Power constrained linear estimation of correlated sources in hierarchical wireless sensor networks

Abstract

We propose a power allocation scheme for estimation in hierarchical wireless sensor networks. The sensors in the network are divided into disjoint clusters and each cluster observes a random source which is correlated with the sources being observed by other clusters. The estimation is performed in two steps: in the first step, the sensors in each cluster send a scaled version of their noisy measurements to their respective cluster-head (CH) which forms a preliminary estimate of the underlying source; and in the second step, the CHs send their partial estimates to a remote fusion center (FC) for final estimation. The estimates are based on LMMSE estimation rule. The communication between the sensors and the CHs, and between the CHs and the FC takes place on orthogonal channels. The proposed power allocation scheme minimizes the estimation distortion subject to constraints on the network power consumption. Effectiveness of the scheme is illustrated with simulation examples.