Novel Classification Schemes for Radar Scenarios: Design and Comparison

This paper addresses the classification of the operating scenario in radar applications in terms of data homogeneity. Specifically, we consider configurations accounting for homogeneous, partially-homogeneous, and clutter edge environments. This is an essential issue since such an information allows us to derive the most appropriate target detection schemes for the current scenario. To this end, we resort to a heuristic design procedure and come up with a classification architecture obtained by cascading two binary hypothesis tests that are solved by means of the generalized likelihood ratio test (GLRT) criterion. At the design stage, we first assume that the rank of the clutter covariance matrix is known and derive the GLRT, then we devise a preliminary stage for rank estimation when it is unknown. Illustrative examples show the effectiveness of both the preliminary stage and the classification architecture.