Derivation of an adaptive algorithm for IIR digital systems realized by power-wave digital filters

Abstract

This paper presents a new efficient algorithm for adjusting the coefficients of an adaptive infinite impulse response power-wave digital filter in order to estimate the discrete-time signal y(k) as the recursively filtered version of the signalx(k). To do this, eight different prediction errors are defined which can be computed recursively with respect to model order and time, which leads to similar recursions as for the lattice algorithm [1], where autoregressive processes are considered. These recursions can be implemented as a digital filter which can be separated in ananalysis and asynthesis part, where the latter can be interpreted as a power-wave digital filter [2], [3], [13]. Hence, the stability of this adaptive network is always guaranteed [4]. Furthermore, it is shown that the proposed algorithm can be considered as an extension of the Levinson algorithm [7] for the efficient inversion of a special class of matrices.