Optimization of polynomial predictive IIR filters using genetic algorithms

Polynomial predictive filters (PPFs) are capable of predicting the future values of a slowly changing signal. The coefficients of the FIR PPFs can be analytically derived, but the use of these structures is limited by their high passband gain-peak and modest stopband attenuation. It has been shown t...

Full description

Saved in:
Bibliographic Details
Main Authors: Harju, P.T., Ovaska, S.J.
Format: Conference Proceeding
Language:English
Subjects:
Feedback
Finite impulse response filter
Genetic algorithms
IIR filters
Laboratories
Optimization methods
Passband
Polynomials
Signal analysis
White noise
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Polynomial predictive filters (PPFs) are capable of predicting the future values of a slowly changing signal. The coefficients of the FIR PPFs can be analytically derived, but the use of these structures is limited by their high passband gain-peak and modest stopband attenuation. It has been shown that by introducing feedback terms to the basic FIR structure, these shortcomings can be effectively remedied. Previously, no well-developed method has existed for the choice of the feedback coefficients, however. We present a method for optimizing the feedback coefficients of the IIR PPFs based on an arbitrary desired (lowpass) magnitude response provided by the user. Simple genetic algorithm is employed in the optimization because of its ability to effectively solve even highly nonlinear problems.
DOI:10.1109/ICSIGP.1996.566974