Backpressure Routing Made Practical

Scott Moeller, Avinash Sridharan, Bhaskar Krishnamachari, Omprakash Gnawali
2010 2010 INFOCOM IEEE Conference on Computer Communications Workshops  
The current design methodology for data collection protocols in wireless sensor networks relies on the proactive construction and maintenance of quasi-static shortest path trees for routing. We consider an alternative highly-agile approach called backpressure routing, in which forwarding decisions are made on a per-packet basis using backpressure link weights that incorporate local queue state and link quality information. In theory, according to recent literature on cross-layer stochastic
more » ... rk optimization, such a dynamic routing approach can provide near-optimal utilization of the available bandwidth. However, it has not been implemented on practical systems to date due to concerns about packet looping, the effect of link losses, and large delays. Addressing these concerns, we present the Backpressure Collection Protocol (BCP), the first ever implementation of dynamic backpressure routing in wireless sensor networks. Through experiments on a 40-node wireless sensor network testbed, we demonstrate that incorporating transmission count minimization in the backpressure weight eliminates packet looping and provides excellent throughput performance. We show that BCP provides more than 50% improvements in throughput compared to a state of the art tree-based collection protocol. We also show that the average end-to-end packet delays in BCP can be drastically reduced (75% under high load, 98% under low load) by using LIFO queues.
doi:10.1109/infcomw.2010.5466634 fatcat:ccon56eaizexzfpw2lzn27uj4a