Frequency and initial phase estimation of sinusoidal signal based on bispectrum

Because of the relation with Fourier transform, the bispectrum also has fence effect, and this brings error to frequency estimation based on bispectrum. To overcome this problem, bispectrum phase difference method is proposed. This method constructs two data sets from the sampled data and makes bispectrum analysis respectively. It uses their peak spectrum lines' phase difference to accurately estimate relative frequency difference. Considering the boundary problem of this method, the bispectrum squared amplitude center method is proposed to form a synthetic algorithm with it. Simulation results show that the proposed synthetic algorithm has good estimation performance with low computational complexity. Moreover, theoretical deduction reveals that initial phase estimation is not affected by fence effect.