Estimating directions-of-arrival of coherent signals in unknown correlated noise via spatial smoothing

We show that any weighted spatial smoothing (WSS) of subarray covariance matrices with nonzero weights will lead to a decorrelation of signals, and we discuss the consequences of using negative weights. We next propose a WSS that will eliminate correlated noise and establish a relationship between the WSS and covariance differencing. Finally, we present numerical results to demonstrate the effectiveness of the proposed WSS.