A Novel Distributed Max-Weight Link Scheduler for Multi-Transmit/Receive Wireless Mesh Networks

Abstract

Multi-transmit-receive capability is fast becoming a significant feature of next-generation wireless mesh networks. It enables routers to transmit or receive distinct packets from multiple neighbors simultaneously. A key problem, however, is designing a distributed link-scheduling algorithm that ensures high network capacity. In this paper, we propose dMaxQ, which is a novel queue-length-aware distributed link scheduler that requires only one-hop neighbors' queue information and uses the celebrated max-weight policy in a distributed manner. We have evaluated the performance of dMaxQ in different network topologies for both single-hop and multihop traffic models and compared it against other approaches, including two queue-length-aware centralized algorithms and state-of-the-art distributed approaches: JazzyMAC and receive-oriented multiple access. The results show that for single-hop and multihop traffic scenarios, dMaxQ obtains, respectively, 100% and 90% of the throughput achieved by the theoretical centralized policy. Other distributed algorithms, such as JazzyMAC, only managed 25% of the theoretical throughput.