Efficient lattice realizations of adaptive IIR algorithms

Abstract

Computationally efficient adaptive IIR-filter algorithms are presented based on lattice realizations allowing the adaptive filter stability to be easily monitored. New simplified recursive-in-order equations relating the parameters of the direct-form realization with the ones of two lattice realizations are presented. These equations lead to a simplified technique to compute the regressor vector and to a general method to implement any adaptive IIR algorithm using lattice realization. Results indicate that the proposed lattice-based algorithms converge to a set of parameters that realize the same transfer function as the corresponding direct-form algorithms.