DOA estimation of closely-spaced and spectrally-overlapped sources based on time-frequency sparse representation
Ling Zheng, Haijian Zhang, Hong Sun
Abstract
For the purpose of dealing with closely-spaced and spectrally-overlapped sources, a direction-of-arrival (DOA) estimation algorithm based on time-frequency (TF) sparse representation is proposed. Firstly a short-time Fourier transform (STFT) based single-source TF points selection method is briefly introduced. On this basis, we extract the STFT values corresponding to the single-source TF points of each source from the STFT values of array outputs to construct received data matrix. We then enforce sparsity by imposing the l1 norm based penalties on TF sparse signal representation and solve the optimization problem. Finally, the DOA of each source can be estimated from over-complete basis according to the peak of TF sparse signal vector. Simulation results demonstrate the advantages of the proposed algorithm in terms of dealing with closely-spaced and spectrally-overlapped sources and the flexibility in underdetermined cases.
BibTeX
@inproceedings{icassp2016_doaestimationofc,
title = {DOA estimation of closely-spaced and spectrally-overlapped sources based on time-frequency sparse representation},
author = {Ling Zheng and Haijian Zhang and Hong Sun},
booktitle = {ICASSP 2016},
year = {2016}
}