Estimation of Always Best Connected Network in Heterogeneous Environment Based on Prediction of Recent Call History and Call Blocking Probability

Abstract

Presently, the emergence of 4G heterogeneous network has attracted most of the user centric applications like video chatting, online mobile interactive classroom besides voice services. To facilitate such bandwidth hungry multimedia applications and to ensure QoS (Quality of Service), Always Best Connected (ABC) network is to be selected among available heterogeneous network. The selection of the ABC network is based on certain design parameters such as cost factor, bandwidth utilization, packet delivery ratio, security, throughput, delay, packet loss ratio and call blocking probability. In this paper, all the above mentioned design parameters are considered to evaluate the performance of Always Best Connected network under heterogeneous environment for mobile users. In addition, to select Always Best Connected network in a heterogeneous environment, a novel parameter namely recent call history-‘rch' is proposed to predict the call blocking probability (Cbp) of a network.The estimation of ‘rch' parameter in terms of Cbp avoids unnecessary hand-off and enhances the effective bandwidth utilization of a cellular network under heavy load condition.A QoS mapper submodule is proposed in the QoS broker module to predict the ‘rch' parameter of a network to select an optimum network. Further, a statistcal model based on Bayesian network using the data on recent call history is applied for Cbp estimation and simulated over various heterogeneous environment condition. The simulated results show improved performance of user centric applications when compared with non-predictive methods.