Specific Emitter Identification of Frequency Hopping Signals Based on Feature Extraction and Deep Residual Network

Abstract

In the modern war with complex and changeable electromagnetic environment, Specific Emitter Identification (SEI) is an important and difficult problem, which is of great significance in obtaining intelligence information, identifying ourselves or foe, and specifying combat plans. In order to solve the problems of low accuracy, poor robustness and difficulty in extracting effective individual features of frequency hopping (FH) radio set, this paper studies the generation of PSK signal constellation to proposes a radio frequency fingerprint(RFF) based on constellation. We also improved the existing RFF extraction method, combined with the deep learning recognition method based on the Deep Residual Network (ResNet), to achieve effective feature fusion. By comparing different input methods, we found that the recognition accuracy is improved, and reaches 96.16% in the outfield experiments of 15 FH radio sets. In addition, we also designed a ResNet structure to compare the recognition accuracy under different signal-to-noise ratio(SNR), different network structure, different number of individuals, different modulation methods and different recognition algorithms, which proved the superiority of our method.

Small Sample Fiber Full State Diagnosis Based on Fuzzy Clustering and Improved ResNet Network

“…The residual network (ResNet) is a kind of deep CNN, due to the introduction of residual blocks, which overcomes the problem of gradient disappearance during training. The literature [15,16] applies the powerful feature extraction ability of ResNets for feature extraction. The literature [17] applied the ResNet for fault diagnosis of turnouts, but the pooling layer could not adapt to the input data size.…”