Research On Advanced Radar Signals Recognition And Parameter Estimation

According to the development of ELINT and electromagnetic spectrum sensing under the complex electromagnetic environment background in future, the thesis make an in-depth study on some issues of complex radar signal recognition and parameter estimation. Because of complex modulation signal application in modern radar system, ELINT, characteristic of non-cooperative receiver, make the received signal in low SNR. So radar signal analysis and identified is carried out in this study based on the "complex modulation" and "low SNR". The main research contents and tasks are as follows:First, the image processing technology is introduced in the radar signal recognition to achieve the radar signal identification by time- frequency image recognition. The core of the problem is to extract stable and efficient image features. Two radar signal recognition method are Proposed: time- frequency image shape features Bas ed and instantaneous frequency extracted from time- frequency image based radar signal recognition. In the first proposed recognition method, first, image enhancement, adaptive threshold binarization and mathematical morphological open ing operation are processed with the SPWVD time- frequency image of radar signals to remove image noise, and then use a kind of orthogonal moments, called Legend moments to efficiently characterize the denoised image shape. The Legend moment features can give an efficient identification to eight kinds typical radar signals when SNR greater than-3dB. While in the second recognition method, the fact that time- frequency energy gather to the direction of instantaneous frequency is used. Adaptive threshold binarization, mathematics morphological opening, skeletonized and deburring operation is processed with Choi-Williams time-frequency image of the radar signals to obtain the instantaneous frequency character. Compared with two-dimensional time- frequency image, the extracted instanta neous frequency feature of radar signals is very stable and is one-dimensional, which can effectively identify the seven current widely used typical radar signal when the SNR is greater than 0dB.Secondly, Pseudo Random Binary Code and Linear Frequency Mod ulation combined signal(PRBC-LFM) is a class of complex radar signal that Low Probability of Intercept(LPI) radar systems have been used. PRBC-LFM signal relates to continuous and pulsed in two ways. Respectively, two parameter estimation methods are proposed. For pulse type PRBC-LFM signal, by analyzing the spectrogram characteristics of the signal, the author propose a parameter estimation method based on spectrogram analysis, and study the parameter estimation performance effect of different spectrogram window function length. Meanwhile, to objectively evaluate the performance effect of estimation method, the CRLB theoretical derivation of the parameter estimation is given. While, for continuous PRBC-LFM signal, the fractional Fourier transform(FRFT) and cumulative integral based parameter estimation method is proposed. In the estimation method, after square processing of the signal, chirp parameters are estimated by FRFT. And then based on the estimated chirp parameters, the composite signal can be tra nsformed into a baseband pseudo-code signal with eliminating chirp component. The cumulative integral can be used to estimate all pseudo-code parameters.Thirdly, using the FRFT matching characteristics for LFM signal and coherent integration ideas in signal processing, the author proposes a new signal processing method, called Period FRFT. Period FRFT can achieve best match to a linear frequency modulated continuous wave(LF MCW) signal. With better advantage in signal processing gain than FRFT, Period FRFT can be used for LFMCW detection and estimation in low SNR environment. According to the period FRFT characteristics of LFMCW signal and without relying on the interception assumption of signal start time offset is equal to zero, the periodic FRFT based signal detection and estimation algorithms is proposed. And the theory analysis shows that the estimation algorithm is maximum likelihood estimation(MLE) and asymptotically optimal estimation to LFMCW signal. In addition, to solve the problem of zero signal start time offset implicit assumption in existing symmetrical triangular linear frequency modulated continuous wave(STLFMCW) signal parameter estimation method, a new FRFT based parameter estimation method for STLFMCW signal in a non-zero start time offset is proposed, which utilizes the "time- frequency symmetry" of STLFMCW signal.Finally, according to similar time- frequency structure of with LFMCW, the Period FRFT based parameter estimation method is further extended to the STLFMCW and polyphase code(Frank, P1-P4 code) continuous wave signal. And the author proves that the estimation algorithm is also maximum likelihood estimation(MLE) and asymptotically optimal estimation.