Efficient Direction Estimation of Both Coherent and Uncorrelated Sources Without Prior Knowledge of Source Count

We propose a novel method for accurately estimating the directions of arrival (DOAs) of both coherent and uncorrelated sources with a linear sensor array. Our method eliminates the need for prior knowledge of the source count, enhancing its practicality. The approach begins by estimating the DOAs of uncorrelated sources using a classical subspace technique, establishing a solid foundation for further analysis. To mitigate the impact of uncorrelated sources, a symmetric linear array system is employed, effectively eliminating their effects and improving overall estimation accuracy. Furthermore, we exploit the inherent structure of the sample covariance matrix, treating each row as a Toeplitz matrix with full rank. This allows us to decorrelate the coherency of sources using an existing method. To estimate the DOAs of coherent sources without source count estimation, we develop a sophisticated cost function that exploits the joint diagonalization structure of the Toeplitz matrices. This innovative approach enables robust estimation and eliminates the need for source count estimation in the coherent DOA estimation. The proposed method offers an efficient and effective solution for DOA estimation, accommodating both coherent and uncorrelated sources. Suppose that a symmetric uniform linear array (ULA) with
$({2}{N}+{1})$
sensors receives
${K}$
narrowband signals from the far field, and the proposed method exhibits an impressive capability to resolve a number of source DOAs that exceeds the number of sensors [
${K} > {(}{2}{N}+{1}{)}$
]. Through extensive evaluations, we demonstrate its superior performance in various scenarios.