Improved coherent source decorrelation by autocorrelation matrix smoothing

Many adaptive beamforming and angle of arrival estimation algorithms perform poorly when the signals arriving at the array are mutually correlated or coherent. We describe an improved technique for coherent signal decorrelation that is based on the idea of autocorrelation matrix smoothing (AMS). We begin with a brief review of the definition and properties of AMS, including a unified interpretation of several previously-proposed decorrelation methods as special cases of AMS. Then we use the AMS approach to formulate an improved decorrelation technique, and conclude with computer simulations that illustrate the performance of the new technique.<>