Abstract:The low order cyclic statistics is widely applied with good results, but it has poor effect of noise suppression. The high-order cyclic statistics can suppress Gaussian noise, handle the cyclostationary, nonlinear and non-Gaussian process, but it has very large amount of calculation. Cyclic autocorrelation slice function can extract modulation frequency from the spectrum, but the effect is not obvious under large noise interference. In order to extract the characteristic frequency of cyclostationary signal in a lower Signal-to-Noise Ratio(SNR) and a more rapid calculation speed, carrier frequency slice spectrum of cyclic bispectrum method is adopted to extract the required frequencies. After simulation in Matlab and experimental research, it is found that the carrier frequency slice spectrum of cyclic bispectrum can still be expressed intuitively when the SNR is -15 dB.