Optimal phase demodulation

Abstract
A new realization method for synthesis of the optimal cyclic loss nonlinear filter using a third generation digital computer is proposed. This method consists of carrying a finite set of Fourier coefficients of the conditional density of signal given the observations, and updating these coefficients to determine the Fourier coefficients of the a posteriori density when new observations are obtained. The filtering problem considered is that of phase demodulation. Monte Carlo evaluation of performance of this Fourier filter is given and compared with previous methods. Further, the accuracy and estimate computation time of this method are determined and compared with other methods.