Research On Feature Extraction And Classifier In Individual Identification Of Radar Emitter

Individual identification of radar emitter is the key to obtain enemy information in modern electronic warfare.It can extract electromagnetic characteristics of radar signal and match it with individual equipment of specific emitter.With the increasing complexity of electromagnetic environment and the gradual diversification of radar system,the traditional method of extracting radar signal pulse descriptors and using template matching method can no longer meet the current requirements of accurate identification of individual radar emitter.Therefore,the key technologies,feature extraction and classifier design in individual identification of radar emitter,are studied by this paper.The main work and contributions are as follows:Firstly,this paper investigated the theoretical basis of individual identification of radar emitter,analyze the basic structure of radar emitter,focused on two kinds of radar transmitter structure,and analyzed the process of radar signal from generation to emission.The modulation mode of radar signal is studied by signal simulation and mathematical modeling,and it is pointed out that the extraction of fingerprint features generated by unintentional modulation is the key to achieve accurate identification of individual radar emitter,and the characteristics of fingerprint features are described,which lays a theoretical foundation for the subsequent effective extraction of radiation source features and classifier design.Secondly,for the feature extraction of the radiation source signal,this paper combines the feature extraction method of signal transformation with the automatic feature extraction method of auto-encoder,and proposes an improved method based on time-domain statistical features,frequencydomain bispectrum features and wavelet packet coefficient energy in timefrequency domain.The proposed multi-domain feature extraction method is applied to the process of individual identification of radiation sources,compared with other feature extraction methods in literature,and the classifier is designed by K-nearest neighbor algorithm,decision tree algorithm and support vector machine algorithm.The experimental results show that when the multi-domain feature extraction method is adopted,and the classifier is a decision tree algorithm,the performance of individual identification of radiation sources is the best,and the identification accuracy rate reaches 94.12%,which is much higher than that of the comparative literature.Finally,in order to further improve the accuracy of individual identification of radar radiation sources,this paper introduces the idea of ensemble learning in machine learning into the classifier design.The selection of base learners in the stacking algorithm is studied,and a classifier design method based on the stacking algorithm is proposed.The random forest algorithm and the extreme gradient boosting algorithm(XGBoost)algorithm are used to design the classifier,and the algorithm model is optimized to improve the accuracy of individual identification of radiation sources.Based on the measured radar radiation source data,the recognition accuracy of the random forest algorithm and the XGBoost algorithm has reached more than 97%,and the recognition accuracy of the Stacking algorithm has reached 99.21%.