Distance-weighted backlog differentials for back-pressure routing in multi-hop wireless networks

Jing Lu, Zuming Huang, Ninghao Liu, Quansheng Guan

Abstract: Back-pressure algorithm for queueing network can lead to maximum network throughput and is robust to time-varying network conditions, which has raised considerable attention recently due to the scarcity of wireless bandwidth resources. However, it may result in large end-to-end (e2e) delay and a waste of network resources, particularly when the network loads are light or moderate. The reason is that back-pressure algorithm may explore and exploit unnecessarily long paths to maintain network stability. In this paper, we propose a distance-weighted back-pressure routing (DW-BPR), which weights backlog differentials with distance gradients to refine packet-forwarding routes into directions towards destinations, to reduce e2e delay. DW-BPR forwards packets to nodes that not only have smaller queue backlogs but also are closer to destinations than the present ones. We further improve DW-BPR by a distance-aware link scheduling policy. We prove that DW-BPR results in strong network stability. Simulation results in NS-2 also show efficiency improvement in terms of e2e delay, throughput and average queue length over the original back-pressure algorithm.