Research On Polynomial Phase Signal Parameter Estimation Algorithm

Polynomial phase signal(PPS)is the best representation model of continuous instantaneous phase signal in limited time interval.It exists in many fields,such as radar and sonar.Estimating the parameters of PPS also plays an important role in many applications,such as target recognition,imaging and motion information acquisition in radar field,and target detection in sonar field.Therefore,the study of PPS parameter estimation algorithms is a research topic with important application value and theoretical significance.Aiming at the shortcomings of PPS parameter estimation algorithms,such as high computational complexity,poor anti-noise performance,spurious peaks and error accumulation,this thesis proposes some solutions.The main research contents of this thesis are as follows:Researches on the second-order PPS(also named linear frequency modulated signal)parameter estimation are studied.In order to solve the problem of high computational complexity caused by two-dimensional frequency domain search,a new algorithm based on double-scale strategy is proposed,which is called double scale two-dimensional frequency domain(DSTFD)algorithm.Based on the scale Fourier transform,a local scale Fourier transform(LSFT)with the ability of local spectrum analysis is proposed.By using the LSFT and the Chirp-Z transform(CZT),double-scale estimation of signal is realized,which can overcome the problem of high complexity caused by two-dimensional search of the parameter-domain algorithms.By changing the search range of LSFT and CZT,the proposed algorithm can be divided into two steps:coarse estimation operation and fine estimation operation.The purpose of coarse estimation operation is to determine the approximate range of signal parameters,thereby reducing the search range.The purpose of fine estimation operation is to search the signal parameters accurately within the range determined by coarse estimation operation.The proposed algorithm can effectively reduce the complexity while guaranteeing the estimation accuracy.The performance of the method is verified by simulation.Researches on the third-order PPS(also named quadratic frequency modulated signal)parameter estimation are studied.In order to solve the problem of high complexity,an improved double-scale(IDS)algorithm is proposed.In order to solve the spurious-peaks problem,a two-dimensional product modified Lv's distribution(2D-PMLVD)algorithm is proposed.The IDS algorithm is improved from the DSTFD algorithm.In order to solve the problem that the DSTFD algorithm cannot be directly applied to estimate the parameters of third-order PPS,the IDS algorithm proposes a multi-linear instantaneous autocorrelation function with order reduction function.By using this autocorrelation function,the third-order PPS parameter estimation problem is transformed into the second-order PPS parameter estimation problem.Then,the double-scale estimation strategy of the DSTFD algorithm can be used to reduce the complexity.Because delay factor is introduced into the correlation function.anti-noise performance of the algorithm is improved.The idea of 2D-PMLVD algorithm is:based on the property that the peaks of third-order PPS can be aligned and the spurious-peaks cannot be aligned,the spurious-peaks and noise can be suppressed by multiplying the peaks of third-order PPS.It is found that the peak position of third-order PPS in the parameter domain is determined by reduction-order factor of autocorrelation function.Therefore,a multi-dimensional instantaneous autocorrelation function is proposed.By changing the reduction-order factor in autocorrelation function,the third-order PPS can be formed peaks at different positions in the parameter domain.According to the relationship between peak position and reduction-factor,these peaks are moved to the same position and multiplied to obtain a higher peak.Because spurious-peaks and noise cannot be aligned when peaks alignment occur,the multiplication operation can effectively suppress them,and further improve the anti-noise performance.The performance of the method is verified by simulationResearches on the high-order PPS parameter estimation algorithm are studied.In order to solve the problem of poor anti-noise performance high-order PPS estimation,a new method based on polynomial regression is proposed.The proposed algorithm uses the adaptive short-time Fourier transform instead of the traditional short-time Fourier transform to enhance energy aggregation of high-order PPS in time-frequency domain.On this basis,the polynomial regression method is used to realize simultaneous estimation of the high-order PPS phase parameters,and to avoid the problems of the differentiation(PD)operation leads to the anti-noise performance deteriorates with the increase of signal order and error accumulation.Finally,the proposed algorithm uses O'Shea optimal estimation strategy to optimize the estimation results.By using the above operations,anti-noise performance of the algorithm is improved.The performance of the proposed algorithm is verified by simulation.