CHAPTER 39 Fast Fixed-Order Filters
The arguments so far do not only provide an interpretation for the columns of
but they also motivate other efficient RLS implementations. These alternative implementations are not in the form of array algorithms, as was the case in Chapter 37, but in terms of explicit sets of recursions. We start with the Fast Transversal Filter (FTF) form.
39.1 FAST TRANSVERSAL FILTER
Define the normalized gain vector
and observe that the RLS recursion (37.2) can be used to time-update the weight vectors
as follows:
where cM,i-i is used in the recursion for
instead of cM,i, since the forward projection problem (38.19) is based on HM,i–1while the backward projection problem (38.10) is based on HM,i
Moreover, writing expression (37.10) in terms of {CM,i,CM,i–1}, and using the low-rank factorization (38.25) at time (i – 1), we obtain
in terms ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access