Adaptive matched subspace detectors and adaptive coherence estimators

In this paper we adapt the matched subspace detectors (MSDs) of [5] and [6] to unknown noise covariance matrices in order to produce adaptive MSDs that may be applied to signal detection in radar, sonar, mobile communication, and DOA estimation. A special case of the adaptive MSD uses the Reed ratio statistic [4], and a special case of the adaptive CFAR MSD uses an adaptive coherence estimator (ACE). We compare and contrast the invariances and performances of the two detectors and discuss extensions of them that make them maximum likelihood MSDs. In a companion paper [3], we apply the adaptive CFAR MSD to simulated data and to data recorded from the Mountaintop radar.