Enhanced Radar Signal Recognition through Attention-Driven Multi-Domain Fusion Mechanism

With the advancement of radar electronic countermeasure technology, the survivability of radars faces significant challenges, leading to the widespread application of Low Probability of Intercept (LPI) radar technology. However, due to the complex technology it employs, recognizing LPI radar waveforms remains a challenging task. Most existing studies extract single-domain features based on traditional convolutional feature extraction methods. This results in poor feature extraction capability and low recognition accuracy, especially in low signal-to-noise ratio (SNR) environments. To address this issue, this work proposes a novel method for radar waveform recognition based on an attention-driven multi-domain fusion mechanism. Firstly, the multi-domain information of LPI radar waveform, including time-frequency image (TFI) and frequency sequence (FS), is introduced to enhance the feature extraction capability. Additionally, the Efficient Channel Attention (ECA) mechanism is introduced to improve the feature extraction ability. Simulation results demonstrate that the proposed method outperforms single-domain networks and feature extraction methods based on traditional convolution.