Distributed detection of moving target using MIMO radar in clutter with non-homogeneous power

Previously, we studied moving target detection (MTD) using a distributed MIMO radar, where the multi-static transmit-receive configuration causes non-homogeneous clutter. By representing the non-homogeneous clutter in a low-rank subspace with different subspace coefficients for different transmit-receive pairs, a generalized likelihood ratio test (GLRT), which is referred to as the MIMO-GLRT, was introduced. The MIMO-GLRT, however, is a centralized detector requiring the distributed receivers to send their local observations to a fusion center, which performs parameter estimation and computes a global test variable. In this paper, we consider distributed detection for the moving target problem. The goal is to reduce the communication overhead as well as power/bandwidth consumptions from the receivers to the fusion center. We consider two distributed implementations of the MIMO-GLRT, with or without local data aggregation. Specifically, the one that performs local aggregation computes a single local test statistic at each receive antenna, by using the outputs of all matched filters (each matched to a waveform unique to one transmit antenna); meanwhile, the one that does not perform local aggregation computes multiple local test statistics, one for each matched filter output. In both cases, the local unquantized test statistics from all receive antennas are forwarded to the fusion center and non-coherently combined to form a final test variable. Simulation results are provided to illustrate the performance loss with respect to the centralized MIMO-GLRT and compare with another distributed MIMO moving target detector based on a homogeneous assumption.