Fast Subspace DOA Algorithm Based on Time-frequency Distributions without Eigen Decomposition

Abstract

The time-frequency distribution (TFD) method has received a considerable amount of attention in Direction of Arrival (DOA) estimation of multi-senor and time space processing, which lead to improved spatial resolution and source separation performances. Based on the characteristic of the different time-frequency distribution in time-frequency domain, a novel approach of Time-frequency DOA estimation is proposed. Through computation of the high order power of spatial time-frequency matrix, the signal subspace or noise subspace is estimated without the eigenvalue decomposition (EVD) of spatial time-frequency matrix. Compared with the conventional approaches of DOA estimation based on EVD, the presented method has lower computational complexity without performance degradation. The effectiveness and high resolution is verified via computer simulations.