A Study of Radar Anti-Jamming Based on Deep Convolutional Mix Separation Network

In modern warfare, active jamming can significantly impact the performance of radar systems. In the case of a low signal-to-noise ratio (SNR), suppressing radar jamming becomes more challenging, especially when dealing with mainlobe jamming. The traditional mix source separation (MSS) algorithm exhibits poor anti-jamming performance or may even fail in such scenarios. In the case of unknown prior information, to address this problem, this article integrates the concept of MSS with deep learning techniques to design a deep convolutional mix separation network based on the attention mechanism. End-to-end feature extraction, feature separation, and time-domain waveform recovery are performed on the received signal to separate the pure echo signal and jamming signal from the received signal. Finally, the effectiveness of the proposed method is verified through simulation. The results demonstrate that the proposed method can effectively suppress various types of mainlobe jamming under low SNR conditions. The correlation coefficient exceeds 0.9, indicating its superiority over traditional anti-jamming methods.