Performance analysis of predictive scalable resource allocation for integrated wireless networks

This paper presents a framework of using predictive regression schemes in single-slot and multiple-slot scalable resource allocation for integrated wireless networks. For predictive scalable resource allocation techniques, the number of channels required in a service is determined in correspondence to the increase or decrease in real-time traffic intensity as well as the traffic load history. Based on the measurement of real-time traffic intensity and the recorded load history database, the traffic intensity at the next sampled time can be forecasted by the regression prediction schemes. The predicted channels for the next sampled time is assigned to the service periodically. In this study, we analyze and compare the performance of the various predictive scalable resource allocation schemes in a sudden changing traffic conditions. The voice service blocking probability, data service session delay, and radio resource utilization are assessed.