Moving Target Detection using Distributed MIMO Radar in Non-Homogeneous Clutter with Limited Training Data

This paper concerns detection of a moving target in compound-Gaussian clutter with inverse-Gamma distributed power using a distributed active MIMO radar network. Adaptive detection requires an estimate of the clutter covariance matrix from a finite set of, possibly non-homogeneous, target-free training data. If the number of available training data are limited, conventional covariance matrix estimators will fail to provide adequate detection performance. This motivates the use of structured estimators which do not rely upon large amounts of training data. In this paper, a clutter covariance matrix estimator based non-negative linear least-squares is applied to the problem of adaptive moving target detection in the context of distributed MIMO radar. For each transmit-receive pair of the distributed radar network, an improved estimate of the clutter covariance matrix is computed using the non-negative least-squares estimator. Then, we apply the improved estimates in conjunction with a detector derived using inverse-Gamma distributed clutter power. Simulation examples are used to demonstrate the performance improvement of the constrained least-squares method as compared to a conventional technique. As a baseline, we a compare the performance of both the conventional and constrained least-squares methods against the clairvoyant detector.