Underdetermined DOA estimation of quasi-stationary signals via virtual array interpolation

Abstract

An underdetermined direction of arrival (DOA) estimation method for quasi-stationary signals (QSSs) using virtual array interpolation is proposed. A second-order difference co-array model based on quasi-stationary signals is first constructed. This model is then interpolated into a uniform linear array (ULA). Instead of processing each time frame individually, a single matrix completion operation is applied across all time frames simultaneously. This method leverages the quasi-stationarity of the signals and the low-rank property of the auto-covariance matrix for matrix completion. An alternating direction method of multipliers (ADMM) based solution is introduced to solve the matrix completion problem, which is more efficient than the commonly used semi-definite programming (SDP) framework. Subsequently, the subspace method is utilized on the completed covariance matrix for DOA estimation. Comparative analysis with the existing interpolation-based QSS DOA estimation method demonstrates that the proposed method achieves superior accuracy and efficiency.