Self-regulating sensor network

Abstract

Power is a critical resource in sensor networks, especially when the nodes are un-tethered and used in long running applications. Many approaches have been developed to exploit the inherent space and time redundancy in the data collected by sensor networks by using selective querying. In selective querying, at any given time, only a percentage of the nodes is queried; the rest are put in sleep mode. The key feature of selective querying is the method by which it decides which nodes are active and which nodes are sleep at every time period. The published literature catalogs a rich variety of selection criteria, yet, it barely addresses the orthogonal problem: what percentage of the nodes should be active. In this paper we describe algorithms we have used to make the sensor network self-regulate, i.e. the network continuously monitors the level of activity in the environment and regulates the level of activity of the network accordingly. We further use the analogy of natural self-regulation by also gauging the level of activity to the amount and distribution of resources available.