THEORY AND METHODS OF SIGNSL PROCESSING Signal Reconstruction by Using Normalized Bispectrum Estimates in Noise

The problem of reconstructing an unknown waveform observed in additive noise by using normalized bispectral density estimates is considered. The proposed approach is based on using a continuous-valued normalized bispectral density estimate instead of the discontinuous biphase function conventionally computed in bispectrum-based signal reconstruction algorithms. The performance and reconstruction accuracy of the developed technique are studied by computer simulations in the presence of additive Gaussian and impulsive noises. Computer simulations results demonstrate that improvement of the signal reconstruction accuracy is achieved by the suggested approach.