Robust Wiener Design of Adaptation Laws with Constant Gains.

Mikael Sternad, , Uppsala University
Lars Lindbom, , Ericsson Infotech and
Anders Ahlén , Uppsala University

IFAC Workshop on Adaptation and Learning in Control and Signal Processing (ALCOSP 2001), Como, Italy, August 29-31 2001. © IFAC 2001

Conference proceedings published in: Sergio Bittanti, ed, Adaptation and Learning in Control and Signal Processing 2001. ISBN: 0 08 043683 8, Elsevier Publishing, Sept. 2002.

Abstract:

Filters can be introduced into LMS-like adaptation algoritms to improve their tracking performance. We here discuss the model-based design of such filters when tracking coefficients of linear regression models.

The parameter variations are modeled as ARIMA-processes which represent prior information. The aim is to provide high performance filtering, prediction or fixed lag smoothing for arbitrary lags.

Since the second order properties of the time-varying parameters are in general not known exactly, a robust design for a set of possible models will be of interest. We present a method that minimizes the average tracking MSE, based on probabilistic descriptions of the model uncertainty.

The method is based on a novel signal transformation that recasts the algorithm design into a Wiener problem with uncertain parameter model, which is to be solved iteratively. The performance is illustrated on the tracking of time-varying mobile radio channels in ANSI-136 systems, based on a model of the time-variations affected by parametric uncertainty.

Related publications:

Design of the general constant-gain adaptation algorithms. (Complete report, with proofs.)

Analysis of stability and performance, for slow and fast variations.

The Wiener LMS adaptation algorithm, a special case with low complexity.

A Case Study on IS-136 channels.

PhD Thesis by Lars Lindbom, 1995.

Source:

Preprint in Postscript, 239K
Preprint in Pdf, 240K