponding authors

Abstract This work proposes a new automatic target classifier,
based on a combined neural networks system, by ISAR image
processing. The novelty introduced in our work is twofold. We
first present a novel automatic classification procedure, and then
we discuss an improved multimedia processing of ISAR images
for automatic object detection. The classifier, composed by a
combination of 20 feed-forward artificial neural networks, is used
to recognize aircraft targets extracted from ISAR images. A
multimedia processing by two recently introduced image
processing techniques is exploited to improve the shape and
features extraction process. Performance analysis is carried out
in comparison with conventional multimedia techniques and
standard detectors. Numerical results obtained from wide
simulation trials evidence the efficiency of the proposed method
for the application to automatic aircraft target recognition.
Keywords- Automatic target recognition; artificial intelligence;
neural classifiers; ISAR image processing; shape extraction.
I. INTRODUCTION
Recently, there has been an explosive growth in the
research area related to inverse synthetic aperture radar
(ISAR) imaging of moving targets [1]. High resolution images
of targets of interest at long range can be acquired from ISAR
images. Moreover, ISAR imaging is becoming an
irreplaceable tool in the task of non-cooperative automatic
target recognition (ATR). There are a lot of different
applications, including detection, imaging, and classification
of ships and aircraft with airborne, maritime, and land-based
radar systems [2], [3]. In the last years, many methods of
ISAR ATR techniques have been developed in the open
literature. Before detecting an object, the image is first
segmented and then the object is recognized [4]-[5]. Image
segmentation is the process of partitioning/subdividing a
digital image into multiple meaningful regions. The
segmentation is usually based on measurements taken from the
image and might be gray level, color, texture, depth or motion.
The result of image segmentation is a set of segments that
collectively cover the entire image. All the pixels of the same
ensemble or region are similar with respect to some
characteristic or computed property, such as color, intensity,
or texture. Edge detection is one of the frequently used
techniques in digital image processing. Object recognition is
the task of finding a given object in an image or in a video
sequence. For any object in an image, there are many features
characterizing the object that can be extracted to provide a
feature description of the object. This description, extracted
from a training image, can then be used to identify the object
when attempting to locate it in a test image containing many
other objects [6]. Image segmentation is usually done using
various edge detection techniques such as Sobel, Prewitt,
Roberts, Canny, and other methods [7]. Then, only some
features characterizing the ISAR images are tested, to identify
what kind of target has been detected [8]-[11]. In fact, the
typical algorithms first detect the edge of an ISAR image, and
then adopt different 1-D descriptors such as Fourier
descriptors (FD) [12] or 2-D descriptors such as Fourier-
wavelet descriptors [13] for feature extraction. This is
computationally more efficient than evaluating the whole
target. Other methods exploit optimal classifiers to determine
the specific kind of target, [14]-[16]. However, in all these
techniques, each target profile is presented as an input feature
vector to the classifier. Since providing real-time performance
in radar target recognition is a crucial issue to be satisfied,
usually neural networks with massive parallel structure and
capacity of learning are used in the classifier [17].
The recognition process must be invariant with respect to
the target position. At least three different techniques for
invariant neural network recognition have been recently
proposed. The first approach, namely the invariance by
training, compensates for the pattern shift taking into account
different targets for different pattern shifts during the training
phase. The main drawback of such an approach is that it is
inapplicable in many operating situations. In fact, the number
of possible variations of patterns makes the training set too
large, increasing at the same time the computational
complexity of the learning system. The second technique,
namely invariance by structure, uses neural networks whose
outputs are always invariant to certain transformations. The
disadvantage of such an approach is that high-order neural
networks are required. Their implementation is too
complicated and their application is limited. Recently, the
combined neural network method is approaching as the most
suitable scheme, to lower the system computational
complexity (see [16] and references therein). Finally, the third
technique uses feature vectors as inputs for the neural
networks. These feature vectors are invariant to the required
transformations and, hence, this method is called invariant
feature space. This kind of recognition system usually uses the
magnitude of the Fourier transform, which is invariant to
linear shifts of its input vector [23]. The advantages of using
Vol. 3, No.8, 2012

97 | P a g e
an invariant feature space are as follows: the number of
features can be reduced to realistic levels, the requirements on
the classifier are relaxed, and the invariance for all input
objects is ensured. On the other hand, the main drawback of
using invariant feature spaces is the processing time (it can be
very long) needed to extract the features before the classifier
can be employed.
The novelty introduced in this work is twofold. We first
present a novel automatic classification procedure, based on
combined neural networks' signal processing. Then, we
discuss an improved multimedia processing of ISAR images
for automatic object detection. In particular, the neural
classifier (NC) is developed as an alternative approach to
those existing in the literature (e.g. the Support Vector
Machine based algorithms are widely used for the patterns
recognition and classification). Designers and users will be
then able to choose the different approaches depending both
on the nature of the problem to be solved and on the used
technology. In our case the NC is composed by combining 20
feed-forward artificial Neural Networks (NNs). Nevertheless,
the number of NNs can be changed to obtain several different
performances depending on the difficulty of classification
problem. Moreover, it is well known that the structure of a
neural network is fixed on the base of the problem to be solved
and the available data. Furthermore, it's clear that a
deterministic way to define the number of hidden layers and
the number of neurons does not exist. In our case, after
performing of several experimental results, the NNs have been
all made by one input layer and two hidden layers made of 168
and 8 neurons, respectively. Then, the output of each NN
consists of one neuron that returns a value characterizing the
class of the related input pattern (Fourier descriptors of the
ISAR image to classify). The ISAR images are first pre-
processed with conventional filters, in order to reduce the
speckle noise. Then, the combination of two image processing
techniques, recently introduced in literature, is exploited to
improve the shape and features extraction process. First, the
Smallest Univalue Segment Assimilating Nucleus (SUSAN)
algorithm [18] is applied to the ISAR image. Then, the output
of the SUSAN method is processed by a modified level set
evolution method, namely distance regularized level set
evolution (DRLSE) [19]. We use the first method (i.e.
SUSAN) as a pre-processing step, in order to segment the
input image into two regions of pixels containing the ensemble
of the target pixels and the ensemble of the background pixels
(i.e. pixels not belonging to the target). Then, the DRLSE
algorithm is applied to the first ensemble (i.e. the target pixels
region) as a linking edge method, whose output is the target
contour. Once the aircraft shape is extracted, the invariant
Fourier descriptors (FD) are computed and used as the input of
the neural classifier.
The remainder of this paper is organized as follows. In
Section II, the proposed neural networks classifier is
described, while the conventional multimedia processing is
illustrated in Section III. Section IV presents the proposed
ISAR images segmentation and shape extraction techniques.
Numerical results and comparisons are outlined in Section V,
and finally, our conclusions are depicted in Section VI

II. NEURAL CLASSIFIER FOR OBJECT DETECTION
In this Section, we discuss the structure of the proposed
ATR scheme composed by a system of 20 feed-forward
artificial Neural Networks (NNs) [20]. The recognition
process must be invariant with respect to the target position.
At least three different techniques for invariant neural network
recognition have been recently proposed. The first approach,
namely the invariance by training, compensates for the pattern
shift taking into account different targets for different pattern
shifts during the training phase [21], [22]. The main drawback
of such an approach is that it is inapplicable in many operating
situations. In fact, the number of possible variations of
patterns makes the training set too large, increasing at the
same time the computational complexity of the learning
system. The second technique, namely invariance by structure,
uses neural networks whose outputs are always invariant to
certain transformations. The disadvantage of such an approach
is that high-order neural networks are required. Their
implementation is too complicated and their application is
limited. Recently, the combined neural network method is
approaching as the most suitable scheme, to lower the system
computational complexity (see [16] and references therein).
Finally, the third technique uses feature vectors as inputs for
the neural networks. These feature vectors are invariant to the
required transformations and, hence, this method is called
invariant feature space. This kind of recognition system
usually uses the magnitude of the Fourier transform, which is
invariant to linear shifts of its input vector [23]. The
advantages of using an invariant feature space are as follows:
the number of features can be reduced to realistic levels, the
requirements on the classifier are relaxed, and the invariance
for all input objects is ensured. On the other hand, the main
drawback of using invariant feature spaces is the processing
time (it can be very long) needed to extract the features before
the classifier can be employed.

Fig. 1. Example of one Neural Network for object detection
In this work, we focus on a neural classifier that uses
Fourier descriptors as inputs for the neural networks. As said
before, a deterministic way to define the number of hidden
layers and the number of neurons does not exist. Hence in our
case, referring to the block scheme of Fig. 1, each NN is made
of 4 layers: (i) one input-layer composed by 168 neurons and
Vol. 3, No.8, 2012

98 | P a g e
equal to the size of the input patterns; (ii) two hidden-layers
composed by 8 neurons; (iii) one output-layer characterized by
only one neuron. Linear activation functions have been
applied to both the input and the output-layer, while non-linear
activation functions (in particular, sigmoid functions) have
been chosen for the hidden-layer. Then, the overall scheme of
the Neural Classifier (NC), obtained by a combined system of
these NNs, is depicted in Fig. 2.
The output value of each NN can range between -1 and 1
depending of the input pattern and each input pattern contains
168 Fourier descriptors referred to a specific ISAR image (i.e.
to a specific target). For example, let us now consider a NN
trained for recognizing the target "TG1": to a more and more
positive value of the network output corresponds a higher and
higher probability that the input pattern belongs to the
(correct) class TG1. Negative values of the output mean that
the input pattern is not an element of the considered class.
Therefore, the proposed NC takes a pattern made of 168
elements as input pattern (i.e. the Fourier descriptors of the
ISAR image) and recognizes the correct class of the target. In
particular, four different classes of targets, named TG1, TG2,
TG3, and TG4, have been used in this paper as a proof of
concept of the proposed combined classifier. When the NC
receives a pattern belonging to one of these four classes, it
returns an output value which is referred to the selected class.
As shown in Fig. 2, the NC is composed by two main boxes:
a) the inner classifier, CLi; b) the Final Decision-Maker,
FDM.

Fig. 2. Neural Classifier
Each CLi is referred to the related class (1, 2, 3 or 4) and
consists of a neural sub-system, composed by five NNs, able
to decrease the error probability of trained NNs. Indeed, each
CL is composed by five different-trained NNs used to classify
the same target class (see Fig. 3). The determining boxes,
DETi, with i = 1, 2, 3, 4, perform a very important task, which
is described in the hereinafter text. The rationale of our NC is
as follows. The neural sub-system CLi contains five NNs
which are separately trained, each having the aim to classify
the target TGi. At this point, for a fixed input pattern
belonging to the class TGi and if at least three NNs return the
correct output, the CLi makes a correct classification of the
input pattern as the TGi pattern. Obviously, the ideal operating
case is that all the five NNs perform the correct target
recognition but, to mitigate the possible output errors made by
one or two NNs, this majority rule is here applied. Therefore,
each CL exploits the majority rule to classify the input
patterns. When the NC receives a pattern to classify, only one
sub-system should be active at a time (OUTi is equal to 1
when CLi is active, -1 otherwise). The AVGi output returns
the average value of the five NNs outputs. The box FDM
simply selects the active input and shows it as the output of the
whole NC system. Nevertheless, it could happen that more
than one CLi is active at the same time anyway. In these cases,
the boxes DETi and FDM play a very important role,
exploiting the average values of the related neural sub-
systems. For example, given a generic input pattern TGx, if
the CL1 and CL2 outputs are both active, DET1 and DET2
look at the average values (AVG1 and AVG2) of the five NNs
both for the CL1 and CL2 neural sub-systems, respectively. At
this point, the box FDM operates as follows. If the average
value AVG1 is bigger than AVG2, the input pattern belongs to
the CL1 class (TG1). In the opposite case, the input pattern
belongs to the CL2 class (TG2). Obviously, for the worst
(limit) case in which all the CLi are active, the box FDM
selects the output referred to the CL that performs the biggest
average value.

Fig. 3 - Block scheme of the generic i-th CL block.
III. CONVENTIONAL MULTIMEDIA PROCESSING
The conventional multimedia processing methods for edge
detection usually exploit the Sobel, Prewitt, Roberts, and
Canny detectors [24]. In particular, the Sobel operator
performs 2-D spatial gradient measurement on an image and
so emphasizes regions of high spatial frequency that
correspond to edges. The Prewitt operator is an approximate
way to estimate the magnitude and orientation of the edge.
Then, the Roberts operator performs 2-D spatial gradient
measurement on an image and highlights regions of high
spatial frequency which often correspond to edges. Finally, the
Canny detector is a method to find edges by isolating noise
from the image without affecting the features of the edges in
Vol. 3, No.8, 2012

99 | P a g e
the image and then applying the tendency to find the edges
and the critical value for threshold.
The classical edge detector proposed almost 20 years ago
by Canny [25] performs remarkably well with its simplicity
and elegance. Cannys edge detector attempts to maximize
simultaneously localization and signal-to-noise ratio. A typical
implementation of the Canny edge detector is as follows [26]:
(i) first, smooth the image with an appropriate Gaussian filter
to reduce desired image details; (ii) determine gradient
magnitude and gradient direction at each pixel; (iii) if the
gradient magnitude at a pixel is larger than those at its two
neighbors in the gradient direction, mark the pixel as an edge.
Otherwise, mark the pixel as the background; (iv) remove the
weak edges by hysteresis thresholding. Indeed, in recent
comparisons of edge detector performances (see for example
[24] and references therein), Canny detector was the best or
one of the best. This is the reason why in the following of this
paper we have decided to compare the results obtained with
the new multimedia processing, described in the next Section,
with the multimedia processing obtained by the Canny
operator.
IV. PROPOSED MULTIMEDIA PROCESSING FOR TARGET
RECOGNITION
A. Database creation
We have used 4 targets, corresponding to 4 different
military aircrafts: one MIG-29, one F-104, one F-22, and one
Eurofighter-Typhoon. The ISAR images of these targets,
provided by the multinational firm MBDA (Rome, Italy) are
represented in Fig. 4. Then, we have created a database of
ISAR images composed of more than 500 images, each one
representing a target with a different azimuth angle, as shown
for example in Fig. 5 for 15 different angles for the
Eurofighter Typhoon target. The training and validation tests
for each target class are then made of 30 and 120 ISAR
images, respectively. All the NNs described in the previous
section have been trained by using the well-know Levenberg-
Marquardt back-propagation algorithm [27].
B. Pre-Processing
ISAR images are usually affected by a multiplicative noise
known as speckle noise. This is due to the interferences
produced by radar waves and results in light and dark pixels in
the ISAR image that drastically reduce the image quality.
Automatic interpretation of the image as well as performing
shape and features extraction become cumbersome issues to be
implemented. Therefore, image pre-processing is the first and
crucial phase to be addressed in order to reduce the speckle
effects, and improve the image quality. A great number of
different filters have been proposed in the open literature, such
as the Frost [28], Lee [29], and median [18] filters. Since our
ISAR images are affected by low speckle noise values,
following the same approach of [18], we have used a linear
filter followed by a median filter to improve the image quality.
It has to be noted that, in case of images highly corrupted by
noise, the median filter has been replaced by a Lee filter [28]
to facilitate the automatic segmentation process.
C. Object Detection
The shape extraction process, i.e. the process by which the
contour plots are extracted, is here performed using the
cascade of two different methods. First, we apply to the ISAR
image the SUSAN algorithm [18], and then the output of the
SUSAN method is processed by a recently introduced level set
evolution (LSE) method, called distance regularized level set
evolution (DRLSE) [19].
Fig. 4 - ISAR images of the target: a) MIG-29; b) Eurofighter-
Typhoon; c) F-104; d) F-22
In particular, the SUSAN method is here used to extract
pixels from the ISAR image belonging to two different
regions: target pixels and background pixels. Then the DRLSE
algorithm is used as an edge linking method, to extract the
target contour.
More in details, each pixel in the input ISAR image is
processed with a circular mask (named also window or
kernel), and the sum of grayscale comparison between the
mask center (Nucleus) and a local mask area (known as the
USAN, Univalue Segment Assimilating Nucleus) is
calculated. The mask is placed at each point in the input
image, and then the brightness of each pixel within the USAN
area is compared with the nucleus (center point), as follows:
( )
0 0
0
0 0
( , ) ( , ) 1
,
( , ) ( , ) 0
i i
i
i i
im x y im x y t if
C P P
im x y im x y t if
<
=

>
(1)
where P
0
and P
i
correspond to pixel of the nucleus and any
pixel of USAN area, respectively.
Then, im(x
i
, y
i
) is the gray level of the pixel that have
coordinate (x
i
, y
i
), while t stands for the brightness difference
threshold. The comparison expressed by eq. (1) is performed
for each pixel within the mask and the sum S of all these
comparisons is evaluated. Finally, the sum S is compared with
a fixed threshold G (namely the Geometric threshold [28])
which is set to of the maximum value which can take S.
Vol. 3, No.8, 2012

100 | P a g e

Fig. 5 - ISAR images of the target Eurofighter-Typhoon with 15
different azimuth angles.
The value of the treated pixel is then replaced by the
following:
0
( )
0
G S if S G
R P
otherwise
<
=

(2)
Like in [18], we consider the application of the SUSAN
algorithm as a pre-processing step, in order to segment the
input image into two regions of pixels, the ensemble
composed by the target pixels and the other one full of
background pixels. Hence, we can we add another condition in
the standard SUSAN algorithm, modifying eq. (2) as follows:
0 0 0 0 0
( ) ( , ) ( , )
0
G S if S G
R P im x y if im x y t
otherwise
<
= >
(3)
where the threshold t has been chosen according to the
following [1]:
1 1
( , )
M N
i j
i j
k
t im x y
MN
= =
=

(4)
and k is a constant that depends on the image size-to-target
ratio. Finally, the DRLSE method of [18] is applied to extract
the target shape from the results of the modified SUSAN
algorithm.
D. Features Extraction
The Fourier descriptors (FD) have been frequently used as
features for image processing, remote sensing, shape
recognition and classification [30]. They are chosen
accordingly to their good performance in recognition systems
and their implementation simplicity and efficiency. In fact,
they are invariant to geometrical transformations, such as
translation, scaling and rotation. The authors in [18] have used
the centroid distance as shape signature. This distance is
expressed by the distance between boundaries. Conversely,
here we have applied a simple discrete Fourier transform
(DFT) on the shape boundaries, extracted by the previously
described methods. In particular, we have identified the
boundary of a target by means of a (complex) vector, whose
elements are the coordinates of the contour points. Then, we
have applied the 1D-DFT to this (complex) vector, obtaining
the FDs of the target. The obtained FDs are invariant to
geometrical transformations.
Finally, in order to decrease the computational complexity
of the entire system, we have constrained the vector length to
168 elements, corresponding to 168 FDs of the targets
contour. These FDs have been used as the inputs of the new
ATR technique, detailed in the following Section.
V. PERFORMANCE ANALYSIS
In this Section, we first discuss our results obtained
through the proposed multimedia processing technique, in
terms of shape and edge extraction.
Then we present the results obtained through our neural
classifier in terms of mean detection probability, comparing
our results with state of the art detectors.
Fig. 6 - Target F22: a) noisy ISAR image; b) filtered ISAR image;
c) new edge-linking method; d) edge-linking method (Canny).
A. Results about shape and edge extraction
Here, we discuss the results of the proposed multimedia
processing versus the conventional Canny detector. In the pre-
processing steps, we have used a SUSAN circular mask
composed by 37 pixels with a radius of 3 pixels. Referring to
Fig. 6, the noisy ISAR images are first pre-processed to reduce
the speckle noise (Fig. 6a), then the SUSAN method (Fig. 6b)
is applied in order to extract the edge of the target (image
segmentation).
Finally, the DRLSE method (Fig. 6c) and the Canny
detector (Fig. 6d) are used as linking-edge techniques and the
FDs are computed form both the images, as explained in
Section IV. It is clearly evident from the figures that the edges
obtained with the Canny method (see also Fig. 7 and Fig. 8)
are characterized by a poorer quality in respect to the ones
obtained with the new multimedia processing.
a) b)
c) d)
Vol. 3, No.8, 2012

101 | P a g e

Fig. 7 - Target Eurofighter Typhoon, edges obtained by means of
the: a) Canny detector; b) new multimedia processing.

Fig. 8 - Target F-104, edges obtained by means of the: a) Canny
detector; b) new multimedia processing.
B. Results about target classification
The 168 FDs, describing the specific target under
investigation, are then passed to the classification step, for the
training and validation phase. As previously detailed, each
ISAR image is characterized by a pattern of 168 Fourier
descriptors.
The confusion matrix for the validation test, for the four
targets examined in this paper, is shown in Tab. 1 and in Tab.
2 for the Canny detector and the new multimedia processing,
respectively.
In particular, we have reported the percentage of correct
detection, indicated by bold numbers, and the percentage of
errors (false recognition) in the tables. For example, the target
TG1 is recognized with a detection probability of more than
93% with the new multimedia processing, while the Canny
method reaches only a percentage of 90.83%.
Then, the target TG4 is detected with lower probability in
both cases and it is automatically identified as TG1 with a
percentage of 5.0% or 33.33%, for the new and Canny
processing respectively. Notwithstanding the last two targets
are characterized by lower detection probabilities, the obtained
results, by the new multimedia processing, seem really
promising since the NC is able to achieve quite large
percentage values of correct object detection. In particular, this
is due to the bad performance of the Canny detector for
different azimuth angles.
See for example Fig. 7 (a, b) and Fig. 8 (a, b) where the
edges extracted with the Canny detector and the new
processing are compared. In particular, two targets of interest
are considered: the Eurofighter Typhoon and the F-104,
respectively. It is clearly visible from the figures that, in the
case of the Canny detector, the smoothing effect due to the
segmentation process do not allow to correctly extract the
edges of the detected object.
Tab. 1. Object detection performance of our neural classifier with the
Canny multimedia processing

TG1 TG2 TG3 TG4
TG1 90.83% 5.81% 0.00% 3.34%
TG2 60.84% 25.82% 0.81% 12.50%
TG3 73.33% 15.01% 0.81% 10.83%
TG4 33.33% 12.51% 0.00% 54.17%

Tab. 2. Object detection performance of our neural classifier with the
new multimedia processing

TG1 TG2 TG3 TG4
TG1 93.33% 0.00% 6.66% 0.00%
TG2 2.55% 96.66% 0.83% 0.00%
TG3 22.50% 0.00% 70% 7.50%
TG4 5.00% 0.00% 28.33% 66.66%
Another advantage of our processing lies in the combined
structure of the proposed classifier. In fact, we always exploit
the most suitable neural sub-system for each target class, i.e.
the inner classifier CL composed by 5 NNs. The CL sub-
system aims at decreasing the error probability with respect to
the conventional case in which only one NN decides about the
class of the target. Moreover, the further use of average values
(performed by the DETi sub-systems) improves the
performances of the proposed classifier when ambiguities
exist at the outputs of the neural sub-systems. Finally, in order
to prove the efficiency of our NC with respect to state of the
art detectors, a comparative analysis is shown in Tab. 3. In
particular, the mean values of the correct classification
probability are reported in tab. 2, for our proposed NC (mean
recognition of 81.6%) and for two classifiers proposed in [18].
In particular, the authors in [18] obtain a mean recognition
percentage of 75.98%, using a K Nearest-Neighbor classifier
(K-NN) and then, they improve the system performances
exploiting the Support Vector Machine (SVM) algorithm
(reaching a mean detection value of 80.37%). However, this
further approach appears less effective than the one here
presented.
Tab. 3. Comparison between our method and the classifiers of [18]

Proposed NC K-NN SVM
Mean Detection 81.60% 75.98% 80.37%
VI. CONCLUSIONS AND FUTURE WORKS
This work has proposed a new automatic target classifier,
based on a combined neural networks system, by ISAR image
processing. The novelty introduced in our work is twofold. We
have first introduced a novel automatic classification
procedure, and then we have discussed about an improved
multimedia processing of ISAR images for automatic object
detection.
Vol. 3, No.8, 2012

102 | P a g e
We have exploited a neural classifier, composed by a
combination of 20 feed-forward artificial neural networks. The
classifier is used to recognize aircraft targets extracted from
ISAR images. The combination of two image processing
techniques, recently introduced in literature, is exploited to
improve the shape and features extraction process. Then,
Invariant Fourier descriptors are computed and used as input
features to our combined system. Performance analysis is
carried out in comparison with conventional multimedia
processing techniques as well as with classical automatic
target recognition systems. Numerical results, obtained from
wide simulation trials, evidence the efficiency of the proposed
approach for the application to automatic aircraft target
recognition. Future works will regard the improvement of the
performances of the single NNs by applying suitable
optimization algorithms to the NNs learning process. Indeed, it
is possible to operate a multivariate function decomposition
with the aim to perform the learning optimization of Multi-
Input-Single-Output (MISO) feed-forward Neural Networks
[31]. Furthermore, applying new powerful search algorithms
(e.g. meta-heuristic algorithms such as those shown in [32]-
[35]) can increase the generalization feature of the Neural
Networks in particular after they are built by using a
partitioning of the domain (see [36]). Finally, other more
elaborate algorithms could be applied to the multimedia
processing by starting from novel concepts already existing in
literature (e.g. [37][38]).
ACKNOWLEDGMENT
The authors acknowledge the support of MBDA, Rome
(Italy), for providing the ISAR images.
REFERENCES
[1] F. L. Lewis, Wireless Sensor Networks, in Smart Environments:
Technologies, Protocols, and Applications, ed. D.J. Cook and S.K. Das,
John Wiley, New York, 2004.
[2] I. F. Akyildiz , T. Melodia, K. R. Chowdhury, A survey on wireless
multimedia sensor networks, Elsevier Computer Networks, vol. 51, pp.
921-960, 2007.
[3] Editorial: Signal processing techniques for ISAR and feature
extraction, IET Sign. Proc., vol. 2, no. 3, pp. 189 191, 2008.
[4] V. Zeljkovic, Q. Li, R. Vincelette, C. Tameze, F. Liu, F., Automatic
algorithm for inverse synthetic aperture radar images recognition and
classification, IET Radar, Sonar & Navig., vol. 4, no. 1, pp. 96 109,
2010.
[5] D. Pastina, C. Spina, Multi-feature based automatic recognition of ship
targets in ISAR, IET Radar, Sonar & Navig., vol. 3, no. 4, pp. 406
423, 2009.
[6] B. Leibe, A. Leonardis, and B. Schiele(2004), Combined Object
Categorization andSegmentation with an Implicit Shape Model, Proc.
ECCV Workshop Statistical Learning in Computer Vision, 2004.
[7] I. Kokkinos, P. Maragos, Synergy between Object Recognition and
image segmentation using Expectation and Maximization Algorithm,
IEEE Trans. on Pattern Analysis and Machine Intelligence (PAMI), Vol.
31(8), pp. 1486-1501, 2009.
[8] M. N. Saidi, B. Hoeltzener, A. Toumi, A. Khecnhaf, D. Aboutajdine,
Recognition of ISAR images: target shapes features extraction, IEEE
3
rd
Int. Conf. on Inf. and Commun. Techn.: from Theory to Appl.
(ICTTA), pp. 16, Apr. 2008.
[9] G. T. Maskall, An application of nonlinear feature extraction to the
classification of ISAR images, Proc. of RADAR 2002, pp. 405408,
2002.
[10] M. Vespe, C. J. Baker, H. D. Griffiths, Outline structural representation
for radar target classification based on non-radar templates, IEEE Int.
Conf. on Radar (CIE '06), pp. 14, 2006.
[11] J. Manikandan, B. Venkataramani, M. Jayachandran, Evaluation of
edge detection techniques towards implementation of automatic target
recognition, Int. Conf. on Computational Intelligence and Multimedia
Appl., vol. 2, pp. 441445, Dec. 2007.
[12] A. Toumi, B. Hoelzener, A. Khenchaf, Using Watersheds segmentation
on ISAR image for automatic target recognition, Int. Conf. on Digital
Information Management, pp. 285-290, 2007.
[13] F. Wang, W. Sheng, X. Ma, H. Wang, Target automatic recognition
based on ISAR image with wavelet transform and MBLBP, Int. Symp.
on Sign. Syst. and Electr. (ISSSE), vol. 2, pp. 1 4, 2010.
[14] K. T. Kim, D. K. Seo, H. T. Kim, Efficient classification of ISAR
images, IEEE Trans. Antennas Propag., vol. 53, no. 5, pp. 16111621,
2005.
[15] M. Martorella, E. Giusti, L. Demi, Z. Zhou, A. Cacciamano, F. Berizzi,
B. Bates, Automatic target recognition by means of polarimetric ISAR
images: a model matching based algorithm, Int. Conf. on Radar, pp. 27-
31, Sept. 2008.
[16] H. Yuankui, Y. Yiming, Automatic target recognition of ISAR images
based on Hausdorff distance, 1
st
Asian and Pacific Conf. APSAR, pp.
477-479, Nov. 2007.
[17] P. T. Dung, Combined neural networks for radar target recognition
from radar range profiles, IEEE Int. Conf. on Adv. Techn. for Commun.,
2008.
[18] M. N. Saidi, A. Toumi, B. Hoeltzener, A. Khenchaf, D. Aboutajdine,
Aircraft Target Recognition: A novel approach for features extraction
from ISAR images, IEEE Int. Radar Conf. - Surveillance for a Safer
World, pp. 1 5, 2009.
[19] C. Li, C. Xu, C. Gui, M. D. Fox, Distance Regularized Level Set
Evolution and Its Application to Image Segmentation, IEEE Trans. on
Image Proc., vol. 19, no 12, pp. 3243 3254, 2010.
[20] C. M. Bishop, Neural Networks for Pattern Recognition, Oxford
University Press, 1995
[21] V. M. Orlenko, A. S. Bityutskiy, Simulation of aerial target radar
recognition using neural networks, Proc. of the IRS, pp. 729-733, 2003.
[22] A. Zyweck, R. E. Bogner, Radar target classification of commercial
aircraft, IEEE Trans. on Aerospace and Electr. Syst., vol. 32, no. 2, pp.
598-606, 1996.
[23] S. Hudson, D. Psaltis, Correlation filter for aicraft identification from
radar range profiles, IEEE Trans. on Neural Networks, vol. 29, no. 3,
pp. 741-748, 1193.
[24] F. A. Pellegrino, W. Vanzella, V. Torre, Edge Detection Revisited,
IEEE Trans. on Systems, Man, And CyberneticsPart B: Cybernetics,
vol.. 34, no. 3, pp. 1500-1518, June 2004.
[25] J. Canny, A Computational Approach To Edge Detection, IEEE
Trans. Pattern Analysis and Machine Intelligence, vol. 8, no. 6, pp.:679
698, 1986.
[26] L. Ding, A. Goshtasby, On the Canny edge detector, Elsevier Pattern
Recognition, vol. 34, pp. 721-725, 2001.
[27] P. R. Gill, W. Murray, M. H. Wright, The Levenberg-Marquardt
Method, in Practical Optimization. London: Academic Press, pp. 136-
137, 1981.
[28] V. S. Frost, J. A. Stiles, A. Josephine, K. S. Shanmu-gan, J. C.
Holtzman, A Model for Radar Images and Its Application to Adaptive
Digital Filtering of Multiplicative Noise, IEEE Trans. on Pattern
Analysis and Machine Intellig., vol. PAMI-4, no. 2, pp. 157-166, 1982.
[29] J. S. Lee, Speckle Analysis and Smoothing of Synthetic Aperture Radar
Images, Computer Graphics and Image Processing, vol. 17, pp. 24-32,
1981.
[30] M. Sarfraz, Object Recognition using Fourier Descriptors: Some
Experiments and Observations, IEEE Int. Conf. on Computer Graphics,
Imaging and Visualisation, pp.281-286, 2006.
[31] F. Riganti Fulginei, A, Salvini and M. Parodi, "Learning optimization of
neural networks used for mimo applications based on multivariate
functions decomposition". Inverse Problems in Science & Engineering
(IPSE). vol. 20, p. 29-39, 2012
Vol. 3, No.8, 2012

103 | P a g e
[32] F. Riganti Fulginei, A. Salvini and G. Pulcini, "Metric-topological-
evolutionary optimization". Inverse Problems in Science & Engineering
(IPSE), vol. 20, p. 41-58, 2012
[33] S. Coco, A. Laudani, F. Riganti Fulginei and A. Salvini, "Shape
optimization of multistage depressed collectors by parallel evolutionary
algorithm". IEEE Transactions on Magnetics, vol. 48, p. 435-438, 2012
[34] F. Riganti Fulginei, A. Salvini, A. Laudani and S. Coco, "Team 22
problem approached by a hybrid artificial life method". Compel, vol. 31,
p. 816-826, 2012
[35] F. Riganti Fulginei, A. Laudani, A. Salvini, S. Coco, G. Pollicino and G.
Pulcini, "TWT magnetic focusing structure optimization by parallel
evolutionary algorithm". Compel, vol. 31, p. 1338-1346, 2012
[36] F. Riganti Fulginei and A. Salvini, "Neural network approach for
modelling hysteretic magnetic materials under distorted excitations".
IEEE Transactions on Magnetics, vol. 48, p. 307 -310, 2012
[37] F. Benedetto and G. Giunta, "A Fast Time-Delay Estimator of PN
Signals". IEEE Transactions on Communications, vol. 59, p. 2057 -
2062, 2011
[38] F. Benedetto, G. Giunta and S. Bucci, "A Unified Approach for Time-
Delay Estimators in Spread Spectrum Communications". IEEE
Transactions on Communications, vol. 59, p. 3421 - 3429, 2011

Vol. 3, No.8, 2012

104 | P a g e
An Effective Identification of Species from DNA
Sequence: A Classification Technique by Integrating
DM and ANN

Sathish Kumar S
Research Scholar
Dr MGR Educational and Research Institute University.
Chennai, Tamil Nadu, India
Tel: +919886372152
Dr.N.Duraipandian, M.E., Ph.D
Vice Principal
Velammal Engineering College
Ambattur Redhills Road, Chennai,
Tamil Nadu, India

Abstract Species classification from DNA sequences remains as
an open challenge in the area of bioinformatics, which deals with
the collection, processing and analysis of DNA and proteomic
sequence. Though incorporation of data mining can guide the
process to perform well, poor definition, and heterogeneous
nature of gene sequence remains as a barrier. In this paper, an
effective classification technique to identify the organism from its
gene sequence is proposed. The proposed integrated technique is
mainly based on pattern mining and neural network-based
classification. In pattern mining, the technique mines nucleotide
patterns and their support from selected DNA sequence. The
high dimension of the mined dataset is reduced using Multilinear
Principal Component Analysis (MPCA). In classification, a well-
trained neural network classifies the selected gene sequence and
so the organism is identified even from a part of the sequence.
The proposed technique is evaluated by performing 10-fold cross
validation, a statistical validation measure, and the obtained
results prove the efficacy of the technique.
Keywords- Pattern Generation; DNA Sequence; Pattern Support;
Mining; Neural Network.
I. INTRODUCTION
Bioinformatics is a rapidly growing area of computer
science [19] that deals with the collection, organization, and
analysis of Deoxyribonucleic acid (DNA) and protein
sequence [18]. Today it addresses the formal and practical
issues that occur in the management and analysis of genomic
and proteomic data because it includes the formation and
development of databases, algorithms, computational and
statistical technique, and hypothesis [1].
Genomic signal processing (GSP) is a relatively new area
in bio-informatics that uses traditional digital signal
processing techniques to deal with digital signal
representations and analysis of genomic data [2] [12]. GSP
gains biological knowledge by the analysis, processing, and
use of genomic signals and translates the gained biological
knowledge into systems-based applications [3]. Integration of
signal processing theories and methods with global
understanding of functional genomics with significant
emphasis on genomic regulation is the main objective of GSP
[4].
The whole DNA of a living organism is known as its
Genome [5]. Genomic signals carry genomic information to
all the processes that take place in an organism [6]. Essentially
DNA is a nucleic acid that has two long strands of nucleotides
twisted in the form of a double helix and its external backbone
is made up of alternating deoxyribose sugar and phosphate
molecules. The nitrogenous bases Adenine, Guanine, Cytosine
and Thymine are present in the interior portion of the DNA in
pairs [13] [9]. DNA and proteins can be mathematically
represented as character strings, where each character is a
letter of the alphabet [6] [10] [11].
One of the vital tasks in the study of genomes is gene
identification [7]. DNA analysis utilizes methods such as
clustering [20], data mining [21] [22] [23], gene identification
[24] and gene regulatory network modeling [25] [26]. These
methods present cutting edge research topics and
methodologies for the purpose of facilitating collaboration
between researchers and bioinformaticians. Mining
bioinformatics data is a rising field at the intersection of
bioinformatics and data mining [14]. Some of them belong to
the category of data mining that decides whether or not an
example not yet noticed is of a predefined type. Increased
availability of huge amount of biomedical data and the
expectant need to turn such data into useful information and
knowledge is the main reason for the recent increased
attention in data mining in the biomedical industry.
Large number of research works that incorporate data
mining in bioinformatics for different purposes are available in
the literature [15] [16] [17]. . A few important such researches
are reviewed in section 2. One important research of this type
is the identification of species or name of an organism from its
gene sequence. Characterization of the unknown
environmental isolates with the genomic species is not easy
because genomic species are especially heterogeneous and
poorly defined [8]. Identifying the species or the organism
from its gene sequence is a challenging task. In this paper, we
propose a classification technique to effectively classify the
species or name of an organism from its DNA sequence. This
technique is detailed with mathematical formulations and
illustrations in section 3. Section 4 discusses the
implementation results and Section 5 concludes the paper.
Vol. 3, No.8, 2012

105 | P a g e
II. RELEATED WORKS
Plenty of research works deals with the mining knowledge
from the genomic sequences. Some of the recent research
works are briefly reviewed here. Riccardo Bellazzi et al. [27]
have discussed that in the past years, the gene expression data
analysis that are aiming at complementing microarray analysis
with data and knowledge of various existing sources has
grown from being purely data-centric to integrative. Focusing
on the evolution of gene expression data mining techniques
toward knowledge-based data analysis approaches, they have
reported on the overabundance of such techniques.
Particularly, latest developments in gene expression-based
analysis methods utilized in association and classification
studies, phenotyping and reverse engineering of gene networks
have been discussed.
The gene expression data sets for ovarian, prostate, and
lung cancer was examined by Shital Shah et al. [28]. For
genetic expression data analysis, an integrated gene-search
algorithm was presented. For making predictions and for data
preprocessing (on partitioned data sets) and data mining
(decision tree and support vector machines algorithms), a
genetic algorithm and correlation-based heuristics was
included in the their integrated algorithm. The knowledge,
which was obtained by the algorithm, has high classification
accuracy with the capability to recognize the most important
genes. To further improve the classification accuracy, bagging
and stacking algorithms were employed. The results were
compared with the literary works. The cost and complexity of
cancer detection and classification was eventually condensed
by the mapping of genotype information to the phenotype
parameters.
Locating motif in bio-sequences, which is a very
significant primitive operation in computational biology, was
discussed by Hemalatha et al. [29]. Computer memory space
requirement and computational complexity are few of the
computational requirements that are needed for a motif
discovery algorithm. To overcome the intricacy of motif
discovery, an alternative solution integrating genetic algorithm
and Fuzzy Art machine learning approaches was proposed for
eradicating multiple sequence alignment process. The results
that were attained by their planned model to discover the motif
in terms of speed and length were compared with the enduring
technique. By their technique, the length of 11 was found in
18 sec and length of 15 in 24 sec, whereas the existing
techniques found length of 11 in 34 sec. When compared to
other techniques, the proposed one has outperformed the
accepted existing technique. By employing MATLAB, the
projected algorithm was put into practice and with large DNA
sequence data sets and synthetic data sets, it was tested.
An interactive framework which is based on web for the
analysis and visualization of gene expressions and protein
structures was described by Ashraf S. Hussein [30]. The
formulation of the projected framework encountered various
confronts because of the variety of significant analysis and
visualization techniques, moreover to the survival of a
diversity of biological data types, on which these techniques
function. Data incorporated from heterogeneous resources, for
instance expert-driven data from text, public domain databases
and various large scale experimental data and the lack of
standard I/O that makes it difficult to integrate the most recent
analysis and visualization are the two main challenges that
directed the formulation of the current framework. Hence, the
basic novelty in their proposed framework was the integration
of the state-of -art techniques of both analysis and
visualization for gene expressions and protein structures
through a unified workflow. Moreover, a wide range of input
data types are supported by it and three dimensional
interactive outputs ready for exploration by off-the-shelf
monitors and immersive, 3D, stereo display environments can
be exported by it using Virtual Reality Modeling Language
(VRML).
A stomach cancer detection system, which is on the basis
of Artificial Neural Network (ANN) and the Discrete Cosine
Transform (DCT), was developed by Ahmad M. Sarhan [31].
By employing DCT the projected system extracted the
classification features from stomach microarrays. The
extracted characteristics from DCT coefficients were applied
to an ANN for further classification (tumor or non-tumor). The
microarray images that were employed were acquired from the
Stanford Medical Database (SMD). Simulation results has
illustrated that a very high success rate was produced by the
proposed system.
The challenging issue in microarray technique which was
to analyze and interpret the large volume of data was
discussed by Valarmathie et al. [32]. This can be made
possible by the clustering techniques in data mining. In
hierarchical and k-means clustering techniques which are hard
clustering, the data is split into definite clusters, where each
cluster has exactly one data element so that the result of the
clustering may be wrong many times.
The problems that are addressed in hard clustering can be
resolved in fuzzy clustering technique. Amid all fuzzy based
clustering, fuzzy C-means (FCM) is best suited for microarray
gene expression data. The problem that is related with fuzzy
C-means was the amount of clusters that are to be generated
for the given dataset and that needs to be notified first. By
combining the technique with a popular probability related
Expectation Maximization (EM) algorithm, it can be solved to
model the cluster structure of gene expression data and it has
offered the statistical frame work. Determining the accurate
number of clusters and its efficient interpretation is the main
purpose of the projected hybrid fuzzy C-means technique.
Explorative studies in support of solutions to facilitate the
analysis and interpretation of mining results was described by
Belmamoune et al. [33]. A solution that was located in the
extension of the Gene Expression Management System
(GEMS) was described, i.e. an integrative framework for
spatio-temporal organization of gene expression patterns of
zebra fish to a framework that supports data mining, data
analysis and patterns interpretation.
As a proof of principle, the GEMS is provided with data
mining functionality which is appropriate to monitor spatio-
temporal, thus generating added value to the submission of
data for data mining and analysis. On the basis of the
availability of domain ontologies, the analysis of the genetic
networks was done which vigorously offers the meaning to the
Vol. 3, No.8, 2012

106 | P a g e
discovered patterns of gene expression data. Grouping of data
mining with the already accessible potential of GEMS
considerably augments the existing data processing and
functional analysis strategies.
III. THE INTEGRATED TECHNIQUE FOR SPECIES
CLASSIFICATION
The proposed species classification technique classifies
species based on the given DNA sequence. The DNA
sequence is comprised of four basic nucleotides, Adenine (A),
Guanine (G), Cytosine (C) and Thymine (T). Every species
has a long DNA sequence, which is formed by the four
nucleotides.
The DNA sequence defines the attributes, nature and type
of the species. The proposed technique is an integration of
data mining and artificial intelligence. In the proposed
technique, firstly, nucleotide patterns are mined from the
sequence. The mined patterns form a nucleotide pattern
database with higher dimension. So, secondly, the dimension
of the pattern database is reduced by MPCA. Finally, the
dimensionality reduced pattern database is used to train the
neural network. The technique is described in the further sub
sections.
The proposed species classification technique classifies a
species based on its DNA sequence. The four basic
nucleotides, Adenine (A), Guanine (G), Cytosine (C) and
Thymine (T) are the building blocks of the long DNA
sequence in every species. A DNA sequence defines the
attributes, nature and type of the species. The proposed
technique is developed by integrating data mining and
artificial intelligence techniques. Firstly, the proposed
technique mines the nucleotide patterns from the sequence and
forms a high dimensional nucleotide pattern database with the
mined patterns.
Secondly, the technique uses MPCA and reduces the
dimension of the pattern database. Finally, the dimensionality
reduced pattern database is used to train a neural network. The
following sub sections elaborately describe this technique.
A. Mining Nucleotide Patterns from DNA sequence
The first and initial stage of the proposed technique mines
the nucleotide pattern from the DNA sequence. At this stage,
patterns formed by different combinations of nucleotides are
mined using a novel mining algorithm. Let be the DNA
sequence, which is a combination of four nucleotides A, G, C
and T. For instance, a sample DNA sequence is given as
CGTCGTGGAA.
From the sequence, the mining algorithm extracts different
nucleotide patterns and their support. The algorithm is
comprised of two stages, namely, pattern generation and
support finding. In pattern generation, patterns with different
length are generated whereas in support finding, support
values for every generated pattern are determined from the
DNA sequence. The basic structure of the algorithm is given
as a block diagram in Fig. 1.
Figure 1. Block diagram of the pattern mining algorithm
1) Pattern generation
In pattern generation, different possible combinations of
nucleotide base pairs are generated. As a reference, a base set
B is generated with cardinality 4 | | = B , which has the
elements } , , , { T C G A . Let,
max
, , 2 , 1 }; { L l P
l
A = , be the
pattern set to be generated, where,
max
L is the maximum
length of a pattern in a pattern set. The pattern set is generated
as follows
{ } { } { } ) ( ) ( ) (
2 1
1
) ( ) (
l
k
l
k
l
a B a B a B P P l l A =

(1)
Where,
l l
B k | | , , 2 , 1
) (
A = , | | , , ,
2 1
B a a a 1
l
s s A
and { } ) ( ) ( ) (
2 1 l
a B a B a B A is a set of different
combinations of nucleotide bases. Eq. (1) operates with the
criterions, { } { } ) ( ) (
1
l l
k
l
k
l
P P _
and
{ } { }
l l
P a B a B a B _ ) ( ) ( ) (
2 1
A . Eq. (1) formulated for
pattern generation is analyzed using two examples.
Example 1: To generate a two length pattern set
2
P , i.e.
2 = l . Here, two indexing variables
1
a and
2
a are
generated. At every k i.e. 16 1 to k = and for its
corresponding
1
a and
2
a , the obtained { } ) ( ) (
2 1
a B a B and
{ }
1
2
) 2 (
k
P are tabulated in Table II.
Hence,
2
P is obtained as
} { } {
2
TT TC, TG, TA, CT, CC, CG, CA, GT, GC, GG, GA, AT, AC, AG, AA, P =
by integrating
2
P obtained from every
th
k iteration. From
max
, , ,
2 1 L
P P P K , a consolidated pattern set P , which is the
required pattern to be generated, is obtained as
max
2 1 L
P P P P A = . The cardinality of P can be
determined as

=
=
max
1
| | | |
L
l
l
P P .

Vol. 3, No.8, 2012

107 | P a g e
TABLE I. DIFFERENT COMBINATIONS AND PATTERN SETS GENERATED FOR EVERY
1
a ,
2
a AND k
k a1 a2 {B(a1) B(a2)} { }
1
2
) 2 (
k
P
1 1 1 AA {}
2 1 2 AG {AA}
3 1 3 AC {AA, AG}
4 1 4 AT {AA, AG, AC}
5 2 1 GA {AA, AG, AC, AT}
6 2 2 GG {AA, AG, AC, AT, GA}
7 2 3 GC {AA, AG, AC, AT, GA, GG}
8 2 4 GT {AA, AG, AC, AT, GA, GG,GC}
9 3 1 CA {AA, AG, AC, AT, GA, GG,GC, GT}
10 3 2 CG {AA, AG, AC, AT, GA, GG,GC, GT, CA}
11 3 3 CC {AA, AG, AC, AT, GA, GG,GC, GT, CA, CG}
12 3 4 CT
{AA, AG, AC, AT, GA, GG,GC, GT, CA, CG,
CC }
13 4 1 TA
CC,CT}
14 4 2 TG
CC,CT,TA}
15 4 3 TC
CC,CT,TA,TG}
16 4 4 TT
CC,CT,TA,TG,TC}

2) Determination of Pattern Support
The support, which has to be determined for every
extracted pattern, describes the DNA attribute. By performing
a window based operation over the sequence g , the support
can be determined. Window of sequences are determined for
different lengths as follows
) 1 , , 1 , ( ) ( + + = l j j j g j w
l
A (2)
Once the window of sequences is extracted support is
determined for the mined patterns. The pseudo code, which is
given below, describes the procedure to determine the support
for every pattern.

Figure 2. Pseudo code to determine support for every mined
Figure 3. ained C for each different length pattern hapattern
The obts the support for all the elements that are present in
the corresponding pattern set. From the mined pattern and its
corresponding support, a constructive dataset is generated.
Vol. 3, No.8, 2012

108 | P a g e
3) Constructive dataset generation
A raw dataset is generated using the aforesaid mining
algorithm. But the dataset is not constructive for further
operation. In this stage, a constructive dataset is generated
from the mined dataset, which comprises of patterns with
different lengths and their support.
To accomplish this, firstly the patterns which have length
2 > l are taken. From the pattern set, the modified and
constructive dataset is generated as given in Table 3.
TABLE II. A GENERAL STRUCTURE OF THE PROPOSED CONSTRUCTIVE
DATASET
A G C T
A C2(1) C2(2) C2(3) C2(4)
G C2(5) C2(6) C2(7) C2(8)
C C2(9) C2(10) C2(11) C2(12)
T C2(13) C2(14) C2(15) C2(16)
AA C3(1) C3(2) C3(3) C3(4)
AG C3(5) C3(6) C3(7) C3(8)
AC C3(9) C3(10) C3(11) C3(12)
AT C3(13) A A A A A A
M
M

In the constructive dataset, all the patterns except single
length pattern are considered. Hence, the dataset is of size
4 4
1
max
L
. The generated constructive dataset belongs to
a particular gene sequence. Similarly, the constructive dataset
for different sequences are generated. Hence, the final dataset
) (z
xy
G ;
1
max
4 , , 2 , 1

=
L
x A , 4 3 , 2 , 1 and y = and
g
N z , , 2 , 1 A = is obtained, which is subjected to further
processing.
B. MPCA-based Dimensionality reduction
In all tensor modes, the multilinear algorithm MPCA
captures most of the variation present in the original tensors by
seeking those bases in each mode that allow projected tensors
and performs dimensionality reduction [35]. Initially, in the
process of dimensionality reduction, the distance matrix for
every
th
z matrix is determined as follows, =
) ( ) ( z z
G D
(3)
Where,
=

=
1
0
) (
1
G
N
z
z
xy
G
xy
G
N
(4)

Using Eq. (3) and by determining the mean matrix for
) (z
G using Eq. (4), the distance matrix can be calculated.
Then with mode 2, tensor representations [34]
) (
1
z
T and
) (
2
z
T
are given to the obtained distance matrix. A projection matrix
+is determined as follows,

( ) = +

=
1
0
) ( ) (
G
N
z
T
z z
T T (5)
For both
) (
1
z
T and
) (
2
z
T , the projection matrix (
1
+ and
2
+ ) are determined using the generalized form of calculation
given in Eq. (5). For
1
+ and
2
+ , the corresponding
eigenvectors
1
E and
2
E and the corresponding Eigen values
1
and
2
are determined by subjecting the projection matrix
to a generalized eigenvector problem. The rows of the
eigenvector are arranged based on the index of the eigenvalues
sorted in the descending order. The modified eigenvector
'
1
E
and
'
2
E are obtained by transposing the arranged eigenvector.
The cumulatively distributed Eigen values for the sorted
eigenvalues are generally determined using the following
equation.
=
=
1 | |
0
'
x
sort
x
cdf
x
x
(6)
The sorted Eigen values
sort
x
and the cumulatively
distributed Eigen values
cdf
x
of Eq. (6), can be determined
as
cdf
x
sort
x
cdf
x
1
+ = (7)
Where,
sort cdf
0 0
= at 0 = x . The new dimension
T
is calculated from the obtained
'
x
, using a dimensional
threshold
T
D . To accomplish this, the indices of all
eigenvalues that satisfy the condition
T x
D >
'
are identified.
Then, by extracting the first
T
rows of
'
1
E and
'
2
E , the
corresponding dimensionality reduced eigenvectors
' '
1
E and
' '
2
E are determined. For the
' '
1
E and
' '
2
E , again tensor
matrices but
) (
1
z
T and
) (
2
z
T times are determined [35]. The
Vol. 3, No.8, 2012

109 | P a g e
process followed for projection matrix is repeated for the
tensor matrices to obtain
new
x
1
and
new
x
2
and
new
E
1
and
new
E
2
. The weight of both the tensor eigen values are
determined as
new
x
new
x
w
x
2 1
= . Then, the dimensionality
reduced matrix
) (z
G of size
T R
N N , is obtained by using
the MPCA projections [35], where,
R
N can be determined as
2 1
.
T T R
N = .
C. Classification using ANN
For
G
N gene sequences, the dimensionality reduced gene
patterns and their support are provided by the MPCA. Using
ANN, the class of the original sequence can be identified
using the dataset. Two classical operations, training and
testing are involved in the classification. The neural network is
trained using the
G
N pattern dataset. Here, the process is
performed using multilayer feed forward neural network,
depicted in Fig. 3.
R
N Input nodes,
H
N hidden nodes and
an output node are present in the network.
Before performing any task, the ANN must be trained.
Once trained, the ANN capably identifies the species by
finding the class of the gene sequence. The training phase and
classification phase of the ANN are described below.
1) Training Phase
Back Propagation (BP) algorithm is used to train the
constructed feed forward network. The step-by-step procedure
utilized in the training process is given below.
1. Assign arbitrary weights generated within the interval
to links between the input layer and hidden layer as
well as hidden layer and output layer.
2. Using Eq. (8), (9) and (10), determine the output of
input layer, hidden layer and output layer respectively
by inputting constructive dataset to the network.

Figure 4. The multilayer feed forward neural network used in the proposed technique
=
+ =
R
N
r
rq q
r G W h
1
) 1 (
) ( o ;
R
N r , , 2 , 1 A = ,
H
N q , , 2 , 1 A = (8)

) 1 (
) 1 (
2
) 2 (
1
1
h
h
e
e
h
=
(9)

) 2 (
h S = (10)

where, Eq. (8) is the basis function for the input layer and
Eq. (9) and (10) are the activation functions for hidden and
output layer, respectively.
1. Determine BP error using
( ) =

=
1
0
1
G
N
p
p T
G
S S
N
e (11)

where, e is the BP error,
T
S is the target output
Vol. 3, No.8, 2012

110 | P a g e
2. By adjusting the weights of all the neurons based
on the determined BP error, obtain new weights
using
W W W
old new
A + = (12)
In Eq. (12), the weight to be changed W A
depends on the rate of network learning and
the obtained network output
p
S for the
th
p
gene sequence and it is determined using the
formula e S W
p
. . = A .
3. Until the BP error gets minimized to a minimum
extent, repeat the process from step 2. The
termination criterion for practical cases, is
1 . 0 < e .
2) Classification Phase
In the classification phase, the network finds the class of a
given or test gene sequence and determines the species to
which it belongs. The same processes performed on the
training sequence are repeated for the test sequence . Using the
mined patterns and their support, the constructive nucleotide
dataset is generated. Subsequent to dimensionality reduction
of the generated dataset they are tested in a neural network.
The neural network decides the class of the species to which
the gene sequence belongs.
IV. IMPLEMENTATION RESULTS
The proposed technique is implemented in the working
platform of MATLAB (version 7.10) and the technique is
evaluated using the DNA sequence of two different organisms,
Brucella Suis and Caenorhabditis Elegans (C. Elegans). The
evaluation process is performed using 10-fold cross validation
test. Here, nucleotide patterns are mined with 5
max
= L . The
nucleotide patterns for 3 and l 2 = and their corresponding
support are given in Table III. In Fig. 5, different length
patterns and their support are depicted and the constructive
dataset that is generated from the pattern set is given in Table
IV.
TABLE III. MINED NUCLEOTIDE PATTERNS FROM THE DNA SEQUENCE
OF BRUCELLA SUIS AND C.ELEGANS (A) l 2 = AND (B) l 3 = (A PART OF
THE PATTERN IS GIVEN)
(a)
Species
S. No Pattern
Support
Brucella suis C-elegans
1 aa 169042 168149
2 ag 56284 59645
3 ac 53509 54824
4 at 100894 101778
5 ga 72354 72651
6 gg 45341 45368
7 gc 46001 43023
8 gt 53423 56002
9 ca 67662 69882
10 cg 47630 40344
11 cc 47205 44205
12 ct 57377 58316
13 ta 70670 73713
14 tg 67864 71687
15 tc 73159 70695
16 tt 171584 169717

(b)
Species
S. No Pattern
Support
1 aaa 86090 83349
2 aag 17627 18623
3 aac 18374 19422
4 aat 46951 46755
5 aga 19089 19871
6 agg 11100 10442
7 agc 11318 12264
8 agt 14777 17068
9 aca 17239 17901
10 acg 10577 9453
11 acc 10491 10676
12 act 15202 16794
13 ata 20142 21368
14 atg 15783 16752
15 atc 17406 16451
16 att 47563 47207

Figure 5. Support obtained for different length patterns

1199994
1199995
1199996
1199997
1199998
1199999
1200000
2 3 4 5
Pattern Length
S
u
p
p
o
r
t
Brucella suis
C-elegans
Vol. 3, No.8, 2012

111 | P a g e
TABLE IV. CONSTRUCTIVE DATASET GENERATED FROM THE MINED NUCLEOTIDE PATTERNS (A) l 2 = AND (B) l 3 = (A PART OF THE PATTERN IS GIVEN).
(A)
Species
S.No
A G C T A G C T
1 A 169042 56284 53509 100894 168149 59645 54824 101778
2 G 72354 45341 46001 53423 72651 45368 43023 56002
3 C 67662 47630 47205 57377 69882 40344 44205 58316
4 T 70670 67864 73159 171584 73713 71687 70695 169717
(B)
Species
S. No
A G C T A G C T
1 AA 86090 17627 18374 46951 83349 18623 19422 46755
2 AG 19089 11100 11318 14777 19871 10442 12264 17068
3 AC 17239 10577 10491 15202 17901 9453 10676 16794
4 AT 20142 15783 17406 47563 21368 16752 16451 47207
5 GA 31140 13379 10257 17578 31355 13670 10422 17204
6 GG 14729 8619 11555 10438 14982 9275 9700 11411
7 GC 13025 9601 12066 11309 13656 7494 9646 12227
8 GT 12136 12780 10458 18049 12976 12650 10000 20376
9 CA 26644 12400 12801 15817 27422 13532 12501 16427
10 CG 16283 10873 9652 10822 13747 9848 7594 9155
11 CC 15108 11450 9503 11144 15133 9631 9315 10126
12 CT 13210 12537 13724 17906 12775 13648 13226 18667
13 TA 25167 12878 12077 20548 26023 13819 12479 21392
14 TG 22253 14749 13476 17386 24051 15803 13465 18368
15 TC 22290 16002 15145 19722 23192 13766 14568 19169
16 TT 25182 26764 31571 88066 26594 28637 31018 83467
The pattern data and constructive dataset given in Tables
III and IV are generated from one of the ten folds of gene
sequence of Brucella Suis. Thus, from all the ten folds of gene
sequence of both Brucella Suis and C. Elegans, the pattern
data have been mined and constructive datasets have been
generated. The generated ten folds of data are used to train the
neural network. The results obtained from network training are
given in Fig. 5.
(a)

Vol. 3, No.8, 2012

112 | P a g e

(b)

(c)
Figure 6. Performance of training and test results from ANN: (a) Network performance, (b) Training evaluation and (c) Regression analysis.
Once the training process has been completed, the technique is validated using the test sequence. The results obtained from
10-fold cross validation are given in Table VI.
TABLE V. PERFORMANCE EVALUATION USING 10-FOLD CROSS VALIDATION RESULTS
Rounds in
cross
validation
Species
ANN Output Classification Result ANN Output Classification Result
1 0.2421 TP 0. 5171 TP
2 0.0769 TP 0. 6272 TP
3 0.0828 TP 0. 6361 TP
4 0.2634 TP 0. 8974 TP
5 0.2493 TP 0. 6063 TP
6 0.2613 TP 0. 0141 TN
7 0.5277 TN 0. 9163 TP
8 0.3616 TP 0. 6714 TP
9 0.5849 TN 0. 5103 TP
10 0.2143 TP 0. 5142 TP
Mean
Classification
Accuracy
80% 90%

Vol. 3, No.8, 2012

113 | P a g e
From the results, it can be seen that when a gene sequence
is given to the proposed technique it identifies the
corresponding species. Here, the technique is evaluated with
the DNA sequence of only two genes. The technique is
developed in such a way that it can be applied to any kind of
DNA sequence. The test results claim that the performance of
the technique reaches a satisfactory level.
V. CONCLUSION
In this paper, we have proposed a species identification
technique by integrating data mining technique with artificial
intelligence. Initially, the nucleotide patterns have been mined
effectively. The resultant has been subjected to MPCA-based
dimensionality reduction and eventually classified using a
well-trained neural network. The implementation results have
shown that the proposed technique effectively identifies the
organism from its gene sequence and so the species.
Moreover, results obtained from 10-fold cross validation have
proved that the organism can be identified even from a part of
the DNA sequence.
Though the technique has been tested with the DNA
sequence of only two organisms, the 10-fold cross validation
results have reached a remarkable performance level. From the
results, it can be hypothetically analyzed that a technique,
which identifies the organism only with a part of gene
sequence, have the ability to classify any kind of organism and
so the species.
REFERENCES
[1] P. P. Vaidyanathan and Byung-Jun Yoon, "The role of signal-processing
concepts in genomics and proteomics", Journal of the Franklin Institute
Genomics, Signal Processing, and Statistics Vo. 341, Issue. 1-2, pp. 111-
135, January-March 2004.
[2] Achuth Sankar S. Nair, T.Mahalakshmi " Visualization Of Genomic
Data Using Inter-Nucleotide Distance Signals", Silico Biology, Issue
Volume 6 , 215-222, March ,2006
[3] Edward R. Dougherty, Ilya Shmulevich and Michael L. Bittner,
"Genomic Signal Processing: The Salient Issues", EURASIP Journal on
Applied Signal Processing, Vol. 1, pp. 146153, 2004.
[4] Michel Tibayrenc "The species concept in parasites and other pathogens:
a pragmatic approach?" TRENDS in Parasitology Vol.22 No.2 February
2006. http://www.th.ird.fr/downloads/TIP.pdf
[5] Swapnoneel Roy, Minhazur Rahman, and Ashok Kumar Thakur
"Sorting Primitives and Genome Rearrangement in Bioinformatics: A
Unified Perspective", World Academy of Science, Engineering and
Technology, Issue 38, 2008.
[6] Mai S. Mabrouk, Nahed H. Solouma, Abou-Bakr M. Youssef, and
Yasser M. Kadah "Eukaryotic Gene Prediction by an Investigation of
Nonlinear Dynamical Modeling Techniques on EIIP Coded Sequences",
International Journal of Biological and Life Sciences 3:4 2007
[7] Jianbo Gao, Yan Qi, Yinhe Cao, and Wen-wen Tung, "Protein Coding
Sequence Identification by Simultaneously Characterizing the Periodic
and Random Features of DNA Sequences", Journal of Biomedicine and
Biotechnology, Vol. 2, pp. 139146, 2005.
[8] C.L. Winder1, E. Carr2, R. Goodacre1 and R. Seviour2 "The rapid
identification of Acinetobacter species using Fourier transform infrared
spectroscopy" Journal of Applied Microbiology 96, 328339,2004
[9] Francielle B. Silva, Sabrina N. Vieira, Luiz R. Goulart Filho, Julien F. C.
Boodts , Ana G. Brito-Madurro and Joo M. Madurro "Electrochemical
Investigation of Oligonucleotide-DNA Hybridization on Poly(4-
Methoxyphenethylamine)",International Journal of Molecular Sciences,
9, 1173-1188,8-july,2008.
[10] Anne Jensen,Guillaume Calvayrac, Bensu Karahalil, Vilhelm A. Bohr
and Tinna Stevnsner "Mammalian 8-Oxoguanine DNA Glycosylase 1
Incises 8-Oxoadenine Opposite Cytosine in Nuclei and Mitochondria,
while a Different Glycosylase Incises 8-Oxoadenine Opposite Guanine
in Nuclei", The journal of biological chemistry, Vol. 278, 19541-19548,
2003, DOI:10.1074/jbc.M301504200
[11] Jeremy D. Volkening, Stephen J. Spatz "Purification of DNA from the
cell-associated herpesvirus Mareks disease virus for 454
pyrosequencing using micrococcal nuclease digestion and polyethylene
glycol precipitation", Journal of Virological Methods, Vol. 157, p.p. 55
61. 2009.
[12] Edward R. Dougherty and Aniruddha Datta "Genomic Signal
Processing: Diagnosis and Therapy", IEEE Signal Processing Magazine,
Vol. 22, No. 1, p.p. 107-112, 2005, DOI: 10.1109/MSP.2005.1407722
[13] TrevorW. Fox,Alex Carreira "A Digital Signal Processing Method for
Gene Prediction with Improved Noise Suppression" ,EURASIP Journal
on Applied Signal Processing, Vol.1, p.p. 108114, 2004
[14] A. Bharathi, Dr.A.M.Natarajan, Cancer Classification of
Bioinformatics data using ANOVA International Journal of Computer
Theory and Engineering, Vol. 2, No. 3, June, 2010
[15] Jayanthi Ranjan "Applications of Data Mining Techniques In
Pharmaceutical Industry", Journal of Theoretical and Applied
Information Technology, 2005 2007.
[16] Paul Hooley, Ian J. Chilton, Daron A.Fincham, Alan T.Burns and
Michael P.Whitehead "Assigning Level in Data-mining Exercises",
Bioscience Education Journal, Volume 9:June 2007
[17] Antony Brownea , Brian D. Hudsonb , David C. Whitleyb ,Martyn G.
Fordb , Philip Pictonc "Biological data mining with neural
networks:implementation and application of a flexible decision tree
extraction algorithm to genomic problem domains" Neurocomputing,
Volume 57, March 2004, Pages 275-293
[18] Gerd Pfeiffer, Stefan Baumgart, Jan Schroder, and Manfred Schimmler
"A Massively Parallel Architecture for Bioinformatics", Lecture Notes in
Computer Science, 2009, Volume 5544/2009, p.p. 994-1003, DOI:
10.1007/978-3-642-01970-8_100
[19] Simon Miles, "Agent-Oriented Data Curation in Bioinformatics", In
Workshop on Multi-Agent Systems in Medicine, Computational
Biology, and Bioinformatics (MAS*BioMed'05), July 2005, Utrecht,
Netherlands
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.124.2202&rep
=rep1&type=pdf
[20] Dhaeseleer P, Liang S, Somogyi R, "Genetic network inference: from
co-expression clustering to reverse engineering", Bioinformatics, Vol.
16, No. 8, pp.707726, 2000.
[21] Kirschner M, Pujol G, Radu A, "Oligonucleotide microarray data
mining: search for age-dependent gene expression", Biochemical and
Biophysical Research Communications, Vol. 298, No. 5, pp. 772778,
2002.
[22] Ponomarenko J, Merkulova T, Orlova G, Fokin O, Gorshkov E,
Ponomarenko M, "Mining DNA sequences to predict sites which
mutations cause genetic diseases", Knowl-based Syst, Vol. 15, No. 4,
pp.225233, 2002.
[23] Oliveira and Johnston, "Mining the schistosome DNA sequence
database", Trends Parasitol, Vol. 17, No. 10, pp.501503, 2001.
[24] Fuhrman, Cunningham, Wen, Zweiger, Seilhamer and Somogyi, "The
application of Shannon entropy in the identification of putative drug
targets", Biosystems, Vol. 55, pp.514, 2000.
[25] Arkin, Shen and Ross, "A test case of correlation metric construction of
a reaction pathway from measurements", Science, Vol. 277, pp. 1275-
1279, 1997.
[26] Cho and Won, "Machine learning in DNA Microarray analysis for
cancer classification. In: Yi-Ping Phoebe Chen, in proceedings of the
First Asia-Pacific Bioinformatics Conference. Australian Computer
Society, pp. 189-198, 2003
[27] Riccardo Bellazzi and Blaz Zupan, "Towards knowledge-based gene
expression data mining", Journal of Biomedical Informatics, Vol.40,
pp.787-802, 2007
[28] Shital Shah and Andrew Kusiak, "Cancer gene search with data-mining
and genetic algorithms", Computers in Biology and Medicine, Vol.37,
pp.251-261, 2007
Vol. 3, No.8, 2012

114 | P a g e
[29] Hemalatha and Vivekanandan, "Genetic Algorithm Based Probabilistic
Motif Discovery in Unaligned Biological Sequences", Journal of
Computer science, Vol.4, No.8, pp.625-630, 2008
[30] Ashraf S. Hussein, "Analysis and Visualization of Gene Expressions and
Protein Structures", Journal of software, Vol.3, No.7, pp.2-11, October
2008
[31] Ahmad M. Sarhan, Cancer Classification Based on Microarray Gene
Expression Data Using DCT and Ann, Journal of Theoretical and
Applied Information Technology, Vol.6, No.2, pp.208-216, 2009
[32] Valarmathie, Srinath, Ravichandran and Dinakaran, "Hybrid Fuzzy C-
Means Clustering Technique for Gene Expression Data", International
Journal of Research and Reviews in Applied Sciences, Vol.1, No.1,
pp.33-37, October 2009
[33] Belmamoune, Potikanond and Fons J.Verbeek, "Mining and analysing
spatio-temporal patterns of gene expression in an integrative database
framework", Journal of Integrative Bioinformatics, Vol.7, No.3, pp.1-10,
2010
[34] Hans knutsson, A Tensor representation of 3-D structure, 5th IEEE-
ASSP and EURASIP Workshop on Multidimensional Signal Processing,
The Netherlands, September, 1987.
[35] Haiping Lu Plataniotis, K.N. Venetsanopoulos, A.N., MPCA:
Multilinear Principal Component Analysis of Tensor Objects, IEEE
Transactions on Neural Networks, Vol.19 No.1, p.p. 18 39, 2008,
ISSN: 1045-9227, DOI: 10.1109/TNN.2007.901277
[36] Emilio Corchado, lvaro Herrero, Neural Visualization of network
traffic data for intrusion detection, Applied Soft Computing, Volume
11, Issue 2, March 2011, Pages 20422056
[37] E.W.T. Ngai, Yong Hu, Y.H. Wong, Yijun Chen, Xin Sun, The
application of datamining techniques in financial fraud detection: A
classification framework and an academic review of literature Decision
Support Systems, Volume 50, Issue 3, February 2011, Pages 559569
[38] Arzu encan ahin, smail lke Kse & Reat Selba, Comparative
analysis of neural network and neuro fuzzy system for thermodynamic
properties of refrigerants Applied Artificial Intelligence: An
International Journal, Volume 26, Issue 7, 2012, DOI:
10.1080/08839514.2012.701427

Vol. 3, No.8, 2012

115 | P a g e
Brainstorming 2.0: Toward collaborative tool
based on social networks

MohamedChrayah
1
,Kamal Eddine El Kadiri
1
, Boubker Sbihi
1
,Noura Aknin
2

1
Laboratory LIROSA,
2
Laboratory LaSIT
Faculty of sciences
Ttouan, Morocco

Abstract Social networks are part of Web 2.0 collaborative
tools that have a major impact in enriching the sharing and
communication enabling a maximum of collaboration and
innovation globally between web users. It is in this context that
this article is positioned to be part of a series of scientific research
conducted by our research team and that mixes social networks
and collaborative decision making on the net. It aims to provide a
new tool open source for solving various social problems posed by
users in a collaborative 2.0 based on the technique for generating
ideas, brainstorming method and social networks together for the
maximum possible adequate profiles to the virtual brainstorming
session. A tool is run by a user called expert accompanied by a
number of users called validators to drive the process of
extracting ideas to the loan of various users of the net. It offers
then the solution to the problem of sending a satisfaction
questionnaire administered by an expert ready for the affected
user to measure the level of his satisfaction and also the success of
the process launched. For its implementation, we propose a
unified modeling using UML language, followed by a realization
using the JAVA language.
Keywords- component: Web2.0, brainstorming, social networks,
UML.
I. INTRODUCTION
Over time, the web has seen many changes starting with
the static web which allowed only to display static pages made
by the directors of the net and whose content was not always
updated followed by collaborative Web 2.0 who proposed the
involvement of users in content creation. Web 2.0 was
proposed in August 2001 by Dale Dougherty of O'Reilly
Media, but the real release of concept 2.0 was published in an
international conference in 2005 by Tim O'Reilly [10]. He
proposed a new vision of the web which consists on a higher
participation of Internet users as producers of information thus
forming communities participating in the communication,
sharing and dissemination of information.
With this concept a lot of software and services are freely
available on the web and therefore the amount of information
has increased which encouraged users to participate and inter
exchange. Social networks have existed since 2003, where
they have grown exponentially up to date [7]. They collect
data on members, and then store this information as data
profiles, these sites represent an appropriate database to search
for suitable profiles to any operation or survey in the web.
Moreover, decision-making has changed a lot with the
emergence of information and communication technologies
(ICT) [6]. Makers become less statically located; on the
contrary, they act in a distributed manner. This change creates
a new set of requirements: collaborative decision-making
based on collaboration using Web 2.0 tools.
In the next section, we present the web 2.0; its principles,
its most used tools and especially social networking the tool
used in our article, and then well present the notion of
collaborative decision based on the method of generating ideas
(brainstorming) and social networks to achieve in the end a
collaborative decision as a result of a series of proposals and
virtual meeting by web users 2.0. Finally, well propose a
design and implementation of the proposed tool using the
UML.
II. WEB 2.0
A. Web 2.0 and its dimensions
Web 2.0 is social, is open, it lets you control your data,
mixing the global with the local. Web 2.0 is new interfaces -
new ways to search and access content. Web 2.0 is a platform
ready to receive the educators, the media, politics, and
communities. Thus, users who contribute to information
exchange can simply interact (share, exchange, etc.) with both
the content and page structure, but also between them,
including creating the social Web. The user becomes, using
the tools at its disposal, an active person of the cloth [9].
Web 2.0 can be viewed in three dimensions as shown in
the following figure:

Figure 1: Dimensions Of Web 2.0.
Vol. 3, No.8, 2012

116 | P a g e
Social dimension: Web 2.0 is a real network of social
interaction based on the participation of Internet users. User
communities are created in this context based on areas of
mutual benefit. Anyone can easily create an information space
accessible by anyone and anywhere in the world where he can
put anything (anything is the one of the boundaries of Web
2.0).
Technical dimension: Web 2.0 is an advanced technique
that makes it simple to access the production and use of
information through the tilting of the software installation to
use online services. Thanks to the use of multiple technologies
(XHTML, CSS and JavaScript for the presentation of the sites,
DOM, Document Object Model, for dynamic and interactive
signage, XML and XSLT for data manipulation).
Economic dimension: Funding Web 2.0 sites is done
mainly through advertisements, commercial offers and
trafficking networks instead of gifts or payments for licenses
to use proprietary software. A project based on an economic
model of Web 2.0 is based on the large mass of users who
consume information mixed with advertising or commercial
content which finances containers [2].
B. The Tools Of Web 2.0
Web 2.0 consists of a set of Internet technologies that
facilitate the open and participatory work. Its main feature is
that Web 2.0 tools allow users to control the network and
interact proactively to improve or transform situations that
affect them.
1) Blog:
The term "blog" is short for weblog, which can be
translated as "Internet newspaper". Frequently defined as a
personal site, this is an individual space of expression, created
to give voice to all Internet users (individuals, businesses,
artists, politicians, associations ...). Blogs are extremely
simplified sites and dedicated to writing, where "the entries
appear in ante-chronological order." The animation of blogs is
initially limited to technophiles capable of creating the
structure of their blog and have it hosted on a server.
The Pew Research Center's Internet and American Life
Project has conducted a survey in 2008 which has resulted in
startling statistics that 40% of adult Internet users in the
United States have blogs.
Blogs have been discussed recently as a innovative
knowledge of sharing technology, knowledge and
management [12].
2) Wiki:
The term is derived from the expression Hawaiian wiki-
wiki, meaning quick. A wiki is a collaborative website whose
content can be edited by visitors on the site, allowing users to
easily create and edit collaborative web pages [8]. In essence,
a wiki is a simplification of the process of creating HTML
web pages combined with a system that records each
individual change that occurs over time, so that at any time a
page can be forced to return to the one of its previous states. A
wiki can also provide tools that allow the users community to
monitor the changing state of the wiki and discuss issues that
arise. Some wikis restrict access to a group of members,
allowing only members to change the page content but
everyone can see it. Others allow unrestricted access, allowing
anyone to both modify the content and display.
3) Social Networks:
A social network is a set of social entities such as
individuals or social organizations linked together by bonds
created during social interactions.
It is represented by a structure or a dynamic form of a
social group its a web space to:
Express them selves
To promote itself
Exchange
Get back in touch
Social networks have as common basis the sharing:
Sharing knowledge
Sharing of professional contacts
Content Sharing
Social networks are social websites that enable people to
form online communities and share content created by these
users. People can be users of the open Internet or restricted to
those who belong to a particular organization (eg company,
university, etc.). [14]
Table 1 gives an idea of the number of users of social
networks, and the classification of these social sites depending
on the number of participants.
TABLE I. SOCIAL NETWORKS USED MOST [15]
Site Name Users (in Million)
Facebook 309
MySpace 253
WindowsLiveSpaces 120
Habbo 117
Friendster 90
Hi5 80
Tagged 70
Orkut 67
Flixter 63

As the table shows, Facebook is positioning itself in first
place with 309 million users. The implication of this number
in a decision-making would be a dream for designers of
collaborative Web 2.0
4) RSS Feeds:
RSS (Really Simple Syndication) is a simple XML syntax
to describe the recent additions of content to a website. These
additions can include elements of news, blog updates, library
acquisitions or any other information. it just facilitates
dynamic sharing of content between a publisher (website) and
a reader (the Internet) by allowing authors and editors of a
website to make available to the community some content that
can be reused for integration into another site [13]. Since RSS
uses XML to disseminate information relevant to user needs,
RSS could well become the universal method for extracting
information from the Internet.
Vol. 3, No.8, 2012

117 | P a g e
III. BRAINSTORMING 2.0:COLLABORATIVE DECISION
MAKING
A. Brainstorming Method:
Brainstorming is a technique for generating ideas that
stimulates creative thinking in finding solutions to a given
problem. This is to produce as many ideas as possible in the
shortest time on a given topic without criticism, without
judgment. This searching method favors quantity, spontaneity
and imagination [11].
Table II shows the essential steps for a successful
brainstorming session:
TABLE II. STEPS BRAINSTORMING
Steps Sequence
Step 1 Presentation of the problem
Step 2 warming period
Step 3 Brainstorming
Step 4 classification / grouping of ideas
Step 5 Final Decision
As shown in Figure 3, the Collaborative Decision Making
Brainstorming consists of four iterative steps starting with the
presentation of the problem and ending with the classification
of ideas.
B. Brainstorming2.0:
1) Towards a tool for collaborative decision making:
The collaborative decision making is based on users
participation as actors for the production and wide
dissemination of the decision subsequently forming
communities. The size and mass of decisions will increase the
quantitative level but still the qualitative decisions suggested
by users.
The users of the system must be ordered according to their
importance and give more privileges. It is not enough to give
any decision, but the right decision, for this we proposed
Brainstorming 2.0 tool to overcome the problems already
discussed above.
Brainstorming 2.0 is an Open Source, free dedicated to all
users of Web 2.0; their goal is to found a topic concerning
decisions. It is not as tools publishers social owner as Google
Plus (G+), which is a social network where there are paid
services, and their only purpose the meeting between friend or
the professional. Brainstorming 2.0 is a social tool that
organizes virtual brainstorming sessions between users of the
web communities.
2) Classification Decisions:
The decision generated in the brainstorming 2.0 system
can be classified into four classes [4][3][5][1] according to
their quality:
TABLE III. CALASSIFICATION DECISIONS
Code libel Weighting
G Good 10
M Meduim 5
L Low 1
E Error -10

Figure2. Brainstorming 2.0 Architecture
According to the above table the user who produces the
right decisions and averages has the chance to become a
validator in a short time; it also proposes to create a virtual
currency that will increase every time someone publishes
validated decisions. Users will then have access to some
opportunity not given to all others. This value will depend on
the turnout of participation in the virtual brainstorming
sessions and also the notation affected by the validators.
3) Classification Of Users:
It is proposed to decompose the system users into three
groups: simple users who consume and produce decisions, the
validators who validate decisions and finally the experts who
make tracking validations, pointing validators for each
problem and its publication when validated by the validators.
A simple user can become a validator if its weight exceeds
1000 pt and is recommended by an expert, a validator must
communicate with other validators and expert in the validation
process. A validator can become an expert if its weight
exceeds 10000 pt. Brainstorming 2.0 users are represented in
the following table [4][3][5][1]:
TABLE IV. CLASSEMENT DES UTILISATEURS
Code Libel Weighting
E EXPERT >=10000
V VALIDATOR >10000And >=1000
U USER <=1000

Regarding the weighting at each decision by a user there is
an increase in the value using the following formula (1):

Vol. 3, No.8, 2012

118 | P a g e
P(User)=P(Decisions) (1)

The following figure shows the use case diagram tool
braistorming 2.0

Figure3 Uses Cases Brainstorming 2.0.

The use case diagram shown in Fig.3 clarified the
interaction between the system and all actors (users), it is clear
from the diagram that the expert is the main user of the
system, he inherits all the actions of other users in addition to
managing the flow of information he can administrate the
platform. The actor system is who is responsible to the
updating of weights and linking with social networking to
select the appropriate profiles.
4) Decision Making Process In The 2.0 Brainstorming:
The process of decision making in the Brainstorming 2.0
runs as follows, figure 4.
A user poses a problem, a system expert considers the
issue and distributes it to all users with an adequate profile to
this subject by searching social networks related to the system,
and in parallel the expert selects validators to validate
decisions proposed by them. After treatment and decisions
classification by the validators, the expert groups the decisions
that have obtained Class B (Good) by the validators to
generate a final decision of the issue to be released in the
portal. Finally a satisfaction questionnaire is sent to the user
who submitted the problem to get an idea about his
satisfaction with the solution proposed.
The figure.5 gives us an idea about the chronological
interactions between all major and minor players in the
system, from the diagram it is the simple user who initiates the
transaction, the one who started the virtual brainstorming
session, then after the analysis of the issue the expert send it to
other users to give these decisions and in the end transfer them
to the validators.

Figure4 Process Brainstorming 2.0.
But the sequence diagram is insufficient to give us an
overview of the system, that is why we use the class diagram
to understand the relationship between the classes of the
system, the Fig6 represents the class diagram of the
brainstorming 2.0 system; According to the diagram the
system is composed of 9 classes, Expert class inherits the
methods of the Validator class that also inherits from the User
class, the class Profil_Network_Social contains information on
existing profiles in social networks interacting with the
system.
It is clear that the class Decision_final is related directly to
the Expert class as the one who distributes the final decision,
Decision_Validate class with the class validator and finally the
classes Decision and Problem with the USER class, because
hes the one who launches the problem and also the one who
gives decisions.
IV. EXAMPLE : A CASE STUDY.
An expert receives a user's question "What is your opinion
on Web 2.0?" And then he launches it into the system, after
consultation with users six users answered the question with
judgment:
User1 : Favourable opinion
User2 : Unfavourable opinion
User5 : Unfavourable opinion
Vol. 3, No.8, 2012

119 | P a g e

Figure5 Sequence Brainstorming 2.0 Diagram

Figure6 Class Brainstorming 2.0 Diagram
The expert selects two system validators to review and
validate user responses by giving them a ranking:
TABLE V. RANKING ANSWERS
Validator1 Validator2
User1 G G
User2 L L
User3 G G
Validator1 Validator2
User4 G G
User5 L L
User6 G G
After ranking the responses, the expert selects the response
that received a Class B by both validators and diffuses it on
the portal (Favourable Opinion) with the judgment of users
who responded favorably, at the end the expert sends a
questionnaire to the user that sent the question to get an idea of
his level of satisfaction with the answer.If the user is satisfied,
it will make him seek help and become part of the community
of this tool and if it is not satisfied, he can restart another
process or ask for help from People experts in the wanted
domain.What is clear is that decisions vary from personal
context to professional staff. To improve performance in a
professional context, it might be thought to pay employees and
to create training sessions to achieve meaningful results at the
expense of a sum of money to take advantage of collaborative
network intelligence.
CONCLUSION
The aim of our proposal is to find a mechanism mixing
between the benefits of Web 2.0 tools and the technique for
generating ideas brainstorming in order to achieve a system of
collaborative decision making. This new system will help find
lot of solutions through their connection with social networks,
which contains adequate profiles and also good decisions
because the raw information has no value in the new system
and as the one who adopts the problem is an expert. The
limitations of this tool are that it contradicts the general
concept of Web 2.0 (the participation of everyone in the
decision) since the expert takes some decision. There is also
the responsibility of the validators in the selection and
classification decisions. In addition, it is necessary to test and
Vol. 3, No. 8, 2012

120 | P a g e
measure the level of satisfaction undefined users to meet the
needs of each context.
This article is a beginning of a series of articles that will
follow and that will be the implementation of this tool in java.
A presentation detailed the tool will be made in future
publications
PERSPECTIVE
As prospects we propose to generalize the use of this tool and
measure the rate of satisfaction of its users. We propose also to
design tools for collaborative semantic decision support which
understands the sense of the decision. Adapt this tool to other
collaborative tools are blogs, wikis and RSS feeds. Impose a
single and secure identification with fingerprints to ensure
good use away from hackers and malicious people.
REFERENCES
[1] Boubker Sbihi, K.E. El Kadiri, and N. Aknin, The Vblogs: Towards a
New Generation of Blogs, International Journal of Computer Science
Issues, Vol. 7, issue 3, No. 2, May 2010, pp. 9-14.
[2] Boubker Sbihi, TOWARDS A NEW VISION OF WEB 2.0, Computer
Sciences and Telecommunications, 2009
[3] Boubker Sbihi, and Kamal Eddine El Kadiri, Web 2.2: Toward
classified information on the Web, International Journal of Web
Applications, Vol. 1, No. 2, June 2009, pp. 102- 109.
[4] Chrayah Mohamed, Kamal Eddine El Kadiri, Boubker Sbihi and Noura
Aknin, Brainstorming +: Towards a tool for decision-making In the Web
2.0, IJCSI International Journal of Computer Science Issues, Vol. 9,
Issue 2, No 2, March 2012
[5] Chrayah Mohamed, Boubker Sbihi and Kamal Eddine El Kadiri
:Towards an instrument for a new generation of blog. EMSCE 2011 May
19_20 2011 ISBN:978-84-694-4025-4.
[6] Claus Rinnera,Carsten Kelerb,Stephen Andrulisa, The use of Web2.0
concepts to support deliberation in spatial decision-making, Computers,
Environment and Urban SystemsVolume 32, Issue 5, September 2008,
Pages 386395.
[7] Danah boyd, Nicole Ellison, Social Network Sites: Definition, History,
and ScholarshipJournal of Computer-Mediated Communication Volume
13, Issue 1, pages 210230, October 2007.
[8] Howard T. Welser ,Patrick Underwood, Wiki Networks: Connections of
Creativity and Collaboration 2011, Pages 247271.
[9] Musser J, OReilly T. OReilly Radar Report: Web 2.0 Principles and
Best Practices: OReilly Media 2006.
[10] OReilly T. What Is Web 2.0? Design Patterns and Business Models
[11] for the Next Generation of Software; 2005.
[12] Patrick C.Shih, Gina Venolia, Gary M. Olson Brainstorming under
constraints: Why software developers brainstorm in groups 2011 Pages
74-83
[13] Sangmi Chaia, Minkyun Kimb, What makes bloggers share knowledge?
An investigation on the role of trust, International Journal of Information
Management 30 (2010) 408415
[14] Yu-Feng Lan,Yang-Siang Sie, Using RSS to support mobile learning
based on media richness theory Volume 55, Issue 2, September 2010,
Pages 723732
[15] WonKim a, Ok-RanJeong a, Sang-WonLee On social Web sites
Information Systems 35 (2010) 215236
AUTHORS PROFILE
Chrayah Mohamedis computer engineer, pHD student and themember of
LIROSA laboratory.
Kamal Eddine El Kadiriis pHD doctor and professor of computerscience at
Faculty of Sciences of Ttouan in Morocco. He is theDirector of the ENSA
School of engineers of Tetouan and theDirector of LIROSA laboratory. He
has published several articleson E-learning and Web 2.0. He is part of many
boards ofinternational journals and international conferences.
BoubkerSbihiis pHD doctor and professor of computer scienceat the School
of Information Science in morocco. He is theresponsible of Department of
Information Management. He haspublished many articles on E-learning and
Web 2.0. He is part ofmany boards of international journals and international
conferences.
NouraAkninis pHD doctor and professor of computer science atFaculty of
Sciences of Ttouan in Morocco. She has publishedmany articles on E-
learning and Web 2.0. She is part of manyboards of international journals and
international conferences. Shehas member of the IEEE and the IEEE
Computer Society

Vol. 3, No.8, 2012

121 | P a g e
A Review On Cognitive Mismatch Between
Computer and Information Technology And
Physicians
Fozia Anwar
CIS Department
Universiti Teknologi Petronas
Perak, Malaysia
Dr. Suziah Sulaiman
CIS Department
Perak, Malaysia

Dr. P.D.D.Dominic
CIS Department
Perak, Malaysia

Abstract Health Information Technology has a great potential
to transform the existing health care systems by making them
safe, effective and efficient. Multi-functionality and
interoperability of health information systems are very important
functions. Hence these features cannot be achieved without
addressing the knowledge and skills of the health care personnel.
There is a great mismatch between Information Technology
knowledge and skills of physicians as this discipline is completely
missing in their educational tenure. So usability of health
information technologies and system as well as evidence based
practice in the future can be improved by addressing this
cognitive mismatch. This will result in persistent partnership in
HIS design between physician and IT personnel to get maximum
usibility of the systems,
Keywords- cognitive mismatch; HIT; usibility.
I. INTRODUCTION
Information technology in health sector is spreading
globally [1]. There are major campaigns in many countries
which are involved not only large expenditures but public
resources as well to boost the use and adoption rate of health
information systems (HIS) and technologies. Therefore for the
successful usability of HIS it is critical to address physicians
views and skills on the use of these technologies [2].
Use of health information technology is offering evidence-
based practice to endorse health and human prosperity. Health
information technology (HIT) consists of a set of technologies
with a great diversion for transmitting and managing health
care data for the use of all stakeholders of health care systems
[3]. Therefore the development of user-centered information
systems is important to get best usability and advantages of the
developed systems. Development of the user centered
information system is like a pendulum that has to be swung
between users and developer. The success of the ICT system
depends on the balance of the pendulum in the form of input
and feedback from the both ends. In most of the cases of the
HIS the balance of the pendulum is disturbed due to lack of
information technology cognitive skills of physicians. Hence
the usability of the particular healthcare information system
could not be achieved at the extent of the expectations [4].
Health care software developers often neglect significant
characteristics, tasks, user preferences, knowledge skills and
usability issues from the physician point of view resulting in
systems that decreases productivity or simply remain unusable
for the end-users [4]. Users of healthcare information system
are dynamic and hence their needs are, so, development of any
solution, system, or technology should be accounted the needs
and requirements of the end users [6] [7] [8].
Definition of the term Cognitive skills varies but American
Psychological Associations (2007) define as all forms of
knowing and awareness, such as perceiving, conceiving,
remembering, reasoning, judging, imagining and problem
solving.[9] Simple definition of cognitive skills could be that
these are the skills that one has learnt and ability to learn
further new knowledge. Cognitive skills and cognitive ability
of people had a profound effect in making a successful
adoption and implementation of an information system with
great usability.
The information technologies have greatly influenced on
medical research, education, and healthcare delivery [10].
Therefore, medical schools should fit into place specialists to
educate both students and tutors simultaneously; hence it is
necessary to make additions and alterations in medical
education accordingly [11].
II. LITERATURE REVIEW
A. Role of Health Informatics
A great philosopher Immanuel Kant said: We see things
not as they are but as we are. Perception and interpretation are
having the same meaning. Anxiety of present situation having
past experience is the perception. Both terms processing and
perception are frequently used interchangeably. The etiology of
misinterpretation is lack of experience or knowledge about the
information technology.
Information plays a crucial role in medicine. All the time
physicians are playing with the information they create, collect,
search, adapt; in fact they drown in information. Medical or
health informatics is all about this. The exact position of
medical informatics is at the intersection of information
technology, cognitive science, artificial intelligence, and
medicine. So this is not a simple field involving only one
aspect like medical computing, telecommunications, or
information engineering, rather it is a dialogue between
Vol. 3, No.8, 2012

122 | P a g e
physicians, patients, and medical informaticians in a medical
information system. Rather it explores and develops new
knowledge, builds new theories, and organizes principles and
solutions. Todays challenge is not to have an access of
hardware rather it is the ability to use the information system.
HIT technologies and information systems develop the way to
positively increase the outcome of clinical care. Computers and
evidence decision makings are two pillars of health
informatics. To take full advantage of the HIT we have to learn
the skills for framing and analyzing and integrating the
healthcare information [12].
It is needed that end-users especially physicians should
participate in healthcare IT development process [13]. Different
models are suggested for better collaboration between users
and IT experts. Parmit K Chilana suggested a model which
emphasizes the persistent partnership of healthcare domain
experts and IT experts throughout the development,
implementation, evaluation and deployment processes. It gives
emphasis on improving interdisciplinary skills on both ends of
the pendulum to achieve a good balance. As a result of which
an ICT system can be made which is not only be able to
support the goals of the complex domain but can clearly outline
the challenges of clinical ICT system [14].
It is essential in the research field of health informatics that
a good fit and faith must persist in an ICT system and clinical
practices [13]. Nevertheless, few numbers of studies have been
done to answer the question of the cognitive aspects and need
of interdisciplinary skills of the users and IT experts of
different domains and the impact of the trainings given to users
of HIS. Terms HIT and HIS are used in the research literature
to cover a wide range of information technology applications
in health care.
Trend of IT applications use is increasing by the Health
Professionals (physicians, nurses and allied Staff), as, some
health professional also develops, deploy or research health
care IT [15]. Consequently, Health Professionals need to
educate themselves for their respective roles along with their
temporal requirements in health informatics [16] as their
cognitive skills about IT and IS are not addressed throughout
their educational tenure. Some examples of IT use in healthcare
sector are use of physician digital assistance (PDAs),
computerized physician order entry (CPOE), electronic health
record (EHR), clinical decision support system (CDSS), picture
archiving and communication system (PACS), radiology
information system (RIS), pharmacology information system
(PIS), hospital information system (HIS), disease early warning
system (DEW), Telemedicine and health management
information system (HMIS) [18].
Shah and Robinson found some key impediments to user
involvement in HIT which are lack of resources, attitudes of
technical developers and healthcare personnel, and lack of
understanding the appropriate interdisciplinary knowledge and
skills. However, a physicians involvement is very crucial for
the success of HIT [19]. A proactive use of information
technology in the health sector empowers consumers of health
services to have an easy access health information and decision
making tool and by the employment of HIT. Health
professionals can collaborate more easily when distance is a
major factor in health care delivery [20]. In past HIT
applications were used for administrative activities and
financial activities rather assisting and delivering health care
[21].
Majumder identifies different trends in medical education.
There is a noticeable change in healthcare environment due to
advancement of ICT and the pervasion of World Wide Web in
medicine. The health sector is investing a lot in the
information systems and health technologies, so , physician
have to depend on ICT and informatics skills for providing
their contributions in research, education and health care
organization. Therefore, knowledge and skill to use ICT
application for self-directed learning is important for medical
professionals. Some international organizations like the World
Bank and WHO have taken initiatives to orient
interdisciplinary skills in the health sector to prepare physicians
to cope with the ever changing dynamic health environment.
Such pedagogical environments require a lot of cognitive skills
and interdisciplinary knowledge [22].
By addressing the IT knowledge cognitive barrier medical
students not only be better exposed to evidence based decision
making but it will foster their capacity towards a new world of
learning. By exposing medical professionals to the information
technology knowledge will prepare them for a better
orientation of electronic health information systems.
Koschmann assessed three different approaches to address
cognitive mismatch, which are learning about, through and
with computers. He concluded that none of the above
mentioned approaches are sufficient alone in fact combination
is required [23].
HMSO Report showed that the failure of The National
Program for IT in the NHS is the lack of IT skills within
National Health Services [24]. Still culture of use of IT in the
health care system is not established yet. Bond, 2006 in his
doctoral dissertation explored the ways of developing
informatics skill and knowledge to fill the knowledge gap .He
used a questionnaire tool to collect data from 132 nurses about
their usability experience and practicing IT skills behavior in
their routine practices [25].
In another study Steven R. Simon and Rainu Kaushal
randomly took 1884 physicians as their sample size with the
response rate of 71.4%. They explored the level of physicians
use of electronic health record among two groups and also
investigated the factors which correlate this use. One group is
that which used EHRs functions and thought that they are a
useful tool for their practice. The other group is low users of
the information system and they even dont aware of the total
available and useful functionalities of the system. Therefore the
adoption of HIT and HIS is very critical and is a key issue in
the health care system. In the past more attention was given on
adoption but little literature is available on physicians
capabilities to use the system [26].
On CDCP (the Centers for Disease Control and Prevention)
panel a report named Electronic Medical Record/Electronic
Health Record Systems of Office-based Physicians: United
States states that physicians are not using the core
functionalities of electronic health record because of two
possible reasons. One is that their system may lack in having
Vol. 3, No.8, 2012

123 | P a g e
these functionalities or the other possibility is that the physician
may not aware or have not had knowledge of the presence of
those functionalities in their system. When it is analyzed that
what is the reason that physicians dont know about the
availability of the system results the largest gap in the
knowledge and cognition mismatch was there [27]. Adherence
to data definitions with the proper involvement of end users in
the development process of HIS are important factors in
gaining the interoperability of the system [28].
B. Knowledge And Skills Of Health Professionals
The health care system is an information intensive domain
in which timely access to quality information with accuracy is
very critical. Medical professionals are one of the major
stakeholders of health information technology. There are lots of
barriers to acceptance of information technology by this group
of stakeholders. Barriers of this knowledge gap are
sociological, cultural, and organizational and technological.
Some researchers reported that physicians are reluctant to
introduce the systems into their practice because of time
constraints [29] [30] [31]. At present healthcare professionals
still seem to be lagging behind in participation in IT
Development [32]. One important factor is knowledge and
cognitive barrier as their level of information technology
literacy is low to use applications of IT [33]. This low level
even indefensibly low level of IT literacy or basic knowledge
and skills is the reason argued by the health professionals. Due
to this reason understanding the concepts and importance of
health information systems and information technology cannot
be justified by medical professionals. Nykanen and Karimaa in
their research stated that the starting point for development of
health information system should be to acquire an insight into
the healthcare domain where the information system is going to
be used [34].
C. Adoption And Use Of Hit
Health information technology (HIT) is a vital element
which can address inefficiencies and discrepancies in
healthcare in an efficient way. So it is necessary to understand
the challenges hurdles and barriers which can limit meaningful
use of HIT. Poorly developed user interface and system design
is a big hurdle in the clinical workflow and can result in wasted
time, poor data collection, misleading data analysis, and
potentially negative clinical outcomes. Decisions on
technology acquisitions and implementations are often made by
individuals or groups that lack clinical informatics expertise
[35]. Hence it results in poor usability of the system.
Developing such a system is a waste of human resources as
well as economics. Proper introductory education of IT is one
of the key antecedents of attitude toward computer use
behavior by humans [36].
A designreality gap concept is introduced by Heeks
arguing the misunderstandings and mismatches between the
current realities and design of healthcare information systems
[37]
III. CONCLUSION
Information technology cognitive skills for physicians are
very critical to address in gaining persistent partnership of
physicians in the HIT Application development and in
achieving the maximum usability of health information systems
and technologies. It will be very helpful in successful
implementation of HIS but also foster the evidence based
practice and increase usability of HIS. Interdisciplinary skills in
medical education will be of great value in future research as
well.
REFERENCES
[1] Fozia Anwar, Azra Shamim Barriers in Adoption of Health Information
Technology in Developing Societies International Journal of Advanced
Computer Science and Applications.Vol 2 Issue 8 August 2011
[2] Norm Archer, Mihail Cocosila, A Comparison of Physician Pre-
Adoption and Adoption Views on Electronic Health Records in
Canadian Medical Practices, J Med Internet Res. 2011 Jul-Sep; 13(3):
e57
[3] Viitanen J, Hyppnen H, Lveri T, Vnsk J, Reponen J, Winblad
I. National questionnaire study on clinical ICT systems proofs:
physicians suffer from poor usability. Int J Med Inform. 2011
Oct;80(10):708-25.
[4] Marc Berg, Patients and professionals in the information society: what
might keep us awake in 2013, International Journal of Medical
Informatics 66 (2002) 31_37.
[5] Susanna Martikainen, Johanna Viitanen, Mikko Korpela, Tinja Lveri.
Physicians experiences of participation in healthcare IT development in
Finland: Willing but not able. International Journal of Medical
Informatics 8 1 (2 0 12) 98113 Hyrinen, K., Saranto, K., Nyknen, P.
(2008) Definition, Structure, Content, Use and Impacts of Electronic
Health Records: A Review of the Research Literature. International
Journal of Medical Informatics 77, 291-304.
[6] J. Bardram, A. Mihailidis, W. Dadong (Eds.), Pervasive Computing in
Healthcare, CRC Press, Taylor & Francis Group, USA (2006)
[7] R. Lenz, T. Elstner, H. Siegele, K.A. Kuhn A practical approach to
process support in health information systemsJ. Am. Med. Inform.
Assoc., 9 (6) (2002), pp. 571585
[8] American Psychological Association. 2007. APA Dictionary of
Psychology. Washington, D.C.:American Psychological Association.
[9] Masys DR. Advances in information technology. Implications for
medical education. West J Med. 1998; 168: 341 347.
[10] Ward JPT, Gordon J, Field MJ, Lehmann HP. Communication and
information technology in medical education. Lancet. 2001; 357: 792
796.
[11] Paperny DM. Computers and information technology: implications for
the 21st century. Adolesc Med. 2000; 11: 183 202.
[12] Pauker SG, Stahl JE. Medical informatics: where the action is
[editorial].West J Med 1997 Feb; 166:148-150
[13] Reuss, E., Rochus, K., Naef, R., Hunziker, S., Furler, L. (2007a) Nurses
Working Practices: What can We Learn from Designing Computerized
Patient Record Systems. In A. Holzinger (eds.) USAB2007, Graz,
Austria, Springer-Verlag, Berlin, Heidelberg, 55-68.
[14] Parmit K. Chilana, Andrew J. Ko, Jacob O. Wobbrock, Tovi Grossman,
George Fitzmaurice. Post-Deployment Usability: A Survey of Current
Practices, CHI 2011
[15] Sebastian Garde, David Harrison, Evelyn Hovenga Australian Skill
Needs Analysis of Health Informatics Professionals Volume 1: Rationale
& Methods, Key Findings & Conclusions, Research Report 1/2005
[16] Shah, S. G. S., Robinson, I. (2007) Benefits of and Barriers to involving
Users in Medical Device Technology Development and Evaluation.
International Journal of Technology Assessment in Health Care 23, 131-
137.
[17] Viitanen J, Hyppnen H, Lveri T, Vnsk J, Reponen J, Winblad
I. National questionnaire study on clinical ICT systems proofs:
physicians suffer from poor usability. Int J Med Inform. 2011 Oct;
80(10):708-25.
Vol. 3, No.8, 2012

124 | P a g e
[18] Innovators and Visionaries: Strategies for Creating a Person-centered
Health System. FACCT: Foundation for Accountability; September
2003.
[19] Shah, S. G. S., Robinson, I. (2006) User Involvement in Healthcare
Technology Development and Assessment: Structured Literature
Review. International Journal of Health Care Quality Assurance 19, 500-
515.
[20] Innovation of health technology, Report by Centre of AHIP center of
policy and research, Effective new solution for Americans health
insurance plans
[21] Audet AM, Doty MM, Peugh J, Shamasdin J, Zapert K, Schoenbaum S,
Information Technologies: when will they make it into physicians black
bags? MedGenMed. 2004; 6:2. [PMID: 15775829](13)
[22] Majumder A, D'Souza U, Rahman S. Trends in medical education:
Challenges and directions for need-based reforms of medical training in
South-East Asia. Indian J Med Sci 2004; 58:369-80
[23] Koschmann T. Medical education and computer literacy: learning about,
through, and with computers. Acad Med. 1995 Sep; 70(9):818-21
[24] Guest Editorial Nurses requirements for information technology: A
challenge for educators International Journal of Nursing Studies 44
(2007) 10751078.
[25] National Audit Office, 2006. Department of Health. The National
Program for IT in the NHS.
[26] Steven R. Simon; Rainu Kaushal; Paul D. Cleary; Chelsea A. Jenter;
Lynn A. Volk; E. John Orav; Elisabeth Burdick; Eric G. Poon; David W.
Bates. Physicians and Electronic Health Records: A Statewide Survey
Arch Intern Med. 2007; 167(5):507-512.
[27] http://www.cdc.gov/nchs/products/pubs/pubd/hestats/electronic/electroni
c.htm. Accessed on February 6, 2012.
[28] Khadzir bin Sheikh Haji Ahmad, Health Data Integration, Malaysian
Journal of Public Health Medicine, Vol. 11(Suppl 1) 2011
[29] Anogianakis, G., S. Maglavera, and A. Pomportsis, ATTRACT--
applications in telemedicine taking rapid advantage of cable television
network evolution. Studies in Health Technology & Informatics, 1998.
50: p. 60-6.
[30] Bielli, E., A Wireless Health Outcomes Monitoring System (WHOMS):
development and field testing with cancer patients using mobile phones.
BMC Medical Informatics & Decision Making, 2004. 4: p. 7.
[31] Muuronen, A. (2008) Lkreill on liikaa paperitit (in Finnish). (The
doctors have to do too much paper work). Helsingin Sanomat
newspaper, an article published in discussion column,published in
December 6th, 2008.
[32] S. De Rouck, A. Jacobs, M. Leys A methodology for shifting the focus
of e-Health support design onto user needs: a case in the homecare field
Int. J. Med. Inform., 77 (9) (2008), pp. 589601
[33] David A. Dzewaltowski, Paul A. Estabrooks, Lisa M. Klesges,
Sheana Bull and Russell E. Glasgow . Behavior change intervention
research in community settings: how generalizable are the results?
Health Promotion International Oxford Journal, Volume19, Issue2
Pp. 235-245.
[34] P. Nykanen, E. Karimaa, Success and failure factors in the regional
health information system design processresult from a constructive
evaluation study, Methods Inf. Med. 4 (2006) 8589
[35] Kadry, Bassama; Sanderson, Iain Ca; Macario, Alexb, Challenges that
limit meaningful use of health information technology, Current Opinion
in Anaesthesiology: April 2010 - Volume 23 - Issue 2 - p 184192
[36] Chuang, Y.H., Chuang, Y.W., 2002. Attitudes of two-year RNBSN
nursing students towards computers, The Journal of Health Science 5
(1), 7184
[37] R. Heeks, Health information systems: failure success and
improvisation, Int. J. Med. Inform. 75 (2006) 125137
AUTHOR'S PROFILE
Fozia Anwar is a research scholar in CIS department of University
Technology Petronas and after completing her Bachelor degree in
dentistry she did MS in Health Informatics from COMSATS Institute of
Information and Technology.

Vol. 3, No.8, 2012

125 | P a g e
Techniques to improve the GPS precision

Nelson Acosta
Institute of Advanced Informatic Technology Research
University Center of the Province of Buenos Aires
Tandil, Argentina
Juan Toloza
Institute of Advanced Informatic Technology Research
University Center of the Province of Buenos Aires
Tandil, Argentina

Abstract The accuracy of a standard market receiver GPS
(Global Positioning System) is near 10-15 meters the 95% of the
times. To reach a sub-metric level of accuracy some techniques
must be used [1]. This article describes some of these procedures
to improve the positioning accuracy by using a low-cost GPS in a
differential relative positioning way. The proposed techniques are
some variations of Kalman, fuzzy logic and information selection.
Keywords- GPS accuracy; relative positioning; DGPS; precision
farming GPS.
I. INTRODUCTION
The GPS operation principle is based on measuring ranges
of distances between the receiver and the satellites [2] [3]. The
GPS has architecture of three segments: spatial, control and
users. The spatial includes 24 satellites over 20 thousand km
away from the Earth, with six orbital levels and a 12-hour
period. The second segment includes the Earth stations to
control the satellites trajectories. Finally, the users segment
includes GPS receivers using two frequencies: L1 at 1575.42
Mhz for civil use, and L2 at 1227.60 Mhz reserved to military
use [4].
The accuracy in longitude and latitude coordinates is of
10-15 meters 95% of the readings [5]. Sometimes, it is more
precise, but it depends on a variety of factors that include from
the deviation or the delay of the signal when cross the
atmosphere, the bouncing of the signal in buildings or its
concealment due to the presence of trees [6], low accuracy of
clocks and noise in the receiver. In altitude the accuracy is
reduced to 50% regarding the obtained in terms of longitude
and latitude (15-23 meters 95%) [7].
Systems that enhance positional accuracy are: the DGPS
(Differential GPS), AGPS (Assisted GPS), RTK (Real-Time
Kinematic), e-Dif (extended Differential), amongst others.
The DGPS corrections service has two hard restrictions: it
must be afforded and the receiver must be close to a DGPS
station (less than 1000 km). The achieved accuracy can be of a
few meters [7, 8, 9, 10]. The correction signal cannot be
received if it is a mountainous zone.
In the case of AGPS, it is necessary to have mobile devices
with active data connection or cell phone like GPRS, Ethernet
or WiFi [11]. It is used in the cases where there is a weak
signal due to a surrounding of buildings or trees; this implies
having a not much precise position. Standard GPS receivers,
in order to triangulate and position, need a certain time of cold
start [12][13].
a) For use the RTK system, it is paid for the service and,
besides, it is very expensive to acquire the infrastructure. This
is a technique used in topography, marine navigation and in
agricultural automatic guidance in the use of measurements of
signals carrying navigators with GPS, GLONASS (Globalnaya
Navigatsionnaya Sputnikovaya Sistema) and/or Galileos
signals, where only one reference station provides correction
in real time, obtaining a sub-metric accuracy [14].
b) The last case, e-Dif system, is autonomous and it
process files with RINEX (Receiver Independent Exchange)
format, which was created to unify data of different receivers
manufacturers [15]. It generates autonomous corrections
regarding a coordinate of arbitrary reference and it
extrapolates them in time [16]. It is a very consistent relative
positioning and its accuracy is of about 1 meter. The systems
objective is to study waste from the initializing process to
isolate the most important systematic errors that introduce the
corresponding equations to each satellite. It is applicable for a
reduced time of 40 minutes approximately; since later the
systematic error changes, in this case a new error must be
calculated again. In regions where differential corrections
aren't available and it is paid for the service, like in South
American, African and Australian, this system become more
interesting.
c) Besides, there are systems that increase accuracy to
sub-metric levels. Those based on satellite SBAS (Satellite-
Based Augmentation System), based on ground GBAS
(Ground-Based Augmentation System) and based on aircraft
ABAS (Aircraft-Based Augmentation System). Most of these
implementations are used in different applications and some of
them are available for users without special permissions. Even
then, costs are high due to the need of certain devices with
special characteristics or some infrastructure in agreement
with the accuracy level desired.
Errors produced by the GPS system affect in the same way
the receivers located near each other in a limited radius. This
implies that errors are strongly correlated among near
receivers. Thus, if the error produced in one receiver is known,
it can be spread towards the rest in order to make them correct
their position. This principle is only applicable to receivers
that are exactly the same, the same methodology of [17];
since, if different, their specifications change so the signal
processed by one individual is not the same to that processed
by another one.
Vol. 3, No.8, 2012

126 | P a g e
All GPS differential methods use the same concept [18].
DGPS requires a base station with a GPS receiver in a precise
known position. The base compares its known position with
that calculated by the satellite signal. The estimated difference
in the base is applicable then to the mobile GPS receiver as a
differential correction with the premise that any two receivers
relatively near experiment similar errors [5].
In this article it is emphasized in the behavior of GPS
errors in time, after the techniques developed here to
determine magnitude and direction of error are applied. In
Section 2 techniques to calculate errors are analyzed. In
Section 3 algorithms used are described and finally the error
behavior is presented in figures. In Section 4, the last one,
conclusions and future work are presented.
II. TECHNIQUES FOR ERROR ANALYSIS
The experiment carried out is based on the principle of the
adopted methodology by the DGPS but with a low cost
standard GPS receiver. In order to get measurements, three
Garmin 18X USB GPS receivers are used connected to two
notebooks. The base station is composed of a notebook and
two of the three GPS receivers; the mobile for the other
notebook and the remaining GPS receiver. The link between
the base station and the mobile one is a point to point wireless
connection.
In this context, in the base system, measurements from the
GPS receivers are obtained and after a certain period of time,
which is necessary for the system stabilization, two positions
are estimated. The positions estimation are carried out with a
Kalman filter, since an estimation problem with so many noisy
redundant data is a natural application for the Kalman filter;
this allows using some of the redundant information to remove
the effects from the error sources. The Kalman filter is used to
eliminate the white Gaussian noise [19].
Receivers are placed at a known distance between
themselves (relative positioning) [20]. At the end of this stage,
a cloud of points from standard GPS system is obtained with
the positions delivered by the receivers and with those values a
position is estimated for each receiver. The estimated point
from GPS 1 is selected as anchor point of the whole
experiment. From this, all necessary calculations are carried
out with the objective of finding the GPS system error.
With both estimated positions, it proceeds to calculate the
distance between them. If the estimated distance is different
from the actual one (more/less a threshold) it is detected that
there is a positioning error. Besides, a circumference with a
radius equal to the actual distance measured with a tape
measure from GPS 1 to GPS 2 with center in the estimated
point for the GPS 1 is drawn. A circumference is chosen, as
GPS 2 can be at that distance but in any point of the
circumference. This is the working principle that the GPS
system uses to get the receivers position.
After calculating the distance of the estimated points and
contrasting it to the actual one, a positioning error is deduced.
Once it is known that there is an error, it must compute its
magnitude and direction. On the one hand, the two estimated
points are learnt with which the straight line is drawn and
which bonds them. Equation 1 belongs to the straight line that
crosses these two points.
( )
( )
( )
( ) 1 2
1
1 2
1
x x
x x
y y
y y
(1)

where the x represents the component of Longitude and the
y that of Latitude.
On the other hand, it is known that the GPS 2 is in some
point of the circumference with center in the GPS 1 and of
radius the distance that was defined at the moment of
positioning the two receivers. Equation 2 belongs to the
circumference with center in (x1,y1) of radius r.
( ) ( )
2 2 2
1 1 r y y x x = + (2)

Knowing about the equations that define the straight line
crossing both estimated points and the radius circumference
equation equal to the GPS 1-GPS 2 distance with center in
GPS 1, it is proceeded to approximate, by means of the
intersection of the straight line and the circumference.
This intersection is presented as a polynomial of second
degree. It is mathematically solved and a, b and c coefficients
are obtained (equations 4, 5 and 6). Equation 3 only presents
an auxiliary estimate in order not to repeat it in the other
operations and to increase legibility in the rest of the
equations. With these coefficients cleared by means of
Bascara (equation 7) the roots are found (two because of being
of second degree) from the polynomial. From the two roots
found, one is chosen and the intersection points are calculated.
2
1 1 2
2
2
2 y y y y divisor + = (3)
( )
divisor
x x x x
a
2
1 1 2
2
2
2
1
+
+ = (4)

( )
divisor
y x y x x y x
y b
1
2
1 1 1 2 1
2
2
1
2 4 2
2
+
+ = (5)
( )
divisor
y x y x x y x
r y c
2
1
2
1
2
1 1 2
2
1
2
2 2 2
1
2 +
+ = (6)
( )
a
ac b b
x
2
4
2

= (7)

After having obtained the two roots, only one is taken into
account. The nearest to the estimated point of GPS 2 in some
of its components is chosen, in this case in Latitude, since the
other one is meaningless due to being too far away (on the
other side of the circumference).
( )
2 1 2
. , . , root roots root roots y rootCloser result
latitude
=
Finally, the other component (Longitude) is obtained
depending on the Latitude found in the previous point as seen
in equation 8.
Vol. 3, No.8, 2012

127 | P a g e
( )
( )
|
|
.
|
\
|

+ =
1 2
1
1 2 1
*
y y
y result
x x x result
latitude
longitude
(8)

Once the error magnitude is known, is necessary to get the
direction in order to have the final correction vector. There are
nine possible cases, that are the combination of two variables
(latitude and longitude) and three values that are: greater,
lesser and equal. In Table I these cases are presented.
TABLE I. DETERMINATION OF ERROR DIRECTION.
Estimated
Latitude Longitude
Greater Greater
Greater Lesser
Greater Equal
Lesser Greater
Lesser Lesser
Lesser Equal
Equal Greater
Equal Lesser
Equal Equal
Latitude Longitude
Actual

The way to read the table is top-down. Estimated latitude
greater than actual latitude and an equal longitude to actual is
presented in the third row. In the case where actual and
estimated components are equal, like in ninth row of the Table
I, there is no correction.
Therefore, in order to get the error direction it is proceeded
to verify which side of the corrected point the estimated point
is. If the straight line that joins both estimated points is
horizontal or vertical, some of the components are null, in the
horizontal case, Latitude is eliminated and in the vertical the
Longitude.
III. ALGORITHMS AND USED TECHNIQUES
For the analysis of data, combinations of different
techniques and algorithms are used in order to find a better
result. In a first processing stage, applied mathematics covers:
- Static Kalman: is a set of mathematical equations that
provide an efficient recursive solution of the method
of least squares. This solution allows calculating a
linear, unbiased and optimum estimator of the state of
a process in each moment of time (t) with base on the
available information at the moment t-1, and update,
with the available information at the moment t, the
estimator value.
- Dynamic Kalman: is the system in which the value of
variable x to be estimated has a value that changes
throughout the time ( ), but these states have
some known relationship with the instant i and i+1.
For example, if an object position is measured, it can
be predicted that the position will be:

where is the passed time and the speed at instant
i. Position can be obtained by a GPS, for instance,
and speed with an additional measurement element
such as an accelerometer.
- Kalman with adjustment of error standard deviation:
the deviation is modified and checked in order to see
which adjust better. This measure is calculated as the
square root of variance, which is at the same time the
sum of the squares of each error (Table II) as shown
in equation 9. It is worth mentioning that from Table
II the only error that is not taken into account is that
of signal P(Y) arrival; since work is carried out
without the precision code.

m m
R
6,7 5 . 0 1 2 5 . 2 5 3
2 2 2 2 2 2
= + + + + + = o

(9)
TABLE II. GPS ERROR SOURCE.
Source Effect
(Meters)
Arrival of signal C/A 3
Arrival of signal P(Y) 0.3
Ionosphere 5
Ephemeris 2.5
Satellite clock error 2
Multipath 1
Troposphere 0.5
Numerical errors 1

Now, error standard deviation ( ) in the receivers
position is estimated, but having into account additionally the
PDOP (Position Dilution of Precision) and the numerical
error; therefore, the PDOP is added to the calculated deviation
from typical errors, since for each measurement taken, this
varies according to the instant geometry of satellites. The
result of standard deviation used for the Kalman filter is
equation 10.
m PDOP PDOP
num R rc
2 2 2 2 2 2
1 6.7 * * + = + = o o o

(10)

The fact of applying Kalman with adjustment of standard
deviation, since it fluctuates for each piece of information
coming from the receivers in each moment as geometry of
satellites varies.
- Points average: one of the media limitations is that it
is affected by extreme values; very high values tent to
increase it while very low values tend to reduce it;
this implies that it may stop being representative of
the population. It is analyzed but not implemented in
the solution. This solution was used in [21].
- Fuzzy logic: to determine the position error degree.
Rules that determine the position error are related to
analyzing some parameters like: PDOP, SNR
(Signal-to-Noise Ratio) and difference of tracked
i
v t x x
i i
*
1
A + =
+
i i
x x =
+1
i
v t A
rc
o
Vol. 3, No.8, 2012

128 | P a g e
satellites. The fuzzy system output weights the
Kalman filter gain, providing more weight to more
precise positions and the other way round.
- Filters allow discarding measurements with much
noise or error that influence over the final result of an
estimation of a position. Thus, measurements having
many errors do not slant the final estimation towards
a position far away from the actual one. The
application of these filters can be made as
measurements are not very far away in time and it is
supposed that the Vehicle in which the mobile
receiver is placed does not move at high speed; this
implies that values do not change radically. High/low
step filters are used in an analogical way to the
electronic filter.
After apply the current techniques, in Figs. 1, 2 and 3 it is
observed the original and estimated errors. The absolute errors
are calculated by means of relative positioning determined by
sets of two receivers. The distance between the pairs of
receivers is known at starting the experiment. In Fig. 1, the
estimated error tends to zero. By the other hand, in Fig. 2,
estimated error fluctuates around the average of the original
error. Finally, in Fig. 3, the estimated error is zero in a
moment, but after a period, begins to oscillate. With this set of
graphics, it is observed an oscillation in the standard system
that doesnt allows having a known error to correct a position.
When the presented techniques are applied it is possible to
obtain an smoothing error, a value that fluctuates less in time,
which allows to correct positions in other receivers in order to
improve the accuracy.

Figure 1. Behavior in time of estimated and original errors for GPS 1 regarding to GPS 2.


Absolute errors GPS 1-2
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1 27 53 79 105 131 157 183 209 235 261 287 313 339 365 391 417 443 469 495 521 547 573 599
Samples
M
e
t
e
r
s
original errors estimated errors
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
1 29 57 85 113 141 169 197 225 253 281 309 337 365 393 421 449 477 505 533 561 589
Samples
M
e
t
e
r
s
Vol. 3, No.8, 2012

129 | P a g e

Since the tool that implements these techniques is
thought to operate in different places of heterogeneous
characteristics, relations and configurations are used in order
to be able to customize the use according to needs. Relations
and configurations used are the following:
- Rate Degree/Meters in Latitude: given the
asymmetry, in different places on Earth, the
distance that a Latitude degree measures varies.
- Rate Degree/Meters in Longitude: ditto to Latitude.
- Cold start time: a start time is considered so the
system can be stabilized. In this time, samples of
the device are taken and, only at its end, estimation
is carried out. The objective is to reduce or soften
systematic and random errors of the GPS system.
- Receivers distance threshold: it can be determined,
in an accurate way, the whole interval of the
distance measurement between the base devices.
As positioning is relative and its distance is known,
it can be added a value, since there exists a
possibility that an element of distance measurement
be not accurate enough. Besides, it reduces the
computational load because of not having to
process data if distance is within the allowed
threshold.
IV. CONCLUSION AND FUTURE WORKS
The techniques developed allow obtaining the magnitude
and direction of error provoked by the GPS system as
presented in [21]. This correction is used by another receiver
to correct its own position and thus increase the positional
accuracy with the aim of measuring the most precise
distances.
The experiments carried out with different sets of data
provide positions that are used to measure distances and
error fluctuates in 1 meter the 95% of measurements and
in some cases in 0.2 meters.
The principle is based in mathematical, geometrical
functions and filters. With the techniques presented in this
article, it is possible to obtain a smoothing error and a value
constant in time, which allows correcting positions in other
receivers in order to improve their accuracy.
As future work, the aim is to increase the accuracy until
reaching a maximum error of positioning of 0.1 meters.
This increase can include the use of another additional
signal. Besides, further measurements will be carried out in
order to analyze data and to deduce its behavior. It is
intended to extend the use in faster vehicles in order to
widen the application field of the DGPS system introduced
here.
ACKNOWLEDGMENT
Agencia Nacional de Promocin Cientfica y
Tecnolgica for supporting since 2009.
REFERENCES
[1] M. S. Grewal, L. R. Weill and A. P. Andrews, Global Positioning
Systems, Inertial Navigation, and Integration. 2
nd
Edition, Wiley,
2007.
[2] G. Xu, GPS: Theory, Algorithms and Applications. 2
nd
Edition,
Springer-Verlag Berlin Heidelberg, 2007.
[3] P. Misra and P. Enge, Global Positioning System: Signals,
Measurements, and Performance. New York, Ganhga-Jamuna Press,
2010.
[4] T. Feldmann, A. Bauch, D. Piester, H. Esteban, J. Palacio, F. J.
Galindo, T. Gotoh, H. Maeno, U. Weinbach and S. Schon, GPS
carrier phase and precise point positioning time scale comparisons
using different software packages, Frequency Control Symposium,
2009 Joint with the 22
nd
European Frequency and Time Forum, IEEE,
pp. 120-125, 2009.
[5] P. A. Zandbergen and L. L. Arnold, Positional accuracy of the wide
area augmentation system in consumer-grade GPS units, Computers
and Geosciences Volume 37 Issue 7, Elsevier, pp. 883-892, 2011.
[6] C. Ordnez Galn, J. R. Rodrguez-Prez, J. Martnez Torres and P. J.
Garca Nieto, Analysis of the influence of forest environments on the
accuracy of GPS measurements by using genetic algorithms,
Mathematical and Computer Modelling Volume 54 Issue 7-8,
Elsevier, pp. 1829-1834, 2011.
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1 22 43 64 85 106 127 148 169 190 211 232 253 274 295 316 337 358 379 400 421 442 463 484 505 526 547 568 589
Samples
M
e
t
e
r
s
Vol. 3, No.8, 2012

130 | P a g e
[7] S. Featherstone, Outdoor Guide to Using Your GPS. Creative
Publishing International, Inc., 2004.
[8] G. Satheesh, Global Positioning Systems: Principles and Applications.
Mc-Graw Hill, 2005.
[9] S. D. Ilev, Global Mobile Satellite Communications for Maritime,
Land and Aeronautical Applications. Springer, 2005.
[10] M. Ghavami, L. B. Michael and R. Kohno, Ultra Wideband Signals
and Systems in Communication Engineering. 2
nd
Edition, John Wiley
and Sons, Ltd., 2007.
[11] C. Ho, An effective approach in improving A-GPS accuracy to
enhance hybrid positioning computation, 17
th
International
Conference on Embedded and Real-Time Computing Systems and
Applications (RTCSA), Toyama, Japan, IEEE, pp. 126-130., 2011.
[12] J. Li and M. Wu, A positioning algorithm of AGPS, International
Conference on Signal Processing Systems, Singapore, IEEE, pp. 385-
388, 2009.
[13] F. Van Diggelen, A-GPS, Assisted GPS, GNSS, and SBAS. Artech
House, 2009.
[14] D. Dardari, E. Falletti and M. Luise, Sattellite and Terrestrial Radio
Positioning Techniques: A Signal Processing Perspective. 1
st
Edition,
Elsevier, 2012.
[15] N. H. M. Hanif, M. A. Haron, M. H. Jusoh, S . A. M. Al Junid, M. F.
M. Idros, F. N. Osman and Z. Othman, Implementation of real-time
kinematic data to determine the ionospheric total electron content, 3
rd

International Conference on Intelligent Systems Modelling and
Simulation (ISMS), Kota Kinabalu, Malaysia, IEEE, pp. 238-243,
2012.
[16] GPS World (serial online), CSI wireless differential software
patented, Volume 13 Issue 8, EDS Foundation Index, Ipswich, MA.
pp. 48, 2002. Accessed June 18, 2012.
[17] M. G. Wing and J. Frank, Vertical measurement accuracy and
reliability of mapping-grade GPS receivers, Computers and
Electronics in Agriculture Volume 78 Issue 2, Elsevier, pp. 188-194,
2011.
[18] V. Di Lecce, A. Amato and V. Piuri, Neural technologies for
increasing the GPS position accuracy, International Conference on
Computational Intelligence for Measurement Systems And
Applications (CIMSA), Istanbul, Turkey, IEEE, pp. 4-8, 2008.
[19] H. Eom and M. Lee, Position error correction for DGPS based
localization using LSM and Kalman filter, International Conference
on Control, Automation and Systems (ICCAS), Gyeonggi-do, Korea,
IEEE, pp. 1576-1579, 2010.
[20] Y. He, H. Yu and H. Fang, Study on improving GPS measurement
accuracy, Instrumentation and Measurement Technology Conference
(IMTC), Otawa, Canada, IEEE, pp. 1476-1479, 2005.
[21] J. Toloza, N. Acosta and A. De Giusti, An approach to determine the
magnitude and direction error in GPS system, Asian Journal of
Computer Science and Informatin Technology, in press, 2012.
AUTHORS PROFILE
Ph.D. Hctor Nelson Acosta was graduated at the
Universidad Nacional del Centro de la Provincia de
Buenos Aires (Argentina), and got the PhD degree at
the Autonomous University of Madrid (Madrid,
Spain). He is the director of the Instituto de
Investigacin en Tecnologa Informtica Avanzada
(INTIA) at the Universidad Nacional del Centro de la
Provincia de Buenos Aires since the 1998. He is
working as a proffesor since 1993 at the Computer and
Systems Department, at the Universidad Nacional del
Centro de la Provincia de Buenos Aires, Tandil (Argentina). He has been a
visiting proffesor of several Universities in Argentine. He has suppervised 6
Ph.D. students in Computer Sciences and 2 MSc in Informatics. His
research insterest includes real-time systems, signal procesing, pattern
recognition, custom architectures, custom processors, robotic navigation
and accuracy positioning systems.
Eng. Juan Manuel Toloza was graduated in
march 2009 at the Universidad Nacional del Centro
de la Provincia de Buenos Aires (Argentina). He is in
the last year of Ph. D. (Computer Science) at the
Universidad Nacional de La Plata (Argentina). He
have a scholarship of the Agencia Nacional de
Promocin Cientfica y Tecnolgica since 2009. He
works in the Instituto de Investigacin en Tecnologa
Informtica Avanzada at the Universidad Nacional
del Centro de la Provincia de Buenos Aires. His
research insterest includes signal procesing, data analysis from diferents
sensors, robotics and accuracy GPS. He is teacher assistant in Computer`s
Architecture and Digital Techniques.

Vol. 3, No.8, 2012

131 | P a g e
M-Commerce service systems implementation

Dr.Asmahan Altaher
Applied Science University
Amman - Jordan

Abstract Mobile commerce supports automated banking
services. However, the implementation of m-commerce services
systems has become increasingly important in todays dynamic
banking environment. This research studied the relationships
between technology acceptance model and m- commerce services.
The results of the survey on 249 respondents in several Jordan
banks revealed that technology acceptance model had a
significant impact on m-commerce services. The results led to the
recommendation that the technology acceptance model is a
success model for support using new services for electronic
commerce. In addition, managers play a significant role in
influencing the mobile services in banks through social
interaction. Managers should focus on relative advantage,
usefulness, and ease of use, in order to develop the mobile
commerce services implementation.
Keywords- M-commerce services; usefulness; and ease of use;
social interaction.
I. INTRODUCTION
The growth in the mobile phone industry has accelerated in
few years due to constant technological development.
Recently, new forms of mobile services have made possible
text messaging, web surfing, digital imaging, payments,
banking, financial instrument trading and shopping [6]. With
the rapid advance of telecommunication technologies, Mobile
Services (MDS) is defined here as wireless access to the
internet through a mobile communication network. However,
mobile services are becoming increasingly important for
companies and consumers because of ubiquitous, universal,
easy-to-access information, and personalized exchange of
information [21]. Hence, it is very important to understand
how individual differences influence the use of mobile
services and the behavioral adoption requirements of these
services, marketers and service providers [4]. Employees may
resist the change and may be a major source of concern for
them. To avoid negative consequences of resistance to change,
system implementers and managers must actively manage the
change process and gain acceptance for new IS. The
Technology Acceptance Model (TAM) suggests that employee
attitudes may change if they think the new system will help
them do more or better work for the same effort, and that it is
easy to use [15]. TAM suggests that managers cannot get
employees to use a system until they want to use it. To
convince employees to use the systems, managers may need to
change employees' attitudes about the system [10]. Employees
attitudes may change if employees believe that the system will
allow them do better work for the same amount of effort
(perceived usefulness), and that it is easy to use. Training,
documentation, and user support consultants are external
variables that may help explain the usefulness of the system
and make it easier to use; in addition, mobility model is very
important to improve the efficiency of secure mobility and
routing discovery [20]. TAM has many variants, for example,
one variant considers subjective norms, whereas another adds
attitudes toward behaviours like social influence (subjective
norms), and facilitating condition (top management support).
Although social influences are important, they are likely to be
important only for yang workers when they are likely to start
using the system. TAM assumes that technology will be
accepted if people's attitudes and beliefs support its use. One
way to make sure that employees attitudes and beliefs are
favourable toward the system is to have them participate in its
design and implementation. When future users of the system
participate in its design and implementation, they may be more
willing to accept the consequences of the trade-off [10] [18].
The objective of this study is to investigate how the
technology acceptance model can involve employees and
allow them to understand the m- commerce services in Jordan
banks. In order to address these questions, the researcher drew
post-adoption components from previous studies in related
fields, and conducted a large-scale survey on mature TAM in
terms of how each component affected the use of m-commerce
services systems. Subsequently, the theoretical background
and hypotheses of this study is presented, after which a
description of the research methods and data collected was
presented. This was followed by an analysis of the research
results. Thereafter, the conclusion was drawn.
II. LITERATURE REVIEW
Kim et al. [4] studied and examined factors that are most
important in converting services to m-commerce services and
whether or not they differ from those that are effective in
maintaining continuers. The researchers conducted an online
survey to compare continuers and discontinuers empirically in
terms of the relative importance of four post-adoption factors
to the behavioural intention to use m-commerce. The results
show that usefulness and social influence were more important
for discontinuers; and ubiquitous connectivity for continuers
[17] find that the mobile phone user types are important to the
mobile phone service companies, and communicative
references for understanding the mobile phones. Mobile phone
user types recently have become a popular subject of
discussion. The researchers classify the mobile phone user
types into four types: guanxi-expanding, illness-phobia,
convenience-oriented and life-interrupting. The users of these
mobile service companies references for understanding the
mobile phone use were adopted. However, Weber et al. [18]
argued that the use of mobile technology for carrying out
Vol. 3, No.8, 2012

132 | P a g e
surveys is very important and are necessary actions which
provide a guideline for realizing the potential of mobile
surveys. Essentially, the benefits are enhancements in the
areas of survey quality, management and technology. The
researchers recommended that the development of mobile
surveys is important to create a new business model in the
market. On the other hand, Xie et al. [10] present some
potential security threats against m-commerce. The security
attacks such as wireless-internet can seriously degrade the
performance of mobile services of users; in order to promote
the widespread deployment of m-commerce, the firms need to
design novel and robust as well as efficient security schemes
to handle these attacks. In this article, the researchers
discussed some important challenges for providing ubiquitous
and secure m-commerce. Finally, Lin [5] agreed that m-
commerce service is being totally satisfied with the system and
their users' values; mobile technology trusting expectations
were very important in the continued m-commerce service
usage behaviors, and the providers might not fulfills the m-
commerce service need for consumers, but satisfied with the
m-commerce service delivered. M-commerce services were
introduced only a few years ago and are distinctly different
from prior services and information systems because m-
commerce services systems are a type of information system;
and many adoption studies [for example, Technology
Acceptance Model (TAM)] have been conducted in the area.
To study acceptance of the m-commerce services systems, the
researcher selected four key components that have been
frequently used in previous adoption studies in these areas:
social influence, usefulness, ease of use and mobility model.
Definitions of each component, as well as accounts of why
they are pertinent to this study are as follow.
III. PROCEDURAL DEFINITIONS
A. Social Influence:
Social influence refers to the effect of interaction among
people in their social context. Social influence helps to
determine whether technologies were adopted and whether
products are purchased [14]. Fisher and Price [3] found that
when people purchased new products, others opinions
influenced their purchasing decisions. Social influence may
also affect the use of MDS. In fact, since M-commerce is a
part of the telecommunications industry an industry
specifically designed to facilitate social interactions social
influence may be an even more important factor in service
choice [4] [13][20].
B. Usefulness:
Usefulness is how helpful a user feels a new product is to
his or her work [10]. When the usefulness of a new product is
high, the product is adopted rapidly in the market. [4] [16]
have established that good quality and good function, as
perceived by users, allow them to adopt a new system easily in
an organizational environment. Usefulness appears to be a key
factor in the adoption of M- commerce [11] [12] especially
when their usefulness differentiates them from traditional
internet services. For example, subjects in one study reported
using mcommerce services only when the usefulness of their
mobility really mattered [6].
C. Ease of Use:
Ease of use, another subjective measure, describes how
easy and comfortable people find a system to learn and use.
Kim et al. [4] suggested that a system is adopted quickly when
a user can easily learn how to use it. Novak [8] proposed that a
better system must be efficient to use. Ease of use will also be
an important factor for the adoption of M- commerce services
[1]. Providers are coming more and more to believe that ease
of use is the key to retaining the greatest barrier to the
adoption of M-commerce. However, compared with the
resources of other systems, the resources available to make M-
commerce services easier are severely constrained [4] [17].
D. Mobile Commerce Services:
Mobile internet services numerously involve mobile
commerce (m-commerce) services for continued and promoted
business profits. Mobile technology trusting expectations were
very important in the continued m-commerce service usage
behaviors; and the providers might not fulfills the m-
commerce service need of consumers, but satisfied with the m-
commerce service delivered. M-commerce provides a
significant value in convenience, efficiency, entertainment,
mobility and location and besides extending the benefits of the
web [5] [6]. Therefore, this study proposes personal values for
the prediction of continuance intention. This may offer
additional information or insights beyond TAM in explaining
why m-commerce services are used. M-commerce service and
usage continuance intentions in terms of personal values lead
to trust expectations about the m-commerce trusting and
perceived performance. Most efficient factor is quick customer
service. Legislation has increased customers rights and
technology and competition have increased their choice of
products and providers. These changes will result in growth in
users with sophisticated needs supported by [2] [9] [19].
Internet technology is rapidly changing the way personal
financial services are being designed and delivered. For
several years, commercial banks in Jordan have tried to
introduce electronic banking (e-banking) systems to improve
their operations and to reduce costs. Despite all their efforts
aimed at developing better and easier e-banking systems, these
systems remained largely unnoticed by the customers, and
certainly were seriously underused in spite of their availability.
In this Internet age, when the customer is having access to
a variety of products and services it is becoming very difficult
for banks to survive. In this situation, when customer enquiries
are not met easily or transactions are complicated, the
customer will ask for new levels services, and only chose
those institutions who are making a real effort to provide a
high level of quality, fast and efficient service through all the
banks touch points, call centers, TAM, voice response
systems, Internet and branches. It is considered as one of the
better financial sectors in the region and generates in total
close to 5% of the Gross Domestic Product (GDP). One of the
weakest points in the financial sector is, with the exception of
mortgage lending, the lack of long-term lending and the
absence of secured loans. It is worth mentioning that the
percentage of Jordanian households who own personal
computer is 15.9%, Internet access is 6%, 1,000,000 regular
telephone lines, around 1.6 million mobile telephony
Vol. 3, No.8, 2012

133 | P a g e
subscribers, 21 licensed Internet service providers, and more
than 500,000. The banking sector is very dynamic and liberal
in Jordan. Moreover, some of the commercial banks in Jordan
are offering electronic services. Samples of these services are:
(1) Internet banking: Arab Bank is the first bank to launch
Internet banking service. This has been started in Jordan in
May 2000.
(2) Internet Shopping Card (ISC): It provides convenient
and easy access to on-line shopping transactions.
(3) WAP banking: Customers can use WAP mobile phone
and access their accounts.
(4) SMS banking: Customers can use a mobile phone to
access their accounts.
(5) Phone bank service: This provides access to customers
accounts.
(6) On-line stock trading: Jordan Kuwait Bank (JKB)
offers this service in collaboration with its affiliate (United
Financial Investment).
(7) Net banker: For performing banking transactions.
(8) Mobile Banking: This service allows the customers to
perform banking transactions by using a mobile.
(9) Pre-paid mobile cards: Customers can buy the mobile
prepaid cards electronically.
(10) Banking via SMS: It enables the customers to receive
information on their transactions through their mobile
telephones [1]. The financial services industry will be used to
explore m-commerce in Jordanian banking services sector, and
to investigate the effects of TAM through carrying out m-
commerce in Jordanian banking sector (e-banking). The
following are the research model.
RESEARCH HYPOTHESES
H
1
: There is a significant positive effect between
technology acceptance model and m-commerce services
implementation.
H
2
technology acceptance model and social influences.
H
3
technology acceptance model and technology usefulness.
H
4
technology acceptance model and technology ease of use.
RESEARCH MODEL
The model depends in Technology acceptance model
pearlson(2009). The researcher tries to investigate the model
statistical in order find new way that can that supports
involved employees and allow them to understand the m-
commerce services in Jordan banks.

Figure.1. Research Model
Recourse: Pearlson (2009)
IV. RESEARCH METHODOLOGY
This research is concerned with making our problems
accessible by directing it in a way that generates precise
answers to precise question. The research methodology can be
derived from two approaches that can be classified into two
main categories: quantitative and qualitative methodology
[13].
A. Data Collection Methods
The data and information will be gathered from two
resources, namely: primary and secondary resources.
i) Primary resources: Individual focus groups, and a panel
of respondents set up by the researcher whose opinions may be
sought on specific issues from time to time are examples of
primary data sources [1]. Data can also be culled from
administering questionnaire. Questionnaires are an efficient
data collection mechanism when the researcher knows exactly
what is required and how to measure the variable of interest
[22]. In this study, questionnaires were sent to respondents in
the senior level, and top managers in management.
ii) Secondary resources: Data can also be obtained from
secondary sources, as the scientific (Books, articles, etc)
sources concerned with the study.
B. Initial Design And Development Of The Survey Instrument:
Many criteria should be considered when designing a
questionnaire survey [22]. On the choice of wording,
questionnaire design and layout were adopted. Items in the
questionnaire were designed to being simple, clear, short,
technical, accurate, bias free, and at an appropriate reading
level [1] [22] were taken into account when designing the
questionnaire, such as those that started with a brief
description on how to answer the questionnaire. An initial
draft of questionnaire was developed based on an extensive
literature review and existing measures.
C. Decisions Related To Population And Sample Selections:
The banks market is huge and has been growing rapidly in
recent years. Banks can use all of the e-services.

Technology
usefulness

Acceptance m-
commerce services

Social
influences
Technology
ease of use
Vol. 3, No.8, 2012

134 | P a g e
However, many large banks receive additional e-services,
the banks was seen as an appropriate business environment
that is particularly suitable to test for the research model and
the determinations of the m-commerce implementations in e-
banking. Jordan has sixteen banks, Arab banks controlling the
local market and, account for more than 40 - 50% of the
countrys total production. Sekaran [22] defined research
population as any exactly defined set of people, or collection
of items, that is under investigation. In the light of this
definition, the research population, and the actual sample are
identified as the four banks that dominate the local market and
account for more than 75% of the countrys total production.
These are Arab Bank, The Housing Bank for Trading Finance,
Jordan Bank, and Standard Chartered Bank.
A questionnaire was sent to respondents in operation,
middle and top management. A stratified random sampling
method will be used, as it is the most convenient, and the most
applicable in the Jordanian context.
The unit of analysis in this study is managers working in
Jordan banks. 200 questionnaires were sent to 320
populations; 263 were returned, and 14 questionnaires were
ignored because they were not returned. The overall response
rate for this study is 82%, while the response rate actually used
is 78%.
This is regarded as relatively high, since the respondents
are managers who are supposed to be too busy to answer
questionnaires. However, it is found that the sample is
sufficient to represent the regression analysis conducted.
D. Operationalisation And Measurement Strategy Of The
Model Variables:
The measures of model variables in this study were
analyzed using statistical procedures starting with internal
consistency test and establishing constructs reliability.
Statistical procedures are common among many researchers
such as Malhotra et al. [23].
1) Internal consistency to assess the reliability of the scale
using Cronbach's alpha.
2) Developing a structural model based on the composite
measures to linking the hypothesized model's constructs.
3) Descriptive analysis of the mean and standard deviation
of the investigated hypotheses.
E. Internal Reliability
The internal consistency measures and Cronbachs alpha
are obtained in order to assess the reliability of the
measurement instruments. Table 1 shows the Cronbachs
alpha value for each scale. It is clear that Cronbach alpha is
acceptable statistically and managerially because values are
greater than the accepted 0.65%.
TABLE 1. Cronbach alpha
Variable Cronbachs alpha
Implementation M-commerce 0.756
Social influence 0.732
Usefulness 0.678
Ease of use 0.692
DESCRIPTIVE STATISTICS
Descriptive statistics such as means frequencies, and
standard deviation, were used to identify the major
characteristics of respondents in terms of their gender, age,
educational level and working experience.
TABLE 2. DESCRIPTIVE STATISTICS
Demographic object Valid item %
Gender
Male 64.7
Female 35.3
Employee age (years)
20-30 41.0
30-40 35.38
40-50 15.17
Above 50 8.45
Education level
College degree 1.5
Bachelor degree 88.2
Postgraduate degree 10.3
Work experience (years)
Less than 6 29.0
From 1 - 2 25.0
5 10 20.37
More than 10 25.63
V. RESULTS
This study attempts to identify the technology acceptance
model effects on implementation of mobile commerce e-
banking services. The statistical results indicated the TAM
effects of mobile services implementation. All analyses were
conducted with SPSS. Frequency and percentage were used to
describe the samples of the study and multiple regression
analysis was conducted to test the research hypotheses.
A. Measuring The Effect Between The Independent Variable
And Dependent Variable Simple Regression
Most of the respondents agreed that there is a good and
close relationship between TAM factor and mobile services
implementations (Mean=3, 69; SD=0.98). The result of the
regression analysis shows that there is a significant positive
effect at the function level ( 0.01) which means that there is
a relationship between TAM and mobile commerce
implementation for the independent variables with a variance
of 47.4%; thus, Hypothesis 1 was supported.
B. The Effect Between Social Influence And Mobile
Commerce Implementation
From the results in Table 3 which relate to correlation
between the independent variable social influence and the
dependent variable mobile commerce implementation, a
positive and significant effect can be found at function level (
0.01) which supports the hypothesis (H
2
), where (r=0.499).
Based on the results in Table 3 which represent simple
regression analysis, Simple regression was used to test the
above hypothesis, and it was found that the calculated t
(8.112) is significant at (p>0.001) level, which means that
there is a relationship between social influence and mobile
commerce implementation. It was observed that the variance
was 41.1%, as such, Hypothesis 2 was supported.
Vol. 3, No.8, 2012

135 | P a g e
TABLE 3. THE EFFECT BETWEEN SOCIAL INFLUENCE AND MOBILE
COMMERCE IMPLEMENTATION
Variable Beta Sig
Social influence and mobile commerce
implementation
0.499 0.000
C. Effect Between Usefulness And Mobile Commerce
Implementation
Based on the results in Table 4 which relate to correlation
relationship between the independent variable usefulness and
the dependent variable mobile commerce implementation, a
positive and significant effect can be found at function level (
0.01) which supports the hypothesis (H
3
). Based on the
results in Table 4 which represent simple regression analysis,
simple regression was used to test the above hypothesis, and it
was found that the calculated t (89.932) is significant at (0.01)
level, which means that there is a relationship between them.
A significant effect can be noticed at function level ( 0.01)
to the independent variable usefulness in the dependent
variable mobile commerce implementation. The variance was
observed as 61.7%. Thus, hypothesis 3 is supported.
TABLE 4. THE EFFECT USEFULNESS AND MOBILE COMMERCE
IMPLEMENTATION
Variable Beta Sig
Usefulness and mobile commerce implementation 0.393 0.000

D. Effect between ease of use and mobile commerce
implementation
Based on the results in Table 5 which relate to correlation
relationship between the independent variable services, ease of
use and the dependent variable mobile commerce
implementation, a positive and significant effect can be found
at function level ( 0.01), which supports the hypothesis
(H
4
) (Table 5). Simple regression was used to test the above
hypothesis and it was found that the calculated t (7.641) is
significant at (0.01) level; a significant effect can be found at
function level ( 0.01). The variance is (0.681). Thus,
hypothesis 4 is supported partially.
TABLE 5. The Effect between Ease of Use and Mobile Commerce
Implementation
Variable Beta Sig
Ease of use and mobile commerce implementation 0.329 0.000
VI. CONCLUSION
The researcher investigated the Technology acceptance
model statistical in order find new way that can that supports
involved employees and allows them to understand the m-
commerce services in Jordan banks. The researcher argued
that developing countries have to apply the technology
acceptance model in order to help employees accept the new
types of e-services.
A technology acceptance model is a success model for
support using new services for electronic commerce. Social
influence plays a significant role in influencing the mobile
services in banks. Managers should focus on relative
advantage of the usefulness and ease of use, in order to
develop the mobile commerce implementation. Banks should
introduce m-commerce service to revise their customer service
need. In addition, banks should focus on service model and its
applications. Managers should be demanding and challenging,
so long as they are consistent in their treatment to usefulness
of mobile commerce services. Finally, bank managers should
make continuous evaluation and honest dialogue about the
ease of use of mobile commerce services.
However, regardless of their limitations, it is useful to
continue analysing the new types of e-services through the
technology acceptance model. By doing so, first it is expected
that deeper insights would be gained into the banking industry
itself and its evolution over time. Secondly, this study hopes to
develop the evaluation framework that was used, which can be
transferred to other types of industries.
ACKNOWLEDGEMENT
The author is grateful to the Applied Science Private
University, Amman, Jordan, for the financial support granted
to cover the publication fee of this research article.
REFERENCES
[1] Alawneh, A., and Hattab, E. (2008b). E-business Value Creation
Jordanian Banking Services Industry: An Empirical Analysis of Key
Factors. International Arab Conference on e- technology. Arab Open
University, Amman-Jordan. October 15-10, 2010.
[2] Deng, H., Li, W. and Agrawal, D.P. (2002a) Routing security in Ad
Hoc networks, IEEE Communications Magazine, Special Topics on
Security in Telecommunication Networks, October 2002, Vol. 40, No.
10, pp.7075.
[3] Fisher, R.J. and Price, L.L. (1992) An investigation into the social
context of early adoption behavior, Journal of Consumer Research, Vol.
19, pp.477486.
[4] Kim, H., Lee, I. and Kim, J. (2008) Maintaining continuers vs.
converting discontinuers: relative importance of post-adoption factors
for mobile data services, Int. J. Mobile Communications, Vol. 6, No. 1,
pp.108132.
[5] Lin, Y-M. and Shih, D-H. (2008) Deconstructing mobile commerce
service with continuance intention, Int. J. Mobile Communications, Vol.
6, No. 1, pp.6787.
[6] Liu, C-C. (2008) Mobile phone user types by Q methodology: an
exploratory research, Int. J. Mobile Communications, Vol. 6, No. 1,
pp.1631.
[7] Miranda, F.J. Corts, R and Barriuso, C. (2006) Quantitative Evaluation
of e-Banking Web Sites: an Empirical Study of Spanish Banks The
Electronic Journal Information Systems Evaluation Volume 9 Issue 2, pp
73 - 82, available online at www.ejise.com
[8] Novak, T. P., and Hoffman, D. (1997) Modeling the Structure of Flow
Experience Among Web Users, Vanderbilt University.
[9] Orr, B. (2004) E-Banking job one: Give customers a good ride. ABA
Banking Journal, Vol. 96, Iss. 5, pp 56-57.
[10] pearlson , C, Saunders ,S (2009) strategic management information
systems , Forth edition , Johan Wiley & sons, INC.
[11] Scornavacca, E., Prasad, M. and Lehmann, H. (2006) Exploring the
organizational impact and perceived benefits of wireless Personal Digital
Assistants in restaurants, Int. J. Mobile Communications, Vol. 4,
pp.558567.
[12] Selz, D. and Schubert, P. (1997): Web assessment: A model for the
evaluation and the assessment of successful electronic commerce
applications. Electronic Markets, vol 7, n 3, pp.46-48.
[13] Stuart, J.B. and Sid, L.H. (2003) Rising sun: iMode and the wireless
internet, Communications of the ACM, Vol. 46, pp.7884.
[14] [Venkatesh, V., Ramesh, V. and Massey, A.P. (2003) Understanding
usability in mobile commerce, Communications of the ACM, Vol. 46,
pp.5356.
[15] venkatesh,,V., Morreis,MG., Davis, F.D. (2003) User acceptance of
information technology :Toward a unifeied view. MIS Quarterly, 27(3),
435- 478.
Vol. 3, No.8, 2012

136 | P a g e
[16] zMiranda, F.J. Corts, R and Barriuso, C. (2006) Quantitative
Evaluation of e-Banking Web Sites: an Empirical Study of Spanish
Banks The Electronic Journal Information Systems Evaluation Volume
9 Issue 2, pp 73 - 82, available online at www.ejise.
[17] Watson, R.T., Pitt, L.F., Berthon, P. and Zinkhan, G.M. (2002) U-
commerce: expanding the universe of marketing, Journal of the
Academy of Marketing Sciences, Vol. 30, pp.333347.
[18] Weber, M., Denk, M., Oberecker, K., Strauss, C. and Stummer, C.
(2008) Panel surveys go mobile, Int. J. Mobile Communications, Vol.
6, No. 1, pp.88107.
[19] Xie, B., Kumar, A. and Agrawal, D.P. (2008) enabling multi-service on
3G and beyond: challenges and future directions, IEEE Wireless
Communication Magazine.
[20] Xie, B., Kumar, A., Zhao, D., Reddy, R. and He, B. (2010) On secure
communication in integrated heterogeneous wireless networks, Int. J.
Information Technology, Communications and Convergence, Vol. 1, No.
1, pp.4-23.
[21] Wells, N., and J. Wolfers, Finance with a personalized touch,
Communication of the ACM, Vol. 43, No. 8: 30-34, 2000.
[22] Sekaran ,U. (1992 ). Research Methods for Business: A skill Building
Approach, Second Edtion, alynow university press, USA p (285-301).
[23] Malhotra, K.and Briks D., (2000). Marketing Research: An Applied
Approach, European Edition , prentice Hall.
Vol. 3, No.8, 2012

137 | P a g e
Clone Detection Using DIFF Algorithm For Aspect
Mining

Rowyda Mohammed Abd El-
Aziz
Department of Computer Science
Faculty of Computers and
Information
Helwan University
Cairo, Egypt

Amal Elsayed Aboutabl
Information
Helwan University
Cairo, Egypt

Mostafa-Sami Mostafa
Information
Helwan University
Cairo, Egypt

Abstract Aspect mining is a reverse engineering process that
aims at mining legacy systems to discover crosscutting concerns
to be refactored into aspects. This process improves system
reusability and maintainability. But, locating crosscutting
concerns in legacy systems manually is very difficult and causes
many errors. So, there is a need for automated techniques that
can discover crosscutting concerns in source code. Aspect mining
approaches are automated techniques that vary according to the
type of crosscutting concerns symptoms they search for. Code
duplication is one of such symptoms which risks software
maintenance and evolution. So, many code clone detection
techniques have been proposed to find this duplicated code in
legacy systems. In this paper, we present a clone detection
technique to extract exact clones from object-oriented source
code using Differential File Comparison Algorithm (DIFF) to
improve system reusability and maintainability which is a major
objective of aspect mining.
Keywords- aspect mining; reverse engineering; clone detection;
DIFF algorithm.
I. INTRODUCTION
In software engineering, it is essential to manage the
complexity and evolution of software systems. Hence,
decomposing large software systems into smaller units is
required. The result of this decomposition is separation of
concerns that leads to facilitating parallel work, team
specialization, quality assurance and work planning [1].
However, there are some functionalities that cannot be
assigned to a single unit because the code implementing them
is scattered over many units and tangled with other units. Such
functionalities are called crosscutting concerns [2]. The
existence of these crosscutting concerns leads to reducing
maintainability, evolution and reliability of software systems.
Aspect Oriented Software Development (AOSD) is a new
programming paradigm that solves the problem of crosscutting
concerns existence in legacy systems. Aspect oriented
programming modularizes such crosscutting concerns in new
units called aspects and introduces ways for weaving aspect
code with the system code at the appropriate places [3]. The
success of aspect oriented programming directs software
engineers to a new research area called aspect mining. Aspect
mining is a specialized reverse engineering process which aims
at discovering crosscutting concerns automatically in existing
systems. This process improves system maintainability and
evolution and reduces system complexity. It also enables
migration from object-oriented to aspect-oriented systems in an
efficient way [4][5][6]. Aspect mining approaches vary
according to the type of crosscutting concerns symptoms they
search for. Code duplication is one of the main symptoms of
crosscutting concerns. It is considered a major problem for
large industrial software systems because it increases their
complexity and maintenance cost. So, many clone detection
techniques are used to find this duplicated code in legacy
systems and will be discussed in details in section 2. In this
paper, we present a clone detection technique to extract exact
clones from object-oriented source code using Differential File
Comparison Algorithm (DIFF).
The basic idea is to find different lines of code between two
source code files using Diff Algorithm. As a consequence, the
remaining lines of code in both files are identical and
considered clones. Clones can then be extracted from files.
Finding clones in source code as a symptom of crosscutting
concerns helps in improving system reusability and
maintainability which is the aim of aspect mining. In section 2,
previous work on clone detection techniques is presented. In
section 3, we describe the basic idea of the used technique to
detect clones in source code. In section 4, experimental work
and results are discussed. Finally, conclusion and future work
are presented in section 5.
II. PREVIOUS WORK
Previous studies report that about 5% to 20% of software
systems contain code duplication which is a consequence of
copying existing code fragments and then reusing them by
pasting with or without minor modifications instead of
rewriting similar code from scratch [7]. Therefore, it is
considered a common activity in software development.
Developers perform this activity to reduce programming time
and effort. However, this activity results into software systems
which are difficult to maintain. The reason is that if a bug is
detected in a code fragment, other similar code fragments have
to be checked for the same bug. Consequently, there is a need
Vol. 3, No.8, 2012

138 | P a g e
for automated techniques that can find duplicated code
fragments in source code such as clone detection techniques.
A. Clone Detection Techniques
Clone detection techniques can be categorized into the
following [8]:
String-based techniques (also called text-based
techniques): at the beginning, little or no
transformation in raw source code is performed; for
example, white spaces and comments are ignored.
Then, the source code is divided into a number of
strings (lines). These strings are compared according
to the used algorithm to find duplicated ones [9].
Token-based techniques: use lexical analysis for
tokenizing source code into a stream of tokens used as
a basis for clone detection.
AST-based techniques: use parsing to represent source
code as an abstract syntax tree (AST) [10]. Then,
clone detection algorithm compares similar sub-trees
in this tree.
PDG-based techniques: use Program Dependence
Graphs (PDGs) to represent source code [11]. PDGs
describe the semantic nature of source code in high
abstraction such as control and data flow of the
program.
Metrics-based techniques: hashing algorithms are
used in such techniques [12]. A number of metrics are
calculated for each code fragment in source code.
Then, code fragments are compared to find similar
ones.
B. Clone Terminology
When two code fragments are identical or similar, they are
called clones. There are four types of clones: Type I, Type II,
Type III and Type IV. Each of these four types of clones
belongs to one of two classes according to the type of similarity
it represents: textual similarity or functional similarity. In this
context, clones of Type I, Type II and Type III are categorized
under textual similarity and Type IV is categorized under
functional similarity [13].
Type I: is called exact clones where a copied code
fragment is identical to the original code fragment
except for some possible variations in whitespaces
and comments.
Type II: a copied code fragment is identical to the
original code fragment except for some possible
variations about user-defined identifiers (name of
variables, constants, methods, classes and so on),
types, layout and comments.
Type III: a copied code fragment is modified by
changing the structure of the original code fragment,
e.g. adding or removing some statements.
Type IV: in this type, clones have semantic similarity
between code fragments. Clones, according to this
type, are not necessarily copied from the original code
because sometimes, they have the same logic and are
similar in their functionalities but developed by
different developers.
III. PROPOSED TECHNIQUE
In this paper, a clone detection technique is presented using
Differential File Comparison Algorithm (DIFF) [14] to detect
exact clones in source code files. Our clone detection technique
passes through three stages:
Source code normalization: this stage acts as a
preprocessing stage. Our clone detection technique is
text-based and, therefore, a little transformation of the
source code is needed. White spaces and comments
are removed at this stage.
Differential File Comparison: This is the main stage
of the proposed technique. The Differential File
Comparison algorithm (DIFF) [14] determines
differences of lines between two files. It solves the
problem of longest common subsequence by finding
the lines that are not changed between files. So, its
goal is to maximize the number of lines left
unchanged. An advantage of the DIFF algorithm is
that it makes efficient use of time and space. So, this
idea is used to find differences in source code lines
between two files.
Extracting exact clones: After finding differences in
source code lines between the two given source code
files using the DIFF Algorithm, the remaining lines of
code in both files are identical and considered clones.
The complement of the difference between 2 files is
determined which results in extracting exact clones
from two given source code files.
The main steps of DIFF algorithm are summarized as
follows [14]:
1. Determine equivalence classes in file 2 and associate
them with lines in file 1. Hashing is used to get better
optimization when comparing large files (thousands of
lines).
2. Find the longest common subsequence of lines.
3. Get a more convenient representation for the longest
common subsequence.
4. Weed out spurious sequences called jackpots.
IV. EXPERIMENTAL WORK AND RESULTS
Our experiment was conducted on a simple case study
consisting of two source code files implemented in the C#
programming language. These files have some differences and
similarities in their lines of code as shown in figure 1. At the
beginning, the two files are normalized by removing white
spaces and comments. Then, they are compared using DIFF
algorithm and the differences in source code lines between both
files are highlighted as shown in figure 2.
Figure1. Two source code files
Vol. 3, No.8, 2012

139 | P a g e
Figure1. Two source code files

Finally, exact cloned lines of code are detected in both files
after removing those differences from source code lines as
shown in figure 3.
Clone Detective tool [15] [16] is a Visual Studio integration
that allows analyzing C# projects for source code that is
duplicated somewhere else. Clone Detective tool is supposed to
detect type I and type II clones but it may miss some clones as
explained in [17].
Figure3. Cloned lines of code
By comparing our results with those obtained from the
Clone Detective tool for Visual Studio 2008 using the same
case study; it is found that the Clone Detective tool cannot
detect all the differences in lines of code whereas our proposed
technique can do that.
Table 1 shows the results of comparing the two tools
regarding the total number of lines in each file and the total
number of cloned lines between two files with setting clone
minimum length equals to one. It is noticed that our proposed
technique can detect all exact cloned lines which are actually
14 lines but Clone Detective tool detects 24 cloned lines and
this is not accurate because only 14 lines are exact clones and
other lines are different.

V. CONCLUSION AND FUTURE WORK
We present a simple clone detector to discover code cloning
which is a symptom of crosscutting concerns existence in
software systems. Detection of code clones decreases
maintenance cost, increases understandability of the system and
helps in obtaining better reusability and maintainability which
is the aim of aspect mining .The technique is experimented on a
simple case study (two source code files) and finally exact
clones are extracted from source code.
We consider this tool as a starting point towards a complete
clone detection system. In the future, this tool can be extended
to detect type II and type III clones and mine source code
written in other programming languages, not only C#. It can
also be extended to work on more than two source code files.
class Program {
public int sumElements(int[] arr){
int sum = 0;
for (int i = 0; i < 5; i++)
{
sum += arr[i];
}
return sum;
}
static void Main(string[] args)
{
Program p = new Program();
int result;
int avg;
int arr = new int[5];
int size = arr.Length;
Console.WriteLine("Enter
numbers:");
for (int i = 0; i < 5; i++)
arr[i]=
int.Parse(Console.ReadLine());
// sum of array elements
result = p.sumElements(arr);
// average of array elements
avg = result / size;
Console.WriteLine("Addition is:"
+ result);
Console.WriteLine("Average is:"
+ avg);
}}
class Prog {
public float sumElement(float[] arr) {
int sum = 1;
for (int i = 0; i < 5; i++)
{
sum += arr[i];
}
return sum;
}
static void Main(string[] args)
{
Prog p = new Prog();
float result;
float avg;
float arr = new float[5];
int size = arr.Length;
Console.WriteLine("Enter numbers:");
for (int j = 0; j < 5; j++)
arr[j] = int.Parse(Console.ReadLine());
// sum of array elements
result = p.sumElements(arr);
// average of array elements
avrg = result / size;
Console.WriteLine("Addition is:" +
result);
Console.WriteLine("Average is:" +
avg);
}}
Comparison
Total number
of lines
Total number
of cloned lines
Proposed
Technique
Source 26 14
Destination 26 14
Clone
Detective
Source 26 24
Destination 26 24
Figure2. Difference between lines of code
Table1.Comparison of results obtained by the proposed technique and the
Clone Detective tool

Vol. 3, No.8, 2012

140 | P a g e
REFERENCES
[1] Arie van Deursen, Marius Marin and Leon Moonen, Aspect Mining and
Refactoring, In Proceedings of the First International Workshop on
REFactoring: Achievements, Challenges, Effects (REFACE03), 2003.
[2] Bounour Nora and Ghoul Said, A model-driven Approach to Aspect
Mining, Information Technology Journal ,vol.5, 2006 , pp. 573-576.
[3] M.Marin, A.vanDeursen and L.Moonen ,Identifying Crosscutting
Concerns Using Fan-In Analysis,ACM Transactions on Software
Engineering and Methodology, Vol. 17, December 2007.
[4] Bounour Nora, Ghoul Said and Atil Fadila, A Comparative
Classification of Aspect Mining Approaches, Journal of Computer
Science,vol. 2 , pp. 322-325, 2006.
[5] Chanchal Kumar Roy, Mohammad Gias Uddin, Banani Royand Thomas
R. Dean,Evaluating Aspect Mining Techniques: A Case Study, 15th
IEEE International Conference on Program Comprehension (ICPC'07),
2007.
[6] Andy Kellens, Kim Mens, and Paolo Tonella, A Survey of Automated
Code-Level Aspect Mining Techniques,In Transactions on Aspect
Oriented Software Development, Vol. 4 (LNCS 4640), pp. 145-164,
2007.
[7] Chanchal Kumar Roy and James R. Cordy, A Survey on Software
Clone Detection Research, Technical Report No.2007-541, School of
Computing,Queen's University, KingstonOntario, Canada, September
2007.
[8] Magiel Bruntink, Aspect Mining using Clone Class Metrics, In
Proceedings of the 1st Workshop on Aspect Reverse Engineering, 2004.
[9] Kunal Pandove,Three Stage Transformation for Software Clone
Detection, Master Thesis,Computer Science and Engineering
Department, Thapar Institute of Engineering and Technology, Deemed
University,May 2005.
[10] Ira D. Baxter, Andrew Yahin,Leonardo Moura, Marcelo SantAnna and
Lorraine Bier,Clone Detection Using Abstract Syntax Trees, In
Proceedings of the 14th International Conference on Software
Maintenance (ICSM'98), pp. 368-377, Bethesda, Maryland, November
1998.
[11] Jens Krinke, Identifying Similar Code with Program Dependence
Graphs, In Proceedings of the 8th Working Conference on Reverse
Engineering (WCRE'01), pp. 301-309,Stuttgart, Germany, October
2001.
[12] Jean Mayrand, Claude Leblanc and Ettore M. Merlo, Experiment on the
Automatic Detectionof Function Clones in a Software System Using
Metrics, In Proceedings of the International Conference on Software
Maintenance (ICSM '96),1996.
[13] Yogita Sharma Hybrid Technique for Object Oriented Software Clone
Detection, Master Thesis,Computer Science and Engineering
Department,Thapar University, June 2011.
[14] J.W.Hunt and M.D.McIlroy, An Algorithm for Differential File
Comparison, Bell Laboratories, Murray Hill, New Jersey, 1976.
[15] http://clonedetectivevs.codeplex.com, last accessed Augest 2012.
[16] Elmar Juergens, Florian Deissenboeck and Benjamin Hummel,
CloneDetectiveA Workbench for Clone Detection Research, In
Proccedings of the 30th International Conference on Software
Engineering (ICSE), 2009.
[17] Chanchal K. Roy, James R. Cordy and Rainer Koschke,Comparison
and Evaluation of Code Clone Detection Techniques and Tools: A
Qualitative Approach, Science of Computer Programming Journal,
February 2009.
AUTHORS PROFILE
Rowyda Mohammed Abd El-Aziz is currently a
Software Developer at the Ministry of Planning, Cairo,
Egypt. She worked as Teaching Assistant in Modern
Sciences and Arts University in Egypt for four years. She
is a Masters Student at the Computer Science Department,
Faculty of Computers and Information, Helwan
University, Cairo, Egypt. Her current research interests
include software engineering and Human Computer
Interaction.
Amal Elsayed Aboutabl is currently an Assistant
Professor at the Computer Science Department, Faculty
of Computers and Information, Helwan University,
Cairo, Egypt. She received her B.Sc. in Computer
Science from the American University in Cairo and both
of her M.Sc. and Ph.D. in Computer Science from Cairo
University. She worked for IBM and ICL in Egypt for
seven years. She was also a Fulbright Scholar at the
Department of Computer Science, University of Virginia, USA. Her current
research interests include parallel computing, image processing and software
engineering.
Mostafa-Sami M. Mostafa is currently a Professor
of computer science, Faculty of Computers and
Information, Helwan University, Cairo, Egypt. He
worked as an Ex-Dean of faculty of Computers and
Information Technology, MUST, Cairo. He worked also
as an Ex-Dean of student affairs and Ex-Head of
Computer Science Department, faculty of Computers and
Information, Helwan University, Cairo, Egypt. He is a
Computer Engineer graduated 1967, MTC, Cairo, Egypt.
He received his MSC 1977 and his PhD 1980 from University of Paul Sabatier,
Toulouse, France. His research activities are in Software Engineering and
Computer Networking. He is awarded supervising more than 80 Masters of Sc.
and 18 PhDs in system modeling and design, software testing, middleware
system development, real-time systems, computer graphics and animation,
virtual reality, network security, wireless sensor networks and biomedical
engineering.
Vol. 3, No. 8, 2012

141 | P a g e
On the Projection Matrices Influence in the
Classification of Compressed Sensed ECG Signals

Monica Fira, Liviu Goras
Institute of Computer Science
Romanian Academy
Iasi, Romania
Liviu Goras, Nicolae Cleju, Constantin Barabasa
Faculty of Electronics, Telecommunications and
Information Technology
Gheorghe Asachi Technical University of Iasi
Iasi, Romania

Abstract In this paper the classification results of compressed
sensed ECG signals based on various types of projection matrices
is investigated. The compressed signals are classified using the
KNN (K-Nearest Neighbour) algorithm. A comparative analysis
is made with respect to the projection matrices used, as well as of
the results obtained in the case of the original (uncompressed)
signals for various compression ratios. For Bernoulli projection
matrices it has been observed that the classification results for
compressed cardiac cycles are comparable to those obtained for
uncompressed cardiac cycles. Thus, for normal uncompressed
cardiac cycles a classification ratio of 91.33% was obtained, while
for the signals compressed with a Bernoulli matrix, up to a
compression ratio of 15:1 classification rates of approximately
93% were obtained. Significant improvements of classification in
the compressed space take place up to a compression ratio of
30:1.
Keywords- ECG; compressed sensing; projection matrix;
classification; KNN.
I. INTRODUCTION
In the last decade, a new concept regarding the acquisition,
analysis, synthesis and reconstruction of signals was
introduced. Known under several equivalent names:
compressed/compressive sampling or sensing
(acquisition/detection by compression), it speculates the
sparsity of various classes of signals with respect to certain
basis or dictionaries. In the following we refer to a signal f
(including a biomedical one) which is a member of a class F c
R
N
of ND discrete signal, in particular 1D temporal or 2D
spatial signals (images). We ask the question of correlating the
properties of the class F to the minimum number of
measurements necessary for coding the signal f with a
Euclidean metric recovery error, c, imposed, respectively ||f-
f
c
||
l2
sc. The compressed sensing concept relies on an important
result obtained by Candes and Tao [1-4] namely that if the
signals of the class F admit representations through a small
number of components in an adequately selected base, i.e. they
are sparse in that basis, it is possible to reconstruct them with
a very good precision from a small number of random
measurements by solving a simple problem of linear
programming. Specifically, it is shown that if the the n-th
component f(n) of a signal in a given base, whose values in
descending order satisfy the relation |f|(n) Rn
-1/p
with R,p > 0
(which represents a constraint on the descending speed of the
components) and K measurements (projections) of the form
yk = <X
k
,f >, k=1,,K,
are performed, where X
k
are N-dimensional Gaussian
independent vectors with normal standard distribution, then
any signal that meets the mentioned constraint for a given p
can be reconstructed with a very high probability in the form
of a f
#
signal defined as a solution of minimum norm l
1
of the
system yk = <Xk,f
#
> with the relationship
||f-f#||l2 C
p
R(K/logN)
-r

where
r = 1/p-1/2.
The result is optimal in the sense that it is generally
impossible to obtain a better precision out of K measurements
regardless of the mode in which these measurements are
performed.
Reformulating the main problem, the situation can be
regarded as the one of recovering a signal feR
N
using a
minimum number of measurements, i.e. of linear functionals
associated to the signal, so that the Euclidean distance l
2

between the initial and the reconstructed signal to be lower
than an imposed value c.
II. METHODOLOGY AND OBJECTIVE
Assuming the existence of a dictionary D of elements
1 } { = k
L
k
d with L>N, each column of the dictionary is a
normalized vector (
1 ,
2
>= =<
k k k
d d d
) belonging to C
N
that
will be called atom. The dictionary contains L vectors and can
be viewed as a matrix of size NxL. An example is the Coifman
dictionary which contains L=NlogN elements consisting of
attenuated harmonic waveforms of various durations and
localizations. Other types of dictionaries are those proposed by
Ron and Shen [5] or the combined ridglet/wavelet systems
proposed by Starck, Candes and Donoho [6].
For a given sparse signal
N
C S e the determination of the
vector of coefficients with the highest number of null
elements belonging to C
L
so that S D = is envisaged.
Vol. 3, No. 8, 2012

142 | P a g e
Formally the problem consists of solving the optimization
problem:
D S P = subject to min ) (
0
0

where the norm l
0
is the number of non-zero elements in .
Unfortunately the problem is rarely easy to solve. Since in
general L>>N the solution is not unique. Determining the
solution of the problem (P
0
) requires enumerating all subsets
of the dictionary and finding the smallest subset which can be
able to represent the signal.
A remarkable result [2] is that for a large number of
dictionaries, the determination of sparse solutions can be
achieved based on the convex optimization, respectively by
solving the problem
D S P = subject to min ) (
1
1

Intuitively, using the norm l
1
can be regarded as a
convexification of the problem (P
0
). The convex optimization
problems are well studied and there are numerous algorithms
and software; as already mentioned, the problem (P
1
) is a
linear programming problem and can be solved by interior
point type methods even for large values of N and L. The
possibility of solving a problem P0 by solving problem (P
1
)
may seem surprising. However, there are results which ensure
in a rigorous manner the fact that, if there is a highly sparse
solution for the problem (P
0
) then it is identical to the solution
of the problem (P
1
). Conversely, if the solution of the problem
(P
1
) is sparse enough, i.e., if the sparsity degree is below a
certain threshold, then it is ensured the fact that this is also the
solution for the problem (P
0
).
In order to obtain the representation of the signals in
overcomplete dictionaries several methods have been
proposed in the past few years, such as the method of
frames, matching pursuit, basis pursuit (BP), as well as
the method of best orthogonal basis [2].
A possibility of improving the results of the reconstruction
when using the concept of compressed sensing is to use
specific dictionaries, constructed according to the nature,
particularities, statistics or the type of the compressed signal.
Thus, there are algorithms [7] which on reconstruction will use
a certain dictionary selected from a series of several available
dictionaries, namely, the dedicated dictionary constructed for
that particular class of signals. These types of reconstruction
algorithms have the advantage of a good reconstruction, but
they require additional information related to the initial signal,
based on which it will be decided on the dictionary used on
reconstruction. A solution to this problem would be the correct
classification of the original signal or of the compressed
signal. For biomedical signals this classification of the signal
involves placing the signal into one of several predefined
pathological classes for which there exist specific dictionaries.
In practical applications, this classification of the original
signal is not possible or it requires an additional effort.
Therefore, the ideal solution (which does not require an extra
effort in the compression stage) is to classify the compressed
signal during the reconstruction stage. In other words, for the
classification of the compressed signal [10], the problem of
classification is moved from the compression stage into the
reconstruction stage [8].
In this paper we investigate the possibility of classification
of the ECG signals after their compression based on the
concept of compressive sensing. In order to obtain good
results both from the classification point of view and from the
point of view of the reconstruction, we will segment the ECG
signal into cardiac cycles which will be further compressed. In
other words, ECG segments will be used (cardiac cycles) and
the ECG signal will be reconstructed by concatenating these
cardiac segments (cycles). According to the algorithm
described in [9] the segmentation of the ECG signal into
cardiac cycles is achieved based on the R waves detection.
Thus, one cardiac cycle is represented by the ECG signal
between the middle of a RR segment and the middle of the
next RR segment, where the RR segment means the ECG
waveform between two successive R waves. Figure 1
represents the segmentation of the ECG signal. After the
segmentation of the ECG signal there is a centering of the R
wave which is made by resampling on 150 samples on both
sides of the R wave. In this way all cardiac cycles will have
size 301 and the R wave will be positioned on the sample 151
[9].

Figure 1. Segmented ECG signal [9]
In order to compress the signals obtained this way, based
on the concepts of compressed sensing, a KxN projection
matrix of measurements has been used. The compression ratio
depends on the value of K. Due to the fact that the original
ECG segments have the size of 301 (because there was a
resampling of the cardiac cycles and all cycles have been
resampled on 301 samples), the projection matrix will have
one of the dimensions 301, N = 301, and the other dimension
of the matrix, K, will represent the number of measurements.
Thus, if the projection matrix has the size 20x301, it means
that for the compression of any cardiac cycle of size 301 only
20 measurements will be taken, resulting a compressed
version of any cardiac cycle of size 20, which means a
compression ratio of 15:1.
For the classification of the compressed cardiac cycles we
used the KNN classifier with an Euclidean distance type, and
Vol. 3, No. 8, 2012

143 | P a g e
the decision of belonging to a certain class was based on the
nearest neighbor.
A data set of 5601 compressed cardiac cycles, 701 cardiac
cycles from each of all the 8 classes (normal and 7
pathological classes) was constructed.
In order to train the KNN classifier we used 1500 cardiac
cycles, the testing being made on the rest of the data from the
database.
We tested several types of projection matrices (Gaussian
random, Fourier, random with elements of -1, 0, 1, etc).
Together with the type of matrices, the number of
measurements was varied from 2 to 60 (equivalent to
compression ratio between 150:1 and 5:1). Thus, using
different types of matrices, an analysis of the classification of
the compressed cardiac cycles for various compression ratios
was performed.
The following types of matrices were used:
- Random projection matrix (marked on graphs with
random): all entries of the K N projection matrix are
independent standard normal distributed random
variables.
- Matrices with zeros and ones, with a predefined
number of ones (3, 5, 7, 10, 50 or 150) randomly
distributed across each measurement (marked on
graphs with V1_3, V1_5, V1_7, V1_10 , V1_50 or
V1_150)
- Matrices with zeros and ones, with a predefined
number of ones (3, 5, 7, 10, 50 or 150) randomly
distributed across each of the N matrix columns
(marked on graphs with V1m_3, V1m_5, V1m_10 or
V1m_15)
- Random projection matrices with values of -1, 0 and 1
uniformly distributed (marked on graphs with
V_1_0_-1 (1/3 1/3 1/3)) i.e. Bernoulli matrix with
constant distribution
- Random projection matrix with values of -1, 0 and 1,
and unequal distribution (marked on graphs with
V_1_0_-1 (1/4 1/2 1/4)) i.e. Bernoulli matrix
- Matrices with 1 and -1, with a predefined (5, 50 or
150) number of 1s randomly distributed across each
measurement (note on graphs with V-1_5, V-1_50 or
V-1_150)
- Random Fourier matrix: The signal is a discrete
function f on Z/NZ, and the measurements are the
Fourier coefficients at a randomly selected set of
frequencies of size K (K < N).
- Random projection matrix with 0 and 1 (marked on
graphs with V_0_1_random): all entries of the N K
projection matrix are independent standard normally
distributed random variables.
III. EXPERIMENTAL RESULTS AND DISCUSSIONS
A number of 24 ECG annotated recordings from the MIT-
BIH Arrhythmia database have been used to test the
possibility of the classification of compressed patterns [30].
The ECG signals were initially digitized through sampling at
360 samples per second, quantized and encoded with 11 bits
and then resampled as described above.
Based on the database annotations, eight major classes
have been identified, namely a class of normal cardiac beats
and seven classes of pathological beats: atrial premature beat,
left bundle branch block beat, right bundle branch block beat,
premature ventricular contraction, fusion of ventricular and
normal beat, paced beat, fusion of paced and normal beat.
For the resampled cardiac cycles, but without compression,
using for training 1500 cycles and using the KNN algorithm,
we found a classification ratio of 91.33%.
In Figure 2 the classification curves for various projection
matrices are represented. Very good results have been
obtained for the Bernoulli matrix, namely for projection
matrices with values of -1, 0 and 1, in equal proportion
(,,) or variable proportions (, , ). Also, very good
results were obtained for the projection matrix containing only
the elements of 0 and -1, in equal proportions ( and ),
which, in fact, is a custom Bernoulli matrix.
From the point of view of the results, the second best
projection matrix is random with independent standard normal
distributed random variables entries.
The weakest results are obtained with the matrix
containing values of 0 and -1, with a number of 5 non-zero
elements. The difference between the results obtained with
this matrix and the next matrix from the classification point of
view are high, namely from 50% in case of a compression of
30:1 obtained with the matrix V-1_5, to approximately 70%
for compression of 30:1 with the Fourier matrix.
In Figure 3 the results for three compression ratios, 20:1,
30:1 and 60:1 are presented.
It is also observed that for a compression ratio lower than
20:1 the results of the classification do not improve
significantly, i.e. one observes a stabilization of the
classification ratio. Also, between the compression of 20:1 and
30:1 the improvement of the classification ratio is small,
therefore choosing the classification ratio will be based on the
sparsity of the signal, which will implicitly influence the
reconstruction errors also.
Another aspect to be mentioned, and which is especially
important for hardware implementations of compressed
sensing devices, is that in the case of projection matrices
which contain only elements of -1, 0 and 1 there is the
advantage of reducing the number of calculations required for
compression. If in the case of random matrices used for
compression a significant number of multiplications is
necessary, for matrices with elements -1, 0 and 1 (Bernoulli
matrices) we need only a small number of additions.

Vol. 3, No. 8, 2012

144 | P a g e

Figure 2. The compression ratio vs. classification% for various projection matrices

Figure 3. The compression ratio of 20:1, 30:1 and 60:1 vs. classification% for various projection matrices
IV. CONCLUSIONS
This paper presents a comparative analysis of the
classification results for compressively sensed cardiac cycles,
using different project matrices and a variable number of
measurements.
The classification of cardiac cycles is made using the KNN
algorithm and the construction of the projection matrices is
varied, including random matrices with real numbers,
Bernoulli matrices, random matrices with elements of -1, 0
and 1 with different probabilities, random matrices with values
of 0 and 1 and normal distribution, etc.
For Bernoulli projection matrices it has been observed that
the classification results for compressed cardiac cycles are
comparable to those obtained for uncompressed cardiac

0
10
20
30
40
50
60
70
80
90
100
1
5
0
.
5
0
5
0
.
1
7
3
0
.
1
0
2
1
.
5
0
1
6
.
7
2
1
3
.
6
8
1
1
.
5
8
1
0
.
0
3
8
.
8
5
7
.
9
2
7
.
1
7
6
.
5
4
6
.
0
2
5
.
5
7
5
.
1
9
compression ratio
C
l
a
s
s
i
f
i
c
a
t
i
o
n

%
V_-1_150
V_1_0_-1 (1/4 1/2 1/4)
V_1_0_-1 (1/3 1/3 1/3)
random
V1_5
V1_3
V1_7
V1_10
V1m_3
V1m_5
V1m_10
V1_50
V1m_15
V_-1_50
V1_150
V_0_1_random
f ourier
V_-1_5
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
C
l
a
s
s
i
f
i
c
a
t
i
o
n

%

compression ratio of 20:1
Vol. 3, No. 8, 2012

145 | P a g e
cycles. Thus, for normal uncompressed cardiac cycles a
classification ratio of 91.33% was obtained, while for the
signals compressed with a Bernoulli matrix, up to a
compression ratio of 15:1 classification rates of approximately
93% were obtained.
Significant improvements of classification in the
compressed space take place up to a compression ratio of 30:1.
ACKNOWLEDGMENT
This work has been supported by CNCSIS UEFISCSU,
project PNII RU - PD 347/2010 (M. Fira)
This paper was realized with the support of EURODOC
Doctoral Scholarships for research performance at European
level project, financed by the European Social Found and
Romanian Government (N. Cleju, C. Barabasa).
REFERENCES
[1] D. Donoho, Compressed sensing, IEEE Transactions on Information,
Theory, vol. 52, no. 4, pp. 12891306, Apr. 2006.
[2] S.S. Chen , D.L. Donoho , M.A. Saunders , "Atomic Decomposition by
Basis Pursuit", SIAM Journal on Scientific Computing, Vol. 43, No. 1,
2005
[3] J. Haupt, R. Nowak, Signal reconstruction from noisy random
projections, IEEE Trans. on Information Theory, 52(9), pp. 4036-4048,
September 2006)
[4] E. Cands, M. Wakin, An introduction to compressive sampling, IEEE
Signal Processing Magazine, 25(2), pp.21 - 30, March 2008)
[5] A. Ron, Z. Shen, Affine systems in L2(Rd): the analysis of the analysis
operator, J. Funct. Anal. 148 (1997) 408447.
[6] J.-L. Starck, M. Elad, D.L. Donoho, Redundant multiscale transforms
and their application for morphological component analysis, Adv.
Imag. Elect. Phys. 132 (2004).
[7] M. Fira, L. Goras, C. Barabasa, N. Cleju, On ECG Compressed Sensing
using Specific Overcomplete Dictionaries, Advances in Electrical and
Computer Engineering, Vol. 10, Nr. 4, 2010, pp. 23- 28
[8] C. Monica Fira, L. Goras, C. Barabasa, N. Cleju, ECG compressed
sensing based on classification in compressed space and specified
dictionaries, EUSIPCO 2011 (The 2011 European Signal Processing
Conference), 29 august 2 septembrie 2011, Barcelona, Spania
[9] M. Fira, L. Goras, "An ECG Signals Compression Method and Its
Validation Using NNs", IEEE Transactions on Biomedical Engineering,
Vol. 55, No. 4, 1319 1326, April 2008
[10] Yi-Haur Shiau, Chaur-Chin Chen, A Sparse Representation Method
with Maximum Probability of Partial Ranking for Face Recognition,
International Journal of Advanced Research in Artificial Intelligence,
Vol. 1, No. 1, 2012

Vol.3, No. 8, 2012
146 | P a g e
An Approach of Improving Students Academic
Performance by using K-means clustering algorithm
and Decision tree
Md. Hedayetul Islam Shovon
Department of Computer Science and Engineering
Rajshahi University of Engineering & Technology
Rajshahi-6204, Bangladesh

Mahfuza Haque

Department of Computer Science and Engineering
Rajshahi University of Engineering & Technology
Rajshahi-6204, Bangladesh
AbstractImproving students academic performance is not an
easy task for the academic community of higher learning. The
academic performance of engineering and science students
during their first year at university is a turning point in their
educational path and usually encroaches on their General Point
Average (GPA) in a decisive manner. The students evaluation
factors like class quizzes mid and final exam assignment lab -
work are studied. It is recommended that all these correlated
information should be conveyed to the class teacher before the
conduction of final exam. This study will help the teachers to
reduce the drop out ratio to a significant level and improve the
performance of students. In this paper, we present a hybrid
procedure based on Decision Tree of Data mining method and
Data Clustering that enables academicians to predict students
GPA and based on that instructor can take necessary step to
improve student academic performance
Keywords- Database; Data clustering; Data mining; classification;
prediction; Assessments; Decision tree; academic performance.
I. INTRODUCTION
Graded Point Average (GPA) is a commonly used
indicator of academic performance. Many universities set a
minimum GPA that should be maintained. Therefore, GPA
still remains the most common factor used by the academic
planners to evaluate progression in an academic environment.
Many factors could act as barriers to student attaining and
maintaining a high GPA that reflects their overall academic
performance, during their tenure in university. These factors
could be targeted by the faculty members in developing
strategies to improve student learning and improve their
academic performance by way of monitoring the progression
of their performance [1]. With the help of clustering algorithm
and decision tree of data mining technique it is possible to
discover the key characteristics for future prediction. Data
clustering is a process of extracting previously unknown,
valid, positional useful and hidden patterns from large data
sets. The amount of data stored in educational databases is
increasing rapidly. Clustering technique is most widely used
technique for future prediction. The main goal of clustering is
to partition students into homogeneous groups according to
their characteristics and abilities (Kifaya, 2009). These
applications can help both instructor and student to enhance
the education quality. This study makes use of cluster analysis
to segment students into groups according to their
characteristics [2]. Decision tree analysis is a popular data
mining technique that can be used to explain different
variables like attendance ratio and grade ratio. Clustering is
one of the basic techniques often used in analyzing data sets
[3]. This study makes use of cluster analysis to segment
students in to groups according to their characteristics and use
decision tree for making meaningful decision for the students.
2. Methodology
A. Data Clustering
Data Clustering is unsupervised and statistical data analysis
technique. It is used to classify the same data into a
homogeneous group. It is used to operate on a large data-set to
discover hidden pattern and relationship helps to make
decision quickly and efficiently. In a word, Cluster analysis is
used to segment a large set of data into subsets called clusters.
Each cluster is a collection of data objects that are similar to
one another are placed within the same cluster but are
dissimilar to objects in other clusters.

a. Implementation Of K-Means Clustering Algorithm
K-Means is one of the simplest unsupervised learning
algorithms used for clustering. K-means partitions n
observations in to k clusters in which each observation
belongs to the cluster with the nearest mean. This algorithm
aims at minimizing an objective function, in this case a
squared error function. The algorithm and flow-chart of K-
means clustering is given below

Fig.1 Traditional K-Means Algorithm [4].
From the algorithm it is easily seen that, initially we have
only the raw data. So, it is clustered around a single point. If
the cluster number K is fixed then we need to cluster around
that point. If the cluster is not fixed then it is continued until
the centered is not changed. Initially the students are all in a
same group. But when K-means clustering is applied on it then
Vol.3, No. 8, 2012
147 | P a g e
it clusters the students into three major categories, one is
good, one is medium, and the other is low standard student.
The flow chart of the k-means algorithm that means how
the k-means work out is given below.
NO

YES

Fig.2 Flow-Chart Of K-Means Clustering.

B. Data mining
Data mining, also popularly known as Knowledge
Discovery in Database, refers to extracting or mining"
knowledge from large amounts of data. Data mining
techniques are used to operate on large volumes of data to
discover hidden patterns and relationships helpful in decision
making. The sequences of steps identified in extracting
knowledge from data are shown in fig.3

Knowledge
Fig.3 Steps Of Knowledge Extraction

I. Decision Tree
Decision tree induction can be integrated with data
warehousing techniques for data mining. A decision tree is a
predictive node ling technique used in classification,
clustering, and prediction tasks. A decision tree is a tree where
the root and each internal node are labeled with a question.
The arcs emanating from each node represent each possible
answer to the associated question. Each leaf node represents a
prediction of a solution to the problem under consideration.
The basic algorithm for decision tree induction is a greedy
algorithm that constructs decision trees in a top-down
recursive divide-and-conquer manner. Decision Tree
Algorithm: generate a decision tree from the given training
data.
1 Create a node N
2 If samples are all of the same class, C then
3 Return N as a leaf node labeled with the class C;
4 If attribute-list is empty then
5 Return N as a leaf node labeled with the most
common class in samples.
6 Select test-attribute, the attribute among attribute-list
with the highest information gain;
7 Label node N with test-attribute;
8 For each known value ai of test-attribute.
9 Grow a branch from node N for the condition test
attribute = ai ;
10 Let Si be the set of samples for which test-attribute =
ai;
11 If Si is empty then
12 Attach a leaf labeled with the most common class in
samples;
13 Else attach the node returned by generate-decision-
tree (Si,attribute-list-attribute);
Each internal node tests an attribute, each branch
corresponds to attribute value, and each leaf node assigns a
classification.

Fig.4 Decision Tree.
Selection
Pre-processing
Transformation
Data mining
Evaluation
Raw data

University
Depatment
Courses
Grade(PSG,CT,quiz,lab,attendance)
High Medium Low
Good Averag
e
Low
START
No of
cluster
Grouping
based on
minimum
distance
Recalculate
centroid
Distance
objects to
centroid

Centroid
Are the
centroid
fixed?
END
Vol.3, No. 8, 2012
148 | P a g e
From 50 training sample here only 20 samples are shown.
Table 1: Training Sample.

II. RESULT AND DISCUSSION
From the training data GPA and the attendance ration of
the student is given below
Graph.1: Shows the relationship between GPA and Attendance ratio.

If we apply K-means clustering algorithm on the training
data then we can group the students in three classes High
Medium and Low according to their new grade. New
grade is calculated from the previous semester grade that
means external assessment and internal assessment. The table
and corresponding graph is given below.
Table 2. Percentage of students according to GPA.
Class
GPA No of student Percentage
1
2
3
4
5
2.00-2.20
2.20-3.00
3.00-3.32
3.32-3.56
3.56-4.00
5
10
17
15
13
8.33
16.67
28.33
25
21.67

Here, I cluster student among their GPA that means, from
GPA 2.00- 2.20 we have 8.33% student. From 2.20-3.00
student percentage is 16.67%.
From 3.00-3.32 we have 28.33%. From 3.32-3.56
percentage is 25% .The percentage is 21.67% between GPA
3.56-4.00.
The graphical representation of GPA and the percentage of
students among the student are given below.
Graph 2: Number and percentage of students regarding to GPA

Table 2

Class GPA No of
student
Percentage
High
Medium
Low
>=3.50
2.20<=GPA <3.5
<=2.20
28
27
5
46.67
45
8.33

After clustering the student, we group the student into
three categories. One is High, second is Medium, and the last
one is Low.
Graphical representation of these three categories is given
below.
Graph 3: Shows the percentage of students getting high, medium and low
GPA

If we apply data mining technique decision tree then it will
help us to make correct decision about the student which is
need to take by the instructor. The decision step is given
below.
0 10 20 30
2.00-2.20
2.20-3.00
3.00-3.32
3.32-3.56
3.56-4.00
Percentage of student
G
P
A

R
a
n
g
e

No of student
Percentage
Percentage's of student
GPA
1
2
3
Vol.3, No. 8, 2012
149 | P a g e
Table 3: Decision based on the student categories:

Step No Grade Effort
S-01 A+ He/She is a good
student. Need not to
take special care.

S-02 A,A- Is not so good. Need
to take care of CT &
Quiz.

S-03 B+,B Is a medium student.
Should take care of
CT,quiz and lab
performance also.

S-04 Below B grade Is a lower standard
student. Need lot of
practice of his/her
lesson and also take
care of all the courses
ct,lab,quiz ,attendance
carefully.

III. CONCLUSION AND FUTURE WORK
In this study we make use of data mining process in
students database using k-means clustering algorithm and
decision tree technique to predict students learning activities.
We hope that the information generated after the
implementation of data mining and data clustering technique
may be helpful for instructor as well as for students. This work
may improve students performance; reduce failing ratio by
taking appropriate steps at right time to improve the quality of
education. For future work, we hope to refine our technique in
order to get more valuable and accurate outputs, useful for
instructors to improve the students learning outcomes.
REFERENCES
[1] Oyelade, Oladipupo & Obagbuwa, Application of K-means clustering
algorithm for prediction of students academic performance.,
IJCSIS2010,vol.7,No.1, pp-292.
[2] Research, ISSN 1450-216X Vol.43 No.1 (Shaeela Ayesha, Tasleem
Mustafa, Ahsan Raza Sattar & M.Inayat Khan, Data Mining Model for
Higher Education System ,European Journal of Scientific 2010), pp.24.
[3] Dr.Vuda Sreenivasaro & Capt.Genetu Yohannes ,Improving academic
performance of student of defence university based on data warehousing
and data mining, Global Journal of computer science and technology,
v.12,Issue2,Version.1,pp-29.
[4] Research, ISSN 1450-216X Vol.43 No.1 (Shaeela Ayesha, Tasleem
Mustafa, Ahsan Raza Sattar & M.Inayat Khan, Data Mining Model for
Higher Education System ,European Journal of Scientific 2010), pp.27.
[5] Dr.Vuda Sreenivasaro & Capt.Genetu Yohannes ,Improving academic
performance of student of defence university based on data warehousing
and data mining, Global Journal of computer science and technology,
v.12,Issue2,Version.1,pp-33.
[6] Kifaya(2009) Mining student evaluation using associative classification
and clustering.
[7] ZhaoHui. Maclennan.J, (2005). Data Mining with SQL Server 2005
Wihely Publishing, Inc.
AUTHORS PROFILE
Md. Hedayetul Islam Shovon was born in 1986 in Bangladesh. He
received his B.Sc Engineering degree from Rajshahi University of
Engineering and Technology (RUET) in 2009. He joined in the department of
Computer Science and Engineering of RUET in October 2009. Since then he
involved in different research oriented activities. His research interests include
Image Processing, Data Clustering, Computer Vision and Optimization etc.
Mahfuza Haque was born in Bangladesh. He received his B.Sc
Engineering degree from the department of Computer Science and
Engineering of Rajshahi University of Engineering and Technology (RUET)
in 2012. His research interest includes Pattern recognition, Data clustering,
Data mining.
Vol. 3, No. 8, 2012
150 | P a g e
Prevention and Detection of Financial Statement
Fraud An Implementation of Data Mining
Framework

Rajan Gupta

Research Scholar, Dept. of Computer
Sc. & Applications, MaharshiDayanand University,
Rohtak (Haryana) India.

Nasib Singh Gill
Professor,Dept. of Computer
Sc. & Applications, MaharshiDayanand University,
Rohtak (Haryana), India.
Abstract Every day, news of financial statement fraud is
adversely affecting the economy worldwide. Considering the
influence of the loss incurred due to fraud, effective measures
and methods should be employed for prevention and detection of
financial statement fraud. Data mining methods could possibly
assist auditors in prevention and detection of fraud because data
mining can use past cases of fraud to build models to identify and
detect the risk of fraud and can design new techniques for
preventing fraudulent financial reporting. In this study we
implement a data mining methodology for preventing fraudulent
financial reporting at the first place and for detection if fraud has
been perpetrated. The association rules generated in this study
are going to be of great importance for both researchers and
practitioners in preventing fraudulent financial reporting.
Decision rules produced in this research complements the
prevention mechanism by detecting financial statement fraud.
Keywords- Data mining framework; Rule engine; Rule monitor.
I. INTRODUCTION
Financial statement fraud is a deliberate misstatement of
material facts by the management in the books of accounts of
a company with the aim of deceiving investors and creditors.
This illegitimate task performed by management has a severe
impact on the economy throughout the world because it
significantly dampens the confidence of investors.
The magnitude of this problem can be evaluated by the
fact that a number of Chinese companies listed on US stock
exchanges have of faced accusations accounting fraud, and in
June 2011, the U.S. Securities and Exchange Commission
warned investors against investing with Chinese firms listing
via reverse mergers. While over 20 US listed Chinese
companies have been de-listed or halted in 2011, a number of
others have been hit by the resignation of their auditors [1].
Association of certified fraud examiners (ACFE) in its
report to the nation on occupational fraud and abuse (2012) [2]
suggests that the typical organization loses 5% of its revenue
to fraud each year. The median loss caused by occupational
fraud cases was $140,000.
This study by ACFE reveals that perpetrators with higher
levels of authority tend to cause much larger losses. The
median loss among frauds committed by owner / executives
was $573,000, the median loss caused by managers was
$180,000 and the median loss caused by employees was
$60,000. The report by the ACFE also measured the common
methods of detecting fraud and found that in more than 43 %
cases tips and complaints have been the most effective means
of detecting frauds.
Prevention and detection of financial statement fraud has
become a major concern for almost all organisations globally.
Though, it is a fact that prevention of financial statement fraud
is the best way to reduce it, but detection of fraudulent
financial reporting is critical in case of failure of prevention
mechanism.
The aim of this paper is to provide a methodology for
prevention and detection of financial statement fraud and to
present the empirical results by implementing the framework.
In this research, we test the applicability of data mining
framework for prevention and detection of financial statement
fraud. As per the recommendations of the framework we apply
descriptive data mining for prevention and predictive data
mining techniques for detection of financial statement fraud.
This paper is organized as follows. Section 2 summarizes
the contribution in the field of prevention and detection of
financial statement fraud. Section 3 implements the data
mining framework for detection of fraud if prevention
techniques have failed followed by conclusion (Section 4).
II. RELATED WORK
Cost of financial statement fraud is very high both in terms
of finance as well as the goodwill of the organization and
related country. In order to curb the chances of fraud and to
detect the fraudulent financial reporting, number of
researchers had used various techniques from the field of
statics, artificial intelligence and data mining.
For instance, Spathis et al [3] compared multi-criteria
decision aids with statistical techniques such as logit and
discriminant analysis in detecting fraudulent financial
statements. Neural Network based support systems was
proposed by Koskivaara [4] in 2004. He demonstrated neural
network as a possible tool for use in auditing and found that
the main application areas of NN were detection of material
errors, and management fraud.
Vol. 3, No. 8, 2012
151 | P a g e
A decision tree was constructed by Koh and Low [5] in
order to predict the hidden problems in financial statements by
examining the following six variables: quick assets to current
liabilities, market value of equity to total assets, total liabilities
to total assets, interest payments to earnings before interest
and tax, net income to total assets, and retained earnings to
total assets. Kirkos et al [6], carry out an in-depth analysis of
publicly available data of 76 Greek manufacturing firms for
detecting fraudulent financial statements by using three Data
Mining classification methods namely Decision Trees, Neural
Networks and Bayesian Belief Networks. They investigated
the usefulness of these techniques in identification of FFS.
In 2007, a genetic algorithm approach to detecting
financial statement fraud was presented by Hoogs et al [7]. An
innovative fraud detection mechanism is developed by Huang
et al. [8] on the basis of Zipfs Law. This technique reduces
the burden of auditors in reviewing the overwhelming
volumes of datasets and assists them in identification of any
potential fraud records. A novel financial kernel using support
vector machines for detection of management fraud was
developed by Cecchini et al [9].
In 2008, the effectiveness of CART on identification and
detection of financial statement fraud was examined by
Belinna et al [10] and found CART as a very effective
technique in distinguishing fraudulent financial statement
from non-fraudulent. Juszczak et al. [11] apply many different
classification techniques in a supervised two-class setting and
a semi-supervised one-class setting in order to compare the
performances of these techniques and settings.
Further, Zhou & Kapoor [12] in 2011 applied four data
mining techniques namely regression, decision trees, neural
network and Bayesian networks in order to examine the
effectiveness and limitations of these techniques in detection
of financial statement fraud. They explore a self adaptive
framework based on a response surface model with domain
knowledge to detect financial statement fraud.
Ravisankar et al [13] applied six data mining techniques
namely Multilayer Feed Forward Neural Network (MLFF),
Support Vector Machines (SVM), Genetic Programming (GP),
Group Method of Data Handling (GMDH), Logistic
Regression (LR), and Probabilistic Neural Network (PNN) to
identify companies that resort to financial statement fraud on a
data set obtained from 202 Chinese companies. They found
Probabilistic neural network as the best techniques without
feature selection. Genetic Programming and PNN
outperformed others with feature selection and with
marginally equal accuracies.
Recently, Johan Perols [14] compares the performance of
six popular statistical and machine learning models in
detecting financial statement fraud. The results show,
somewhat surprisingly, that logistic regression and support
vector machines perform well relative to an artificial neural
network in detection and identification of financial statement
fraud.
The review of the existing literature reveals that the
research conducted till date is solely in the field of detection
and identification of financial statement fraud and a very little
or no work has been done in the field of prevention of
fraudulent financial reporting.
Therefore, in the present research we implement a data
mining framework for prevention along with detection of
financial statement fraud.
The major objective of this research is to test the
applicability of predictive and descriptive data mining
techniques for detection and prevention of fraud respectively
by implementing a data mining framework. In order to feel the
sense of fraud, we implement association rule mining and to
detect fraudulent financial reporting we apply three
classification techniques namely decision trees, nave
Bayesian classifier and Genetic programming.
III. THE METHODOLOGY: APPLICABILITY & ITS
IMPLEMENTATION
The methodology applied in this paper is a data mining
framework of Gupta & Gill (2012) [15]. The framework is
presented as Fig 1.
The first step of the framework is feature selection. We
selected 62 financial ratios / variables as features to be used as
input vector in further analysis.
These features represent behavioural characteristics along
with measures of liquidity, safety, profitability and efficiency
of the organisations under consideration. Table 1 present the
list of 62 features.
Figure 1: A data mining framework for prevention and detection of financial
statement fraud.

Feature Selection
Data Collection
Data Preprocessing
Performance Evaluation
Predictive Data Mining Fraud Detection
Descriptive Data Mining (Association Rules)
Opportunity
Fraud
Risk
Rationalization Motivation
Rule Engine
Rule Monitor
Fraud Prevention
Data Mining

FRAUD ALERT
CAPITAL STRUCTURE
CONDITION
S
CHOICES
Vol. 3, No. 8, 2012
152 | P a g e
Table 1: Features For Prevention & Detection Of Financial
Statement Fraud

S.No. Financial Items / Ratios
1 Debt
2. Total assets
3 Gross profit
4 Net profit
5 Primary business income
6 Cash and deposits
7 Accounts receivable
8 Inventory/Primary business income
9 Inventory/Total assets
10 Gross profit/Total assets
11 Net profit/Total assets
12 Current assets/Total assets
13 Net profit/Primary business income
14 Accounts receivable/Primary business income
15 Primary business income/Total assets
16 Current assets/Current liabilities
17 Primary business income/Fixed assets
18 Cash/Total assets
19 Inventory/Current liabilities
20 Total debt/Total equity
21 Long term debt/Total assets
22 Net profit/Gross profit
23 Total debt/Total assets
24 Total assets/Capital and reserves
25 Long term debt/Total capital and reserves
26 Fixed assets/Total assets
27 Deposits and cash/Current assets
28 Capitals and reserves/Total debt
29 Accounts receivable/Total assets
30 Gross profit/Primary business profit
31 Undistributed profit/Net profit
32 Primary business profit/Primary business profit of last year
33 Primary business income/Last year's primary business income
34 Account receivable /Accounts receivable of last year
35 Total assets/Total assets of last year
36 Debit / Equity
37 Accounts Receivable / Sales
38 Inventory / Sales
39 Sales Gross Margin
40 Working Capital / Total Assets
41 Net Profit / Sales
42 Sales / Total Assets
43 Net income / Fixed Assets
44 Quick assets / Current Liabilities
45 Revenue /Total Assets
46 Current Liabilities / Revenue
47 Total Liability / Revenue
48 Sales Growth Ratio
49 EBIT
50 Z Score
51 Retained Earnings / Total Assets
52 EBIT / Total Assets
53 Total Liabilities / Total assets
54 Cash return on assets
55 Interest expense / Total Liabilities
56 EBIT / sales
57 Age of the company (Number of years since first filing available
from provider)
58 Change in cash scaled to total assets
59 Change in current assets scaled by current liabilities
60 Change in total liabilities scaled by total assets
61 Size of company on the basis of assets
62 Size of company on the basis of revenue
During the second step of Data Collection, all the
financial ratios of Table 1 have been collected from financial
statements namely balance sheet, income statement and cash
flow statement for 114 companies listed in different stock
exchanges globally. The dataset used in this study has been
collected from www.wikinvest.com. The companies accused
of fraudulent financial reporting has been identified by
analysing Accounting and Auditing Enforcement Releases
published by S.E.C. (U.S. Securities and Exchange
Commission) for the period of five years starting from 2007.
All the incidents of violation of the Foreign Corrupt Practices
Act (FCPA) have been removed from the sample, because
FCPA prohibits the practice of bribing foreign officials and
most of the AAERs issued because of FCPA do not reflect
which financial statement viz. balance sheet or income
statement, is affected.
We identified 29 organisations with charges of issuing
fraudulent financial statements and hence termed as fraudulent
in this study. 85 organisations out of total of 114 have been
marked as non fraudulent since no indication or proof of
falsifying financial statement has been reported. However,
absence of any proof does not guarantee that these firms have
not falsified their financial statements or will not do the same
in future.
In order to make dataset ready for mining, data need to be
pre - processed. Data has been transformed in to an
appropriate format for mining during the step of Data
preprocessing. Dataset is cleaned further by replacing
missing values with the mean of the variable. Each of the
independent financial variables has been normalized by using
range transformation (min = 0.0, max = 1.0).
We compiled all the 62 input variables given in Table 1.In
order to reduce dimensionality of the dataset we applied one
way ANOVA. The variables with p value <=0.05 are
considered significant and informative and with high p value
are deemed to be non informative. Informative variables are
tested further using descriptive data mining methods. The
input variables which are considered significant are given in
Table 2 along with respective F- values and p values.
TABLE 2: LIST OF INFORMATIVE VARIABLES
S.No. Financial Items / Ratios F - value
P
value
1
Debt 1.345 .028
2
Inventory/Primary business income 3.031 .001
3
Inventory/Total assets 17.468 .000
4
Net profit/Total assets 3.035 .001
5
Accounts receivable/Primary business
income
6.099 .018
6
Primary business income/Total assets 3.038 .001
7
Primary business income/Fixed assets 3.055 .001
8
Cash/Total assets 2.918 .001
9
Inventory/Current liabilities 6.744 .001
10
Total debt/Total assets 2.851 .001
11
Long term debt/Total capital and
reserves
4.266 .014
12
Deposits and cash/Current assets 2.932 .001
Vol. 3, No. 8, 2012
153 | P a g e
13
Capitals and reserves/Total debt 2.213 .003
14
Gross profit/Primary business profit 3.847 .008
15
Accounts Receivable / Sales 1.702 .021
16
Working Capital / Total Assets 2.906 .001
17
Sales / Total Assets 12.818 .003
18
Net income / Fixed Assets 3.038 .001
19
Quick assets / Current Liabilities 1.839 .050
20
Revenue /Total Assets 12.818 .003
21
Capital and Reserves / Total Debt 1.130 .049
22
Retained Earnings / total assets 3.039 .001
23
EBIT 4.363 .023
24
EBIT / Total Assets 3.043 .001
25
Z score 3.054 .001
26
Total liabilities / Total Assets 3.154 .002
27
Cash flow from operations 1.720 .018
28
Cash return on assets 3.940 .002
29
Interest Expenses 1.806 .010
30
Interest exp / Total Liabilities 1.440 .042
31
Size of the company on the basis of
assets
1.179 .043
32
Change in cash scaled by total assets 2.967 .001
33
Current Liabilities of the previous
year
1.391 .028
34
Total Liabilities of the previous year 1.346 .022
35
Change in Total Liabilities scaled by
Total Assets
3.188 .001
The step of data preprocessing is followed by selection of
an appropriate data mining technique. The framework
suggests the use of descriptive data mining technique for
prevention and predictive methods for detection of financial
statement fraud. Therefore, we first apply association rule
mining for preventing fraudulent financial reporting at the first
place.

We implement association rules by using RapidMiner
version 5.2.3. All the informative variables have been
converted into nominal variables. Nominal variables further
converted into binomial variables because it is the preliminary
requirement for rule engine. In the next stage of the
framework, Rule engine generates the required association
rules.
In the process of rule generation, frequent itemsets is being
generated using FP Growth. The minimum support for FP
Growth has been set to 0.95. The frequent itemsets generated
has been used for creating the association rules. The minimum
confidence for generating rules is 0.8. Table 3 lists the
association rules generated by rule engine.
Now, the rule monitor module will monitor the financial
ratios of each organisation and compare the values of the
ratios with the values given in the association rules for
indicating the anomaly. Anomalies detected by rule monitor
are reflected as number of non fraud companies identified as
fraud in Table 3. The results generated by rule monitor are
able to raise an alarm regarding fraud.
In view of the whistle blown by rule monitor,
organisations should consider the presence or absence of
conditions which refers to certain financial pressures exhibited
by the management. Such organisations should think in terms
of providing employees the working environment that values
honesty because irresponsible and ineffective corporate
governance could increase the chances of financial statement
fraud. The absence of effective corporate governance may
provide enough opportunity to the managers / employees for
selecting an option of fraudulent financial reporting. Hence,
this unlawful practice of fraudulent financial reporting could
be prevented by checking or taking away the opportunity to
commit fraud and by avoiding the combination of opportunity,
pressure and motive in an organisation.
TABLE 3: ASSOCIATION RULE
S.N Association Rule Support Confidence Lift Conviction Number of non-fraud
companies identified
as fraud
1. Inventory / total assets > 0.033 Fraud 43% 86% 1.153 1.812 30
2. Cash / Total Assets <0.198 Fraud 42.1% 84.2% 1.129 1.611 36
3. Inventory / Current Liabilities >0.190 Fraud 43.9% 87.7% 1.176 2.071 31
4. Deposits and cash / Current Assets <0.408
Fraud
43.9% 87.7% 1.176 2.071 33
5. Sales / Total Assets >.553 Fraud 44.7% 89.5% 1.200 2.417 23
6. Revenue / Total Assets >.553 Fraud 44.7% 89.5% 1.200 2.417 25
7. Inventory / current Liabilities >0.190 &&
Deposits and cash / Current Assets < 0.408
Fraud

43.9% 87.7% 1.176 2.071 15
Vol. 3, No. 8, 2012
154 | P a g e
Once the prevention mechanism has failed to prevent fraud
then the framework suggest the usage of predictive data
mining for detection and identification of financial statement
fraud. In this study three data mining techniques namely
CART, Naive Bayesian Classifier and Genetic Programming
have been used for detection of fraudulent financial statements
and differentiating between fraud and non fraud reporting.In
order to have better reliability of the result, ten fold cross
validation has been implemented.
A decision tree (CART) has been constructed in this study
by using SIPINA Research edition software version 32 bit.
The complete dataset has been used as training data for
constructing the tree given as Figure 2. The confidence level
was set to 0.05. CART manages to classify 95 % cases. This
method well classifies 98 % non fraud cases and misclassifies
only 4 fraud cases. The percentage of classification for fraud
cases is 86 %.
The financial ratio namely Deposits and cash to current
assets has been used as the first splitter by the decision tree
constructed in this research. This ratio is an indicator for the
measurement of capability of a company in converting its non
liquid assets into cash. At second level of the tree, retained
earnings / total assets (t2) and net profit / total assets has been
used as a splitter. The ratios used by tree are given in Table 4.
TABLE 4
S. No. Financial Ratios / Items
1
Net profit/Total assets
2
Size of the company on the basis of assets (Total
Assets P)
3
Deposits and cash/Current assets
4
Capital & Reserves / Total Assets
5
Inventory / Current Liabilities
6.
Cash return on Assets
7.
Cash / Total Assets
8.
Inventory / Total Assets
9.
Retained earnings / Total Assets (t2)
We applied Nave Bayesian Classifier, the second method
of classification by using SIPINA Research edition software
version 32 bit. The method correctly classifies 88% cases.
Third method of classification, Genetic programming has
been implemented using a data mining tool Discipulus version
5.1. The process begins with division of dataset in to two
datasets namely training data and validation data. The training
data set has been used to train the sample and validation
dataset is used exclusively for the purpose of validation. In
this study, 80% of the whole dataset is designated as training
data for training the sample, whereas, rest 20% is assigned
exclusively for the purpose of validation. Since our dependent
variable (target output) is binary, we select hits then fitness
as a fitness function. Every single run of Discipulus has been
set to terminate after it has gone 50 generations with no
improvement in fitness.
Performance evaluation, the final step of the framework
is used for measuring the performance and judging the
efficacy of data mining methods. Performance of association
rules generated in this study has been measured with the help
of support, confidence, lift and conviction (Table 3). The rules
generated by rule engine have support of more than 40% and
confidence more than 80%.
Sensitivity and specificity have been used as a metrics for
performance evaluation of classification techniques used in
this research. The confusion matrix for Decision trees, Nave
Bayesian classifier and Genetic programming is given below.
TABLE: 5 (CONFUSION MATRIX FOR DECISION TREE)
Label Non Fraud Fraud
NF (Non Fraud) 83 2
F (Fraud) 4 25

TABLE: 6 (CONFUSION MATRIX FOR NAIVE BAYSIAN CLASSIFIER)
NF 79 6
F 8 21

TABLE: 7 (CONFUSION MATRIX FOR GENETIC PROGRAMMING)
NF 84 1
F 13 16
Performance matrix indicating the sensitivity (type 1 error)
and specificity (type II error) of the three methods used in this
study is given in Table 8.
Table: 8 (Performance Matrix)
S.No. Predictor Sensitivity
(%)
Specificity
(%)
1 Decision Tree 86.2 97.7%
2 Nave Bayesian
Classifier
84 92.9
3 Genetic
Programming
53 99.2

Figure: 2 Decision Tree (CART)

Vol. 3, No. 8, 2012
155 | P a g e
Decision tree (CART) classifies 25 fraud cases as fraud
from a total of 29 such cases correctly therefore, produces best
sensitivity. The following are the decision rules generated by
using decision tree (Figure 2).
1. If ((Deposits and cash / Current assets >=0.16) &&
(Retained Earnings / Total Assets> = -0.46) &&
(Inventory / Total Assets > = 0.23)) then Fraud
(Inventory / Total Assets < 0.23) && (Size of the
company on the basis of assets >=45.05)
&&(Inventory / Current Liabilities >=1.22)) then
Fraud
(Inventory / Total Assets < 0.23) && (Size of the
company on the basis of assets >=45.05) &&
(Inventory / Current Liabilities >=1.22) && (Capital
and Reserves / Total Assets > = 5855.00)) then Fraud
4. If ((Deposits and cash / Current assets < 0.16) &&
(Retained Earnings / Total Assets> = -1.88) && (Net
profit / Total Assets < 0.09)) then Fraud
(Retained Earnings / Total Assets< -0.46) && (Cash
return on Assets < 0.01)) then Fraud
Genetic programming outperforms the other two
techniques by correctly classifying 84 cases out of 85 non
fraud organisations, hence produces best specificity. Table 9
represents the input impact of various input parameters on the
model generated by Genetic Programming.

TABLE: 9 IMPACT OF INPUT VARIBALES (GENETIC PROGRAMMING)
S.No. Variable Frequency Average Impact Maximum
Impact
1 Debt 0.06 00.00000 00.00000
2 Inventory/Primary business income 0.35 22.52747 53.84615
3 Inventory/Total assets 0.35 09.70696 20.87912
4 Net profit/Total assets 0.06 02.19780 02.19780
5 Cash/Total assets 0.29 03.84615 05.49451
6 Total debt/Total assets 0.12 00.00000 00.00000
7 Fixed assets/Total assets 0.00 00.00000 00.00000
8 Deposits and cash/Current assets 0.18 06.59341 06.59341
9 Working Capital / Total Assets 0.06 00.00000 00.00000
10 Sales / Total Assets 0.00 00.00000 00.00000
11 Net income / Fixed Assets 0.41 07.69231 09.89011
12 Revenue /Total Assets 0.29 09.01099 14.28571
13 EBIT 0.06 05.49451 05.49451
14 Z score 0.06 19.78022 19.78022
15 Accounts receivable/Primary business income 0.29 00.54945 01.09890
16 Primary business income/Total assets 0.18 02.74725 03.29670
17 Primary business income/Fixed assets 0.41 03.29670 08.79121
18 Capitals and reserves/Total debt 0.00 00.00000 00.00000
19 Gross profit/Primary business profit 0.53 05.65149 09.89011
20 Accounts Receivable / Sales 0.00 00.00000 00.00000
21 Retained earnings / Total Assets 0.18 02.93040 04.39560
22 EBIT / Total Assets 0.24 03.29670 05.49451
Since Decision trees are capable of identifying type 1 error
in more than 86% and Genetic programming correctly detect
type II error for almost all the cases present in the dataset,
therefore, we arrive at a conclusion that data mining
techniques used in this study are capable enough for
identification and detection of financial statement fraud in case
of failure of prevention mechanism.
IV. CONCLUSION
Prevention along with detection of financial statement
fraud would be of great value to the organizations throughout
the world. Considering the need of such a mechanism, we
employ a data mining framework for prevention and detection
of financial statement fraud in this study. The framework used
in this research follow the conventional flow of data mining.
We identified and collected 62 features from financial
statements of 114 organizations. Then we find 35 informative
variables by using one way ANOVA.
These informative variables are being used for
implementing association rule mining for prevention and three
predictive mining techniques namely Decision Tree, Nave
Bayesian Classifier, Genetic programming for detection of
financial statement fraud. Rule Engine module of the
framework generated 7 association rules. These rules are used
by rule monitor module for raising an alarm regarding fraud
and hence preventing it at the first place.
The three data mining methods used for detection of
financial statement fraud are compared on the basis of two
important evaluation criteria namely sensitivity and
specificity. Decision tree produces best sensitivity and Genetic
programming best specificity as compared with other two
methods. These techniques will detect the fraud in case of
failure of prevention mechanism. Hence, the framework used
in this research is able to prevent fraudulent financial reporting
and detect it if management of the organization is capable of
perpetrating financial statement fraud despite the presence of
anti fraud environment.
Vol. 3, No. 8, 2012
156 | P a g e
REFERENCES
[1] Atkins Matt, Accounting Fraud in US listed Chinese companies
(September 2011). Available at: http://www.financierworldwide.com
[2] ACFE, 2012 ACFE Report to the nations on ocupational fraud and
abuse, Technical report- Global fraud survey 2012, 2012.
[3] C. Spathis, M. Doumpos, C. Zopounidis, Detecting falsified financial
statements: a comparative study using multicriteria analysis and
multivariate statistical techniques, European Accounting Review 11 (3)
(2002) 509535.
[4] E. Koskivaara, Artificial neural networks in auditing: state of the art,
The ICFAI Journal of Audit Practice 1 (4) (2004) 1233.
[5] H.C. Koh, C.K. Low, Going concern prediction using data mining
techniques, Managerial Auditing Journal 19 (3) (2004) 462476.
[6] Efstathios Kirkos, Charalambos Spathis &Yannis Manolopoulos
(2007), Data mining techniques for the detection of fraudulent financial
statements. Expert Systems with Applications 32 (23) (2007) 9951003
[7] Hoogs Bethany, Thomas Kiehl, Christina Lacomb and DenizSenturk
(2007). A Genetic Algorithm Approach to Detecting Temporal
Patterns Indicative Of Financial Statement Fraud, Intelligent systems in
accounting finance and management 2007; 15: 41 56, John Wiley &
Sons, USA, available at: www.interscience.wiley.com
[8] S.-M. Huang, D.C. Yen, L.-W.Yang, J.-S.Hua, An investigation of
Zipf's Law for fraud detection.Decision Support Systems 46 (1) (2008)
7083.
[9] M. Cecchini, H. Aytug, G.J. Koehler, and P. Pathak. Detecting
Management Fraud in Public
Companies.http://warrington.ufl.edu/isom/docs/papers/DetectingMana
gementFraudInPublicCompanies.pdf
[10] BelinnaBai, Jerome yen, Xiaoguang Yang, False Financial Statements:
Characteristics of china listed companies and CART Detection
Approach, International Journal of Information Technology and
Decision Making , Vol. 7, No. 2(2008), 339 359
[11] Juszczak, P., Adams, N.M., Hand, D.J., Whitrow, C., & Weston, D.J.
(2008). Off-the-peg and bespoke classifiers for fraud detection,
Computational Statistics and Data Analysis, vol. 52 (9): 4521-4532
[12] Wei Zhou, G. Kappor, Detecting evolutionary financial statement
fraud. Decision Support Systems 50 (2011) 570 575.
[13] P.Ravisankar, V. Ravi, G.RaghavaRao, I., Bose, Detection of financial
statement fraud and feature selection using data mining techniques,
Decision Support Systems, 50(2011) 491 500
[14] Johan Perols, Financial Statement Fraud Detection: An Analysis of
Statistical and Machine Learning Algorithms, A Journal of Practice &
Theory 30 (2), 19 (2011), pp. 19-50
[15] Gupta Rajan& Gill Nasib S. A data mining framework for prevention
and detection of financial statement fraud, International Journal of
Computer Application, 50(8): 7 - 14, July 2012. Published by
Foundation of computer science, New York, U.S.A.
AUTHORS PROFILE
Rajan Gupta obtained masters degree in computer
application from Department of Computer Science &
Application, Guru Jambheshwar University, Hisar,
Haryana, India and Master Degree of Philosophy in
Computer Science from Madurai Kamraj University,
Madurai, India. He is currently pursuing Doctorate degree
in Computer Science from Department of Computer
Science & Application, Mahrshi Dayanand University, Rohtak, Haryana,
India.

Dr Nasib S. Gill obtained Doctorate degree in computer
science and Post-doctoral research in Computer Science from
Brunel Univerrsity, U.K. He is currently working as
Professor and Head in the Department of Computer Science
and Application, Mahrshi Dayanand University, Rohtak,
Haryana, India. He is having more than 22 years of teaching
and 20 years of research experience. His interest areas
include software metrics, component based metrics, testing, reusability, Data
Mining and Data warehousing, NLP, AOSD, Information and Network
Security.

Vol. 3, No. 8, 2012
157 | P a g e
Review of Remote Terminal Unit (RTU) and
Gateways for Digital Oilfield delpoyments

Francis Enejo Idachaba
Department of Electrical and Information Engineering
Covenant University Ota.
Ogun state
Nigeria

Ayobami Ogunrinde
SPDC Nigeria

Abstract The increasing decline in easy oil has led to an
increasing need for the optimization of oil and gas processes.
Digital oilfields utilize remote operations to achieve these
optimization goals and the remote telemetry unit and gateways
are very critical in the realization of this objective. This paper
presents a review of the RTUs and gateways utilized in digital
oilfield architectures. It presents a review of the architecture,
their functionality and selection criteria. It also provides a
comparison of the specifications of some popular RTUs.
KeywordsDigital Oilfield; Gateway; HMI; i-fields; RTU;
Smartfields.
I. INTRODUCTION
The advent of Digital Oilfields, Smartfields or i-fields has
led to an increase in the need to monitor, control and automate
various systems at remote oil and gas production sites to
increase production, reduce overall production cost, and
reduce employee exposure. Control systems such as SCADA
(Supervisory Control and Data Acquisition), or DCS
(Distributed Control System) using Remote Telemetry Units
and Gateways are deployed to achieve these control functions.
The RTUs and gateways comprises of various components
such as
HMI (Human Machine Interface)
RTU (Remote Terminal Unit): This collects the site
data and sends it to a station via a communications
system.
Supervisory systems/ Master station: this collects the
information from the process and control the process.
This is usually a computer.
Communication system that provides a means by
which all components communicate securely without
loss of data and information.
Digital oilfield installations require bidirectional
transmission of data from the sensors located in the field and
control signals from the control room or the office domain to
these sensors and devices located in the field. The data from
the sensors are transmitted at defined intervals or by exception
while the control algorithms used for the field devices can
either be the on/off control or a variable control. This paper
Ease of Use
II. RTU
Remote Terminal Unit (RTU) is a microprocessor-based
device connected to sensors, transmitters or process equipment
for the purpose of remote telemetry and control.
RTUs find applications in oil and gas remote
instrumentation monitoring, networks of remote pump
stations, Environmental monitoring systems, Air traffic
equipment etc. [1]
RTUs with the aid of appropriate sensors, monitors
production processes at remote site and transmits all data to a
central station where it is collated and monitored. An RTU can
be interfaced using serial ports (RS232, RS482, and RS422) or
Ethernet to communicate with the central stations. They also
support various protocol standards such as Modbus, IEC
60870, DNP3 making it possible to interface with 3rd party
software.
III. RTU ARCHITECTURE
The RTU architecture comprises of a CPU, volatile
memory and nonvolatile memory for processing and storing
programs and data. It communicates with other devices via
either serial ports or an onboard modem with I/O interfaces. It
has a power supply with a backup battery, surge protection
against spikes, real-time clock and a watchdog timer to ensure
that it restarts when operating in the sleep mode.[2]
Figure 1 shows the block diagram of a typical RTU
configuration. A typical RTU hardware module includes a
control processor and associated memory, analog inputs,
analog outputs, counter inputs, digital inputs, digital outputs,
communication interfaces and power supply [3]
A. Central Processing Unit (Cpu)
Current RTU designs utilize a 16bit or 32 bits
microprocessor with a total memory capacity of 256kbytes
expandable to 4 Mbytes. It also has two or three
communication ports (RS232, RS422 and RS485) or Ethernet
link. This system is controlled by a firmware and a real-time
clock with full calendar is used for accurate time stamping of
events. A watchdog timer provides a check that the RTU
program is executing regularly. The RTU program regularly
resets the watchdog timer and if this is not done within a
Vol. 3, No. 8, 2012
158 | P a g e
certain time-out period the watchdog timer flags an error condition and can sometimes reset the CPU. [3]

Figure. 1 RTU Hardware Structure [3]

B. Analog Input Modules
An analog input signal is generally a voltage or current that
varies over a defined value range, in direct proportion to a
physical process measurement. 4-20 milliamp signals are most
commonly used to represent physical measurements like
pressure, flow and temperature. [4]. Five main components
that makes up the analog input module are as follows:
Input multiplexer: This samples several analog inputs
in turn and switches each to the output in sequence.
The output goes to the analog digital converter.
Input signal amplifier: This amplifies the low-level
voltages to match the input range of the boards A/D
converter
Sample and hold circuit
A/D converter: This measures the input analog voltage
and output a digital code corresponding to the input
voltages.
Bus interface and board timing system.
Typical analog input modules features include:
8, 16, or 32 analog inputs
Resolution of 8 to 12 bits
Range of 4-20 mA
Input resistance typically 240kohms to 1 Mohms
Conversion rates typically 10 microseconds to 30
milliseconds.
C. Analog Output Module
Analog Output modules function is to convert a digital
value supplied by the CPU to an analog value by means of a
digital to analog converter. This analog representation can be
used for variable control of actuators.
Analog output modules features are as follow:
8, 16 or 32 analog outputs
Resolution of 8 or 12 bits
Conversion rate from 10 seconds to 30 milliseconds
Outputs ranging from 4-20 mA/0 to 10 volts
D. Digital Input Modules
These are used to indicate status and alarm signals [5]
E. Digital Output Modules
These modules are used to drive an output voltage at each
of the appropriate output channels with three approaches
possible.
Triac Switching: Triacs are used to achieve AC power
control. The Triac responds primarily as a switch, the
AC energy source for a portion of each alternation can
be controlled [6]
Read Relay Switching
Vol. 3, No. 8, 2012
159 | P a g e
TTL voltage outputs
F. Power Supply Module
RTUs need a continuous power supply to function, but
there are situations where RTUs are located at quite a distance
from an electric power supply. In these cases, RTUs are
equipped with alternate power source and battery backup
facilities in case of power losses.
Solar panels are commonly used to power low-powered
RTUs, due to the general availability of sunlight. Thermo
electric generators can also be used to supply power to the
RTUs where gas is easily available like in pipelines. [4]
G. Communication interfaces
Modern RTU are designed to be flexible enough to handle
multiple communication media such as
RS 232/RS 442/RS 485
Ethernet
Dial up telephone lines/dedicated landlines
Mircrowave/MUX
Satellite
X.25 packet protocols
Radio via trunked/VHF/UHF/900 Mhz
IV. GATEWAY
A gateway is a device with dedicated hardware and
software that translates between two different protocols,
making communication possible between networks of
different architectures and protocols. The job of a gateway is
much more complex than that of a network router or switch
due to this conversion functions. Gateways in digital oil fields
collate data from the RTUs in the field and remote sites and
integrate these data into the Companies IT network
Gateways are necessary for communication between
terminals connected to heterogeneous networks using different
protocols and having different network characteristics. They
provide the connectivity between systems at remote locations
with the target system to enable different network applications
A gateway can function as a protocol gateway which
converts between protocols; an application gateway which
accepts inputs in one format translates it and then sends it, or a
security gateway which basically acts as a firewall securing
and filtering packets. [7]
Components of a gateway
Microprocessor
Motherboard
RAM
Flash
Interface boards as I/O ports
V. SELECTING THE RIGHT EQUIPEMENT
The selection of RTUs and Gateways are based on the
specifications of the implementation in terms of data type,
capacity and transmission rate. These parameters include [8]
1) Capacity: The RTU must be able to support the data
transmission frequency and the data rates. The RTU must also
have sufficient spare IOs to allow for expansion.
2) Environment Factor: The RTU must be able to withstand
the environmental factors and be designed to the required
ingress protection ratings and installed in the appropriate
hazardous area classification.
3) Control: RTUs are also used for applications requiring
different control schemes such as on/off and variable control.
Control relays serve as the control element and are connected
to the RTU and activated remotely to achieve the desired
control. During RTU selection, the type of control required
must be defined to ensure that the selected RTU supports the
control system.
4) Connectivity: The network connectivity requirements
and data formats must defined before the RTU and Gateways
are selected. The data exchange format must also be defined
before the selection of the RTU and gateways are finalized.
5) Upgradeability: The ease of firmware upgrade is also
another key parameter as it will be desired for the RTU and
gateway to be upgradeable over the air without the
requirement of un-installation and office based upgrade.
6) Transmission range: The RTU frequency and range are
to be confirmed to be suitable for the application and this is
usually confirmed by network plan. The equipment are
expected to operate within the approved frequency band and
with the required licenses.
7) Power: The power supply requirements should also be
within the required capacity and quality and these power
supply systems must be able to with stand the environmental
factors, cost and weather conditions and also be resistant to
vandalization. Options include solar panels, batteries or other
power source. Also putting into consideration cost as well.
VI. RTU MANUFACTURERS
There are various manufacturers of Remote Terminal Unit
for various functions and industries. A list of some RTU
manufactures and their products are presented in the Table 1
1) Vmonitor iX-S8 Wireless RTU: An intelligent remote
terminal unit and wireless technology with low power
consumption to provide a reliable and cost effective means of
remotely monitor and automate your applications in the oil
and gas fields. [9]
2) ControlWave Micro Hybrid RTU/PLC: A highly
programmable controller that combine the unique capabilities
of a programmable logic controller (PLC) and a remote
terminal unit (RTU) into a single hybrid controller. [10]
3) Zetron Model 1732 RTU: A cost-effective solution for
applications that need to connect widely distributed remote
sites to a central control program using radio, telephone and
wire line communications media. [11]
4) Brodersen RTU32: Brodersen RTU32 RTU, PLC and
controller series based on a 32-bit platform provides RTU/PLC
with power and leading edge functionality. [12]
Vol. 3, No. 8, 2012
160 | P a g e
5) Siemens Vicos RTU: A telecontrol with standard
SIMATIC S7 Programmable Logic controller [13]
6) Oleumtech Wireless RTU/Modbus Gateway: Wio
wireless RTU products are low cost remote terminal units that
combine traditional remote IO functionality of a standard [14]
REFERENCES
[1] Borin Manifacturing. Borin Manifacturing, Inc. Borin [online]
http://www.borin.com/remote-terminal-unit/. [Accessed 24
th
July 2012]
[2] TheWater Environment Federation, Automation of waste water
treatment facilities. WEF Press McGraw-Hill Companies, 2007.
[3] Clarke, G.R., Reynders,D and Edwin W, SCADA protocols: DNP3,
60870.5 and related systems. Elsevier, 2004. pp.15 - 25
[4] Shaw, W.T. Cybersecurity for SCADA systems. Pennwell books,
2006. pp.25 - 32
[5] Sumathi S. and Surekha..P. LabVIEW based advances instrumentation
sustems. Springer, 2007. pp 242 - 246
[6] Patrick, D.R. and Fardo, S.W. Electricity and electronics fundamentals.:
Fiarmont Press, Inc, 2008.
[7] Zhang, P and William A. Advnced Industrial Control Technology.
2010. pp 378 379
[8] DPS Telecom. [online] http://www.dpstele.com. [Accessed 24
th
July
2012]
[9] VMonitor. VMonitor.[online] http://www.vmonitor.com. [Accessed 24
th

July 2012]
[10] Emerson Process Mangement. [Online]
http://www.documentation.emersonprocess.com [Accessed 24
th
July
2012]
[11] Zetron. [online] http://www.zetron.com. [Accessed 24
th
July 2012]
[12] Borderson. [online] http://brodersensystems.com. [Accessed 24
th
July
2012]
[13] Siemens. [Online] https://concert.siemens.com. [Accessed 24
th
July
2012]
AUTHORS PROFILE
Francis Idachaba ([email protected]) a lecturer with Covenanrt
University Ota is currently with on a Fellowship with Shell Petroleum
Development Company in Nigeria.
Ayobami Ogunrinde ([email protected] ) is currently with
SPDC in Nigeria.
TABLE 1. RTU MANUFATURERS SPECIFICATION

Vmonitor iX-S8 Control Wave Micro
Hybrid
Zetron 1732
RTU
Broderson RTU 32 Siemens
Vicos RTU
Processor
8051 Micro-
Controller running at
12.58Mhz 64KB
Data Memory
32-bit ARM 9 Processor, 33 MHz
CPU Module (Low power) 150
MHz Module
PLC 32-bit 500 Mhz CPU
with 128K L2 Cache
SIMATIC S7
300 Central Units
Serial Port RS232
RS232/RS485
I2C
RS232
RS485
Ethernet
RS232 Dual Ethernet
RS232
RS232/RS422/RS485
USB

RS232
Ethernet
Comm. Protocol Vmonitor Proprietary
Protocol
MODBUS
MODBUS, Foundation Fieldbus.
HART, DFI, CIP, DNP3, Serial
ASCII
MODBUS MODBUS, DNP3 Suite,
ProfiNET Client,
PROFIBUS DP Master,
COMLI, IEC 61400-25
IEC60870-5,
SINAUT8-FW
PCM,
Profibus,
ProfiNET
Operation Modes Continuous, Stand
alone and Sleep
Mode
Run Mode, Remote Mode, Local
Mode and Sleep Mode for Low
power application
Fail-Safe Mode
Radio 900 MHz or 2.4 GHz 902-928MHz 902-928MHz
Distance Up to 5 Miles
Operating
Temperature
(-40 to 85)C
Humidity (5 90%)
(-40 to 70)C Humidity (15 95%) (0 to 50)C
Humidity (0
90%)
(-40 to 70)C (0 to 60)C
Humidity (5
95%)
Power Supply Input Power 8-
15VDC (Battery
Powered)
9 30 VDC 16VDC 2A
Max, 120 VAC
24 48VDC
115-230 VAC/DC
115-230 VAC/DC with
UPS and Battery
24VDC, 230
VAC
Integrated
Analogue/Digital
Input
(AI): 1-5 V, 4-20mA
(DI): 3-24VAC
(AI): 1-5 V, 4-20m A
(DI): 12-24VDC
(AI): +/- 10
VDC, 4-20 mA
(AI): 0-10V, 0-5V, +/-
5V, +/- 10V, 0-20mA, 4-
20mA
(DI): 10-30VDC

IJACSA - Volume 3 No. 8, August 2012

Uploaded by

Copyright:

Available Formats

IJACSA - Volume 3 No. 8, August 2012

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

IJACSA - Volume 3 No. 8, August 2012

Uploaded by

Copyright:

Available Formats

THE SCIENCE AND INFORMATION ORGANIZATION

www. t hes ai . or g | i nf o@t hes ai . or g

are independent random numbers uniquely

You might also like