Spatially Smoothed TF-Root-MUSIC for DOA Estimation of Coherent and Non-Stationary Sources under Noisy Conditions

This paper proposes a method for efficient Direction-of-Arrival (DOA) estimation of coherent and non-stationary sources under adverse noise conditions. The method consists of three main parts: 1) derivation of Spatial Time-Frequency Distribution (STFD) matrix; 2) application of the forward-backward spatial smoothing technique; 3) estimating the angles of arrival by solving for the roots of the polynomial. The key significance of the proposed method is that the combination of existing methods and techniques allows an estimation of DOA angles for both coherent and non-stationary source signals under noise. Whereas the individual use of the existing methods does not show adequate performance under these conditions. The experiments allow studying the performance of the proposed method for 1) both coherent and non-coherent clean sinusoidal signals; 2) noisy non-stationary chirp signals; 3) coherent and non-stationary signals under noise. Furthermore, extensive simulations have been carried out to compute the root mean square error (RMSE) performance of the proposed method in comparison with the existing ones. The experiments have been designed for varying number of microphones, level of noise, and value of the time-frequency threshold. As a result of the experiments, we observe the efficacy of the proposed method in comparison with the conventional Root-MUSIC and Time-Frequency MUSIC (TF-MUSIC) methods.