Enabling Accurate Node Control in Randomized Duty Cycling Networks

In this paper, we propose a novel duty cycling algorithm for a large-scale dense wireless sensor networks. The proposed algorithm is based on a social behavior of nodes in the sense that individual node's sleep/wakeup decision is influenced by the state of its neighbors. We analyze the behavior of the proposed duty cycling algorithm using a stochastic spatial process. In particular, we consider a geometric form of neighborhood dependence and a reversible Markov chain, and apply this model to analyze the behavior of the duty cycling network. We then identify a set of parameters for the reversible spatial process model, and study the steady state of the network with respect to these parameters. We report that our algorithm is scalable to a large network, and can effectively control the active node density while achieving a small variance. We also report that the social behavior of nodes has interesting and non-obvious impacts on the performance of duty cycling. Finally, we present how to set the parameters of the algorithm to obtain a desirable duty cycling behavior.