design and implementation of user context aware
TRANSCRIPT
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
47
Design and Implementation of User Context aware
Recommendation Engine for Mobile using Bayesian
Network, Fuzzy Logic and Rule Base
Thyagaraju GS Research Scholar (VTU),
Dept Of CSE, SDMCET, Dharwad -580 002, Karnataka, India.
Umakant P Kulkarni Professor
Dept of CSE, SDMCET, Dharwad -580 002, Karnataka, India.
ABSTRACT
Context-aware computing refers to a general class of mobile
systems that can sense their physical environment, and adapt
their behavior accordingly. Such systems are a component of
a ubiquitous computing. Context aware computing makes
systems aware of situations of interest, enhances services to
users, automates systems and reduces obtrusiveness and
customizes and personalizes applications. Mobile phones and
PDAs are converging into mobile lifestyle devices that offer a
wide range of applications to end users. Many of these
applications will have the ability to adapt themselves to the
user’s situation, commonly referred to as context awareness.
Context-aware services have been introduced into mobile
devices, such as cellular phones. Context aware service
recommendation engine for mobile is designed to
automatically adopt its behavior to changing environment. To
achieve this, an important issue to be addressed is how to
effectively select services for adaptation according to the
user’s current context. In this paper, we propose an intelligent
service recommendation model. We formulate the service
adaptation process by using artificial intelligence techniques
like Bayesian Network, fuzzy logic and rule based reasoning
.Bayesian Network to classify the incoming call (high priority
call, low priority call and unknown calls), fuzzy linguistic
variables and membership degrees to define the context
situations, the rules for adopting the policies of implementing
a service, fitness degree computation and service
recommendation. In addition to this we have proposed
maximum to minimum priority based context attributes
matching algorithm for rule selection based on fitness degree
of rules. The context aware mobile is tested for library and
class room scenario to exemplify the proposed service
recommendation engine and demonstrate its effectiveness
General Terms
Context Aware Computing, Ubiquitous Computing, Service
Recommendation Engine.
Keywords
Context Aware MOBILE, User Context, Socialization,
Personalization, Bayesian, Fuzzy logic, Rule Base.
1. INTRODUCTION The users nowadays are mobile dependent. Services provided
by the existing mobiles with minimum functionalities are not
up to the mark. Context Aware Mobile is in high demand.
Mobiles are one of the most popular consumer products all
over the world, and have evolved such that they can now
provide personalized and adaptive services to users in many
ways. The existing technologies allow users to move around
with computing power and network resources at hand (say
portable computers and wireless communications). Due to
their popularity and easy access and varies functionalities,
various technologies have been developed that contribute to
making the mobile even more context aware. Mobile internet
services enable access to information in a more flexible
manner. These changes have increasingly enabled people to
access their personal information, corporate data, and public
resources ―anytime, anywhere‖. There are already many
wireless handheld computers available, running different
operating systems such as Palm OS, Microsoft Pocket PC
(Windows CE), and Symbian EPOC. Contextual presentation
is an emerging technique that has huge commercial
possibilities .The theory behind the applications is complex
and this makes the implementation non trivial.
With the appearance of mobile devices such as cellphones,
PDAs or laptops, context-aware applications are becoming
prevalent. Context-aware systems provide relevant
information, and services based on information to the user,
depends on the users’ situation. Mobile computing imposes
new challenges in designing computer hardware and software
due to user mobility, the diverse types of devices used,
resource constraints, and the dynamic nature in execution
context. Context-aware mobile computing middleware
provides abstraction and support for application programmers
to ease the task of developing mobile applications, ensuring
acceptable QoS and allowing for adaptation to changes in the
operating environment. An important issue to address in
designing a context aware middleware is how to effectively
recommend services for adaptation according to the user’s
current context. However, this issue has not been adequately
addressed in existing work which has been focused either on
the software realization of services configuration or on a
specific scenario or domain [1,2,3]. This paper is concerned
with the formulization and development of a service
recommendation engine for context-aware mobile computing
middleware. We propose the design and implementation of
user context aware recommendation for mobile using artificial
intelligent tools like Bayesian Network, Fuzzy logic and Rule
base.
The recommender makes mobile to adapt to dynamically
changing personal, social, environmental and physiological
states. To list some of the services (but not limited) provided
by recommender are as follows:
1. Provide the callers with the ability to communicate the high
priority calls irrespective of his situation and location.
2. It goes to silent mode in the class room/meeting room
automatically.
3. It goes to the vibrating mode automatically in the Library
and also provides services like book search.
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
48
4. It provides notifications whenever required.
5. It provides Context based desktop applications.
A number of sensors including accelerometers, temperature,
time, location, etc are embedded in the mobile to provide data
about the user’s context. For experiment purpose we are using
sensor board (embedded with sensors like accelerometer,
temperature, humidity, Bluetooth transceiver) and mobile with
Bluetooth enabled. In addition the recommender augments its
contextual knowledge by making use of applications such as
electronic calendars, policy/rule base, address books, context
repository, action repository and task lists. It alleviates
cognitive load on user and reduces the application searching
time. Current commercial mobile phones impose additional
cognitive load on their users by requiring them to be
conscious of their phone’s states. Examples include
remembering to turn the ringer on and off, handling missed
calls, determining call priority, and worrying about inaudible
ringer volume in a loud environment.
The motivation behind of our paper is to reduce the user’s
cognitive load, to reduce the user’s service searching time, to
increase user services.
The remainder of the paper is structured as follows. Section 2
describes related work. Section 3 presents Modeling and
design of recommender. Section 4 presents architecture of
recommendation engine. Section 5 discuses the experiments
and result analysis. Section 6 concludes the paper with future
work.
2. LITERATURE SURVEY Socialization and Personalization of mobile is an active
research topic. A general definition of socialization is to make
someone behave in a way that is acceptable to society.
Personalization is ―understanding the needs of each individual
and helping satisfy a goal that efficiently and knowledgeably
addresses each individual’s need in a given context―.
Personalization and Socialization has really gained
importance with always connected services in the context
aware applications. Context aware applications and services
use context information to provide relevant services to the
user and task at hand [4-10]. Recommender systems are
intimately related to personalized services. In theory
recommender systems provide the underlying implementation
of personalization in practice, recommendation and
personalization often combine to one. The recommendation
systems developed for mobile earlier are either content or
content boosted collaborative. The context aware
recommender utilizes context data as an additional input to
the recommendation task, alongside information of users and
items [11-22].
The proposed recommendation engine for mobile utilizes the
social locations like college campus, library and class room.
The engine recommends the appropriate services like book
search in library, appropriate notifications in class room and
shopping in outdoor to mention few. The Engine makes use of
Bayesian network to determine the social affinity of incoming
call by classifying the incoming call as high priority, low
priority and unknown calls by considering users mobile usage
history. In addition the system utilizes fuzzified values of user
context in order to improve the performance.
Over the last decade, most research, aimed on context aware
mobile phone has been done. Let us go through some of the
existing context aware mobile phones.
1. SENSAY [23] is a context-aware mobile phone that adapts
to dynamically changing environmental and physiological
States. The drawbacks of SenSay is a number of sensors
including accelerometers, light, and microphones has to be
mounted at various points on the body to provide data about
the user’s context.
2. Proactive and Adaptive Fuzzy Profile Control for
Mobile Phones[24]—Here the adaptation is based on
recognizing patterns of human practices, which may change
over time. The control system is implemented with a fuzzy
controller that supports reinforcement learning. The operation
of the system is demonstrated with a mobile phone that is
controlled by a PC. The PC lets a user to simulate the context
parameters, and the phone works as a user interface for profile
selection and display.
3. CAESAR[25]: A Context-Aware, Social Recommender
System for Low-End Mobile Devices. Here the
recommendation is based on the factors like Social affinity
computation from call data records and, user address books.
In addition it makes use of Feedback based Tuning to find
whether the recommendation made was useful or not.
4. Collaboration in Context-Aware Mobile Phone
Applications[26] The research work presents role of context
information in improving the collaboration of mobile
communication by supplying relevant information to the
cooperating parties, one being a mobile terminal user and the
other either another person, group of people, or a mobile
service provider.
5. A Framework for Context-Aware University Mobile
Organizer [27] .The research work discusses some essential
principles and technologies for developing and implementing
context aware applications.
6. Context Management and Reasoning for Adaptive
Service Provisioning [28] The research work presents the
architecture components related to context acquisition through
the reasoning and context management. It presents a
framework for modular multi domain context detection and
shows how required context can be obtained by acquiring a
variety of source data and applying reasoning mechanisms for
aggregation. Furthermore, a user interface for easy and fast
extension of the context model is introduced.
7. Intelligent Agent based Hotel Search & Booking System
[29] uses an intelligent agent (instead of the human agent) to
perform searching and booking activities that can improve the
speed of the search and reduce cost significantly.
8. Intelligent Agent based Mobile Shopper[30] This
research focuses on the use of mobile devices for shopping.
9. Service Adaptation Using Fuzzy Theory in Context-
aware Mobile Computing Middleware [3] This research
proposes a Fuzzy-based Service Adaptation Model (FSAM)
that can be used in context aware middleware.
10.UbiPhone: Human-Centered Ubiquitous Phone
System[31] UbiPhone automatically connects using the most
appropriate phone system based on current context
information, such as caller and contact’s location, presence
status, network status, available phone systems, calendars, and
social relationships.
In comparison with the previous works the major contribution
of this paper can be summarized as follows:
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
49
1. Reduction in application searching time in different
context. For example if user enters into the library , the
proposed mobile will gets adapted to the library situation
automatically by configuring its desktop and internal settings
to facilitate the library services like book search ,web link
,silent mode and friends search .
2.Design of Recommendation engine utilizing the contextual
parameters like Location (Class Room , College Campus,
House ,etc) Personal(Age, Name) ,Temporal (time ,date,) ,
Physical (Fall ,Normal) , and Schedule Agendas.
3. Exploitation of hybrid Fuzzy system, Bayesian Networks
and the utility theory (usage history and context history) for
modeling and implementation.
4. Executing Actions using simple if then else rules base.
3. MODELING OF THE PROPOSED
RECOMMENDATION SYSTEM
3.1 Preliminaries To introduce the proposed modeling of recommendation
engine we first give the definition of the concepts and
terminologies used in the proposed system. The concept,
algorithm, definitions (although trivial for some readers) and
modeling, are needed as a basis for the subsequent sections
where the architectural and operational features of the
complete system are described.
Fuzzification: The Conversion of Crisp input into fuzzy
values represented by linguistic terms using membership
functions is called fuzzification. There are different forms of
membership functions such as triangular, trapezoidal,
piecewise linear, Gaussian, or singleton(fig1).
Fig1: Different types Fuzzy Membership functions
Table1: Fuzzy information of primitive context related to user.
Context
Attributes
Priority Fuzzy Set Values Fuzzy Linguistic
terms
Fuzzy Membership
Function Location 2 Home, College Campus, Library, Classrooms ,
Outdoors
low , fair , good,
Excellent
Trapezoidal
Time 3 Sleepy Hours, Early Morning , Morning ,
Afternoon , Forenoon , Evening , Night , Late
Night , Free Hours , Class Hours , Meeting Hours,
Break Time Library Hours , Sports Time
Yes ,No Singleton Function
WeekDay 4 MonDay ,Tuesday, Wednesday ,Thursday ,
Friday, Saturday , Sunday , Holiday, Working
Day
Yes ,No Singleton Function
Physical 1 Fall , Normal Yes, No Singleton Function
Temperature 5 Current temperature in degree Celsius Very Cold, Cold,
Warm, Hot, Very
Hot
Trapezoidal function
Recommendation Process: As shown in the figure 2 the
recommendation process involves different steps like
fuzzification and action recommendation. Action
recommendation involves context generation, rule matching
and popping actions (Services and settings). The complete
process is given in the form of algorithm1 (Fig3).
Triangular Trapezoidal Singleton
Gaussian Piecewise Linear
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
50
Fig2: Fuzzification process in Recommendation System.
Fig3: Algorithm to recommend services and settings based on the user context
Actions: An action is a set values of services and settings
provided by the middleware and invoked by a mobile
recommendation engine. Let A = {A1, A2,A3, ----An} be the
set of actions provided by the middleware where
Ai(1<=i<=m) represent the service / settings provided by the
middleware. Ex: A= {Volume, Call Settings, Desktop
Applications, Profile}
Rules: It represents a method to deliver the settings and
services with a certain conditions. Each rule is a triplet
(Rid,C,A) . Whereas Rid Rule Number, C is set of conditions
and A is a set of actions.
Context: Context is used to represent the user’s situation
with respect to mobile applications. It is a vector of vectors. In
our proposed work context is set of fuzzified instances of
primitive context like location, time, day, userid, temperature,
fall and incoming call .Let C= {CLocation, CTime, CDay,
CFall, CIncommingCall, CTemperature } be a set of context
attributes whose values are monitored by context aggregator
of the recommendation engine.
Rule Fitness Function: Let FD(Rj) be the fitness degree
for the Rule Rj under current context situation.
FD(Rj)=
𝑁𝑜 𝑜𝑓 𝑀𝑎𝑡𝑐 𝑖𝑛𝑔 𝐶𝑜𝑛𝑑 𝑖𝑡𝑖𝑜𝑛 𝐴𝑡𝑡𝑟𝑖𝑏𝑢𝑡𝑒𝑠 𝑜𝑓 𝑅𝑢𝑙𝑒 𝐵𝑎𝑠𝑒
𝑇𝑜𝑡𝑎𝑙 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐶𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 𝐴𝑡𝑡𝑟𝑖𝑏𝑢𝑡𝑒𝑠 𝑜𝑓 𝑅𝑢𝑙𝑒 𝐵𝑎𝑠𝑒 (1)
Rj will be recommended only if FD(Rj) ≥ τ, where τ is
threshold function and its value will be 0.2 (based on the
knowledge).
Input
Crisp Linguistic
Recommendation
Engine Fuzzification
Membership
Actions
Situation history
Rule Base
Input
Mobile-GUI
Algorithm1: Recommending services and settings based on the user context: While (mobile! =Switched_OFF)
{
for (sn=1; sn<=#_of_Sensors; sn++)
Sensor_Readings[sn][ ]= Get_Sensor_Readings(sn);
end_for for(ca=1;ca<=#of_context_attributes;ca++)
Current _Context[ca][ ]=FuzzyValue (Sensor_Readings[sn++] );
end_for Rule_ID = Get_Rule_ID(Current _Context); //using algorithm 3
for(ac=1;ac<=#of_action_attributes;ac++)
Current _Action[ac][] = Get_Action(Rule_ID(Action(ac))
end_for Recommend_Action(Current _Action);
end_for Previous_Context = Current_Context;
While(Current_Context== Previous _Context) { for(sn=1;sn<=#_of_Sensors ;sn++)
Sensor_Readings[sn][ ]= Get_Sensor_Readings(sn);
end_for for(ca=1;ca<=#of_context_attributes;ca++)
Current_Context[ca][ ]=FuzzyValue (Sensor_Readings[sn++] );
end_for if(Current_Context!= Previous _Context)
{ Rule_ID = Get_Rule_ID(Current _Context); //using algorithm 3
for(ac=1;ac<=#of_action_attributes;ac++) Current _Action[ac][] = Get_Action(Rule_ID(Action(ac))
end_for Recommend_Action(Current _Action);
Previous_Context = Current_Context;
} end_if
}end_While
}end_While
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
51
Bluetooth Access Point: it is a device that allows wireless
communication devices to connect to the wireless network
through them by using the Bluetooth technology and any
other related standards
MAC Address: stands for Media Access Control address. It
is a unique identifier assigned by the manufacturer to
Bluetooth Access Point, network interface cards (NICs) or
network adapters, and it is known as the physical address.
3.2 System Modeling:
Our main objective is to embed the user’s context and social
awareness in the recommendation system. Here the purpose of
mobile service refers to the subjective reasons why certain
services are used and certain services are rejected in certain
social situations. Our hypothesis is that Mobile applications
have different purpose for different situations and to truly
personalize the mobile applications offering, the Mobile
service should model the purpose of Mobile Applications to
its user. In real-time situations users preferences are truly
influenced by the personal and social situations .Therefore we
postulate that social and personal context or situations encode
additional structure that can be utilized to improve qualitative
recommendation performance.
Modeling three domains: The problem is associated with
three domains of data: user’s data, context data, and mobile
content data. The immediate research question arises as to
how three domains should be best combined for learning of
mobile applications and settings purpose. Figure4 illustrates
the three different domains of input features for context ware
personal mobile services:
Fig4: Three Conceptual Domains
1. Users domain has information about user profiles, social
roles and relationships between caller and callee.
2. Context domain represents the situations that link users
with Mobile; and
3. Mobile content domain encapsulates Mobile profiles,
Applications, Menu Items and settings metadata and their
descriptors.
The three domains are conceptually orthogonal and as such
act as independent sources of data for the Mobile service
recommendation problem. From these three data sources the
mobile service recommender aims to predict purposeful
service selections, given the past behavior of the member in
different situations, based on the user and context features.
Here we emphasize the fact that the recommendations are not
only to be personalized but also to be situationalized
according to the learned purposes of mobile profiles and
settings in similar past situations. In practice, to avoid
requiring to store and process the full feature set for each
prediction the recommenders learn a context aware and
personal service purpose model, which is more compact than
the full dataset but retains desired prediction accuracy.
3.2.1. User Domain:
The user is categorized into the Caller and Callee. Each user is
represented as a vector of attributes like type (Caller, Owner),
MobileNo, Role (Father, Mother, Friend, Boss, Colleague,
Unknown, etc), and Name. Whenever the user makes a call
the call will be classified into High Priority Call, Medium
Priority Call, Low Priority Call and Unknown call. Each user
in his mobile will create separate clusters of contacts as High
Priority, Normal Priority, Low Priority and Unknown
contacts. Table 2 gives the details of service provide for each
type of caller.
Table2: Accessibility Options
The priority can be assigned explicitly by the user. Implicitly
the system can assign the priority of the user based on the
social affinity between the caller and callee. The social
affinity between two users depends on different factors as
follows:
1. Call Acceptance
2. Call Rejection
3. Talk time between pairs of users.
4. Number of call between the users and hit ratio.
Caller
Priority
Services Provided
High All time – Accessible
Predefined Notification to owner
Medium Free Time – Accessible
SMS Notification to Owner
Free time slot announcement to caller
Diverting call to other number
specified by the owner
Low Restricted Time– Accessible
SMS Notification to Owner
Restricted Time slot announcement to
caller
Diverting call to other number
specified by the owner
Unknown Restricted Time– Accessible
SMS Notification to Owner
Restricted Time slot announcement to
caller
Users -
Social Affinity,
Role, calendar,
planner.
Context - location,
activity, day, time, incoming
call
Mobile Service - Volume, Call
Settings, Profile,
Applications
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
52
Fig5: Algorithm to assign the priority to incoming call
using Bayesian Probability (Fig 6 illustrates the
algorithm).
Whereas P is the Bayesian Probability. The recommendation
makes use of Bayesian theorem to predict the appropriate
service in a given context i.e
P(Acceptance/Context) =
𝑷
𝑪𝒐𝒏𝒕𝒆𝒙𝒕
𝑨𝒄𝒄𝒆𝒑𝒕𝒂𝒏𝒄𝒆 𝑿 𝑷(𝑨𝒄𝒄𝒆𝒑𝒕𝒂𝒏𝒄𝒆)
𝑷(𝑪𝒐𝒏𝒕𝒆𝒙𝒕) (2)
P(Rejection/Context)
= 𝑷
𝑪𝒐𝒏𝒕𝒆𝒙𝒕
𝑹𝒆𝒋𝒆𝒄𝒕𝒊𝒐𝒏 𝑿 𝑷(𝑹𝒆𝒋𝒆𝒄𝒕𝒊𝒐𝒏)
𝑷(𝑪𝒐𝒏𝒕𝒆𝒙𝒕) (3)
3.2.2. Context domain
The system is designed to recommend the services based on
the users situation. User’s situation is derived based on the
values of primitive contexts like location, time, weekday;
temperature, incoming call and physical fall. For experimental
purpose we considered the locations like Home, College
Campus, Library, Class Rooms and outdoors.
3.2.2.1:Location Context
The location of user is identified using fuzzy linguistic
variable very low, low, good and excellent. For example the
position of users can be represented by a linguistic variable
xuser whose linguistic values from the following domain {very
low library, excellent admin, good classroom}.
The location is determined using fuzzification process. The
input data are pre-processed so that they are represented as a
fuzzy membership vector. To identify the user’s location, the
algorithm retrieves from the database the MAC addresses of
the AP(Access Point) in the vicinity of the user with their
corresponding signal strengths
μexcellent/μgood/μfair/μlow =
0 𝑖𝑓 𝑎𝑝𝑠𝑠 ≤ 𝑎𝑎𝑝𝑠𝑠 −𝑎
𝑏−𝑎 𝑖𝑓 𝑎 ≤ 𝑎𝑝𝑠𝑠 ≤ 𝑏
1 𝑖𝑓 𝑏 ≤ 𝑎𝑝𝑠𝑠 ≤ 𝑐𝑑−𝑎𝑝𝑠𝑠
𝑑−𝑐 𝑖𝑓 𝑐 ≤ 𝑎𝑝𝑠𝑠 ≤ 𝑑
0 𝑖𝑓 𝑎𝑝𝑠𝑠 ≥ 𝑑
(4)
Fig 6: Illustration of assigning priority to Incoming Call based on the Bayesian Probability
Incoming Call
Caller Identity Previous Settings
Unknown δP = PAcceptance – PRejection
Yes No
Unknown
Known
δP >0 δP ==0
Set Priority = High
Set Priority = Medium
Set Priority = Low
δP <0
Set Priority = Undefined
Algorithm2: Assigning the priority to user call
(incoming call) using Bayesian Probability
If (user is unknown)
Assign Priority as Undefined
else if (user is known)
{ S1 : PAcceptance = P(Acceptance/Context)
S2 : PRejection = P(Rejection/Context)
S3: δP = PAcceptance – PRejection
Priority =
𝑯𝒊𝒈𝒉 𝒊𝒇 𝛅𝐏 > 0 𝑴𝒆𝒅𝒊𝒖𝒎 𝒊𝒇 𝛅𝐏 == 𝟎
𝑳𝒐𝒘 𝒊𝒇 𝛅𝐏 < 𝟎
}
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
53
For example the user is situated in position where the
membership functions say μexcellent, μgood , μfair and μlow for the
locations like library, administrator office and College campus
Outdoor will have the linguistic values as follows:
Table3 : Example Fuzzy Membership values
Location μexcellent μgood μfair μlow Library 1 0.0 0.0 0.0 Administrative
Building 0.5 1.0 0.0 0.0
College Campus Outdoor 0.0 0.5 0.5 0.75
The CAM system recognizes the position of user is at library.
For a given Location μexcellent, μgood , μfair and μlow is calculated
. For example if we consider the three locations l1,l2,l3 the
position of user will be associated with any one of the location
based on their linguistic values for the membership functions
μexcellent, μgood , μfair and μlow.
The rule base for identifying the location is as follows:
If(μexcellent(l1)!=0||μexcellent(l2)!=0|| μexcellent(l3)!=0)
{ If (μexcellent (l1)> μexcellent (l2) && μexcellent (l1) > μexcellent (l3))
then the location of the user is l1
If (μexcellent (l2)> μexcellent (l1) && μexcellent (l2) > μexcellent (l3))
then the location of the user is l2
If (μexcellent (l3)> μexcellent (l2) && μexcellent (l3) > μexcellent (l2))
then the location of the user is l3
}
else If( For all locations μexcellent is zero)
If(μgood(l1)!=0 || μgood(l2)!=0 || μgood(l3)!=0)
{ If (μgood (l1)> μgood (l2) && μgood (l1) > μgood (l3)) then the
location of the user is l1
If (μgood (l2)> μgood (l1) && μgood (l2) > μgood (l3)) then the
location of the user is l2
If (μgood (l3)> μgood (l2) && μgood (l3) > μgood (l2)) then the
location of the user is l3
}
else If( For all locations μexcellent and μgood is zero)
If(μfair(l1)!=0 || μfair (l2)!=0 || μfair (l3)!=0)
{ If (μfair (l1)> μfair (l2) && μfair (l1) > μfair (l3)) then the
location of the user is l1
If (μfair (l2)> μfair (l1) && μfair (l2) > μfair (l3)) then the
location of the user is l2
If (μfair (l3)> μfair (l2) && μfair (l3) > μfair (l2)) then the
location of the user is l3
}
else If( For all locations μexcellent and μgood and μfair is zero)
If(μlow (l1)!=0 || μlow (l2)!=0 || μlow (l3)!=0)
{ If (μlow (l1)> μlow (l2) && μlow (l1) > μlow (l3)) then the
location of the user is l1
If (μlow r (l2)> μlow r (l1) && μlow (l2) > μlow (l3)) then the
location of the user is l2
If (μlow (l3)> μlow (l2) && μlow (l3) > μlow (l2)) then the
location of the user is l3
}
The proposed system makes use of Bluetooth technology for
indoor location of college campus of a mobile device or user.
Bluetooth access points of a network are used for the location.
Location is made by means of the signal strength received
from the access points in the college campus. The location is
determined using the Received Signal Strength Indicator
(RSSI) and MAC address of Bluetooth Access Point .The
signal strength will be measured by the mobile device and it
calculates its location. With the RSSI we build a access point
and RSSI map of the environment. As the location technique
is based on the Received Signal Strength Indicator (RSSI) of
bluetooth nodes. The system works in a similar way than the
RADAR system [32-35], where, first, a server must store a
map of the RSSI at different coordinates. To build the map of
the RSSI in a closed environment (i.e. college campus), a
fixed number of access points will be considered. To create
and conform the map, a mobile device should move through
all the coordinates of interest. From each coordinate, this
device will notify some parameters to store with the map: its
location, information of the signal power that receives from
each access point. Mobile device notifies these parameters by
sending information in one tuple similar to the shown in
equation (5) .
[(BTAP1,SS1), (BTAP2,SS2), . . . , (BTAPn,SSn)] - ( 5)
Fig 7: Example access points at different locations and range of signal strength
Library
Class Room
R
Admin
College Campus
College Campus
College Campus
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
54
Sl.NO Device Mac Address Location
1 BTAP1 00:0B:85:7356:8A Outdoor
2 BTAP2 00:23:B4:A2:CO:A5 Admin
3 BTAP3 00:1B:24:2D:28:B8 CSE-Class Room
4 BTAP4 00:19:7E:5D:17:F5 Seminar Hall
5 BTAP5 00:4E:04:43:52:A2 CSE-Staff Room
6 BTAP6 00:1A:74:4D:63:B9 CSE-Seminar Hall
7 BTAP7 54:9B:12:01:1E:D3 Library
Figure 8: Example of Bluetooth Access points (BTAP) with their associated physical Locations
3.2.2.2: Time Context: The system reads a time from the
system clock .Each time will be represented using a pair
(L(t),TT). Whereas L(t) is the linguistic terms
(AM,PM,EG,NT,MNT,and EAM) and TT is the Time
Types(FreeHours , Working Hours , Break Time , Lunch
Time ,Sleeping Hours). The time will be mapped into its
corresponding Linguistic terms using equation below. From
the user profile and calendar information the time type will be
obtained.
L(t)=
AM if t ε [6am, 12pm]
PM if tε [12pm, 5pm]
EG if t ε 5pm, 7pm
NT if t ε 7pm to 12pm
MNT if t ε 12pm to 3am
EAM if t ε [3am to 5am]
(6)
3.2.2.3Temperature: The temperature was measured using
the sensor embedded .The temperature measured in degree
Celsius was mapped into linguistic terms Very Cold, Cold,
Warm, Hot and Very Hot using equation below :
L(Temp) =
Very Cold if temp < 10Cold if temp ε [10,20]
Norm if temp ε 20,27
Warm if tempε 27,32
Hot if tempε 32,35
Very Hot if temp > 35
(7)
3.2.2.4: Fall Detection: Using the Accelerometer the
acceleration of the mobile is measured. Based on the
accelerometer values one can determine the activity of the
user like walking, running, sitting and fall. A fall typically
starts with a short free fall period. This causes the
acceleration’s amplitude to drop significantly below the
threshold .This represents the period of time when the actual
fall is taking place and it causes a spike in the graph. If a
person is seriously injured in a fall they usually remain on the
ground for a period of time. This is characterized by the flat
line at the end of the graph as discussed in (23).
Table 4: Context based Mobile services and Settings
3.2.3. Mobile content domain:
Mobile is modeled in terms of its services and settings .The
Contents of mobile Service and settings can be classified into
incoming call, call settings, volume settings, profile, desktop
applications and hidden applications. User usually prefers
services and settings based on her context. For experiment we
are considering services and settings related to the user’s
situation in college campus.
Context Services and Settings Provided
Location Location (Class room , Library ,Outdoor , etc ) discovery
Location specific settings and service as predefined by owner
For example when owner visits library mobile goes to silent mode and provides a library related services like
books (journal/ newspaper /digital library) search , friends search ,internet link ,notifications and accessibility
to high priority call in silent mode to mention few.
Time Identifying time as working hours ,free hours ,class hours ,break hours ,lunch hours ,etc
Time Specific settings and service.
For example when owner is in classroom the services and settings changes when time changes from class hours
to free hours
Weekday Classifies the day as working day or holiday
Physical Determines whether the user is in normal or fall condition
If he/mobile is in fall condition it provides the necessary notification as predefined by the owner.
Temperature Identifying the temperature as cold, very cold ,hot and very hot
Temperature specific settings and service as predefined by owner
For example if the temperature is very cold the mobile invokes the application providing the details of coffee/ hot
snacks availability in the college campus.
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
55
Table5: Mobile Services and Settings with allowable choices
4. PROPOSED LAYERED
ARCHITECTURE OF
RECOMMENDATION ENGINE The proposed architecture of context aware mobile is layered
architecture as shown in the figure9. The input provided by
different embedded physical and logical sensor like system
clock, temperature sensor , accelerometer, calendar and user
profile is fuzzified into respective linguistic terms .The
fuzzified sensors values will be processed and context will be
generated. In this step all the aggregated value undergo
preprocessing and fuzzification resulting in the Social Context
.The resultant context will be stored in the form of vector
containing the value for the elements Day, Time, Location,
Temperature and Fall. The vector is then matched with all
possible conditions in the predefined rule base.
Fig9: Generic Architecture of the proposed recommendation Engine
The rule base is build making use of the log file of one month
usage history of mobile. The rule base presented in this paper
is designed making use of the tool e2grulewriter, v1.01, 2010
by expertise2Go.com. The snapshot of few rules with
condition and actions is as illustrated in the figure 10. Figure
gives some of the rules generated in the knowledge base.
Services/Settings Allowable Choices/types
Incoming Call High Priority Call , Low Priority Call, Unknown Call ,No Calls.
Call Settings Call Ringing , Call Vibrating , Answer the call , Reject the call , Call Divert.
Volume Settings High, Medium , Low , Silent.
Profile General , Silent .
Desktop Applications Messaging, My_Shopping , My_Social_Site , My_Library, My_Groups, My_College, Radio, TV, Camera,
Video , My_Music, Games, Voice_Recorder, My_Entertainment, Organizer, Alarm, Calendar, Internet,
Hidden_Applications, Notifications
Hidden Applications Context info, Settings, Notifications, All applications which are not on the desktop.
Users Environmental Space (House, Indoor, Outdoor, College Campus, Library, Class Room , Bus Stand, Railway Station, etc.,)
S1 S2 S3 ----
-
---- ---- Sn
S1_Data S2_Data S3_Data Sn_Data
Fuzzification
Context –Generation
Input / Sensor Layer
Fuzzif-ication
Layer
Context
Layer
Actions Recommendation
Actions Repository
A1 A2 A3 A4 An A(n-1)
Actions
(Volume Adjustment, Call Settings, Incoming Call Settings, Desk Top Applications -)
Recom-mendati-on
Layer
Output
Layer
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
56
Rule1 Rule2 Rule3 Rule4 - - - - - Rule20
CONDITIONS
Fall 0 0 0 0 1
Temp Hot Warm Warm Norm Very Hot
Time Type WorkingHours WorkingHours WorkingHours FreeHours FreeHours
Time PM AM AM AM AM
Day WorkingDay WorkingDay WorkingDay WorkingDay WorkingDay
Location CCAdmin CClib CCCR Outdoor House
Incoming Call HighPrioirity HighPriority HighPriority LowPriority Undefined
ACTIONS --
Profile Silent Silent Silent General General
Volume Settings Silent Silent Silent High High
Call Settings Answer Answer the Call Answer the Call Call Ringing Answer the Call
Messaging Yes No No Yes Yes
Contacts True True False True True
Log Yes Yes No No Yes
Settings No No No No No
Gallery No No No No No
Media No No No No No
AllHiddenApplications Yes Yes Yes Yes Yes
MyEntertainment Yes No No No No
MyShopping No No No No No
Notifications Yes Yes Yes Yes Yes
Fig 10: Rule Base for context aware mobile
RULE [no] IF THEN
RULE [2] [Fall] : "0" and [Temp] : "Warm" and [Time Type] : "Working Hours" and [Time] : "AM" and [Day] : "Working Day" and [Location]: "CCLib" and [Incoming Call] : “High Priority “
[Profile] = "Silent" and [Volume Settings] = "Silent" and [Call Settings] = "Answer the Call" and [Messaging] = "yes" and [Contacts] = "TRue" and [Log] = "Yes" and [Settings] = "no" and [Gallery] = "no" and [Media] = "no" and [AllHiddenApplications] = "yes" and [MyEntertainment] = "no" and [MyShopping] = "no" and [MyNotifications] = "yes" and [MyCollege] = "yes" and [MyLib] = "yes" and [Organizer] = "yes" and [Web] = "yes"
RULE [3] [Fall] : "0" and [Temp] : "Warm" and [Time Type] : "Working Hours" and [Time] : "AM" and [Day] : "Working Day" and [Location] : "CCCR and [Incoming Call] : “High Priority”
[Profile] = "Silent" and [Volume Settings] = "Silent" and [Call Settings] = "Answer the Call" and [Messaging] = "no" and [Contacts] = "False" and [Log] = "No" and [Settings] = "no" and [Gallery] = "no" and [Media] = "no" and [AllHiddenApplications] = "yes" and [MyEntertainment] = "no" and [MyShopping] = "no" and [MyNotifications] = "yes" and [MyCollege] = "no" and [MyLib] = "yes" and [Organizer] = "yes" and [Web] = "yes"
RULE [5]
[Fall] : "0" and [Temp] : "Norm" and [Time Type] : "Free Hours" and [Time] : "EAM" and [Day] : "Working Day" and [Location] : "House" and [Incoming Call] : “Low Priority”
[Profile] = "General" and [Volume Settings] = "High" and [Call Settings] = "Call Ringing" and [Messaging] = "yes" and [Contacts] = "TRue" and [Log] = "No" and [Settings] = "no" and [Gallery] = "yes" and [Media] = "no" and [AllHiddenApplications] = "yes" and [MyEntertainment] = "yes" and [MyShopping] = "no" and [MyNotifications] = "yes" and [MyCollege] = "yes" and [MyLib] = "no" and [Organizer] = "yes" and [Web] = "yes"
RULE [10]
[Fall] : "1" and [Temp] : "Very Cold" and [Time Type] : "Free Hours" and [Time] : "EAM" and [Day] : "Working Day" and [Location] : "OutDoor" and [Incoming Call] : “Low Priority”
[Profile] = "General" and [Volume Settings] = "High" and [Call Settings] = "Answer the Call" and [Messaging] = "yes" and [Contacts] = "TRue" and [Log] = "No" and [Settings] = "no" and [Gallery] = "no" and [Media] = "no" and [AllHiddenApplications] = "no" and [MyEntertainment] = "no" and [MyShopping] = "no" and [MyNotifications] = "yes" and [MyCollege] = "no" and [MyLib] = "no" and [Organizer] = "no" and [Web] = "no"
RULE [18] If [Fall] : "1" and [Temp] : "Very Hot" and [Time Type] : "Free Hours" and [Time] : "AM" and [Day] : "Working Day" and [Location] : "House" and [Incoming Call] : “Undefined”
Then [Profile] = "General" and [Volume Settings] = "High" and [Call
Settings] = "Answer the Call" and [Messaging] = "yes" and [Contacts]
= "TRue" and [Log] = "Yes" and [Settings] = "no" and [Gallery] = "no"
and [Media] = "no" and [AllHiddenApplications] = "yes" and
[MyEntertainment] = "no" and [MyShopping] = "no" and
[MyNotifications] = "yes" and [MyCollege] = "no" and [MyLib] = "no"
and [Organizer] = "no" and [Web] = "no"
Fig 11: Example knowledge based Rules extracted from the Rule base
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
57
For selecting the appropriate rule the Maximum to minimum attributes matching algorithm described below (Fig12) is utilized. The
algorithm initially searches for rule where all the condition attributes and context attributes will be same .If not found the algorithm
searches the rule wherein the n-1 attributes(Deleting the least priority attribute).At each time the algorithm searches for best rule by
deleting the least priority attribute and thus maintaining the best fitness degree (>=0.2)
Fig12: Maximum to Minimum attributes matching algorithm (MMAM )
5. EXPERIMENTAL AND RESULT
ANALYSIS The Experiment was conducted under two phases: Phase1:
Using Simulator Phase2: Using Experimental Set with
devices and sensors.
Phase 1: In this phase the proposed system was tested
using simulator developed inJ2ME .Following is one of the
several scenarios tested using simulator.
Scenario (User is in Central Library): In this scenario the
user was allowed to carry her mobile to Library Hall. A soon
as the user enters into the library hall Mobile switches into the
Library Context from the previous context (Fig13.1 and
Algorithm3: Maximum to Minimum attributes matching algorithm (MMAM ) Input : 1) Current Context Attributes
2) Rule Base (R,C,A) 3) The attributes are sorted in descending order of priority Priority(C1)> Priority(C2) Priority(C2) > Priority (C3) ------------------------------- ------------------------------ Priority(Cn-1) > Priority(Cn)
Output : 1) Matched Rule with Condition attributes and Action Attributes.
2) Fitness degree of the Rule.
Algorithm : for(Rule#=1; Rule#<=m; Rule#++) {If (∃Rule#|∀𝐢=𝟏
𝐧 (Rule#Ci==CCj)) Then return { Rule# ; //Rule ID ∀𝐢=𝟏
𝐧−𝟏 Rule#Ci ; // Context Conditions Values
∀𝐣=𝟏𝐩
Rule#A ; // Action Values
FD(Rule#); // Dependency degree/ Fitness Degree } Else If (∃Rule#|∀𝐢=𝟏
𝐧−𝟏(Rule#Ci==CCj)) Then return { Rule# ; ∀𝐢=𝟏
𝐧−𝟏 Rule#Ci ;
∀𝐣=𝟏𝐩
Rule#A ;
FD(Rule#); } ---------------------------------------- ----------------------------------------- Else If (∃Rule#|∀𝐢=𝟏
𝐧−𝟏(Rule#Ci==CCj)) Then return { Rule# ; ∀𝐢=𝟏
𝐧−𝟏 Rule#Ci ;
∀𝐣=𝟏𝐩
Rule#A ;
FD(Rule#); } }
Algorithm Complexity Best Case Complexity: O(1) Worst Case Complexity: O(m(n!))
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
58
Fig13.2). The Mobile adapts the settings and services as per
the requirement of the situation (Fig13.3). The user was
allowed to use the library service (say book search) which is
illustrated in the figures (Fig13.4 to Fig13.9).
Fig 13: Different states of Context aware Mobile in Library.
Fig13.1 Fig13.2 Fig13.3 Fig13.4 Fig13.5
Fig13.6 Fig13.7 Fig13.8 Fig13.9
Fig13.1: Acquiring the current context, Fig13.2: Display of current context Info, Fig13.3: Details of settings and services for the current context Fig13.4: Library Services Provided by the device Fig13.5: Search Options Fig13.6, Fig13.7: Book Search Fig13.8: Details about the title entered Fig13.9: Details about the book selected
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
59
Phase 2: The experimental set up for the realization of the
concept is as shown in the figure14. The system consists of 5
major components which are described in the following
sections.
Fig14 : Experimental set up of context aware mobile
1. Context aware Mobile (CAM ): is enabled with
Bluetooth, GPRS and supporting Symbian Operating System ,
which supports J2ME. On the mobile side, the application was
developed and implemented using Java 2 micro edition
(J2ME) .The J2ME application runs on any Symbian OS
based phone. Note that the application can only operate on a
Bluetooth enabled mobile phone
2. Sensor Board: The sensor board was designed specifically
for the concept demonstration .Sensors like temperature
sensor and accelerometer was embedded in the sensor
(however the latest mobile will have inbuilt temperature and
accelerometer) .Bluetooth transceiver was used to establish
the communication between sensor board and Mobile. Also
some of the sensors like Humidity and Noise sensor was not
used because of their role in college campus in not
considerable.
3. Database: The server uses a MySQL database. MySQL is
an open source relational database management system which
uses Structured Query Language (SQL). MySQL was chosen
because of its reliability, speed and flexibility. The server
receives requests from the application program. The request
can be either to register a new user, update user information,
or locate an existing user. The server tokenizes the user
requests, and issues the appropriate SQL statement to perform
the required action.
4. Server : The Netbeans IDE6.9.1 was used to develop
server . In addition Wamp server was installed in a system so
that database can be accessible for the Server.
5. Client :The client application was developed in J2ME and
installed in Bluetooth Enabled Cell.
Experiment 1: This experiment was conducted to determine
the Bayesian Priority assignment accuracy (BPAA) of
proposed algorithm in classifying the Incoming call as High
priority call, Low Priority Call and Unknown Call. Three
different user’s one month mobile usage history database with
sample size of 1000, 500 and 100 respectively was used as a
training database for the algorithm. As illustrated in the figure
15 below the performance was better for database with more
number of samples per month.
Fig15 : Priority Assignment Accuracy of Bayesian Network
Experiment 2: This experiment was conducted to determine
the amount of time required to recognize the different location
with respect to Bluetooth access point signal strength. For
each location three different trials were made with different
signal strength .As it is illustrated in the Fig the response time
was less (50-100 milli seconds) for excellent signal strength
as compared to low signal strength (1000-1500milliseconds).
00.10.20.30.40.50.60.70.80.9
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
BPAA
DAY #
Mobile with 1000 Sample data set per month
Mobile with 500 Sample data set per month
Mobile with 100 Sample data set per month
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
60
Fig16:Time to recognize the different location based Bluetooth Signal strength
Experiment 3: This experiment was conducted to determine
the average service popup time with respect to different
location .The system took more time when user moved to
outdoor as compared to that when the user is in class room.
This is due to the fact that the number of services to be
invoked in outdoor situation is more as compared with the
class room situation.
Fig17: Service popup time
Experiment 4: In this experiment, we aimed to evaluate the
Precision of recommending the appropriate services and
settings based on the current context of the user. About 20
users (students and faculty) were allowed to use the mobile in
college campus in different locations. The result is highly
subjective. Most users agree that the precession rate of 76-82
percent is useful.
Precision = 𝑁𝑢𝑠𝑒𝑓𝑢𝑙
𝑁𝑡𝑜𝑡𝑎𝑙 ----(8)
Where Nuseful = No of services (Expected by the user ⋂ Recommended by the system) Ntotal = No of Services (Recommended by the system ⋃ Expected by the user)
Fig18 : Precision of service recommendation in different context
0
500
1000
1500
Recogniton time in milli
seconds
Signal Strength
trail1
trial2
trial3
0100200300400500600700800900
Service PoPupTime(ms)
Location
Trial1
Trial
Trial3
Trial4
Trail5
0
0.2
0.4
0.6
0.8
1
Precision
User
OutDoor
Class Room
Meeting Room
Library
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
61
Experiment 5: In this experiment we aimed to evaluate the
overall performance of the recommendation system in terms
of time.
Table6 : Overall Performance of Recommendation
System
Experiment 6: To get the proper assessment of our
application, we used the cognitive walkthrough strategy. We
did a survey on a group of 20 people on the usability and
usefulness of our application in the college campus. First we
showed the prototype application demo. The distribution of
the participants is as follows: 8 undergraduate students, 8 post
graduate students and 4 faculty members. We handed
questions about the application over to each participant and
requested them to answer them on a scale of 1 to 5of
satisfaction level. The questionnaire for the survey is given
below:
Overall, how would you rate the following services provided
by the context aware mobile in terms of satisfaction level? (1
= Below Average, 2= Average 3= Good 4 = Very Good 5 =
Excellent)
1. Location based Services
2. Time based services
3. Incoming Call based Services
4. Fall based services
5. Temperature based services
6. System Performance
Fig19: Time to recognize the different location based Bluetooth Signal strength
From the graph, it is evident that participants were
enthusiastic about the application and its usability.
6. CONCLUSION The paper presents the design and implementation of the
proposed context aware mobile. The service adaptation
process is formulated using artificial intelligence techniques
like fuzzy logic, rule based reasoning and Bayesian networks.
Bayesian Network to classify the incoming call (high priority
call, low priority call and unknown calls), fuzzy linguistic
variables and membership degrees to define the context
situations, the rules for adopting the policies of implementing
a service, fitness degree computation and service
recommendation. The intelligent context aware mobile is
tested for library scenario to exemplify the proposed service
recommendation engine and demonstrate its effectiveness.
Most system users we interviewed agree that service and
settings recommending precession of average 79 percent is
acceptable. In particular, many users feel that system
performance should be improved and more number of
meaningful social services should be replaced in the place of
unnecessary services. Our feature work includes improvement
in the rule matching by applying rough set theory, betterment
of services considering the user’s personal and social
activities in addition to physiological as well as addressing
privacy and security issues.
7. REFERENCES [1] Capra, L.; Emmerich,W.; Mascolo,C.;"CARISMA:
context-aware reflective middleware system for mobile
applications"; IEEE Transactions on Software
Engineering, Volume:29, Issue:10, pp.929 – 945,
Oct.2003
[2] Baochun Li; Nahrstedt, K.;"A control-based middleware
framework for quality-of-service adaptations" IEEE
Journal on Selected Areas in Communications, Volume:
17 , Issue: 9 pp. 1632 – 1650 , Sept. 1999
[3] Jiannong Cao1, Na Xing1,2, Alvin T.S Chan1, Yulin
Feng2 , Beihong Jin2 ,‖ Service Adaptation Using Fuzzy
Theory in Context-aware MobileComputing
Middleware‖ , Proceedings of the 11th IEEE
International Conference on Embedded and Real-Time
Computing Systems and Applications (RTCSA’05)
[4] Choonsung Shin and Woontack Woo, Member, IEEE
―Socially Aware TV Program Recommender for
0102030405060708090
Satisfaction Level
Service Type
Excellent
Very Good
Good
Average
Below Average
Tasks Time
Context Discovery 1s – 3s
Rule Matching 10 ms - 1s
Settings adjustment 1s – 2s
Services invoking 100ms -1s
Total 2.11s to 7s
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
62
Multiple Viewers‖ Downloaded on August 9, 2009 at
01:42 from IEEEXplore..
[5] Doug Williams, Marian F Ursu,, Ian Kegel , ―An
Emergent Role for TV in Social Communication‖.
[6] Dr Daniel Chandler , Matthew Ruckwood, Mediated
Communication ― Will Interactive Television Change
the Relationship Between the Viewer and the Television
Set? ― , MC10120 10/01/2005.
[7] Doug Riecken , ―Perosonalized views of personalization
,‖ Comm.ACM ,vol 43,n08,2000.
[8] Muhammad Ashad Kabir ,Jun Han and Alan Colman ,
―Modeling and Coordianting Social Interactions in
Pervasive Environments ―,
[9] N.Datia ,J Moura-Pires .M.Cardoso ,H.Pita , ―Temporal
Patterns o fTV Watching For Portugues Viewers ―.
[10] Joe Jeffrey ,‖High Fidelity Mathematical Models of
Social Systems ―, Northen Illinois Unversity.
[11] Ray van Brandenburg , Master Thesis Faculty of
Electrical Engineering, Mathematics and Computer
Science, Design and Analysis of Communication
Systems (DACS),University of Twente‖ Towards multi-
user personalized TV services,introducing combined
RFID Digest authentication‖ ,Dec2007,Master thesis
[12] Paul Resnick ,Neophytos Iacovou ,Mitesh Suchak Peter
Bergstorm ,and John Riedli ,‖Grouplens : An open
architecture for collaborative filtering of netnews , ― in
Proc.ACM Conf.Computer Supported Cooperative Work
,Chape Hill ,NC,USA ,1994,pp.175 -186.
[13] Annie Chen ,―Context aware collaborative filtering
system : Predicting the user’s preference in the
ubiquitous computing environment ,‖ in Location and
Context Awareness,pp.244-253 .Springer ,2005
[14] Manos Papagelis, Dimitris Plexousakis,‖ Qualitative
analysis of user-based and item-based prediction
algorithms for recommendation agent ―,Engineering
Applications of Artificial Intelligence 18 (2005) 781–789
―www.elsevier.com/locate/engappai
[15] Manos Papagelis1, 2, Ioannis Rousidis2, Dimitris
Plexousakis1, 2, Elias Theoharopoulos3 , ―Incremental
Collaborative Filtering for Highly- Scalable
Recommendation Algorithms‖, Work conducted at ICS-
FORTH
[16] SongJie Gong , ―A Collaborative Filtering
Recommendation Algorithm Based on User Clustering
and Item Clustering ―,JOURNAL OF SOFTWARE,
VOL. 5, NO. 7, JULY 2010, © 2010 ACADEMY
PUBLISHER,doi:10.4304/jsw.5.7.745-752
[17] Kyusik Park, Jongmoo Choi, and Donghee Lee , ―A
Single-Scaled Hybrid Filtering Method for IPTV
Program Recommendation ― , INTERNATIONAL
JOURNAL OF CIRCUITS, SYSTEMS AND SIGNAL
PROCESSING , Issue 4, Volume 4, 2010
[18] J.J. Sandvig and Bamshad Mobasher and Robin Burke ,
―A Survey of Collaborative Recommendation and the
Robustness of Model-Based Algorithms ― , Bulletin of
the IEEE Computer Society Technical Committee on
Data Engineering , Copyright 2008 IEEE.
[19] Michael Hahsler , ―Developing and Testing Top-N
Recommendation Algorithms for 0-1 Data using
recommenderlab ― , February 27, 2011 ,
[20] Amancio Bouza, Gerald Reif, Abraham Bernstein ,
―Probabilistic Partial User Model Similarity for
Collaborative Filtering‖.
[21] SongJie Gong , ― A Collaborative Filtering
Recommendation Algorithm Based on User Clustering
and Item Clustering‖,Journal Of Software ,Vol .5.No
7,July2010.
[22] Manos Papgelis ,Dimitris Plexousakis , ―Qualitative
analysis of user based and item based prediction
algorithms for recommendation agents‖,Engineering
Applications of Artificial Intelligence 18(2005) 781-789
,www.elsevier.com/locate/engappai
[23] Daniel Siewiorek, Asim Smailagic, Junichi Furukawa,
Neema Moraveji, Kathryn Reiger, and Jeremy Shaffer ,‖
―SenSay: A Context-Aware Mobile Phone‖, Proceedings
of the Seventh IEEE International Symposium on
Wearable Computers (ISWC’03)
[24] Miika Valtonen, Antti-Matti Vainio, Jukka Vanhala ,
―Proactive and Adaptive Fuzzy Profile Controlfor
Mobile Phones‖ , 978-1-4244-3304-9/09/$25.00 ©2009
IEEE
[25] Lakshmish Ramaswamy1, Deepak P2, Ramana
Polavarapu2 Kutila Gunasekera3, Dinesh Garg4, Karthik
Visweswariah2, Shivkumar Kalyanaraman , ‖ CAESAR:
A Context-Aware, Social Recommender System for
Low-End Mobile Devices‖, 2009 Tenth International
Conference on Mobile Data Management: Systems,
Services and Middleware, 978-0-7695-3650-7/09 $25.00
© 2009 IEEE DOI 10.1109/MDM.2009.66
[26] Jonna Häkkilä, Jani Mäntyjärvi,‖ Collaboration in
Context-Aware Mobile Phone Applications‖,
Proceedings of the 38th Hawaii International Conference
on System Sciences – 2005, 0-7695-2268-8/05/$20.00
(C) 2005 IEEE
[27] Seyed Hadi Mirisaee, Abdullah Mohd Zin,‖ A
Framework for Context-Aware University Mobile
Organizer‖, 2009 International Conference on Future
Computer and Communication, 978-0-7695-3591-3/09,
DOI 10.1109/ICFCC.2009.84.
[28] Boris Moltchanov1, Michael Knappmeyer2, Omri
Fuchs3, Elio Paschetta1,‖ Context Management and
Reasoning for Adaptive Service Provisioning ―,9781-
4244-3941-6/09/$25.00 ©2009 IEEE
[29] Courtney McTavish, Suresh Sankaranarayanan,‖
Intelligent Agent based Hotel Search &Booking
System‖, 978-1-4244-6875-1/10/$26.00 ©2010 IEEE
[30] Ryan Anthony Brown , Dr. Suresh Sankaranarayanan,‖
Intelligent Agent based Mobile Shopper‖, 978-1-4244-
3474-9/09/$25.00 ©2009 IEEE
[31] Ren-Hung Hwang, Shang-Yi Tsai, and Chiung-Ying
Wang,‖ UbiPhone: Human-Centered Ubiquitous Phone
System‖, Published by the IEEE CS n 1536-
1268/09/$25.00 © 2009 IEEE,pervasive computing
journal ,, April–June 2009 PERVASIVE
computing,pp40-47
[32] Symbian OS. Available at http://www.symbian.org.
International Journal of Computer Applications (0975 – 8887)
Volume 40– No.3, February 2012
63
[33] Paramvir Bahl and Venkata N. Padamanabhan,
‖RADAR: An In-Building RF-based User Location and
Tracking System‖, Microsoft Research, 1999
[34] F. Barcelo-Arroyo et al, ―Indoor Location for Safety
Applications using Wireless Networks,‖ in Proc of the
1st ERCIM Workshop on Mobility, Portugal, 2007
[35] S. Thongthammachart and H. Olesen, ―Bluetooth
Enables indoor Mobile Location Services,‖ in Proc. of
the 57th IEEESemiannual Vehicular Technology
Conference, 2003.
[36] Sina Zulkernain, Praveen Madiraju1, Sheikh Iqbal
Ahamed, Karl Stamm (Marquette University,
Milwaukee, Wisconsin, USA ,‖ A Mobile Intelligent
Interruption Management System ― , Journal of Universal
Computer Science, vol. 16, no. 15 (2010), 2060-
2080submitted: 31/1/10, accepted: 28/7/10, appeared:
1/8/10 © J.UCS
[37] Han-Saem Park, Keunhyun Oh, and Sung-Bae Cho ,
―Bayesian Network-Based High-Level Context
Recognition for Mobile Context Sharing in Cyber-
Physical System‖ , Department of Computer Science,
Yonsei University, 262 Seongsanno, Sudaemoon-gu,
Seoul 120-749, Republic of Korea, Hindawi Publishing
Corporation ,International Journal of Distributed Sensor
Networks Volume 2011, Article ID 650387, 10 pages
doi:10.1155/2011/650387.
8. AUTHORS PROFILE
Thyagaraju.GS received the M.Tech Degree in Computer
Science And Technology From University Of Mysore ,India
in 2002.He has got ten years of experience in academics
,fours years of Research Experience . He is a member of
IETE.He has guided many students at UG and PG level.He is
pursuing Ph.D in Computer Science Engineering. His
Research Interests are Context Aware Computing in
Ubiquitous and Intelligent Systems He is now working as a
Senior Lecturer in Dept Of CSE, SDM College Of
Engineering, Karnataka, Dharwad.
Dr. Umakant Kulkarni obtained his BE Degree from
Karnataka University, Dharwad in the year 1989, ME Degree
from PSG College of Technology, Coimbatore in the year
1991 and PhD from Shivaji University, Kolhapur in the year
2007. He has published many papers at International Journal
and IEEE conferences in the areas of Pervasive and
Ubiquitous Computing, Distributed Data Mining, Agents
Technology and Autonomic Computing. He is Member of
IETE and ISTE. He served as Head of Department and Chief
Nodal Officer- TEQIP a World Bank funded project. He has
guided many students at PG level and five research scholars
are pursuing their PhDs. Currently he is serving as professor
in the Department of Computer Science & Engineering, SDM
College of Engineering & Technology, Dharwad, Karnataka
State, India.