Improving performance of adaptive radar detectors in nonhomogeneous environment

Nonhomogeneous environment with unknown statistics can severe disturb detection performance of adaptive radar detectors, due to the lack of sufficient amount of independent and identically distributed (iid) data from which the detector can estimate the statistics of the environment. Significant performance improvement can be achieved by employing pre-processing algorithms to detect the outlier samples and reject them. The usual approach for nonhomogeneity detection is Generalized Inner Product (GIP). In this paper, a new outlier detection algorithm based on Automatic Censored Mean Level (ACML) algorithm is proposed which does not require any prior knowledge about the background environment. Simulation results demonstrate that the new proposed algorithm is more effective in finding nonhomogeneous samples and thus improves the performance of adaptive detector better than GIP algorithm.