| With the rapid development of modern electronic technology and its widely used in themodern war, the modulation of the new radar signal source is more flexible, diverseparameters, and the electromagnetic signals environment in modern battlefield is complicated,so that the radar signal recognition technology based on the traditional parameters could notmeet the real needs. Therefore, exploring new recognition methods of identification in theelectronic warfare, in order to improve the radar countermeasures equipment technologicallevel, makes the radar signal source individual recognition technology emerge as the timerequire, and develop rapidly. The theory and method of intentional modulation andunintentional modulation of radar signal recognition is analysed and researched deeply. Themain research work of this paper is summarized as follows:Considering two typical intentional modulation of radar signal recognition, Phase ShiftKeying (PSK) signal and Frequency Modulation (FM) signal, a classified recognition methodfrom rough to the detail is proposed in this paper. Considering the significantly differentcharacteristics of the3dB bandwidth of frequency spectrum between PSK signal and FMsignal, rough classification and recognition are applied firstly. Based on this, for the differentwithin-class characteristics of PSK signal and FM signal, a method based on wavelet ridgefrequency characteristics is proposed in this paper, PSK signal is classified and recognized asthe Binary-Phase Shift Keying (BPSK) signal and Quarternary-Phase Shift Keying (QPSK) indetail, FM signal is classified and recognized as Linear FM (LFM) signal and Nonlinear FM(NLFM), thus all the classification and recognition process are completed.In the process of intentional modulated radar signal recognition, instantaneous frequencyprobability density distribution model of additive white Gaussian noise (AWGN) channelcarrier signal is studied in this paper, and verified the correctness of distribution model byhypothesis testing. The instantaneous frequency probability of density distribution is analyzedon different sampling rate and SNR. According to the conclusions of the analysis, a methodbased on extraction and reconstruction of the instantaneous frequency estimation algorithm isproposed, and the performance of the algorithm is analyzed in detail. In this paper, themulti-carrier signal instantaneous frequency aliasing as well as unable to distinguish is alsoanalyzed, and derived the limit formula of the instantaneous frequency aliasing completely.Finally through the Monte Carlo simulation verified the correctness of estimation algorithm advanced.In the case of the completed type identification signal parameter estimation, the parametersestimation algorithm of LFM radar signal based on modified cubic phase function is proposed.Cubic phase function (CPF) algorithm can be used to achieve parameters estimation of LFMand NLFM signals, this method can extract the phase function coefficients of such signals.CPF conducts parameter estimation with parameter space maximum value by second ordernonlinear transformation, which has high precision and acquires the ability of resisting noise.The computation of this algorithm is very large, for this question, through extracting the pulseleading edge and pulse width information by using channelizing, FM slope of the searchrange setting criterion is proposed. In addition, a method called dual-scale FM slope search isbrought to reduce the computation, at the same time, this method can ensure the accuracy ofthe algorithm in low SNR.Considering conventional pulse repetition interval(PRI) characteristics of radar, time ofarrival (TOA) difference method recognizing radar signals source individuals is applied in thispaper. Considering carrier frequency offset characteristics of radar signal, this paper appliesfast fourier transform (FFT) method for recognizing radar signals source individuals.Considering waveform characteristics of the radar pulse signal is used to recognize radarsignals source individuals, based on the analysis of the actual measured pulse, By the analysisof the most representative signals source individual characteristics, the pulse signal risingedge of the envelope waveform characteristic is extracted, the time domain characteristics andfrequency domain characteristics are extracted to recognize using the Hausdorff distancemethod in this paper. the result which is mentioned above is given in the correspondingprobability of correct recognition in this paper.Considering unintentional modulation of radar signal characteristics caused by the phasenoise, the double spectrum analysis method based on transform domain is proposed in thispaper, and the signal double spectrum calculation method and its physical significance is alsogiven. Surrounding-line integral method was applied to optimize the calculation results ofdouble spectrum analysis, centroid distance method and FCM clustering method forsubsequent identification. Finally, the results of computer simulation and actual data verifythe correctness and effectiveness of the method in this paper. |
