An extended inverse \(QR\) adaptive filtering algorithm (Q1341387)

In a previous work [A state-space approach to RLS filtering, IEEE Signal Process. Mag. 11, 3 (1994)] \textit{A. H. Sayed} and \textit{T. Kailath} suggested the approach of reducing the study of adaptive filtering problems to that of equivalent state-space estimation. Here, such an approach is revisited. It is shown that the inverse QR algorithm of \textit{S. Alexander} and \textit{A. Ghirnikar} [IEEE Trans. Signal Process. 41, 20- 30 (1993; Zbl 0774.93075)] is a special case of a covariance square-root to the state-space problem, which in fact immediately gives a multi- channel version of the algorithm. Moreover, an extended square-root algorithm gives an extended inverse QR adaptive filtering algorithm with some additional nice features.