Research On The Algorithm Of Clutter Simulation With Specified Bispectrum

Considering the application of the random noise series with the specified bispectrum structure in radar s ystem simulation is increasing. In this paper, the modeling algorithm for the random sequence of t he clutter with the specified bispectrum is studied. Thus, the random sequence with the specified bispectrum structure will be simulated more accurately.First of all, this paper introduces using higher order spectral decomposition method to build a specific bispectrum of random sequence with the input of non Gaussian three-order white noise sequence. When the input is the three-order white noise random process of non Gaussian distribution, put a non Gaussian distribution three-order white noise passing through a linear time invariant system which can make the output signal has a specified bispectrum structure. Well, the focus of this part will be how to solve the linear system to yield the sequence with specified bispectrum. Considering the application of Cepstrum in the field of the minimum phase identification, this paper presents the application of cepstrum projection technology as the method of solving linear systems with the input of non Gaussian three-order white noise sequence. By projecting the cepstrum into Descartes coordinates, the corresponding linear time invariant system is obtained, and the purpose of simulation of random process with specified bispectrum has been done.Then, this part introduces the method of constructing non Gauss ian random sequence with the specified bispectrum using the SWM-K+1 system. This part first introduces the constructing method of random process with the specified bispectrum structure by using the SWM-K system with the K inputs of non gaussian three-order white noise random sequence, the essence is using bispectrum decomposition technique. Then, the non-Gaussian random sequences with the specified bispectrum and the specified power spectral density are constructed by using SWM-K+1 system, among the K+1 inputs, the first K inputs is non Gaussian three-order white noise random sequence, and the last one is white Gauss random sequence. Because the bispectrum of Gaussian white noise random sequence is zero, SWM-K+1 system utilizes SWM-K system with K inputs of nongauss third-order white noise to construct specified bispectrum structure, using an input is white Gaussian distribution system to construct specified power spectral density. So the problem of the simulation sequence with specified the bispectrum and specified power spectrum density is solved.Finally, the two-order Volterra filter is introduced to construct the clutter sequence with the spectrum of the specified bispectrum and the specified power spectrum. When the input is second-order white Gaussian random, if the clutter modeling system is linear time invari ant system, according to the relevant knowledge, the output will also obey the Gaussian distribution random sequence, the bispectrum(i.e.three order specturm) will be zero. Considering this, in t his part, the system will use the nonlinear system. In the nonlinear system, the two order Volterra filter(SVF) is adopted by the majority of scholars, which mainly uses the nonlinear filtering of the signal. In this part, considering the effect of Volter ra filter on the nonlinearity, the Volterra filter is used to construct the modeling with specified bispectrum. The output of the second-order Volterra filter(SVF) will have the specified bispectrum structure with the input of Gauss white noise sequence. So, the main research of this part is how to solve the specified Volterra filter which will yield non Gaussian random process with specified bispectrum. At the same time, the power spectrum of the signal can be constructed by white Gauss noise due to the characteristic which the bispectrum(three-order spectrum) of white Gauss noise is zero.