Parameter Estimation Of LFM Signal In Alpha Stable Noise

Linear frequency modulation(LFM)signals are a class of non-stationary signal generated by pulse compression technology,with large time-bandwidth product and low load probability which have been widely used in radar,communication,biomedicine,seismic survey.Therefore,the parameters estimation of LFM signals has always attracted much attention.However,most of the research on parameters estimation methods for LFM signals are carried out under Gaussian noise.In recent years,studies have found that complex noise environments in practical applications are more suitable being described by alpha stable distribution.Signal processing methods based on Gaussian assumptions are no longer be applicable.Therefore,the research on the parameter estimation methods of LFM signals under alpha stable distribution noise has great significance.The traditional time-frequency analysis methods significantly degrade the performance under the alpha stable distribution noise,which in turn affects the signal parameter analysis.In this thesis,based on the principle of M estimation,a robust cost function is constructed using an approximate alpha stable probability density function,and an adaptive ASR robust filtering method is proposed.In this method,the robust cost function can remove impulse outliers in the observed data,and the influence function can be used to evaluate the robustness of the robust cost function.In addition,the selection of different linearity parameters in the robust cost function will achieve different filtering performance.This method uses the PCK information of the observed data to realize the adaptive selection of linearity parameters.Finally,for the problem of parameters detection and estimation of LFM signals under impulsive noise,the ASR method is extended to the time-frequency two-dimensional plane,and a robust AS-FT time-frequency analysis method is proposed to perform time-frequency analysis on the LFM signals.The AS-FT time-frequency distribution of the LFM signals is a straight line,and the signal parameters can be extracted in the time-frequency plane using Hough transform to obtain the parameters estimation of the LFM signals.Simulation experiments show that the method can adaptively suppress the alpha stable noise under different characteristic indexes,and then obtain effective estimation of the parameters with good robustness.Correntropy is a new tool used in impulse noise suppression in recent years.It uses the exponential decay characteristic of the Gaussian kernel function to suppress large-value pulses,and its Taylor series expansion contains correlation function information,so correntropy can be used for the time-frequency analysis of LFM signals,but the correntropy cannot suppress similar amplitude pulses.To solve this problem,in this thesis the cosine kernel function is used to construct the CBC function,and then the exponential function weighting process is utilized to propose a WCBC function with similar amplitude pulse suppression capability.The cosine kernel function in the WCBC function has similar attenuation characteristics to the Gaussian kernel function,which can suppress large-value impulse noise.In addition,the Taylor series expansion of the WCBC function also contains correlation function information.Therefore,the WCBC function can be used for time-frequency analysis of LFM signals under impulsive noise.On this basis,the short-term stationary characteristics of the LFM signals is used to propose a SWFT time-frequency analysis method based on the WCBC function,which is used for the time-frequency analysis of LFM signals under impulsive noise.The SWFT time-frequency distribution of the LFM signals is a straight line,and the Hough transform of the SWFT can obtain the estimated value of each parameter of the LFM signals.The SWFT method uses the WCBC function to process the LFM signals in sections,calculates the WCBC spectrum of each section of the signal,and obtains the joint function of time and frequency.It can perform time-frequency analysis on the LFM signals under the background of impulsive noise,and has strong anti-noise performance.Non-linear amplitude transformation is a commonly used impulse noise suppression method.In view of the problem that the traditional time-frequency analysis methods significantly reduce the performance under impulsive noise,this thesis proposes a Meridian nonlinear amplitude transformation function from the perspective of limiting large-value pulses.It is compared with the common "S" type nonlinear amplitude transformation function and fractional lower-order function,and their nonlinear transformation characteristics are analyzed.On this basis,a WBM parameter estimation method based on Meridian function is proposed for the detection and estimation of the parameters of LFM signals under impulsive noise.The simulation experiment proves that the WBM method can be used for the detection and estimation of the parameters of the LFM signals under various impulse noises,especially under the strong impulse noise environment,it has excellent impulse noise suppression ability and can realize the effective LFM signals parameters estimation.