Smart sleep: Sleep more to reduce delay in duty-cycled wireless sensor networks

A simple random walk (SRW) has been considered as an effective forwarding method for many applications in wireless sensor networks (WSNs) due to its desirable properties. However, a critical downside of SRW - slow diffusion or exploration over the space, typically leads to longer packet delay and undermines its own benefits. Such slow-mixing problem becomes even worse under random duty cycling adopted for energy conservation. In this paper, we study how to overcome this problem without any sacrifice or tradeoff, and propose a simple modification of random duty cycling, named Smart Sleep, which achieves more power-saving as well as faster packet diffusion (or smaller delay) while retaining the benefits of SRW. We also introduce a class of p-backtracking random walks and establish its properties to analytically explain the fast packet diffusion induced by Smart Sleep. We further obtain a necessary condition to achieve an optimal performance under our Smart Sleep, and finally demonstrate remarkable performance improvement via independent simulation results over various network topologies.