A distributed scheduling algorithm for quality of service support in multiaccess networks

Abstract

This thesis presents a distributed scheduling algorithm for the support of quality of service in multiaccess networks. Unlike most contention-based multiaccess protocols which offer no quality of service guarantee and suffer the problems of fairness and low throughput at high load, our algorithm provides fairness and bandwidth reservation in an integrated services environment and at the same time achieves high throughput. Moreover, while most reservation-based multiaccess protocols require a centralized scheduler and a separate channel for arbitration, our algorithm is truly distributed in the sense that network nodes coordinate their transmissions only via headers in the packets. We derive theoretical bounds illustrating how our distributed algorithm approximates the optimal centralized algorithm. Simulation results are also presented to justify our claims.