Research On Fingerprint Features Extraction And Identification Method Of Radar Emitter Signal

Radar Emitter Individual Identification(REII)is a key technology in modern electronic reconnaissance system.It aims to identify the specific radar emitter individual that transmit the signal based on the fingerprint features that reflect individual differences attached to the signal.With the increasing complexity of electromagnetic environment and the upgrading of intelligence and automation level of electronic reconnaissance technology,radar emitter signal waveform changes are complex,parameters are changing rapidly,and features are more concealed,which brings new difficulties and challenges to REII.The key technologies in REII are the focus of this thesis,the theory and method of signal fingerprint features extraction are studied,the application of deep learning technology in REII is explored,and the good performance of the proposed identification scheme is verified through measured data.The main research contents and contributions of this thesis are as follows:Firstly,the related researches on intra-pulse modulation and fingerprint features of radar emitter signal are fully investigated in this thesis.The definition of fingerprint features is given,the intentional modulation and unintentional modulation of radar emitter signal are analyzed,and the unintentional modulation of the signal is simulated.Furthermore,the main sources and characteristics of fingerprint features are elaborated in detail,and a comprehensive analysis of the feasibility of extracting fingerprint features from each source in real environment is made.Secondly,the fingerprint features extraction method of radar emitter signal is studied,and Empirical Mode Decomposition(EMD)algorithm is improved to reduce the computational complexity of fingerprint features extraction process.By designing simple and effective feature formulas,the Hilbert-Huang Transform(HHT)time-frequency spectrum and bispectral features are extracted to construct a signal multi-domain fingerprint feature set.The extracted features are analyzed,and the visualization results show that the extracted features have good discrimination and can reflect the differences of different radar emitter individuals.Experimental results show that the optimal identification accuracy for signals from different radar emitter individuals reaches 96.44%,which is more than 10%higher than the comparison method.Finally,the application of deep learning in REII is discussed in this thesis,and then a REII method based on deep learning is proposed.The structure and parameters of the CNN model used were optimized in order to enhance the network's ability to extract signal fingerprint features,improve identification accuracy,and reduce identification time.Experimental results show that the identification accuracy of the radar emitter individual signals reaches 99.56%by the optimized CNN.Moreover,the average identification time of each signal is only about 0.1 s,which can meet the needs of practical applications.The deep learning-based identification method proposed in this thesis has strong generalization ability and adaptability,which provides a new,simple and effective solution for REII.