Open set recognition of radar active jamming signals based on relative entropy

作者：Zhongyi Guan, Shengqi Zhu, Lan Lan, Ximin Li, Yuxiang Gao, Xiangqun Zhang, Yanxing Wang · 发表于：IET conference proceedings. · 年份：2024 · DOI：10.1049/icp.2024.1703 · 被引用次数：2 · 研究领域：Wireless Signal Modulation Classification、Radar Systems and Signal Processing

It is of great significance to recognize radar active jamming signals in complex electromagnetic environment. However, the existing methods are mainly based on expert knowledge and a closed set, which are affected by environmental factors and unrealistic. In this paper, an open-set recognition method for active jamming signals is studied based on relative entropy. First of all, after performing the short-time Fourier transform (STFT) and moving target detection (MTD), the time-frequency and Range-Doppler (RD) map of the active jammers are obtained. Then, two deep learning-based jamming recognition branches, including the time-frequency and the RD, are trained and validated, where recognition probabilities are obtained for both branches with the same input. Furthermore, the relative entropy between the recognition probability and the ideal probability is utilized as the confidence level, and the open set recognition is realized through a voting mechanism which recognizes the probability and relative entropy. Numerical simulations show that the recognition probability can reach over than 90% at jammer-to-noise ratio (JNR) higher than 10dB.