A super-resolution DOA estimation method for low SNR and small-sized arrays

We address the super-resolution DOA estimation of closely spaced sources at low SNR with a small-sized array. In our method, a downsampling technique (DT) is implemented in the spatial frequency domain and then we can choose the downsampling factor (DF) with more flexibility. Then we use IDFT to realize a nonlinear variable-space transformation (NVST), based on which the estimated covariance matrix is formed. Finally, DOAs are estimated based on the MUSIC algorithm. The source separation is artificially increased through DT in the spatial frequency domain and NVST. Simulation results show the proposed method performs significantly better than the beamformer (BF) and MUSIC algorithm for closely spaced sources at low SNR with small-sized arrays.