A systolic array for recursive least squares computations: mapping directionally weighted RLS on an SVD updating array

摘要

A systolic algorithm/array is described for recursive least squares (RLS) estimation, which achieves an O(n/sup 0/) throughput rate with O(n/sup 2/) parallelism. The array is also useful for several other applications, such as, e.g., SVD updating and Kalman filtering. An additional advantage is that unlike with other RLS-arrays, it is now possible to incorporate alternative data weighting strategies, such as directional weighting, without sacrificing speed.