Deep Networks for Direction of Arrival Estimation With Sparse Prior in Low SNR

This work introduces direction of arrival (DOA) estimation considering the sparsity prior in low signal to noise ratio (SNR) using deep learning (DL). The representation can be prompted realize super-resolution DOA spatial spectrum. undersampled noisy linear measurements spectrum, which is formulated by columns array covariance matrix, given as input neural network for estimation.The convolutional layers enable feature extraction and then nonlinear relationship between spectrum corresponding learned training network. After offline phase, proposed method performs well response signals any amount snapshots SNR. results experiments have demonstrated superior performance compared state-of-the-art methods including traditional superresolution existing DL-based methods.