Reduced-dimensional Coarray-based decomposition for efficient direction-of-arrival estimation

To address the low computational efficiency caused by the high-dimensional second-order difference coarray matrix in large-aperture sparse arrays, this paper proposes an efficient direction-of-arrival (DOA) estimation method termed the 2^N-th order reduced-dimensional coarray-based eigenvalue decomposition framework (2^N-th RCED). The proposed framework performs N recursive sequential combinations of forward-backward averaging and second-order unitary transformations to partition the high-dimensional coarray-based covariance matrix into 2^N submatrices. A reduced-dimensional eigenvalue decomposition (EVD) is subsequently applied to these submatrices to obtain 2^N corresponding noise subspace matrices. A novel reconstruction scheme is then introduced to combine these matrices into an equivalent noise subspace. Theoretical analysis shows that the proposed method significantly reduces computational complexity in both the EVD stage and the spectral peak search. Simulation results verify that the 2^N-th RCED achieves notable improvements in computational efficiency and estimation accuracy compared with classical coarray-based subspace algorithms.