Adaptive IIR filtering using input/output orthogonalization

Author

Beex, A.A. ; Sankaran, Sundar G.

Author_Institution

Syst. Group-DSP Res. Lab., Virginia Polytech. Inst. & State Univ., Blacksburg, VA, USA

Volume

1

fYear

1998

fDate

1-4 Nov. 1998

Firstpage

503

Abstract

An adaptive IIR filter structure is proposed based on orthogonal representations of the input and the (noisy) output. The orthogonalization of the input comes from the backward prediction errors in the lattice predictor. The input orthogonalization is then extended using the output, so that a single orthogonal decomposition results. Consequently the Hessian approximation becomes diagonal. The adaptive IIR filter weights converge rapidly compared to their direct form counterparts.

Keywords

Hessian matrices; IIR filters; adaptive filters; adaptive signal processing; convergence of numerical methods; digital filters; filtering theory; least squares approximations; noise; prediction theory; adaptive IIR filter structure; adaptive IIR filtering; backward prediction errors; diagonal Hessian approximation; direct form weights; filter weights convergence; input/output orthogonalization; lattice predictor; least squares lattice predictor; noisy output; orthogonal decomposition; orthogonal representations; Adaptive filters; Convergence; Cost function; Equations; Filtering algorithms; Finite impulse response filter; IIR filters; Noise measurement; Poles and zeros; Transfer functions;

fLanguage

English

Publisher

ieee

Conference_Titel

Signals, Systems & Computers, 1998. Conference Record of the Thirty-Second Asilomar Conference on

Conference_Location

Pacific Grove, CA, USA

ISSN

1058-6393

Print_ISBN

0-7803-5148-7

Type

conf

DOI

10.1109/ACSSC.1998.750914

Filename

750914