Robust covariance matrix estimation in radar array processing

Recently, several robust covariance matrix estimation techniques were proposed, such as normalized sample covariance matrix (NSCM) and fixed point matrix (FPM). They were claimed to be able to provide better performance for non-Gaussian clutter or in the presence of disturbance. In this paper, the performance of these robust covariance matrix estimators were evaluated and compared with the conventional sample covariance matrix (SCM) technique using both simulated and experimental data, in terms of target direction estimation accuracy and clutter suppression capability in space-time adaptive processing (STAP).