Wireless Network Selection Algorithm for Multimedia Services in Heterogeneous Environment

Abstract

The Next Generation Networks (NGNs) consist of different mobile and wireless technologies with varied operating characteristics. Due to complementary characteristics of different wireless networks, it is proposed to combine them to provide ubiquitous wireless access for users. The integration of Heterogeneous Wireless Networks (HWNs) requires the design of intelligent network selection algorithm to ensure seamless communication, and provide high QoS for different multimedia applications. The objective is to offer seamless multimedia services to users accessing all kinds of infrastructures through heterogeneous access technologies. The major focus of the present research work is on the network selection criteria for multimedia services in heterogeneous wireless network environment. The research work aims to design and implement novel network selection algorithms in heterogeneous environments for multimedia services. The thesis is broadly divided into three parts to select network in various heterogeneous environment based on different criteria used. In first part, network selection based on QoS parameters in heterogeneous wireless networks is proposed by considering a heterogeneous environment consisting of UMTS, WLAN, GPRS, and WiMAX. It utilizes TOPSIS- a MADM algorithm based on AHP for network selection. Here, streaming is considered as a generic service. High weight has been assigned to delay whereas the packet loss and total bandwidth have been assigned lower weights and the least weight has been assigned to cost per byte to choose the appropriate connection out of four alternatives in context of seamless communication. The main constraint of the case study is the choice of weights of attributes of various networks. In order to resolve this issue, a new network selection algorithm which is based on weight estimation of the representative set of the network attributes is proposed using entropy and TOPSIS approach. The proposed model is effectively implemented to select the desired network in a heterogeneous environment of UMTS, WLAN, GPRS, and WiMAX employing triple-play services (Voice, Data and Video). The model selects the desired network in heterogeneous environment in accordance with the required QoS attributes (delay (D), total bandwidth (BW), packet loss (L) and cost per byte (CB)) for the application under consideration. On altering the attribute values of delay by 10% for voice users and 30% for video and data users, change in network selection occurs significantly. In addition to above, a novel method is proposed that takes into consideration user preferences (requested data rate, velocity, tolerable data loss), network conditions (available capacity, coverage, expected data loss due to network overloading/network selection delay) and QoS (network selection rate, RSS, delay, jitter) in order to select the optimal network between WiMAX and 3G and achieves the best balance between user’s requirement and network performance. It incorporates the use of parameterized network and user profiles in order to model diverse QoS flexibility for different real time (RT) and non-real time (NRT) applications. In the RT traffic case, the proposed scheme performs better as compared to RSS based scheme and Hungry scheme by at least 10% in terms of the data loss. The solution is realistic and not very complex to implement on mobile user units and other network elements. In second approach, network selection in wireless heterogeneous environment based on available bandwidth estimation has been presented. We consider available bandwidth as a dynamic parameter to select the network in heterogeneous environment. We propose novel network selection algorithms capable of adapting to prevailing network conditions in heterogeneous environment of 3G & WLAN networks, 3G & WiMAX and WLAN & WiMAX in real time. We utilize a bootstrap approximation based technique to estimate available bandwidth and compare it with hidden Markov model based estimation to determine its precision. It is implemented in temporal and spatial domains to verify its robustness. Estimation error, overhead, estimation time with varying size of traffic and reliability are used as the performance metrics. Analysis of heterogeneous system reveals that the proposed network selection methods can effectively choose the suitable network by negotiating trade-offs among network dynamic conditions and multimedia services. In third part, we propose algorithms for network selection based on the link parameters, assuming that network conditions are dominant in network selection. Network selection function (NSF) consists of averaged RSS, distance and outage probability parameters to perform network selection. It comprises of two stages. In first stage, overlapping region is identified through distance estimation. Network selection algorithm based on averaged received signal strength plus outage is invoked in second stage to select the optimum network between two different networks- GSM and UMTS. The predicted overlapped distance is utilized to make a network selection in order to minimize the probability of network selection failures or unnecessary selections from one cellular network to another. Significant reduction of 68% in network selection rate has been obtained with the application of proposed algorithm as compared to conventional method. In addition to above, another network selection algorithm utilizing signal strength, available bit rate, signal to noise ratio, achievable throughput, bit error rate and outage probability metrics as coefficient of cost function NSDF (Network Selection Decision Function) for network selection has been proposed. The selection metrics are hybridized with PSO for relative dynamic weight optimization. The proposed algorithm is implemented in a typical heterogeneous environment of EDGE (2.5G) and UMTS (3G). The proposed model based on dynamic metric weights optimized by modified PSO resulted in significant reduction of the network selection rate, computational complexity and time. Minimization of network selection rate while maintaining QoS for multimedia services is recorded, which in turn reduces the overhead on the MS. Proposed algorithm maintains QoS while selecting ABC network in heterogeneous environment by keeping utility function at least 0.5. The best possible QoS can be achieved when achievable throughput is greater than guaranteed required bandwidth or utility function is equal to 1. When network selection is performed by using proposed algorithm then utility function is always greater than or equal 0.5. It is very useful and applicable in supporting multimedia applications over wireless environment due its high convergence capability and simplicity. In continuation, another network selection algorithm is proposed on the same platform for a heterogeneous environment consisting of WiMAX and LTE standard. It is based on received signal strength, signal to noise ratio, available bit rate, achievable throughput and bit error rate. Relative weights of the decision making attributes are optimized employing particle swarm optimization approach. Satisfaction decision function (SDF) utilizes dynamic optimized weights to select optimal network in heterogeneous environment of cellular networks. The number of satisfied users calculated as per their demand with wireless network selected by proposed algorithm is optimized by modified PSO and affirmed by Monte Carlo method. It is observed that network selection rate is dependent on user motion. SDF based algorithm yielded 50% better performance in terms of the satisfied users as compared to existing techniques/algorithms. The results of proposed approaches demonstrate that more parameters considered can influence the results in network selection and related QoS. It reduces unnecessary network selections and network selection failures, computational complexity, time and balances the network load with prefect utilization factor. At the end this could lead to the network, where users will be always and anywhere best connected with the desired QoS. Research findings show that the proposed schemes are suitable for implementation on smart mobile terminal available in the market. In future, there are possibilities to make profit from better utilized networks by the operators with more number of smart users/ multimedia users.