A Unitary MUSIC-Like Algorithm for Coherent Sources

This paper proposes a method for direction of arrival (DOA) estimation which can be applied in case of both non-coherent and coherent sources. In comparison to the well-known subspace algorithms such as MUSIC, the proposed method has several advantages. First, in contrast to MUSIC, no forward/backward spatial smoothing for the covariance matrix is needed in the case of coherent sources. Second, the proposed method is more suitable for realtime implementation since it only requires one or a few snapshots in order to provide an accurate DOA estimation, whereas MUSIC requires a large number of snapshots. Third, the proposed method exploits the eigenvalue decomposition (EVD) of a real-valued covariance matrix thereby reducing the computational cost by at least a factor of four. Simulation results show that the proposed method can estimate the DO As of the incident sources with high accuracy even when the sources are coherent.