Research on Radar Communication Jamming Recognition based on Neural Networks

The identification of communication jamming is a pivotal aspect of cognitive electronic warfare, which is of paramount importance for the advancement of jammer and radar capabilities. Traditional methodologies are significantly influenced by human variables, are highly subjective, and are challenging to adapt to the contemporary battlefield. In this work, we propose a convolu-tional neural network (CNN)-based method for recognising interference in radar communication. The method focuses on the automatic extraction of in-terference features through deep learning models, thereby overcoming the limitations of traditional manual feature engineering. Convolutional neural networks are capable of directly processing the two-dimensional time-frequency map of radar signals, thereby enabling the extraction of salient features in frequency and time distributions without the necessity for manual preprocessing. The superposition of multiple convolutional layers enables the model to progressively extract more sophisticated high-level features, thereby facilitating the automatic identification of pivotal signal characteris-tics within intricate interference patterns. The proposed method exhibits a high degree of generalisation ability across multiple types of interference scenarios and is capable of effectively dealing with complex scenarios such as frequency hopping and modulation interference. The experimental results demonstrate that the proposed CNN model displays superior recognition abil-ity in a variety of interference signal environments, particularly in the pres-ence of high noise and nonlinear interference. Furthermore, the recognition accuracy is significantly superior to traditional methods.