A scenario based approach for dealing with

International Journal on Computational Sciences & Applications (IJCSA) Vol.4, No.2, April 2014
DOI:10.5121/ijcsa.2014.4204 31
A Scenario Based Approach For Dealing With
Challenges In A Pervasive Computing
Environment
Divyajyothi M G1
, Rachappa2
and Dr. D H Rao3
1,2
Research Scholar, Department of Computer Science, Jain University, Bangalore
3
Principal and Director, Jain College of Engineering, Belgaum
ABSTRACT
With the surge in modern research focus towards Pervasive Computing, lot of techniques and challenges
needs to be addressed so as to effectively create smart spaces and achieve miniaturization. In the process of
scaling down to compact devices, the real things to ponder upon are the Information Retrieval challenges.
In this work, we discuss the aspects of multimedia which makes information access challenging. An
Example Pattern Recognition scenario is presented and the mathematical techniques that can be used to
model uncertainty are also presented for developing a system that can sense, compute and communicate in
a way that can make human life easy with smart objects assisting from around his surroundings.
KEYWORDS
Pervasive computing, Multimedia retrieval, Modelling uncertainty, Fuzzy theory
1. INTRODUCTION
Tremendous growth and research contributions are going on towards Mark Weiser’s vision [4][5]
on developing a system that can sense, compute and communicate in a way that can make human
life easy with smart objects assisting from around his surroundings. In this process of migrating
towards a smart environment, the real challenges to ponder upon are the performance issues, data
management, software maintenance, energy efficiency, trust, security and privacy of the
computing device to be designed[1][2][3][6][7]. The Demand for pervasive multimedia services
has widely increased with Ubiquitous computing being expected in almost all areas of health care,
entertainment, digital libraries, hotels, class rooms, Smart campuses, automobiles, streets,
airports, social networks. With the amount of multimedia data we have, data analysis, indexing,
retrieval, distribution, and management becomes even more challenging subject to the fact of
adopting all these requirements that suits the human being mode of thinking, expectations and
vision [9].This is where context aware multimedia computing becomes extremely important. In
this paper, the aspects that make the multimedia retrieval challenging is discussed. An example
scenario is presented with explanation on tools that can be readily used to model uncertainty are
described.
2. MULTIMEDIA RETRIEVAL CHALLENGES
2.1. Real –Time Constraints
Multimedia involves very large amounts of data. Multimedia retrieval refers to extracting
semantic information from this large amount of data available in various forms. Thus multimedia

32
retrieval needs efficient techniques and algorithms for redundancy elimination, feature extraction
and categorization.
2.2. Bridging the Semantic Gap
There is a semantic gap between the semantics the searcher attaches to the visual or any
multimedia information and the semantics which is extracted from the digital information stored
for the multimedia objects. The semantic content needs to be structured and summarized and
therefore is a daunting task.
2.3. “Multi” in Multimedia
Analysing characteristics of all the possible myriad forms of media and retrieving it in a standard
way from these vast amounts of multimedia information for all of them is quite a daunting task.
Some forms of multimedia: Text, Sketches, Videos, Graphical Images, Speech, Sound, Movies.
2.4. Effective Multimedia Search Engines
Need an effective search mechanism while answering a query to a document database. The most
common problems, which may occur in this process, are synonymy and polysemy.
Synonymy – Search Engine detecting any subject S that may not exactly present in any article A.
Polysemy -- some words may have many meanings.
2.5. Protocols for Multimedia Networks
For Instance, in order to constantly receive unobtrusive connectivity and response from network
devices embedded in the environment, the computing speed should be invariably good and this
can be made possible if parallel computation can take place.
2.6. Multimedia languages for Multi-Channel Content
We need more possible ways to distribute our content video, audio, Speech, images, Text, for
excellent ubiquitous behaviour. Efficient scalable codecs needs to be proposed for effective
universal multimedia access.
2.7. Multimedia Services for Intelligent Pervasive Computing
It’s necessary to address the security and privacy issues involved while performing secure
transactions in a ubiquitous environment. More hybrid security measures have to be implemented
keeping in mind the performance of devices that are used in pervasive computing environments.
2.8. Scalable Algorithms / Techniques
It’s time we think about whether the existing retrieval, indexing, mining, streaming, delivery,
personalization algorithms are easily feasible with compact pervasive devices. If not new
algorithms have to be formulated based on the existing foundations we have. In such cases
quantum mechanics techniques may provide useful results.

33
3. EXAMPLE SCENARIO
3.1. Complete Pattern Recognition System
3.1.1 Sensor for Feature Extraction
Requires a sensor for Feature Extraction mechanism, requires a Classification / Description
Scheme which is based on learning strategy/paradigms.
a. Supervised Learning – based on availability of a training set (set of patterns).
b. Unsupervised Learning – based on statistical regularities of patterns Decision
Theoretic.
c.
3.1.2 Classification Scheme
Requires a Classification / Description Scheme which uses one of the following:
a. Statistical (Decision theoretic) is based on statistical characterizations of patterns which
are generated by a probabilistic system. For a Probabilistic system, naive Bayes classifier
can be very effectively used because it works on the availability of a particular training
set irrespective of other features and its results are based on maximum likelihood
property. Example for document classification: Documents can be classified based on
many ways by their content, subject, Text and Graphical portions, Images and plain text,
mathematical equations and numbers, noun sorters, mass dividers. In general, each
document classification has its own classification challenges.
b. Syntactic (Structural) – based on structural interrelationships of features, should have a
clear structure of the patterns. An appropriate grammar is the core of any type of
syntactic pattern recognition process. One must make sure that the Grammars are
established from a priori knowledge about the objects or scenes to be recognized.
Syntactic Classifiers: Representing structural information in images can be effectively
done using Fuzzy set Theory. All imprecise relationships between objects can be defined
as spatial fuzzy sets. Formal language can be readily used to represent such structures.
Fuzzy grammars can be generated for the objects to be recognized. All differences in the
structures of the classes must be encoded as different grammars. Another example can be
Diagnosis of the heart using ECG measurements.
c. Neural Classifiers - based on bionics-related concepts in recognizing patterns. Bionics
refers to the science of applying biological concepts to electronic machines. This concept
is made use by the neural approach that applies these biological concepts to machines in
order to recognize patterns. As an outcome of this effort that field of artificial neural
networks has emerged and we can see some interesting results.
Neural Classifiers: Neuro Excel Classifier is one of the most efficient, quick, powerful
and an easy-to-use neural network software tool that is most widely being used for
classifying data in Microsoft Excel. The main objective of this classifier is to aid experts
in the design process of real-world data mining and pattern recognition tasks. One of the
major benefits of this classifier is its ability to hide the underlying density, Thickness,
complexity of neural network processes by providing graphs and statistics for the user so
that the results can be easily understood. The algorithms and techniques used in Neuro
Excel Classifiers are only those which are reliable and which are proven to be efficient.
Another feature is its ability to amalgamate flawlessly with Microsoft Excel. [8]
Apply appropriate Algorithms for the pattern recognition based on the target system.

34
4. MODELLING UNCERTAINTY
Every Situation in the world around us can be represented as a mathematical model. All these
models are established using the building blocks of Set Theory. Set theory is a branch of
mathematics that deals with the properties of sets. According to classical set theory, the
membership of elements in a set is based on a bivalent condition, stating that an element may
either belong or may not belong to the set.
Generalization of the classical set theory gives us the fuzzy set theory, in which the membership
of elements in a set is described with the aid of a membership function valued in the real unit
interval [0, 1]. The method of assessing the membership of elements in classical set theory is
based on a crisp condition of whether the element belongs to or does not belong to any given set
S. Therefore the entire process is being done in binary terms.
4.1. Fuzzy Set Theory
Fuzzy sets are considered to be an extension of classical set theory. Such imprecise concepts,
classical set theory fails to handle since it uses the principle of bivalent condition. So when it
comes to expert systems and recognition system, Fuzzy set theory serves better as it has the
ability of handling effectively the most inherent and imprecise concepts. It is most widely used
mathematical method in modern research because it is much more organised and can handle
imperfect knowledge and vagueness in an intelligent manner. We should be aware that not all
concepts can be converted in the form of an equation. For example, consider the problem of
expressing the term “hotness” in the form of a mathematical equation. Since, “hotness” is not a
quantity; it cannot be expressed in the form of an equation. But still if one asks common people
they have an idea of what is "hot”, and agree that there is no sharp cut-off between "hot" and "not
hot", where something is "hot" at N degrees but "not hot" at N-1 degrees.
Thus we can define a fuzzy set on a classical set Χ is defined as follows:
In the above equation, the membership function μA(x) enumerates whether the elements x belong
to the fundamental set X or not. Based on the bivalent condition, the elements in the set can either
take the value of 1 or 0. If the element is seen mapping to the value 0, it means that the element
does not belong to the given set X. Likewise if the element is seen mapping to 1, it is fully a
member of the set. The elements that range in between are said to be the fuzzy members.
Consider a fuzzy set C, where C = {(3,0.3), (4,0.7), (5,1), (6,0.4)}. Using standard fuzzy set
theory notation, this would be computed as C = {0.3/3, 0.7/4, 1/5, 0.4/6} From the above
notation, it can be observed that any value with a membership grade of zero does not appear in
the expression of the set. Finding the membership grade of the fuzzy set C at 6 using the standard
notation is μB(6) = 0.4.

35
Figure1: Fuzzy Set and Crisp Set
4.1.1 Degrees of Truth
The real numbers in the interval [0, 1] are usually used to represent the Degrees of Truth.
If the values are the extreme points (0 and 1), then they represent absolute falsity and absolute
truth respectively whereas the values ranging in between the extreme points (0 and 1), represent
the intermediate truth degrees.
Hence, when a system of logic uses these degrees of truth it is called fuzzy logic. The logical
operations in such systems are typically defined as follows:
¬P = 1 - P
P ∨ Q = max(P, Q)
P ∧ Q = min(P, Q)
As a result, such systems employing degrees of truth allow us to assess sentences involving more
than one vague and myriad property such as warm, old, strong, fit, happy, bright, sad, cold, and
coldest and so forth. As a result we have a technique to potentially tackle vagueness. [11][12][13]
4.1.2 Logic based on Fuzzy Sets
Logic based on the concept of fuzzy sets, in that any membership that is expressed using varying
probabilities or degrees of truth with values ranging from 0 (does not occur) to 1 (definitely
occurs).
There are a lot of practical uses that Fuzzy logic offers in making a considerable impact in
engineering control systems, computing, medicine and healthcare, where pervasive computing is
allowing self-care rather than professional care [14].
It is more convenient to programming a series of logical conditions and corresponding actions
into a machine. Using fuzzy logic for the same allows the values of the propositions involved to
come straight away from the machine's sensors.
Few Examples are the usage of sensors in devices such as thermometers, motion detectors,
refrigerators, washing machines.

36
Commercial applications of Fuzzy Logic came up as early as in the 1990s. Many other products
using fuzzy logic include camcorders, microwave ovens, Toasters, Vacuum cleaners,
dishwashers. Other applications include expert systems in a pervasive environment using multiple
sensors, interactive devices, and computerized speech- and handwriting-recognition programs
with Interfaces for modes of interactions between people and pervasive computing devices.
5. SET THEORY APPLICATION AREAS
5.1. Electronic Commerce
Rough set theory is used nowadays in Electronic Commerce (EC) for data mining. Everyday large
volumes of data are collected by the EC sites. This data comprises a lot of secure, valuable
information about customers, products, transactions, communication address and so on. This large
amount of data can be better used by the site management to extract the unknown knowledge or
trends hiding in the data, and to arrange their products according to the buyers' preference and
take appropriate selling, security and authentication policies.
5.2. Pattern Recognition
Rough set theory is widely used in combination with neural network theory for pattern
recognition. Using All Set Theory and Set Pair Analysis (SPA) many new methods of pattern
recognition have been proposed. These methods have proven to be much more valid and effective
when compared to existing conventional ways of pattern recognition.
5.3. Computer Networks
Rough Set theory has also found its wide application in computer network fault diagnosis,
anomaly intrusion detection in computer networks. It has served as an excellent tool for dealing
with vagueness and imperfections.
5.3. Pervasive Computing
Lot of Service Match Making Algorithms for Pervasive Computing are based on Rough Set
Theory. Also Rough Set theory is being used for creating a user aware TV program and Settings
in a pervasive environment [10].
6. CONCLUSION
Most of the time, probability is being confused with Degrees of truth. This should not be the case.
Consider the task of flipping a coin. It would be incorrect to say that this task has a 50%-50%
chance of being or not being in F. Though flipping a coin definitely results either in a heads or
tails, giving it a 50% chance is incorrect. So this random event must be given the value 1 for its
degree of truth because irrespective of the situation, one of the sides either heads or tails
definitely appears. Degrees of truth should also never be confused with an unknown or varying
truth value.
For Example, Consider the sentence-“The month of July is usually a monsoon day in some parts
of the world”. Though the degree of truth value does not fall on the extreme points 0 or 1, it can
be still considered as a definite value. Repeated observations made on the same day do not give
us different values. Thus an alternative to probability theory can be the mathematical theory
known as possibility theory which can be used to deal with certain types of uncertainty. This

37
theory was first introduced in 1978. Basically this theory was developed as an extension to the
fuzzy logic and fuzzy set theory by Prof Zadeh. This theory makes use of the necessity and
possibility of any random event whereas the probability theory makes use of only probability to
decide on the likeliness of an event to take place.
Choosing the ELECTIC VIEW can be an alternate solution, which agrees / accepts both the
interpretations: depending on the situation, one has to select one of the two interpretations for
pragmatic, or principled, reasons.
7. ACKNOWLEDGMENTS
We would like to express our gratitude to the person who made completion of this paper possible.
We are deeply indebted to our Prof. Dr. D. H. Rao for his constant help, suggestions and
motivation to study and work. His tremendous knowledge and stimulating suggestions has helped
us a lot in completing our work effectively.
8. REFERENCES
[1] S.K. Das, K. Kant, N.Zhang. Handbook on Securing Cyber-Pysical Critical Infrastructure: Foundation
and Challenges, Morgan Kauffman, 2012.
[2] N.Roy, A.Misra, C.Julien, S.K.Das, J.Biswas. An Energy efficient Quality Adaptive Multi Modal
Sensor Framework for Context Recognition. Proceedings of IEEE International Conference on
Pervasive Computing and communications, pp.63-75, 2011.Tavel, P. 2007. Modeling and Simulation
Design. AK Peters Ltd., Natick, MA.
[3] R.Rajkumar and I.Lee. NSF workshop on cyber-Physical systems.DOI =
http://varma.ece.cmu.edu/cps/, October 2011
[4] M. Weiser. The computer for the twenty-first century. Scientific American, 265(3): 94-104, 1991.
[5] Juan Ye and Simon Dobson. “Pervasive Computing needs better situation –awareness” ,
doi:10.2417/3201201.003943
[6] G.Hayes, E.Poole, G.lachello, S.Patel, a.Grimes, G.Abowd, and K.Truong. Physical, social and
experimental knowledge of privacy and security in a pervasive computing environment. IEEE
Pervasive Computing, 6(4):56-63, 2007
[7] A.T. Campbell, s.B. eiseman, N.D. lane, E.Miluzzo, R.A.Peterson, H.Lu, X.Zheng, M.Musoles,
K.Fodor, and G.S. ahn. The rise of people-centric sensing. IEEE Internet Computing, 12(4):12-21,
2008
[8] Neural Network classifiers , DOI= http://www.ozgrid.com/Services/neuro-excel-classifier.htm
[9] Chris Wellekens, “Special issue on Multimedia Semantic Computing” DOI=
http://www.iscaspeech.org/iscaweb/iscapad/iscapad.php?module=article&id=1132
[10] Thyagaraju G.S, Thyagaraju G.S, “Rough Set Theory Based User Aware TV Program and Settings
Recommender”, 48 International Journal of Advanced Pervasive and Ubiquitous Computing, 4(2), 48-
64, April-June 2012
[11] Novak, V., Perfilieva, i., And mocker, j."mathematical principles of fuzzy logic dodrecht", Kluwer
academic. ISBN 0-7923-8595-0.
[12] “Corn starters”, DOI= http://www.agtest.com/articles, 2000.
[13] Fuller, R., Carlsson, C., “Fuzzy multiple criteria decision making”, Recent developments, Fuzzy
Sets and Systems 78(2) 139- . 153 , 1996
[14] Bingchuan Yuan, John Herbert, Fuzzy CARA - A Fuzzy-Based Context Reasoning System For
Pervasive Healthcare, The 3rd International Conference on Ambient Systems, Networks and
Technologies , Procedia Computer Science 10 ( 2012 ) 357 – 365

38
Authors
Mrs DivyaJyothi M.G. is currently working as Lecturer at the Department of
Information Technology, Al Musanna College of Technology, Sultanate of Oman.
Her teaching interests include Pervasive Computing, Firewalls and Internet
Security Risks, E-Commerce, Computer Networks, Intrusion detection System,
Network Security and Cryptography, Internet Protocols, Client Server Computing,
Unix internals, Linux internal, Kernel Programming, Object Oriented Analysis and
Design, Programming Languages, Operating Systems, Image Processing, Web
Design and Development, etc. Her most recent research focus is in the area of
Pervasive Computing. She received her Bachelor and Master Degree in Computer Science from Mangalore
University, She bagged First Rank in Master’s Degree at Mangalore University. She has been associated as
a Lecturer of the Department of Information Technology since 2007. She has worked as Lecturer at ICFAI
Tech., Bangalore, T John College for MCA, Bangalore, Alva’s Education Foundation Mangalore. She has
guided many project thesis for UG/PG level.
Mr. Rachappa is currently working as Lecturer at the Department of Information
Technology, Al Musanna College of Technology, Sultanate of Oman. His
teaching interests include Computer Security, Pervasive Computing, E-
Commerce, Computer Networks, Intrusion detection System, Network Security
and Cryptography, Internet Protocols, Client Server Computing, Unix internals,
Linux internal, Kernel Programming, Object Oriented Analysis and Design,
Programming Languages, Operating Systems, Web Design and Development, etc.
His most recent research focus is in the area of Security Challenges in Pervasive
Computing. He received his Bachelor Degree in Computer Science from Gulbarga University, Master of
Science Degree from Marathwada University and Master of Technology in Information Technology Degree
from Punjabi University (GGSIIT). He has been associated as a Lecturer of the Department of Information
Technology since 2006. He has worked as Lecturer at R.V. College of Engineering, Bangalore. He has
guided many project thesis for UG/PG level. He is a Life member of CSI, ISTE.
Dr. D H Rao is currently working as a Dean, Faculty of Engineering, VTU,
Belgaum. Principal and Director, Jain College of Engineering, Belgaum.He is the
Chairman, Board of Studies in E & C Engineering, VTU in Belgaum. He is a
Member, Academic Senate in VTU Belgaum. He has over 100+ publications in
reputed journals and conferences. He obtained B.E. (in Electronics from B.M.S.
College of Engineering), M.E. (from Madras University), M.S. (University of
Saskatchewan, Canada) Ph.D. (Univ. of Saskatchewan, Canada).

A scenario based approach for dealing with

More Related Content

What's hot (20)

Viewers also liked (8)

Similar to A scenario based approach for dealing with (20)

Recently uploaded (20)

A scenario based approach for dealing with