An alternating service model with self-similar input to provide guaranteed QoS in wireless internet

Abstract

Over the past few years, we have witnessed a growing popularity of new wireless architectures such as 3G, WiFi and Wi-Max due to the increase in demand for wireless Internet access. The all-IP based future mobile and wireless network model is expected to be the most dominant architecture for QoS provisioning in next-generation wireless networks, mainly due to its scalability and capability of inter-working heterogeneous wireless access networks. Recently, the rapid growth of various wireless infrastructures and the interesting mixture of wireless traffic generated by large number of devices (PDAs, Laptops and cell-phones) have diverted the attention of wireless research community towards understanding the nature of traffic carried by different wireless architectures. A series of recent studies on GPRS aggregated traffic, WAP and Web traffic has proven that wireless traffic exhibits strong long-range dependency. However, much of the current understanding of wireless traffic modeling builds on classical Poisson distributed traffic, which can yield misleading results and hence poor wireless network planning. In this paper, we contribute to the accurate modeling of wireless IP traffic by considering two different types of traffic that exhibit long-range dependency and self-similarity. We consider a model of two queues based on G/M/1 queueing system and analyze it on the basis of 1-limited polling service and find exact bounds on packet delay.