Quality Of Service (QOS) Management In Mixed Wireless Networks Using The Power Performance Measure

Abstract

The explosion of new wireless services is driving the need for coverage beyond traditional cellular networks to other services such as Wi-Max, wireless Local Area Networks (WLAN) and Ad-Hoc or mesh networks . A mixed network as presented in this work, combines fixed/cellular networks with Ad-hoc networks and reflects the future of wireless networks. There are two focus areas in this dissertation. First, it uses a new clustering algorithm that expands a modified k-means clustering algorithm to organize the mixed network. Second, it uses the Power Performance Measure to establish the appropriate operating point in the mixed network in order to provide Quality of Service (QoS) guarantees for time critical applications as a function of the traffic level. The results show the potential to significantly enhance QoS performance of the mixed network and maintain that performance level under operating conditions. Performance in mixed networks is especially difficult to characterize. Serving users with real time applications is a challenge for mixed networks as multiple elements complicate QoS guarantees. This work successfully develops a model for QoS in a mixed network. This work shows that a multistage clustering scheme is an effective method for organizing mixed networks and introduces a clustering scheme that jointly optimizes network performance for Signal to Noise Ratio (SNR) and for Delay using multiple distance measures. This work also demonstrates that a Power Performance Measure, which combines throughput and delay into a single performance measure is an effective tool for managing QoS in Mixed Networks.