Parameter Estimation For Linear Frequency Modulation Signal

Linear frequency modulation (LFM) signal is one of the most popular time-varied signals. It has widespread application in radar, communication, sonar and seismic exploration, etc. As a mature low probability of intercept (LPI) signal, LFM signal plays an important role in most kinds of radar. Consequently, it has great practical significance and remarkable application value for parameter estimation of non-cooperative LFM signal or LFM signal without prior information.Scholars at home and abroad have done intensive research on LFM signal processing for years. They have achieved many fruits on the theory of LFM signal parameter estimation. It is still in the process of continually improvement and development. There have been many methods of parameter estimation of conventional mono-LFM or multi-LFM signals at present. However the methods still have some weakness for the new kind of radar signals, such as complex modulation signals, wideband (ultra-wideband) signals, etc. The study of this paper carries on with the research foundation under this background. The study is concerning on the parameter estimation of complex modulation LFM signals and wideband (ultra-wideband) LFM signals. The main contributions are illustrated as follows:1. The concentration performance depends on the choice of the window functions in the smoothed pseudo Winger-Ville distribution (SPWVD). In order to obtain the optimal time-frequency performance of the SPWVD for a signal without prior information, a new feature value for the determination of the SPWVD maximum concentration is proposed. The proposed feature value can eliminate the limitation of the existing concentration measures. The genetic algorithm (GA) is applied to search the optimal SPWVD performance automatically based on the feature value.2. Convolution signal combined linear frequency modulation and pseudo random binary phase code (LFM-PRBC) is one kind of complex modulation signals with the advantage of both LFM signal and PRBC signal. In order to estimate the parameters of such signal, a novel method based on the SPWVD transform and correlation detection is proposed. The SPWVD characteristics of LFM-PRBC signal are analyzed. First, the above method is used to obtain the optimal SPWVD performance for the LFM-PRBC signal without prior information. Then the SPWVD is segmented by fixed threshold to estimate LFM parameters. Finally, the correlation detection is used to decode the binary phase codes.3. Since the theoretical threshold binarizing the time-frequency distribution (TFD) has not been evaluated in most of the methods so far proposed, a method of TFD binarization by a dynamic threshold based on Otsu algorithm is proposed. The new method is proposed from the viewpoint of the discriminant analysis. It can do the parameter estimation of Multi-LFM signals without prior knowledge effectively.4. Restricting by the current A/D sampling technology, parameter estimation of wideband (ultra-wideband) LFM signal has significant research value. A novel parameter estimation method based on wavelet transform and compressed sensing (CS) theory is proposed. Since the wideband LFM signal has approximate rectangular spectrum, the wavelet transform is used for edge detection. The edges of LFM signal spectrum are sparse relative to the wideband. Then we can obtain a sparse representation of inertest by the sparse transformation. The compressed sampling matching pursuit (CoSaMP) algorithm is introduced to reconstruct the edge information of LFM signal spectrum from the sub-sampling LFM signals. The simulation results prove that the initial frequency and final frequency of wideband LFM signal can be estimated by the proposed method with high estimation precision.5. Inspired by the above parameter estimation method based on CS theory, the differential spectrum of the LFM signal is processed by the orthogonal matching pursuit (OMP) algorithm It is analyzed that the maximum related information between sub-samp ling series and the sensing matrix locates on the edges of the LFM signal spectrum. Based on that the OMP algorithm can extract the maximum related information, a novel method of parameter estimation for the wideband (ultra-wideband) LFM signal is proposed. The proposed method can relieve the pressure of high-speed sampling and estimate the initial frequency and final frequency of the LFM signal effectively.