Do WLS Adaptive Filters Provide Better Tracking Performance than LMS Filters?

• Autorzy: Niedźwiecki M. , Kłaput T.
• Języki publikacji: EN

Abstrakty

EN

Sensitivity of the speed of convergence of gradient algorithms in their startup phase to the distribution of eigenvalues of the regression matrix created same sort of a bad reputation of least mean squares (LMS) filters as devices which can be used to track system changes. The opinion that the tracking performance of LMS algorithms is inferior to that of weighted least squares (WLS) algorithms is widespread but partially unsubstantiated. The point is that the poor initial convergence behaviour must not imply poor steady-state tracking performance. The paper presents both theoretical and simulation evidence confirming this statement.

Słowa kluczowe

EN

system identification; parameter tracking

• Wydawca: Polska Akademia Nauk, Wydział IV Nauk Technicznych
• Czasopismo: Bulletin of the Polish Academy of Sciences. Technical Sciences
• Rocznik: 2001
• Tom: vol. 49, nr 3
• Strony: 517--525
• Opis fizyczny: 2 rys., bibliogr. 3 poz.

Twórcy

autor

Niedźwiecki M.

autor

Kłaput T.

• Department of Automatic Control, Faculty of Electronics, Telecommunication and Computer Science, Technical University of Gdańsk (Wydział Elektroniki, Telekomunikacji i Informatyki Politechniki Gdańskij)