Research On Parameter Estimation Of Frequency Hopping Signals Based On New Bilinear Time-Frequency Analysis

Frequency-hopping signal is one kind of non-stationary signal,the instantaneous frequency of frequency-hopping signal is not constant.The traditional Fourier Transform can not reflect the time-varying law of the instantaneous frequency,and the time-frequency analysis method can effectively describe the time domain and frequency domain related features of the frequency-hopping signal.The Short-Time Fourier Transform of the signal is subject to the contradictory influence between time frequency resolution,and it is difficult to describe the time-frequency distribution more accurately.The Wigner-Ville distribution of signal can effectively improve the time-frequency resolution,but its inherent cross-term problem will lead to the deterioration of the distribution.The bilinear time-frequency distribution can solve the cross-term problem.The kernel function of bilinear distribution determines the performance of time-frequency resolution of the bilinear distribution.Therefore,according to the characteristics of the signal time-frequency distribution,the bilinear distribution of the kernel function is studied in this thesis,which has a good research value.First,this thesis is based on the analysis of the research status of frequency hopping signal parameter estimation.The non-stationary characteristics of frequency-hopping signal are described.Then,the Short-Time Fourier Transform and the Uncertainty Principles is analyzed.The computer implementation of the Short-Time Fourier Transform is studied.Finally,the advantages and disadvantages of the Short-Time Fourier Transform and bilinear time-frequency distribution are analyzed.Second,the parameter estimation of uniform frequency-hopping signal based on optimized kernel function is studied in this thesis.After exploring the mathematical form and graphical representation of the ambiguity function of a uniform frequency hopping signal,a kernel function adapted to a frequency-hopping signal is given based on its ambiguity function.Two parameters of kernel function is used to control its width of direction of delay and frequency shift.Then the method of instantaneous frequency estimation based on time-frequency analysis is described,that is,it extracts the approximate instantaneous frequency according to the time-frequency distribution of the signal.The main parameters of the uniform frequency-hopping signal are estimated according to the bilinear time-frequency distribution corresponding to the kernel function.In the environment of Gauss White noise,fixed-frequency interference,burst interference,linear sweep-frequency interference and nonlinear sweep-frequency interference,the performance of proposed kernel function distribution and Smooth Pseudo Wigner-Ville frequency-hopping signal parameterestimation are compared with the MATLAB programming,and the above parameter estimation process is achieved with the C++ language programming.Finally,the parameter estimation of non-uniform frequency-hopping signal is explored.Firstly,the characteristics of non-uniform frequency-hopping signal are analyzed.Then the difference between non-uniform frequency-hopping signal and uniform frequency-hopping signal is studied.The mathematics form and graphics performance of non-uniform frequency-hopping signal's ambiguity function is explored.Then,the parameters of kernel function of the non-uniform frequency hopping signal are determined firstly.In the environment of Gaussian White noise,fixed-frequency interference,burst interference,linear sweep-frequency interference and nonlinear sweep-frequency interference,the MATLAB programming simulation is used to study and compare performance of the proposed kernel function distribution and the Smooth-Pseudo Wigner-Ville distribution for non-uniform frequency-hopping signal parameter estimation.