Adaptive Signal Processing Based on High-Order Statistics

摘要

The project is aimed at the development of some fundamental results on parameter estimation based on high order statistics. These results provide the basic tools needed for the development of adaptive processing algorithms for non-Gaussian signals. Project accomplishments include: (1) Adaptive algorithms for estimating the parameters of AR, MA, and ARMA processes from their third and fourth order moments have been derived and tested; (2) The performance of parameter estimation algorithms based on third order moments has been thoroughly analyzed; (3) A class of parameter estimation algorithms whose performance is asymptotically minimum variance was developed and shown to provide significantly better estimates than some high-order algorithms suggested in the literature.