Low-Complexity Implementation of a Data-Reuse RLS Algorithm

The recursive least-squares (RLS) family of adaptive algorithms is an attractive solution for adaptive systems due to their capacity of mitigating the correlation of input signals. The associated forgetting factor parameter is used to compromise between convergence speed/tracking and accuracy. This paper proposes a data-reuse methodology for a low-complexity RLS adaptive algorithm based on the dichotomous coordinate descent iterations, which offers improved tracking performances and an acceptable overall arithmetic workload, suitable for hardware implementations.