A Recognition Method For Radar Emitter Signals Based on EEMD and EfficientNet
Bing Luo, Lihua Wu, Yuan Yuan, Rui Lü · 2020 5th International Conference on Mechanical, Control and Computer Engineering (ICMCCE) · 2020
Aiming at the weak representation ability under low signal noise ratio (SNR) in radar emitter signals recognition, a recognition method based on Ensemble Empirical Mode Decomposition (EEMD) noise reduction and EfficientNet is proposed. First, the appropriate EEMD parameters to denoise the time domain signals and Ambiguity Function Contour Map of Tangent Matrix (AFCMTM)' datasets are created. Then, a Deep Network based on EfficientNet-B0 model is established and Transfer Learning is used to complete the training of labeled data. Finally, the network is used to achieve radar emitter signals identification. The simulated experiments show that the average recognition accuracy rate of six kinds of complex modulated signals, i.e., BPSK, BFSK, FMCW, QPSK, LFM-BC and MSEQ, by proposed method keeps above 96.83% in fixed SNR environment above -5dB with good generalization ability and strong robustness.