Adaptive sleep scheduling protocol in wireless sensor networks

Energy efficiency in wireless sensor networks is a major issue, since the sensors usually have limited and irreplaceable power sources. Sleep scheduling solutions proved to be exceptionally effective strategies to achieve this goal. Numerous such algorithms have been proposed and examined, but virtually without any considerable support for dynamic systems. In this paper we propose and analyze an adaptive, soft-state, fully distributed and robust sleep scheduling method that can easily cope with frequent node failures. The proposed scheme can dynamically eliminate the redundancy and estimate the deficient data based on learned relations in a way to ensure low and balanced energy consumption. This is done without the need for offline pre-computations, dedicated phases, time synchronization, localization, or base station assistance. We compare our technique with deterministic clustering methods, provide parameter sensitivity analysis and discuss the simulation results.