Radar Artificial Modulation Signal Recognition Based on Multi-Source Feature Fusion Network
Xinyuan Su, Sinong Quan, Zhihao Cai, Weize Meng, Shiqi Xing · 2024
The recognition of radar artificial modulated signals, achieved through comprehensive analysis and precise identification of the modulation characteristics inherent to radar signals, holds critical importance for differentiating enemy radar emissions and enhancing the anti-jamming capabilities of radar systems. Nevertheless, current recognition methodologies, including support vector machines and long short-term memory network (LSTM) networks, exhibit certain limitations in terms of accuracy. To address this issue, we propose a multi-source feature fusion (MSFF) network aimed at augmenting the recognition performance of artificially modulated radar signals. This network employs short-time Fourier transform (STFT) techniques to process the artificially modulated radar signal and utilizes phase spectrum, real part, and imaginary part as input features to effectively extract essential signal characteristics, thereby significantly improving recognition accuracy. Simulation experiments demonstrate that this network achieves a remarkable recognition accuracy of 99.15%, showcasing its superior effectiveness compared to six typical neural networks.