CFAR Matched Detector of Signals in Autoregressive Noise

Abstract

The generalized likelihood ratio test (GLRT) for detecting a target signal in autoregressive (AR) noise whose covariance matrix is unknown is presented. We proposed a new constant false alarm rate (CFAR) matched detector whose structure does not depend on noise covariance matrix and level and its performance penalty is small. The evaluation of its performance shows that the loss measured with respect to the corresponding optimum structure with known noise covariance matrix is kept within a few dBs.