A Unified theory of scheduling, routing, and flow control in wireless networks

Abstract

A new approach to joint scheduling, flow control and routing in wireless networks, based on formulation as a convex optimization problem is presented. This approach is novel in that it integrates optimal scheduling and flow control with a modified version of minimum delay routing, resulting in significant performance advantage over alternative approaches. We come up with a distributed algorithm for implementation of the proposed scheme and show, by analysis and simulation, that the algorithm achieves fairness and/or priorities among users in accordance with pre-assigned user parameters. In sharp contrast to alternative algorithms that perform scheduling and packet routing based on per-session queue differential between adjacent nodes, our algorithm uses a complete multi-hop view of network conditions for packet routing and retains the desirable properties of minimum delay routing. As the result, it achieves queue sizes and end-to-end delays that are several times smaller, without compromising the throughput.