A fast multichannel adaptive filtering algorithm

In this paper we present a new QR decomposition (QRD) based fast multichannel adaptive filtering algorithm. This algorithm is based on a block reduction technique which leads to a substantial reduction of complexity compared to other QRD-based algorithms. It has a complexity of O (N) where N is the sum of the channels orders which may be different. It, also, has good numerical properties and is amenable to systolic architecture. Simulation shows a better robustness than the QRD-RLS in the context of highly intercorrelated channels inputs.