Covariance Matrix Differencing for Coherent Source DOA Estimation under Unknown Noise Field

In this paper, we propose an algorithm to estimate the one- and two-dimensional direction of arrival angles (DOAs) from the coherent incident sources. We can apply our proposed algorithm for more practical situations even when the unknown noise covariance matrix is in a complex symmetric Toeplitz form; whereas the Prasad's method requires that the unknown noise should be in a real symmetric Toeplitz form. Our proposed method is based on the difference between the forward/backward spatial smoothing for the data covariance matrix and the Hermition of the backward spatial smoothing. This covariance matrix difference is introduced to eliminate the noise components from the array structure. The numerical results verify that the proposed method gives more accurate estimation and superior performance than the conventional root MUSIC of the forward/backward spatial smoothing.