DOA Estimation: LSTM and CNN Learning Algorithms

In the domain of array signal processing, impulsive noise represents a pervasive challenge that undermines the accuracy and reliability of the direction of arrival (DOA) estimation techniques. To address the need to increase the performance of DOA estimation across various application scenarios, a new deep learning-based algorithm is proposed. This algorithm consists of two parts: impulsive noise suppression and DOA estimation. A long short-term memory (LSTM) network is proposed to suppress impulsive noise. The inputs are the array output signals containing impulsive noise, and the outputs are the impulsive noise separated from the input signals. Since the convolutional neural network (CNN) can then learn the spatial features of the signals and perform advanced feature extraction, a new CNN model is proposed to obtain the DOA estimation. To assess the performance of the proposed algorithm, simulation experiments are conducted. The results demonstrate that, compared with existing algorithms, the proposed algorithm substantially improves the accuracy and robustness of DOA estimation under impulsive noise environments.