A Two-Stage Detector for Operation in Outlier-Dense Scenarios

In this paper we devise a two-stage receiver for detecting targets embedded in heterogeneous interference environment. Precisely, the receiver is composed of a training data selector which excises outliers from the available secondary dataset and an adaptive matched filter (AMF) which performs the decision as to the target presence. The selector estimates the outlier subset resorting to the generalized likelihood function (GLF) and removes vectors belong to the estimated outlier subset, the remaining vectors are used as input of the AMF. At the analysis stage, the performance of the derived receiving structure is evaluated on simulated data. The results show that the proposed detector can ensure satisfactory detection performance while preserving the constant false alarm rate (CFAR) property.