The University of New South Wales
Radar target detection requires effective clutter suppression, which is usually achieved by estimating the clutter covariance matrix using training data obtained from surrounding range gates to the cell under test (CUT). This requires that the training data be statistically homogeneous with the CUT. However, this assumption is usually violated in practice leading to significant degradation in performance. This topic will describe various approaches, including recent developments that aim to mitigate this problem and achieve robustness to clutter heterogeneity. The topic will deal both with land and maritime environments.