DOA estimation of LFM signals based on time-frequency points clustering

The multiple signal classification (MUSIC) algorithm based on spatial time-frequency distribution (STFD) is investigated for direction-of-arrival (DOA) estimation of non-stationary signals, and its key step is to select the appropriate time-frequency points. Aiming at the problems that traditional time-frequency MUSIC (TF-MUSIC) algorithm can not extract the time-frequency points of each source and its poor performance in the case of small angle spacing, this paper proposes a novel DOA estimation algorithm for line frequency modulation(LFM) signals based on time-frequency point clustering. Firstly, the algorithm whitens the array receiving signals, and constructs the STFD matrix using the whitened receiving signals, which can suppress the cross-terms and give prominence to the auto-terms. Then, the algorithm extracts the time-frequency points of each signal by utilizing K-means-clustering. Finally, the MUSIC algorithm is used to estimate the DOA. The root mean square error(RMSE) of three different algorithms in different angle interval and different signal-to-noise ratio(SNR) are simulated respectively. Compared with two classical time-frequency music algorithms, this algorithm has better estimation performance at small angle intervals and low SNR.