Analysis of a Distributed Detection Scheme for Finite WSNs

Abstract

We analyze a binary hypothesis test-ing problem built on a wireless sensor network(WSN). We compute a tight approximation of themiss-detection probability of a distributed scheme fordetecting a correlated circularly-symmetric complexGaussian process under the Neyman-Pearson frame-work. Using an analog scheme, the sensors transmitseveral scaled versions of their measurements througha multiple access channel (MAC) to reach the fusioncenter (FC), whose task is to decide whether the pro-cess is present or not. In the analysis, we considerenergy and bandwidth constraints on each node. Weshow that the proposed distributed scheme exploit thecorrelation of the process beneficially, achieves thecentralized detector for high signal-to-noise ratio inthe communication channel and performs much bet-ter that the energy detector, where the correlation isnot considered. We validate the theoretical resultsthrough the method of Monte Carlo.