Improved Simultaneous Perturbation Stochastic Approximation-based Consensus Algorithm for Tracking.

In this paper, we consider a distributed stochastic optimization problem where the goal is to cooperatively minimize a non-stationary mean-risk functional. Such problem is an integral part of many important problems in wireless networks, transportation systems, sensor networks, and others. In particular, we focus on the reduction of computational effort needed to achieve a certain level of accuracy. Thus, we propose an improved Simultaneous Perturbation Stochastic Approximation-based consensus algorithm that achieves better accuracy in contrast to an existing solution over the same time horizon and provide its theoretical analysis. We also show the convergence to a bound for mean-squared errors of estimates. The simulation validates the new algorithm in a multi-sensor multi-target problem.