Power Allocation Policy for Distributed Estimation in Wireless Networks

We consider a sensor network monitoring a stochastic process. The sensors exchange information over wireless communication channels affected by noise. The goal of the sensors is to compute accurate estimates of the state of the stochastic process under limited energy available for communications. We introduce a distributed algorithm which computes a power allocation scheme aimed at ensuring accurate state estimates and at saving communication energy. The distributed algorithm arise from solving an approximate convex optimization problem with constraints. We present how the power allocation scheme can be used for performing distributed estimation in the case where measurements are shared between sensors.