Two-dimensional direction estimation for a mixture of noncoherent and coherent signals

This paper deals with the two-dimensional (2-D) direction-of-arrival (DOA) estimation of a mixture of noncoherent (including uncorrelated and partially correlated) and coherent (i.e., fully correlated) narrowband signals impinging on a planar sensor array composed of two parallel uniform linear arrays (ULAs). An oblique projection based approach for 2-D direction estimation (OPADE) is proposed by using some cross-correlations between the received array data. In the proposed OPADE, the oblique projection is utilized to isolate the coherent signals from the noncoherent ones and the effect of additive noise is alleviated, while the computationally intensive eigendecomposition is avoided, and the estimated elevation and azimuth angles are paired automatically. Further, an iterative alternating scheme is presented to improve the estimation accuracy of the oblique projector and hence that of the DOAs of coherent signals. The Cramér-Rao lower bound (CRB) for the mixture of noncoherent and coherent signals is also derived explicitly, where the prior knowledge of the signal correlation is incorporated into the 2-D DOA estimation of noncoherent signals. Finally the effectiveness of the OPADE and the theoretical analysis are substantiated through numerical examples.