An improved multiple-signal-subspace eigenvalue-decomposition method for coherent-signal direction-of-arrival estimation

This letter proposes an improved method based on multi-column signal eigenvector reconstruction for direction-of-arrival (DOA) estimation of coherent and uncorrelated mixed sources. The equivalent data-covariance matrix of our method is reconstructed by rearranging the elements of all multi-column signal eigenvectors that contain all the information of coherent and uncorrelated signals. The dimension of the reconstructed data matrix can be varied to estimate more signals than traditional eigenvector-reconstruction algorithms and currently existing matrix-reconstruction algorithms with fixed estimated signals. Furthermore, by utilizing complete data matrix information for reconstruction, the algorithm effectively solves the performance degradation problem of the classical matrix-reconstruction estimation of signal parameters via a rotational invariance (ESPRIT)-like algorithm under a specific phase difference and angles, and the estimation performance significantly improves as the matrix degrees of freedom increase. We performed a simulation to evaluate the proposed algorithm, and the results verify its effectiveness.