Hi, I am planning to use a lattice adaptive filter for system identification. I am unable to find method to update the km�s in the lattice. What have I missed? We can convert any FIR to a lattice and can adapt a FIR as needed. Why can�t we only adapt the lattice ? Thank you, Lalin
Adapting a lattice filter
Started by ●May 23, 2006
Reply by ●May 23, 20062006-05-23
Lalin wrote:> Hi, > I am planning to use a lattice adaptive filter for system identification. > I am unable to find method to update the km's in the lattice. What have I > missed?Google -> adaptive lattice filter
Reply by ●May 23, 20062006-05-23
Hi Andor, I am unable to find it in google as well as in well known text books!. I am little puzzled about this conundrum. We can adapt a FIR as needed, can convert the FIR to a unique lattice filter and why can�t we only adapt the lattice? Is it because we can't find a error criteria at intermediate stages ? BTW: GAL algo's use the lattice only as an orthogonalizer so they don't count.
Reply by ●May 24, 20062006-05-24
Check out this link. http://scholar.lib.vt.edu/theses/available/etd-4698-201138/unrestricted/etd.pdf Bob Adams
Reply by ●May 24, 20062006-05-24
>Check out this link. > > >http://scholar.lib.vt.edu/theses/available/etd-4698-201138/unrestricted/etd.pdf > >Bob Adams > >Thank for the link, again this does not help me since, it is Gradient Lattice Algorithm(GAL)these algorithms work on joint estimation principle. The lattice filter is used as a pre-processing(Orthogonalizing unit) to do basic FIR LMS on the backward coefficients. This improves the convergence speed. I am looking for LATTICE ONLY adaptation.
Reply by ●May 24, 20062006-05-24
> >>Check out this link. >> >> >>http://scholar.lib.vt.edu/theses/available/etd-4698-201138/unrestricted/etd.pdf >> >>Bob Adams >> >> >Thank for the link, again this does not help me since, it is Gradient >Lattice Algorithm(GAL)these algorithms work on joint estimationprinciple.>The lattice filter is used as a pre-processing(Orthogonalizing unit) todo>basic FIR LMS on the backward coefficients. This improves theconvergence>speed. I am looking for LATTICE ONLY adaptation. >Have you looked at "Adaptive Filter Theory" by Simon Haykin? I have the thrid edition and it appears that he has a good selection of various adaptive lattice algorithms. You may be able to cast your problem to fit one of the models in the book.
Reply by ●May 24, 20062006-05-24
>Have you looked at "Adaptive Filter Theory" by Simon Haykin? I have the >thrid edition and it appears that he has a good selection of various >adaptive lattice algorithms. You may be able to cast your problem tofit>one of the models in the book. >Yes, I have read/have Simons book, Sayed and Honig and many more Adaptive DSP books. This is section has been conveniently omitted from the books without giving any reason. The problem sounds very trivial but for some reason no one have solution. Thanks, Lalin