摘要

This paper investigates a family of simple recursive algorithms based on feedback around a weight vector or matrix with a multivariable discrete integrator in the forward path. It is shown that, by using simple changes, estimates of the covariance matrix, inverse-covariance matrix and their corresponding square-root values can be found. As a special vector case of the more general matrix algorithm we show that least mean squares (LMS) can also be derived.

  • 出版日期2011-10