Receding horizon consensus estimation of sensor networks with packet dropouts

This paper is concerned with the consensus estimation problem of wireless sensor networks (WSNs) subject to random packet dropouts. In the WSN, the sensor nodes are of general linear dynamics, and the packet dropouts are modeled as Bernoulli processes. The distributed receding horizon estimation (RHE) scheme is proposed to achieve consensus for state estimate. To design such a consensus-based RHE scheme, an optimization problem containing consensus term is first formulated, and then the explicit solution to the optimization problem is provided as the estimation protocol. The sufficient conditions on ensuring the stochastic consensus for state estimates are developed. It is shown that, if the cost functions are designed appropriately and the packet dropout processes fulfill certain conditions, the consensus gain can be designed such that the stochastic consensus is reached.