Estimation over Wireless Sensor Networks

Remote estimation problems are critical to many novel applications enabled by large-scale dense wireless sensor network. Individual sensors simultaneously sense, process and transmit measured information over a lossy wireless network to a central base station, which processes the data and produces an optimal estimate of the state. In this paper, we investigate the tradeoff between the estimation performance and the number of communicating nodes with respect to the major MAC protocols used in wireless sensor networks. We first construct a Markov model of the node behavior to study the correlation between packet reception probability and the number of communicating nodes. We then develop a multi-sensor measurement fusion model. This is used to feed a multi-sensor Kalman filtering algorithm to assess the impact of MAC protocols on estimation performance. We offer a target tracking example to illustrate our approach.