Adaptive Reduced-Rank Interference Suppression: Algorithms, Performance, and Low Power VLSI

Northwestern University, Evanston IL

Investigators

Abstract

This research is concerned with algorithmic, performance, and hardware issues related to reduced-rank adaptive filtering. Reduced-rank filters project the incoming received signal onto a lower dimensional subspace, which reduces the amount of training data needed relative to a conventional full-rank algorithm. Algorithmic techniques will be studied initially within the context of interference suppression for Direct Sequence (DS)-Code-Division Multiple Access (CDMA), although they can be applied to any adaptive linear filter. The focus of the research is on a recently developed class of reduced-rank adaptive algorithms based on the multi-stage Wiener filter of Goldstein and Reed. This technique can achieve full-rank performance with a very low filter rank, which enables rapid convergence and tracking. Furthermore, these algorithms do not rely on an explicit estimate of the signal subspace. The project is multi-disciplinary in that it combines the expertise of the two co-PIs in the areas of adaptive signal processing and low power VLSI design. The main objective of the research is to build a low-power special purpose hardware prototype, which can serve as the computational engine for reduced-rank filtering in a variety of applications. Algorithmic issues to be studied include selection of filter rank, performance in different adaptive filtering applications, such as equalization, and numerical stability and dynamic range problems.