Delay-Optimal Back-Pressure Routing Algorithm for Multihop Wireless Networks

Abstract

Although the back-pressure (BP) algorithm has been proven to be a throughput-optimal policy for traffic scheduling and routing in wireless networks, the optimal control of delays in the BP algorithm remains an open problem, especially for conventional multihop networks. To enhance the delay performance, we proposed an improved BP algorithm called sojourn-time-based BP (STBP) by introducing a novel delay metric called the sojourn time backlog (STB). The STB considers the queue length and accumulated packet delays comprehensively. It provides more pressure to push forward flows suffering from greater delays. Based on this new metric, the calculation of the routing weight for each packet is determined for maximizing the difference of the STB in the routing process. The proposed routing algorithm is robust and distributed, and does not require any prior knowledge of network connections and load conditions. We analyze the network stability with delay considerations and prove the throughput optimality of the STBP in multihop networks. Simulation results reveal that the enhanced algorithm effectively improves the end-to-end delay while ensuring throughput optimality.