Balancing network connectivity and the life-time of sensors through percolation and consensus

Abstract

Due to replacement infeasibility, methods to extend the life-time of sensors have been an issue in Wireless Sensor Networks (WSNs) and these should consider network connectivity simultaneously. Controlling the sleep/awake of sensors is one simple way to reduce their energy consumption. However, this causes a network connectivity degradation by varying network connection topology. For this reason, we propose a simple and autonomous sensor sleep/awake method to achieve their balance. The size of clusters can be a metric to measure network connectivity in that a path exists among any cluster node. From percolation theory, we observe that a cluster size suffers a sharp transition based on edge connection patterns. This allows us to design a sensor sleep/awake algorithm which has an immense simplicity, but still requires global topology information. In many cases, sensors are not aware of the global topology. Further, managing the information becomes challenging under physical topology changes such as sensor add/drop. We show that the global knowledge requirement can be resolved by using a consensus algorithm. Through several graph tests, we show that our method achieves a network balancing between connectivity and life-time with preserving its simplicity. Also, the balancing is autonomous even under physical topology variations.