Trust Based Modeling and Prediction of Socio-Technical Attack in Cellular Environment

rufftartΤεχνίτη Νοημοσύνη και Ρομποτική

29 Οκτ 2013 (πριν από 3 χρόνια και 11 μήνες)

133 εμφανίσεις


A Synopsis


On



Trust Based Modeling and Prediction of
Socio
-
Technical Attack
in Cellular

Environment









For the Award of Ph.D. Degree in

Cyber Laws and Information Security





Submitted to


Indian Institute of Information
Technology

Allahabad



Submitted by

Preetish Ranjan

RS105



Under the
Supervision of

Dr. Abhishek

Vaish


Assistant Professor


IIIT
-
Allahabad


Introduction:


We know that mobile has become an integral part for every individual and now people prefer
multiple

mobile set
s

or dual SIM set that lead to a revolutionary transformation

in every aspect
of human life and

has

taken
-
away our need to use our memory in some aspects. Th
e phone has
become our notepad,
calendar, calculator, alarm clock,

reminder, spouse and even
our
best
friend
.

India is the fifth nation to join
the “100 million mobiles club”

along with United States,
China, Russia and Japan. This mind
-
boggling growth has no doubt facilitated international
commerce, trade and travel but
they

are
being
heavily used in the
planning and coordination of
criminal activities

and hence pose challenges for the investigating agencies to corroborate
evidences.
Since the mobile phones
have become an important tool for

modern human daily
life, teleco
mmunication patterns may reflect d
ifferent human relationships

and
behaviors
while

the

changes in the telecommunication pattern may expose signs of social relationships
and
behavior changes such as

the calling patterns of a person with his/her friends diff
er from those
with spammers.


Mobile phones are
main cause

for socio
-
technical attack

which may create chaos in society
within

few min
utes and may be proved even more

destructive than
that of
a bomb
. It does not
attack physically but it
attacks

on our faith and trust which is the basic pillar of our social
structure.


Socio
-
technical is an organized approach which is defined by
the interaction among people with

technology either in society or in workplace. This may be disrupted through misuse of

technology by some of the malicious intent people to attack
on
the social structure based on
trust and faith. Wrong interpretation of message and manipulating the integrity of information
may create
pandemonium

in the society and can be one of the reasons

for social riots, political
mis
balance

etc.

Now even the hardware and software developers are trying to develop
components by keep
ing

in the mind all the changes they are going to bring about in the
information flow. Social

technical attack includes:



Defamation of person, organization, group



Attack on religious belief



Awful advertiseme
nt over internet and mobile phon
e



Mass spread of wrong or manipulated information through mass mail and SMSs


Trust
is
believ
ing
that

the person who

is trusted will do what is expected and it starts from the
family which grows to
build over society
. Trust for information may be established if it either
comes from genuine source or information is validated by authentic body so that there is always
a fe
eling of security and optimism. There is always a sense of insecurity when we perform online
transaction of money that
the
site or portal we are using is authentic or not

and there is always a

dilemma that SMSs we are receiving is coming from authentic sou
rce.

Telecommunication

network and c
yberspace

may be used interchangeably

as both are

very
widespread and a
lmost exhaustive which
the most easily accessible platform for masses
. If we
want to express something to common people it is one of the very comfort
able areas to explore
with but it is vulnerable if it is being used maliciously. This illegal use of cyber space comes under
socio
-
technical attack as cyber technology is being used to affect the society.
There is
always a
paucity of authenticity about the

matter
and
socio
-
technical attacker enjoys this vulnerability to
affect the trust in information. Due to DNS poisoning or phishing, many fake sites are there in
internet which publish wrong information such as information about any product, free gift, fre
e
tour and most dangerous sometime they ask for username and password of the bank account.
These are ways by which
common users

of our society are

easily duped and finally they lost their
confidence, trust and faith over the internet

technology
.

There are
varied platforms which are vulnerable such as

1.

Spam Attack

2.

Social Phishing Attack

3.

Automated Social Engineering Bot

4.

Sending bulk SMS to mobile phones

5.

Worms and Virus

The sub
-
component of socio
-
technical attack may be as follows:



Social attack by accessing th
e profile information, finding his list of friends, prior
communications and shared events



Sending message to victim’s friends to defame his social status



Maintain the cycle of attack by accessing the information of victim’s friend cluster



Bulk SMSs throug
h website or mobile phone

It has brought along with number of problems that have an undeniable impact on crime.
Mobile
phones are heavily used in the planning and coordination of criminal activities and hence pose
challenges for the investigating agencies to corroborate evidences. In order to maintain national
security as well as law & order, it is necessary to inve
stigate the call data records and its details
e.g. location, calling number, called number, IMEI number, frequency, mode (voice or SMS), BTS
address etc and on the basis of analysis, some concrete predictions can be made which can
relate to the behavior of

the real malefactor and their related parties.

Literature Review
:


The Reality Mining Project at Massachusetts Institute of Technology has made significant studies
in this field as they have used a large set of data of different mobile users. In this
study
,

they
inferred the user behavior on the basis on his daily routine activity and they attempted to
quantify the amount of predictable structure in an individual's life using entropy metric.

Entropy
quantifies the uncertainty involved in predicting the

value of a random variable
.

They

use
d

multiple probability and statistical methods for quantifyin
g social groups, relationships and
communication patterns
for detecting human
-
behavior changes.
In this paper the foremost
emphasis was on the user’s daily ac
tivity such as time of sleeping, office hours, and use of
different features of the cell phone as calendar, clock, and browser.

In paper titled as “Behavioral Entropy of a Cellular Phone User” the behavior of cellular phone
users and their behavior signatu
res based on their calling patterns has been addressed.
Quantification and inferences of a person’s randomness level
has been projected
using
information entropy based on the location of the user, time of the

call etc by
utilizing

the
correlation coefficie
nt and factor analysis.

The problem area and
its solution defined in

“Predicting Social Ties in Mobile phone Networks”
by Huiqi Zhang and Ram Dantu was very inspiring. In this paper they
proposed an affinity model
for quantifying social
-
tie strengths. A
reciprocity index

is integrated to measure the level of
reciprocity between users and their communication partners using Hellinger distance to
calculate the affinity ratio among the population size.

This
same
problem is also

being

discussed
in “The Applica
tion of ARIMA Model in Chinese Mobile User Prediction” by Xu Ye who
implemented ARIMA model to predict future network among different mobile users.

Below table is very brief
but exhaustive
description

of work in for

analyzing
and predicting
the
pattern fro
m
the call detail record
.

S
.
No

Title of paper

Year

Publication

Findings

1
.

Predi cti ng Soci al Ti es i n Mobi l e
Phone Networks

2010

I EEE

Thi s paper i nvesti gated

evol uti on of person
-
to
-
person
social relationship, quantify and predict social tie
strengt
hs based

on CDR of mobile phone users
.

2
.

Quantifying Reciprocity in Social
Networks

2009

I EEE

Thi s paper proposed reci proci ty i ndex for quanti fyi ng
soci al rel ati onshi ps based on CDR and Twi tter bl ogs.

3
.

Soci al Network Reci proci ty as a
Phase Transi ti on i n
Evol uti onary
Cooperati on

2008

I EEE

Thi s

paper focused

on the so
-
cal l ed “Latti ce
Reci proci ty”

mechani sm that enhances evol uti onary survi val of the
co
-
operati ve phenotype i n the Pri soner’s Del i mma game
.


4.

A regressi on based approach for
mi ni ng user moveme
nt patterns
from random sampl e data

2010

El sevi er

Thi s paper propose
d

al gori thm to extract i nformati on
regardi ng frequent user movement behavi or.

5.

Acti vi ty Recogni ti on from Cal l Detai l
Record

2010

I EEE/ACM

D
ai l y behavi or of
user i s di vi ded i n to 48
parts wi th every
hal f an hour as basi c el ement

wi th one acti vi ty mode.

6.

Turni ng tel ecommuni cati on cal l
detai l s to churn predi cti on

2002

El sevi er

Thi s paper

eval uate
d

a churn predi cti on techni que that
predi cts churni ng and cal l pattern from CDR.

7.

Mobi l i ty
-
based predi cti ve cal l
admi ssi on control and bandwi dth
reservati on i n wi rel ess cel l ul ar
network

2001

El sevi er

Thi s

paper propose
d

to predi ct user mobi l i ty based on
the mobi l i ty hi story of users

on the basi s of

computati onal l earni ng theory based on

data
compressi on.

8.

Effi ci ent data mi ni ng for cal l i ng path
pattern i n GSM network

2002

El sevi er

Thi s paper expl ored

a new data mi ni ng capabi l i ty that
i nvol ves mi ni ng cal l i ng path pattern i n GSM.

9.

Cal l i ng communi ti es anal ysi s and
i denti fi cati on using
machi ne l earni ng
techni ques

2008

El sevi er

Thi s paper focused on
i denti fyi ng the cal l i ng
communi ti es and demonstrate how cl uster anal ysi s can
be used to effecti vel y i denti fy communi ti es usi ng CDR.

10.

Appl yi ng data mi ni ng to tel ecom
churn management

2006

El sevi er

Thi s
compares vari ous data mi ni ng techni que that can
assi g
n a ‘propensi ty
-
to
-
churn’ score
to each subscriber
of a mobile operator.

11.

Anomal y Detecti on form Cal l Data
Record

2009

ACM

I t proposed cl usteri ng based al gori thm

for

anomal ous
users

based on fuzzy attri bute val ue.

12.

Bayesi an i nference for l ocal izati on i n
cel l ul ar network

2010

ACM

Thi s paper present a general techni que based on
Bayesi an i nference to l ocate mobi l es i n cel l ul ar network.

13.

Al gebrai c Vi sual Anal ysi s: The
Catal ano
Phone Cal l Data Case Study

2009

ACM

I t proposed

an al gebrai c model capabl e of representi ng a
l arge cl ass of vi sual anal ysi s operati ons on graph data.

14.

Spati al probabi listi c model i ng of cal ls
to busi ness

2010

ACM

I n t
hi s paper, CDR i s bei ng used as

source data and
i nvesti gated i ts rel evance to l ocal search.

15.

Rol e defi ni ng usi ng behavi or based
cl usteri ng i n tel ecommuni cati on
network

2011

ACM

Thi s paper i denti fi es the user behavi or i n the network.

16.

Soci al Network i denti fi cati on and
anal ysi s
usi ng CDR

2009

ACM

Thi s paper emphasi zed

on the data generated by the
tel ecom
i ndustry may be useful for the l
aw enforci ng
agenci es to fi nd out the network of some cri mi nal s.

17.

Expl oi ti ng ti me
-
varyi ng rel ati onshi ps
i n stati sti cal rel ati onal model s

2007

ACM

They presented an i ni ti al approach of model i ng dynami c
rel ati onal data graph
i n predi cti ve model s

of attri butes.

18.

Detecti on of Outl i er Patterns i n Cal l
Record Based on Skel eton Poi nts

2010

ACM

Thi s paper presents an outl i er detecti on al gor
i thm based
on patterns
formed from skel eton poi nts of ti me seri es.

19.

Model i ng cal l detai l records fro
m
mobile telecommunication network

2007

ACM

Thi s paper provi ded model s based on the behavi or of a
sampl e of 250 subscri bers over 52 week peri od.

20.

Cal l i ng Communi ti es anal ysi s and
i denti fi cati on using machi ne l earni ng
techni que

2009

ACM

Thi s paper pursues i denti fyi ng the cal l i ng communi ti es
and demonstrates how cl uster anal ysi s can be used to
i denti fy communi ti es

effecti vel y
.

The above matrix is
screening undoubtedly

that most of the paper
s

explore the

use of call data
record
for
speculating current
marketing strategy
, future marketing trends for the sale of
product
s

and services. Churn prediction of the mobile phone users is very
common in

which
they
tries to predict
all the conditions and situation which
compel

the customers to change their
service providers. Most of the papers focused on identifying social tie strength among mobile
phone users th
r
ough
different methods,
data mining algorithm
ic

a
pproach and

few using
machine learning approach.

There is also a
survey
papers
based
on the behavio
r of a sample of
250 subscriber
s over 52 week
period

disperse the load on the telecom network and
track the
trajectory of the user
.

Some paper concluded that CDR may be used to infer most favorable
transportation mode used by the user and few papers talked about the improvement
in the
quality of CDR for having better analysis.

Through the literature

survey
,

we came across the fact tha
t no paper deals with
socio
-
technical
attack and very
few
paper
s

actually focuses on
the usage of mobile phones by criminals.
.

However, no journal evidence manifests the development

in the direction of crime control
through mobile data log.

There has been

no any methodology discussed to track the anti
-
social
elements of society with minimum error report.

These research efforts lack comprehensive
analysis for one
-
to
-
one or one
-
to
-
many relationships and behaviors in detail

necessary
to
unearth

few
special gr
oups or clusters of people. These detailed features of human relationships
are more important
for detecting terrorists, spam and user preferences.

Hence,

d
ue to

human

s
social behavior

diversities a
nd complexities,
one technique will not

be satisfactory

to
detect
the

different features of human social behaviors.


Objective:


To model and predict the point
of socio
-
technical

at
tack

on

the
trust

and belief

of

people
,

established
over
long
period of
time
. We will try to
attain

our
goal

by



creating

a
social
network



measur
ing

the social
tie
strength
and
affinity



d
esign
ing

algorithm to predict the human behavior

and socio
-
technical attack






Research Questions:


1.

What are the different ways to model social structure in our society

on the basis of
technology they are using?

2.

What is trust

in society

and how can

it
be established

and standardized
?

3.

How can we predict the
likelihood ways that can affect the trust in information?

4.

What could be
the active
component of socio
-
technical
attack?

5.

What are the features of socio
-
technical attack

that

boost up

the attacker

to plan the
conspiracy
?



Motivation
s
:




Use of mobile phone
s

in 26/11 Mumbai attack



Exte
nsive use of mobile phone for corroborating

kidnapping



Use of mobile phones in MMS S
candals such as DPS R K Puram MMS Scandals, Ragini
MMS scandals
and number of bollywood
actors and
actresses

involved in MMS
controversies



Socio
-
technical attack
by means of

mobile phones such as case of rumors about
earthquake in Allahabad


Research
Design:

Our
aim is to design
an
algorithm

based on some mathematical calculation, artificial intelligence
and soft computing

for the

prediction of human behavior using call data record on the basis of
certain heuristics such as behavioral entropy and socia
l tie strength among mobile users. We are
trying to develop certain standard and threshold value to calculate social network affinity so that
we can locate criminals and predict future course of action with minimum error
.
Ther
efore, we
will try to integrat
e
SQL queries, probability, statistical methods and dat
a mining algorithms to
develop our own algorithm
for
analysis of
human
-
behavior

and social network
s from macro level
to micro level.

The res
earch lay down

is segregated into six

part
s
:

1.

Collection of da
ta

from different source

2.

Purification of data
and put data in certain format

3.

Analysis of data from different
prospective

4.

Defining heuristics from purified data

5.

Design of algorithm

6.

Testing and validation of result

Tentative O
utcome:


Final outcome of research work will be
the
emergence of certain

model and
algorithm to
analysis human behavior, social network and to protect society from socio
-
technical attack from
anti
-
social element
s
. P
rotection of our society
from

socio
-
technical atta
ck will lead to develop
and maintain the

some sense of trust and belief which is build over the

past long

period of time.
Hence, this work may be
proved valuable

to bring about
peace and
harmony in society

in
technological era
.


References:

1.

Huiqui

Zhang and Ram Dantu “Predicting Social Ties in Mobile Phone Networks” 2010
IEEE, Canada

2.

Tain Zhu, Bai Wang “Role defining using behavior based clustering in telecommunication
network” ACM
-
2011

3.

Chih
-
Chieh Hung, Wen
-
Chih Peng “A regression
-
based approach fo
r mining user
movement patterns from random sample data” Elsevier
-
2011

4.

Chen Zhou, Zhengguang Xu, Benxiong Huang “Activity Recognition from Call Detail
Record: Relation Between Mobile Behavior Pattern And Social Attribute Using Hierarchial
Conditional Rando
m Fields” 2010 IEEE/ACM International Conference on Green
Computing & 2010 IEEE/ACM International Conference on Cyber, Physical and Social
Computing

5.

Hui Zang, Jean Bolot “Bayesian inference for localization in cellular network” ACM
-
2010

6.

Ramaswamy Hariharan
, Ji Meng Loh “Spatial probabilistic modeling of calls to business”
ACM
-
2010

7.

Muhammad Usman Khan(National University of science and Technology, Pakistan)
“Social Network identification and analysis using CDR” ACM,2009

8.

Andrew G. Miklas, Kiran K. Gollu, Kelv
in K. W. Chan, Stefan Saroiu, Krishna P. Gummadi
and
Eyal de Lara “Exploiting Social Interaction in Mobile Systems” Springer
-
Verlag Berlin
Heidelberg 2007, pp. 409
-
428

9.

Fang Cheng, Lei Tian, Jinfeng Xie “Research on Path and VPI/VCI relating guess solution
for TD
-
SCDMA
network centralized monitoring system


Seventh International
Conference on Intelligent Hiding and Multimedia Signal Processing, 2011 IEEE

10.

Giuseppe Bianchi,
Noco dOHeureuse, Saverio N
iccolini “On
-
demand Time
-
decaying
Bloom Filters for Telemarke
ter Detection” ACM SIGCOMM Computer Communication
Review, Oct 2011

11.

Hui Li, Young
-
Chan Lee, Yan
-
Chun Zhou, Lie Sun “The random subspace binary logit
(RSBL) model for bankruptcy prediction” Elsevier 2011

12.

Xu Ye “The Application of ARIMA Model in Chinese Mobil
e User Prediction” IEEE
-
2011

13.

Chen Zhou, Zhengguang Xu, Ben
xiong Huang “Activity Recognition from Call Detail
Record: Relation Between Mobile Behavior Pattern And Social Attribute Using
Hierarchical Conditional Random Fields” IEEE
-
2010

14.

Huayong Wang, Frances
co Calabrese, Giusy Di Lorenzo, Carlo Ratti “Transportation Mode
Inference from Anonymized and Aggregated ”

2010
-
IEEE