Deep learning (DL) based direction of arrival (DOA) estimation has drawn increasing attention recently because of its significant potential advantages. However, existing DL-based DOA estimation algorithms may have the problems of network training difficulty and low estimation accuracies for multiple signal sources. To address these issues, a DOA estimation algorithm using an interpretable spatial filtering network (SF-Net) and parallel small-scale two-dimensional (2D) convolutional neural networks (CNNs) is proposed. Based on matrix spatial filtering (MSF) theory and single-layer fully connected neural networks, an interpretable SF-Net with the same physical foundation to MSF is constructed to divide the array received data into different angle subregions. Then, parallel 2D CNNs are used to process the output data of SF-Net, obtaining the spatial spectrum estimation of each angle subregion. Finally, spatial spectra of different subregions are synthesized to achieve accurate DOA estimation of multiple sources. Experiment results show that the proposed algorithm can achieve higher multiple source DOA estimation performance with reduced dataset size and network training difficulty in comparison with the state-of-the-art DL-based estimation algorithms. Compared to the conventional estimation algorithms, the proposed algorithm has the advantages in the case of low signal-to-noise-ratio, small number of snapshots, and small source angle separation.

Multiple-input multiple-output (MIMO) radar three-dimensional (3D) imaging is widely applied in military and civil fields. However, MIMO is easily affected by wideband interference (WBI). To solve this problem, in this study, we propose a sparse recovery imaging method with WBI prediction based on the predictive recurrent neural network (PredRNN) and the tensor-based smooth L0 (TSL0) algorithm. Firstly, we extract the time-frequency (TF) feature of the historical measured WBI via the short-time Fourier transform (STFT) operation. In this way, we can use PredRNN to exploit the spatiotemporal correlation of the WBI in the TF domain to predict the TF feature of the WBI in the future. Then, we adaptively design the random sparse stepped frequency waveform by selecting non-overlapped frequencies with the WBI according to the predicted WBI TF feature. Finally, we apply the TSL0 algorithm to reconstruct the 3D high-resolution target image from the sparse signal cube. Simulation results show the high performance and robustness of the proposed imaging method in the presence of different WBIs.