An Improved Method of Radar Emitter Fingerprint Recognition Based on GS-SVM
Xinyue Wang, Chang Su, Songlin Sun · 2019
Radar emitter recognition is a subject of great importance in modern electronic warfare. To realize radar emitter fingerprint recognition, we propose a method of extracting fingerprint features containing statistic characteristics, carrier frequency and wavelet packet transform (WPT). Furthermore, we discuss the selection of kernel functions of the support vector machine classifier (SVM). The grid search algorithm (GS) is applied to optimize the SVM classifier. Finally, the real data experiment results show that the accuracy of test radar signals achieves 99.92% based on GS-SCM which performs better than the simple SVM model in the case of small samples to adapt to the battlefield environment.