An Adaptive Fir Filtering Based on Balanced Realization

Abstract

Balanced realization is an attractive candidate to design state-space adaptive filter structure due to its least parameter sensitivity. In this paper, based on the balanced realization, an adaptive finite impulse response (FIR) filtering algorithm is proposed to minimize the output-error using the coefficients of the transfer function as the adaptive filter parameters. This algorithm is an internally balanced realization form and guarantees that the designed adaptive FIR filtering always minimizes the ratio of maximum-to-minimum eigenvalues of the Grammian matrices.