A STAP algorithm for radar target detection in heterogeneous environments

Traditional STAP processors for radar target detection, such as the GLRT and AMF, require an estimate of the noise covariance matrix. In practice, this estimate is obtained from a training data set that is usually constructed from range gates surrounding the test gate. The training data must be target free and statistically homogeneous with the test data. In heterogeneous and target rich environments, these assumptions do not necessarily hold and degradation in the detection performance results. In this paper, we propose a new detection algorithm, which we call the maximum likelihood estimation detector (MLED), and that operates only on the test data. We show that the new detector has the highly desirable CFAR property. We give the expressions for its probabilities of false alarm and detection and show that it has a performance that is comparable with the traditional algorithms