Novel implementation technique of RLS algorithm for improving throughput of adaptive filters

This paper proposes a method for increasing the throughput of recursive least squares (RLS) adaptive filters when only a few processing elements (PEs) are available. Throughput can be traded against the rate of convergence by varying a parameter. Because the conventional methods for improving the throughput of an RLS filter require the use of more or faster PEs, the proposed technique will expand the area of applications for the RLS algorithm.