Research And Implementation For Algorithms Of Identification And Estimation For Pulse Compression Radar Signals

Pulse compression radar with low probability of intercept brings a new challenge to the modern ARM seeker and other passive surveillance receivers. In present study, the intra-pulse modulation feature analysis and processing of pulse compression radar signals in ARM seeker were carried out, taking BPSK and QPSK signals, LFM and NLFM signals as main objects. And the recognition of modulation types and their corresponding further processing were studied, including the parameter estimation for PSK signals, detection and parameter estimation for frequency modulation signals. Additionally, some hardware experiments were done for the processing of pulse compression radar signals, which supplied necessary parameters for sorting, intercepting, matching and tracking of pulse compression radar signals in ARM seeker.This study was shown around the block diagram of processing system of pulse compression radar signals. For the recognition of modulation types, the intra-pulse modulation feature of signals was analyzed, and a recognition method from the rough to the detailed was proposed. It firstly classified signals roughly into PSK signals and FM signals according to the feature of spectrum band, and then got the detailed classification in each group. The method is easy for computation and valid for performance, which is very valuable in engineering application. Besides, as to statistical pattern recognition, a full recognition algorithm based on adaptive resemblance coefficient was proposed, which realized the modulation type recognition by constructing joint feature distribution and setting up a decision rule. These methods provided the necessary prerequisite for later choosing corresponding and efficient algorithms of parameter estimation.On the parameter estimation of PSK signals, using wavelet theory, a double-scale wavelet transform was proposed for signals with carrier frequency which could locate singularities precisely by combining rough location and accurate location. However, this method is difficult to choose scales and has poor performance of lower SNR. Thus, an estimation algorithm of symbol rate for MPSK signals without carrier frequency was put forward based on product multi-scale wavelet transform, which could extract singularities and estimate symbol rate by FFT transform in lower SNR, and it was easier and rule-based to choose scales. Finally, a parameter estimation method based on instantaneous autocorrelation with time-domain addition was proposed, which could extract singularities through the cumulative time-domain waveform and could also estimate symbol rate by FFT transform. It is easy and practical, and has good estimation performance, which is applied in the following hardware platform.On the detection of PPS, a method of wavelet reassignment and Radon transform was proposed to detect multi-component LFM signals, which improves the aggregation of time-frequency distribution effectively, inhibits noise and enhances the detection performance obviously. Due to the disadvantage of WVD transform in the detection of multi-component LFM signals and NLFM signals, the detection methods of PWVD and LWVD transform were discussed, and a detection method called Product Spectrogram-WVD(PSWVD) transform for multi-component PPS was proposed. This new detection method integrated the superiors of spectrogram and WVD, which removed cross-terms of WVD and held good frequency aggregation and anti-noise performance. On the parameter estimation of PPS, the CPF method was mainly discussed. And an estimation algorithm and its fast implementation were proposed for multi-component LFM signals based on weighted mean CPF. In order to be used in practical discrete sampling system, CPF's FFT algorithm was deduced and improved by rounding the nearest sampling point.Additionally, the hardware implementation of processing of pulse compression radar signals was put into practice. On the hardware platform of processing of radar signals constructed by 4 chips of 300MHz ADSP-TS101S, the recognition of modulation types was realized according to the feature of spectrum and band, the parameter estimation of BPSK signal was achieved by instantaneous autocorrelation with time-domain addition, and the parameter estimation of LFM and NLFM signals was implemented based on CPF. Joint test between the wideband digital channelized receiver and the parameter estimation hardware platform was done successfully.