Hierarchical Organizations of Sensors for Decentralized Parameter Estimation

Abstract

In this paper, we address the problem of decentralized parameter estimation via hierarchical organizations of sensors. In this setup, the nodes are organized in clusters, and a sensor is designated as a cluster-head depending on its channel conditions. The task of the network is to estimate an unknown parameter with the minimum possible distortion, while ensuring a prescribed total power consumption. To this aim, we consider analog transmissions and, further, we decompose the problem into smaller subproblems, which can be autonomously solved for each cluster-head. We show that by balancing the total amount of power between the cluster-heads and the sensors, one can increase the estimation accuracy, and we derive a closed-form expression of the optimum balancing for the Uniform Power Allocation(UPA) case. Next, we propose some hybrid solutions which combine UPA with optimal WF schemes. Finally, we assess the performance of the proposed schemes by means of computer simulations, and we carry out a comparison with the non-hierarchical strategy as a baseline.