2D DOA estimator for coherently distributed sources based on fourth order cumulants

In this paper, a new method to estimate the nominal two-dimensional (2D) direction-of-arrivals (DOAs) of multiple coherently distributed (CD) sources using two uniform linear arrays (ULAs) is proposed. Because the fourth-order cumulants of Gaussian random processing is equal to zero, the proposed method firstly constructs three fourth-order cumulants matrices for Gaussian noise suppression, which can result in the improvement of the performance of DOA estimation. Then the rotational invariant structure is exploited among the three fourth-order cumulants matrices. Finally, the nominal DOAs of coherently distributed sources are obtained from the eigenvalues of the rotational invariant matrices without spectrum peak searching. Compared with some existing methods, the proposed method exhibit a better estimation performance for CD sources even at the conditions of low signal-to-noise ratio (SNR), small number of snapshots or different angular spread degrees. Simulation results validate the effectiveness of the proposed method.