Study of Bispectrum Based on Fractional Lower  Order Statistics and Its Estimation

Stable processes can better model the impulsive random signals and noises in physical observation. In order to suppress the non-Gaussian noise， the bispectrum based on fractional lower order statistics is defined， and its estimation method was proposed. Simulation results shows that nonparametric bispectrum based on fractional lower order statistics is much effective for identifying signal compared with traditional bispectrum， and signal amplitude and phase are kept， but it still has big estimated variance. The new algorithm based on the parameter model of AR processes has maximal spectral flatness， and is much robust and effective for suppressing the fractional lower order noise.