Deep Learning and Time-Frequency Analysis Based Automatic Low Probability of Intercept Radar Waveform Recognition Method

As a critical component of modern electronic warfare, radar plays an essential role in target detection. Due to the characteristics of low interception, the low probability of interception (LPI) radar has better stealth; on the other hand, it challenges the accurate identification of its signal waveform simultaneously. Existing LPI radar waveform recognition approaches are considered inadequate under sufficient channel conduction. In this paper, we selected 13 radar waveforms under 31 signal-to-noise ratios, utilized the time-frequency analysis method of CWD (Choi-Williams Distribution) transform for signal processing, and used our proposed convolutional neural network LeeNet for classification. The results show that the proposed method can achieve an accuracy higher than 96% when the signal-to-noise ratio is 0 dB while performing better at all signal-to-noise ratios than other deep learning architectures.