Recognition of radar in-pulse modulation based on radar de-noising Linknet

Intra-pulse modulation pattern recognition of radar signals is an important aspect of modern electronic reconnaissance warfare. For radar intra-pulse modulated signals with low signal-to-noise ratio (SNR), traditional machine learning and deep learning networks do not preprocess the signals properly, resulting in low recognition accuracy. To solve this problem, this paper proposes a semantic segmentation network-based method to improve the recognition rate of modulation patterns by preprocessing the semantics of signal time-frequency maps. The experimental results show that the method can effectively recognize 12 radar modulation patterns at low SNR, and the pattern recognition accuracy is as high as 98.6% for signals with SNR of -10db.