Automatic tuning of the step size in WLMS algorithms: applications to EDGE

Adaptation laws that estimate time-varying communication channels can be tuned for efficient tracking, by adjusting filters and gains within their structure. However, this requires knowledge of statistical properties that may be poorly known or time-varying. A possible approach is then to adjust the gains of the adaptation law adaptively. The Wiener LMS adaptation algorithm (WLMS) attains efficient tracking by incorporating filters that are attuned to the fading statistics. It includes a gain parameter that will be adjusted adaptively on line. A WLMS channel predictor with adaptively adjusted gain is then evaluated as a component of receivers for EDGE systems, which use delayed decision feedback sequence estimation. The channel variations for such cases have a wide variety of properties that are tracked by gain adaptation.