Time-frequency Aliasing Separation Method of Radar Signal Based on Capsule Neural Network

To solve the time-frequency aliasing problem of various radar signals in a complex electromagnetic environment, a radar signal separation method based on capsule neural network is proposed. In this paper, the capsule neural network (Caps Net) is briefly introduced as preliminary. In the proposed method, the radar signal is firstly preprocessed to build a deep feature space, then pulses are separated based on a Density-Based Spatial Clustering of Applications with Noise (DBSCAN), and used the capsule neural network to separate the aliasing points in the cluster. The performance of the proposed method is verified by numerical simulations. The results show that it can achieve an effective separation of 8 mixed radar signals. Compared with traditional method, the accuracy is significantly improved, and the recognition probability is over 95% with SNR=15dB.