Radar Emitter Signal Fingerprint Feature Extraction And Recognition

With the rapid development of Electronic countermeasure technology,the faint fingerprints feature of radar emitter put forward more requirements for the technique to analysis radar emitter signals.How to analyze intrinsic characteristics in pulse of radar emitter signals and extract fingerprint characteristics for identification using more effective ways have been become hot subjects in the research field of Radar emitter recognition.For this problem,this paper analyzed the cause of fingerprint characteristics of individual radar emitter and then extracted fingerprint characteristics for identification,then Made a in-depth research on methods for fingerprint characteristics extract and identification by theoretical simulation analysis and experimental data processing.According to the diversity of radar emitter intentional modulation,this paper design three different type of emitter with five different signals(LFM?NLFM?BPSK?QPSK?CW),building SSB phase noise power spectral density model,and proving that while the radar emitter with different specification transmit signals,they had tiny different fingerprint which caused by the phase noise that the individual itself produced,so this results provided theoretical support for extract and identification of the radar emitter fingerprint characteristics.Respect to radar emitter signal with phase noise,this paper extract feature using modified Rife algorithm(M-Rif algorithm)through simulation,comparing it with the feature extracted by traditional Rife algorithm,and get the result that M-Rife algorithm has more advantages than traditional one.In the meanwhile,we verified that different radar emitter individual which transmit the same signal have different fingerprint characteristics.Through builting signalphase noise spectrum density model on the basis of the feature BPSK and QPSK signal,the feasibility of M-Rife algorithm to extract fingerprint characteristics of BPSK and QPSK signal had been testified by theoretical model simulation and the measured data analysis.This paper also analyzed the advantages of the application of bispectral analysis on radar emitter fingerprint characteristics extract and identification,and proposed using local contour integral algorithm to further simplify the calculation in the feature extraction of bispectrum estimation and extracted energy entropy,waveform entropy and singular value entropy fromthe bispectrum to constitute three feature vector for emitter individual identification.This paper designed the SVM based on refactoring kernel function,combined with frequency shift feature extracted by M-Rife algorithm and by the three dimensional feature vector extracted by bispectral method,respectively,correctly identified and classified different radar emitter individual which launch the same waveform.Meanwhile,this paper also discussed the influence of signal modulation type to the fingerprint characteristics when different modulation signals were launched by the same one radar emitter.