Convergence Properties of Subband Identification

DAMIÁN MARELLI; MINYUE FU

Singapur

Conferencia; Asian Control Conference (ASCC); 2002

Asian Control Association

The so-called subband identification method has been introducedrecently as an alternative method for identification of finite-impulseresponse systems with a large tap size. It is known that this method can bemore numerically efficient than the classical system identification method,while having compatible performance. In this paper, we assume aprobabilistic framework, and we deal with the following two problems: (1)whether or not the identification result depends on the particularrealization of the random process under consideration, and (2) whether ornot, the identification algorithm converges to the minimum of the errorfunction. We study these properties by considering both the error functionsin individual subbands and a combined error function. We study thecritical-sampling and the oversampling cases. We show that optimumconvergence is not always guaranteed in the oversampling case. Amodification in the identification algorithm will be proposed to fix thisproblem.