Distributed target detection in centralized wireless sensor networks with communication constraints

Distributed inference is an important and challenging problem in wireless sensor networks (WSNs). In this paper we consider distributed detection of a target in centralized WSNs (i.e. WSNs with a fusion centre) subject to communication constraints. We focus on the parallel network topology, where the sensors can only exchange information with the fusion centre, and consider conditionally dependent observations. We develop two types of local decision rules for the sensors (binary and binary with abstention), based on the Neyman-Pearson criterion, and a fusion rule based on a support vector machine (SVM). Under these circumstances we show empirically that, even when individual sensors with very poor performance are used, both local configurations are able to provide very good detection rates as the number of nodes increases.