Adaptive AR model based detection in nonhomogeneous environment

Abstract

Performance of an AR model based adaptive radar detector called ARGLR is investigated in nonhomogeneous environments. It is shown that the performance of this detector will degrade in such situations. An effective form of generalized inner product (GIP) test which uses the assumption of being AR process for clutter is proposed and used to find outliers and improve the performance of detector. Results are presented using computer simulations which demonstrate the new proposed form of GIP is more effective in finding homogeneous training data and thus improves the performance of this detector better than traditional form of GIP.