An Adaptive Filtering Algorithm using Mean Field Annealing Techniques
Blekinge Institute of Technology, Department of Telecommunications and Signal Processing2002 (English)Conference paper (Refereed) Published
We present a new approach to discrete adaptive filtering based on the mean field annealing algorithm. The main idea is to find the discrete filter vector that minimizes the matrix form of the Wiener-Hopf equations in a least-squares sense by a generalized mean field annealing algorithm. It is indicated by simulations that this approach, with complexity O(M^2) where M is the filter length, finds a solution comparable to the one obtained by the recursive least squares (RLS) algorithm but without the transient behavior of the RLS algorithm. Further advantages of the proposed algorithm over other methods such as the recursive least-squares algorithm are that the filter coefficients are always limited and that it facilitates fast recovery after an abrupt system change.
Place, publisher, year, edition, pages
San Diego: IEEE , 2002.
adaptive filters, filtering theory, integral equations, least squares approximations
IdentifiersURN: urn:nbn:se:bth-10226ISI: 000179203300020Local ID: oai:bth.se:forskinfo1CDEA7D9BE685D6BC1256C6A005489A7ISBN: 0 7803 7587 4OAI: oai:DiVA.org:bth-10226DiVA: diva2:838301
IEEE Workshop on Signal Processing Systems