Decentralized Array Processing with Application to Cooperative Passive Radar

In this paper we propose a passive RADAR system for a distributed sensor array. Our system is partitioned in two sets of sensors: the set of surveillance sensors and the set of reference sensors, connected through a meshed communication network topology. Rather than using a fusion center, our target detection algorithm uses near-neighbors communications. Messages are therefore exchanged within the surveillance and reference networks and across the two networks. Our decentralized detection algorithm computes the singular values of the coherence matrix between the surveillance and reference measurements, which is necessary to perform a generalized likelihood ratio (GLR) test at each sensor. Simulation results show that as the number of gossip round increases, the proposed decentralized method detection performance are comparable to those of the centralized counterpart.