Design and Implementation of User Context aware ......System for Low-End Mobile Devices. Here the recommendation is based on the factors like Social affinity computation from call

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.



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:



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



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



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


Restricted Time slot announcement to

caller

Diverting call to other number

specified by the owner

Unknown Restricted Time– Accessible


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



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

}



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))


If (μexcellent (l3)> μexcellent (l2) && μexcellent (l3) > μexcellent (l2))


}

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


If (μgood (l3)> μgood (l2) && μgood (l3) > μgood (l2)) then the


}

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


If (μfair (l2)> μfair (l1) && μfair (l2) > μfair (l3)) then the


If (μfair (l3)> μfair (l2) && μfair (l3) > μfair (l2)) then the


}

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


If (μlow r (l2)> μlow r (l1) && μlow (l2) > μlow (l3)) then the


If (μlow (l3)> μlow (l2) && μlow (l3) > μlow (l2)) then the


}

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



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.



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



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



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!))



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



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


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





60

Fig16:Time to recognize the different location based Bluetooth Signal strength


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



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



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.



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.

Design and Implementation of User Context aware ......System for Low-End Mobile Devices. Here the recommendation is based on the factors like Social affinity computation from call

Documents