USER CENTRIC NETWORK SELECTION IN WIRELESS HETNETS

International Journal of Wireless & Mobile Networks (IJWMN) Vol. 5, No. 3, June 2013
DOI : 10.5121/ijwmn.2013.5304 45
USER CENTRIC NETWORK SELECTION IN
WIRELESS HETNETS
L.Nithyanandan1
, V.Bharathi2
and P.Prabhavathi3
1,2,3
Department of Electronics and Communication Engineering,
Pondicherry Engineering College, Puducherry, India
nithi@pec.edu
bharathime@rediffmail.com
prabhatech@pec.edu
ABSTRACT
The future generation wireless networks are expected to beheterogeneous networksconsisting of UMTS,
WLAN, WiMAX, LTE etc. A heterogeneous network provides users with different data rate and Quality of
Service (QoS). Users of future mobile networks will be able to choose from different radio access
technologies. These networks vary widely in service capabilities such as coverage area, bandwidth and
error characteristics. Network selection is a challenging task in heterogeneous networks and will
influence the performance metrics of importance for both service provider and subscriber.This paper
analysesuser centric network selection based on QoE (Quality of Experience) which include both
technical and economical aspects of the user. WLAN-WiMAX-UMTS networks are integrated and the
network selection for the integrated network is performed using game theory based network selection
algorithm.
KEYWORDS
UMTS, WLAN, WiMAX, QoS, QoE and Heterogeneous network
1. INTRODUCTION
There is a rapid development in the wireless communication nowadays because there is an
increase in number of mobile and internet users. One such new technology is the heterogeneous
network which refers to the integration of different Radio Access Technology (RAT) such as
WLAN, WiMAX, UMTS etc. These wireless networks differ in its bandwidth, coverage area,
data rate, mobility, technology etc. WLAN network is basically 802.11 standardwhich provides
less coverage and high data rates (54Mbps). WiMAX is 802.16 standard with data rates up to
75Mbps and coverage nearly 30Km. UMTS is 3GPP standard which offers less data rate
(2Mbps) and more coverage.Heterogeneous network provides benefits such as seamless
connectivity, ubiquitous availability of multimedia services and effective utilization of available
bandwidth to meet user requirements. In a wireless environment user’s mobility, characteristics
and availability of a network will change in time. So, the mobility management process
considers the dynamic reselection of network as its major task.Network selection is necessary
when users want to migrate between heterogeneous networks. Hence multi mode mobile
terminal is used to select a network from the converged network.
Network selection is a quite difficult process in wireless environment since network condition
varies randomly based on variation in user demands, random activity of users and vagueness of
system parameters. In case of network centric approach, base station consists of a centralized
controller which admits the user to a particular network and allocates bandwidth to the user to
maximize network utility. The user’s needs and optimum results are not achieved through this
approach. Hence, to have a better performance, network selection can be done using user centric
approach. In user centric approach, game theory based network selection algorithms are

46
implemented at the user terminal. Future heterogeneous network will be consisting of different
RANs at same place and network conditions vary according to channel condition,random
requirement of users etc. In this realistic environment network selection using Game theory
produces optimum result for users.
The rest of this paper is structured as follows. Section 2 gives the literature survey. Network
selection process is explained in section 3. Section 4 describes the network selection decision
making. Mapping of game theory to network selection is given in section 5. Section 6 explains
the simulation results and finally section 7 concludes the paper.
2. LITERATURE SURVEY
J.Park, et al., [1] has proposed the user subscription dynamics, revenue maximization and
equilibrium characteristics in two different markets (i.e., monopoly and duopoly). A fuzzy-
logic-based multiple-criteria decision making system to perform access network selection in a
heterogeneous network environment is described in [2]. A simple policy enabled handoff across
heterogeneous wireless networks is presented in [3], which allows users to express policies
which selects best wireless network at any moment, and that makes trade-offs among network
characteristics and dynamics such as cost, performance and power consumption.A network-
selectionstrategy that only considers mobile users power consumption was introduced by
M.Namet.al [4]. Another network selection algorithm was proposed by Q.Y.Songet.al [5] in
which Analytical Hierarchy Process (AHP) and the Grey Relational Analysis (GRA) were
discussed. The AHP algorithm divides the complex network-selection problem into a number of
decision factors, and the optimal solution can be found by integrating the relative dominance
among these factors. The GRA was also proposed for selecting the best network for a mobile
user. This mathematical network selection model is suitable for static environment. Later in
2012 Manzoor Ahmed Khan et.al,[6] introduce da dynamic and uncertain environment for
network selection where the users and operators have only a numerical value of their own
payoffs as information, and constructed various heterogeneous COmbined fully Distributed
Payoff and Strategy Reinforcement Learning (CODIPAS-RL) in which the users try to learn
their own optimal payoff and their optimal strategy simultaneously. In [6] Bush Mosteller based
CODIPAS-RL for interworking two networks is discussed. In this paper Bush Mosteller and
less iterative Boltzmann-Gibbs CODIPAS-RL are discussed and it is implemented in three
network interworked architecture simulated using OPNET.
3. NETWORK SELECTION PROCESS
Network selection process is defined as the process of selecting a specific network from the
converged network which meets the user’s requirements. To introduce the concept of Always
Best Connected (ABC) anytime and anywhere, network selection mechanisms are facilitated in
heterogeneous networks. Different Radio Access Networks (RAN) differs in coverage, data rate,
QoS, pricing scheme etc., hence it is difficult for a user to select a particular network [7]. From
the user’s perspective, the variety of portable devices (such as smart phones, net books, or
laptops) with support for multiple radio network interfaces, enable the option of connecting to
the Internet anywhere and anytime. Users are able to freely migrate from one Radio Access
Technology (RAT) to another or from one service provider to another. A network selection
decision is made at call setup and subsequently the decision is re-made in the case of a handover
trigger. The major phases in network selection process are Monitoring, Network selection
and Call setup. In monitoring phase the mobile terminal monitors the network conditions, lists
the available RAN’s and predicts the characteristics of each RAN. By using these monitored
data the mobile terminal triggers a Hand Off (HO) decision.

47
In network selection phase network selection process is initiated either by automatic
trigger for HO for existing call or by a request for a new connection [8]. The best network is
decided based on decision criteria provided by the device, application and monitoring process.
When the target network is selected the call is set up on the candidate network. But
conventionally this decision was made by network operators mainly based on Received Signal
Strength. When the target network is selected, the connection set up on the target candidate
network is executed. But in case of an existing network connection, HO is executed and the
original connection is removed and the call is re-routed to the new connection. If first choice
network is unavailable, then the next listed candidate is chosen as target network. Connection
setup and connection release are handled by Mobile Internet Protocol Version 6.
4. NETWORK SELECTION DECISION MAKING
Every decision making mechanism requires essential and relevant input information in
order to choose the best network [9]. The decision criteria that may be used in the network
selection process are network metrics, device related information, application requirements and
user preferences. Network metrics includes information about the technical characteristics or
performance of the access networks such as technology type, coverage, security, pricing
scheme, monetary cost, available bandwidth, network load, latency, Received Signal Strength,
blocking probability, network connection time, etc. Device related information refers to
information about the end-users’ terminal device characteristics like supported interfaces,
mobility support, capacity, capability, screen-size and resolution, location-based information,
remaining battery power, etc. Application requirement refers to information about the
requirements (minimum and maximum thresholds) needed in order to provide a certain service
to the end user: delay, jitter, packet loss required throughput, Bit Error Rate, etc. User
preferences comprise information related to the end-users satisfaction such as budget
(willingness to pay), service quality expectations, energy conservation needs, etc. The user
preferences play an important role in the decision mechanism and they may be used to weight
the other parameters involved.
5. MAPPING OF GAME THEORY TO NETWORK SELECTION
Game theory is a mathematical tool used in understanding and modelling competitive situations.
In the wireless environment, game theory has been used in order to solve many distributed
power control, resource management and dynamic pricing related problems[10].The main
components of a game are: the set of players, the set of actions, and the set of payoffs. The
players seek to maximize their payoffs by choosing strategies that deploy actions depending on
the available information at a certain moment. Each player chooses strategies which can
maximize their payoff. The combination of best strategies for each player is known as
equilibrium. The mapping of game theory to network selection is given in Table 1.
Table 1. Mapping of game theory to network selection
Game Component Network Selection Environment Correspondent
Players The agents who are playing the game : Users or networks
Strategies A plan of actions to be taken by the player during the game: available/
requested bandwidth, subscription plan, offered prices, available APs. etc.
Payoffs The motivation of players represented by profit and estimated using utility
functions based on various parameters: monetary cost, quality, network load,
QoS, etc.
Resources The resources for which the players involved in the game are competing:
bandwidth, power, etc.

48
6. NETWORK SELECTION ALGORITHMS
There are different strategies in network selection such as select the economic RAN, random
network selection, select the preferred operator’s network. But these strategies do not provide
optimum results in a dynamic wireless network. An intelligent approach for network selection
provides optimum results. Hence, to get optimum results network selection algorithm based on
game theory was introduced[11]. COmbined and fully Distributed Payoff and Strategy
Reinforcement Learning (CODIPAS-RL) were introduced in [6]. In this paper UMTS-WIMAX-
WLAN interworked environment is simulated. Bush-Mosteller based and Boltzmann-Gibbs
based CODIPAS-RL is implemented for UMTS-WIMAX-WLAN interworked environment.
6.1. Bush Mosteller based CODIPAS-RL
Bush Mosteller based CODIPAS-RL is a stochastic model of reinforcement learning where
users decide which action to take stochastically: each user’s strategy is defined by the
probability of undertaking each of the two actions available to them. After every user has
selected an action according to their probability, every user receives the corresponding utility
and revises their strategy [6, 11, 12]. The action of each user is determined based on (1)
, = , +  , (1 − , ) (1)
Where λj is the user j’s learning rate (0 <λ<1) and Sj,t denotes stimulus of user which is given in
(2) and its values ranges from [-1, 1]. When stimulus magnitude or learning rate increases, then
the change in probability also increase.
, = ,
| ( ) |
(2)
where , denotes the perceived utility at time t of player j and is an aspiration level of player
j. The payoff estimation for the experimented actions by the users is given in the eq. (3).
û , = û , + ( ) ∗ ( , − û , (3)
Bush Mosteller based CODIPAS-RL considers present action of user j as well as the actions of
other users. So it requires more memory and processing time is also high. Hence the number of
iterations or time required for network selection process is more.
6.2. Boltzmann-Gibbs based CODIPAS-RL
Since Bush Mosteller based CODIPAS-RL takes more number of iterations for network
selection, Boltzmann-Gibbs based CODIPAS-RL was used because it considers the previous
action and present action of user j alone and requires less memory. So, less number of iterations
only are required in this algorithm [13, 14].The action of the user is based on Boltzmann
distribution which is given in [6,11] as
,(û , ) =
 û ,
∑
 û , `
` 
,  ,  (4)

49
Where

denotes the rationality level of the user j. û , denotes the estimated payoff of user j
at time t and denotes the action of user j at time t. Based on action of user in (4), the payoff
for each user is calculated using eqn. (5).
û , = û , + ( ) ∗ ( , − û , (5)
The steps for selecting a network are given as follows:
Step 1: Assume users are initially connected to a network and calculate the payoff in current
network
Step 2: Check for the alternate networks where users can connect
Step 3: Determine the payoff and action of the users based on Boltzmann-Gibbs CODIPAS-RL
Step 4: Determine Network Selection Probability (NSP) based on the action of users
7. SIMULATION RESULTS
For effective network selection process three networks such as WLAN, WiMAX and UMTS are
integrated using OPNET modeller [15] and the network selection process is carried out for users
using MATLAB.Fig.1 shows the architecture of WLAN-UMTS-WiMAX integration. The
architecture is based on hybrid coupling in which the data packets between WLAN and
WiMAX reaches the UMTS network directly through Gateway GPRS Support Node (GGSN)
or through the IP core network based on data type(real time or non real time). The IEEE
(WLAN, WiMAX) and 3GPP (UMTS) networks are integrated using IMS (IP Multimedia
Subsystem) which is based on SIP (Session Initiation Protocol) and terminal mobility is
provided using MIPV6. SIP is an application layer protocol which is used to establish, maintain
and terminate the session. The basic components of SIP are P-CSCF (Proxy Call Session
Control Function), I-CSCF (Interrogating Call Session Control Function) S-CSCF (Serving Call
Session Control Function) and Home Subscriber Station (HSS). Within the coverage of UMTS
two networks (WLAN and WiMAX) are present and within the coverage of WiMAX, WLAN is
present. The architecture consists of three users with three different applications such as video,
voice and web browsing.
Figs.2, 3 and 4 show the throughput, delay and load in WLAN network. The maximum
throughput of the network is found to be 0.2 Mbps this is because only less number of users is
considered. Throughput of the network increases as the number of users increase.

50
Fig.1 WLAN-WiMAX-UMTS architecture
Fig.2 Throughput in WLAN

51
Figs.5, 6 and 7 show the throughput, delay and load in UMTS network. Since the number of
users considered in the architecture is minimum, the throughput obtained is less compared to the
actual network throughput. The network throughput decreases based on loading conditions also.
The delay in UMTS is less than WLAN network because UMTS is a circuit switched network
whereas WLAN is a packet switched network.
Fig.3 Delay in WLAN
Fig. 4 WLAN Load

52
Fig.5 Throughput in UMTS
Fig.6 Delay in UMTS

53
Fig.7 Load in UMTS
Fig.8 Throughput in WiMAX
Figs.8, 9 and 10 show the throughput, delay and load in WiMAX. Load created by users are less
since three users are considered in this work that produces less throughput.

54
Fig.9 Delay in WiMAX
Fig.10 Load in WiMAX
Based on the metrics obtained in the integrated network architecture, network selection is
performed. Initially user 1 (Voice) connected to WiMAX checks for the alternate networks
(WLAN and UMTS) and calculates payoff in two networks. Fig.11 shows the payoff of user 1
in WLAN and UMTS. It is found that the payoff for user 1 in UMTS is maximum compared to
WLAN. User 2 (Video) initially connected to WLAN checks for alternate networks and
calculates payoff. The payoff for user 2 in WiMAX and UMTS is given in Fig.12. The figure
shows that the payoff for user 2 in WiMAX is higher compared to UMTS. This is because user
1 already gets maximum payoff in UMTS.

55
Fig.11 Payoff for User 1
User 3 (Web user) initially connected to UMTS checks for the alternate networks such as
WLAN and WiMAX. Fig. 13 shows the payoff for user 3 in WLAN and WiMAX. The payoff
for user 3 in WLAN is high when compared to WiMAX. This is because user 1 and 2 already
gets maximum benefit in UMTS and WiMAX respectively.
0 10 20 30 40 50 60 70 80 90 100
0
50
100
150
Iterations
Payoff
User 1
wlan
umts
0 10 20 30 40 50 60 70 80 90 100
-20
0
20
40
60
80
100
120
140
160
180
Iterations
Payoff
User 2
wimax
umts

56
Fig.14 NSP for User 1
The NSP (Network Selection Probability) is determined based on the payoff as shown in Fig.14.
Since the payoff for user 1 in UMTS is high, the NSP for user 1 in UMTS also high.Figs.15 and
16 show the NSP for user 2 and 3 respectively. Since payoff for user 2 in WiMAX is high so
NSP for user 2 in WiMAX is also high. The maximum payoff for user 3 is attained in WLAN so
NSP for user 3 in WLAN is high.
To implement the Quality of Experience (QoE) concept, users are classified as good and fair
users. A good user is one who is ready to pay more for a particular service and expects high
Quality of Service. On the other hand a fair user pays less for a service and compromises on the
service quality. Fig.17 shows user 1 and user 3 are fair users and user 2 is a good user. Even
though user 1 gets maximum payoff in UMTS since the user is fair user, so user 1 selects
WLAN. User 2 is a good user and selects WiMAX. User 3 is a fair user already connected in
WLAN and remains in WLAN.
0 10 20 30 40 50 60 70 80 90 100
0
20
40
60
80
100
120
140
160
180
Iterations
Payoff
User 3
wimax
wlan
0 10 20 30 40 50 60 70 80 90 100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Iterations
NetworkSelectionProbability
NSP for User 1
WLAN
UMTS

57
Fig.17 NSP for Users based on their budget
0 10 20 30 40 50 60 70 80 90 100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Iterations
NSP for User 2
WiMAX
UMTS
0 10 20 30 40 50 60 70 80 90 100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Iterations
NSP for User 3
WLAN
WiMAX
0 10 20 30 40 50 60 70 80 90 100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Iterations
user 1 in WLAN
user 2 in WiMAX
user 3 in WLAN

58
7. CONCLUSION
Heterogeneous networks require intelligent network selection process for seamless connectivity.
In this paper WLAN, UMTS and WiMAX networks are integrated using IMS based on SIP.
Network selection was performed using the proposed Boltzmann Gibbs based CODIPAS-RL
because this algorithm performs network selection faster than the existing Bush Mosteller based
CODIPAS-RL. i.e., Boltzmann Gibbs based CODIPAS-RL takes 90 iterations whereas Bush
Mosteller based CODIPAS-RL requires 200 iterations for selecting a network. Three users with
three different applications such as voice, video and web were considered and the network
selection was performed. The payoff for each user in three different networks are determined
and based on the payoff, network selection probability was obtained.
REFERENCES
[1] D.E.Charilas and A.D. Panagopolous, “Multi access Radio Network Environments”, In IEEE
Vehicular Technology Magazine, vol. 5, no. 4, pp. 40-49, Dec. 2010
[2] Q.T.Nguyen-Vuong, Y.Ghami-Doudane and N.Agoulmine, “On utility models for access
network selection in wireless heterogeneous networks,” In IEEE Network Operations and
Management Symposium (NOMS), 2008.
[3] H.Tembine and A.P.Azad, “Dynamic Routing Games: An Evolutionary Game Theoretic
Approach,” IEEE conference on decision and control European Conference, pp.4516-4521,
Dec.2011.
[4] M. Nam, N.Choi, Y.Seok and Y.Choi, “WISE: Energy-efficient interface selection on vertical
handoff between 3G networks and WLANs,” Proceedings of IEEE International symposium
PIMRC, pp.692-698,2004.
[5] Q.Y. Song and Abba S Jamalipour, “ Network Selection in an Integrated Wireless LAN and
UMTS environment using mathematical modelling and computing techniques”, IEEE Wireless
Communications Magazine, pp.42-48, June 2005.
[6] Manzoor Ahmed Khan, Hamidou Tembine and Athanasis V. Vasilakos, “ Game Dynamics and
cost of Learning in Heterogeneous 4G Networks”, IEEE Journal on Selected Areas in
Communications, vol. 30, no. 1, pp.198-214, Jan. 2012.
[7] Ramona Trestian, Olga Ormond and Gabriel-MiroMuntean, “ Game Theory- Based Network
Selection: Solutions and Challenges”, IEEE communications surveys and tutorials ,pp.1-20, May
2011.
[8] F.Zhu and J.McNair, “Optimizations for vertical handoff decision algorithms,” Proceedings of
IEEE WCNC, pp.867-872, 2004.
[9] Mohammad SazidZaman Khan, ShaifulAlam and Mohammad RezaulHuque Khan, “A Network
Selection Mechanism for Fourth Generation Communication Networks”, Journal of Advances in
Information Technology, vol. 1, no. 4, pp.189-196, Nov. 2010.
[10] D.Niyato and E. Hossain, “Dynamics of network selection in heterogeneous wireless networks:
An Evolutionary Game Approach,” IEEE Transactions on Vehicular Technology, vol.58,no.4,
pp.2008-2017, May 2009.
[11] Samson Lasaulse, H.Tembine, Game theory and learning for wireless networks: Fundamentals
and applications, Elsevier,2011.
[12] P.Prabhavathi and L.Nithyanandan, “Network selection in wireless heterogeneous networks,”
Proceedings of IEEE International Conference on Communication and Signal Processing – 2013,
pp. 1536-1539, April 2013.

59
pp.2008-2017, May 2009.
[14] H.Tembine, “Dynamic robust games in MIMO systems,” IEEE Transactions on systems Man
and Cybernetics, vol.99, no.41, pp.990-1002, Aug. 2011.
[15] www.opnet.com
Authors Biography
L.Nithyanandan received Bachelor of Engineering from University of Madras in
1992, Master of Technology in 1999 and Ph.D. degree in 2006 from Pondicherry
University. He is working as an Associate Professor of Electronics and
Communication Engineering, Pondicherry Engineering College, Puducherry,
India. He is a gold medallist in PG and has been awarded with chief minister
Medal for his outstanding performance in PG. He has more than 22 publications in
National /International conferences and journals. His areas of interest include
Sensor Networks, Telemedicine, Spread Spectrum Techniques and wireless
communication
V.Bharathi received Bachelor of Engineering from University of Madras in
1999, Master of Technology in 2003 and pursuing Ph.D. in Pondicherry
University. She is working as an Associate Professor of Electronics and
Communication Engineering, Sri Manakula Vinayagar Engineering College,
Puducherry India. She has more than 13 publications in National/International
conferences and Journals. Her areas of interest include Wireless Networks,
Wireless Communicationand Signal Processing.
P.Prabhavathi received her Bachelor of Technology Degree from Pondicherry
University in 2011 and Master of Technology degree in 2013 in Electronics and
Communication Engineering. She is a gold medallist in UG. Her area of interest
include spread spectrum wireless communication and networking.
59
pp.2008-2017, May 2009.
[15] www.opnet.com
Authors Biography
communication
59
pp.2008-2017, May 2009.
[15] www.opnet.com
Authors Biography
communication

60
INTENTIONAL BLANK

USER CENTRIC NETWORK SELECTION IN WIRELESS HETNETS

More Related Content

What's hot (15)

Viewers also liked (18)

Similar to USER CENTRIC NETWORK SELECTION IN WIRELESS HETNETS (20)

Recently uploaded (20)

USER CENTRIC NETWORK SELECTION IN WIRELESS HETNETS