LINEARLY CONSTRAINED GENERAL PARAMETER-BASED ADAPTIVE FILTERS

摘要

General parameter-based filters are reduced-rank adaptive filters, where typically only few adaptive parameters are used. Thus, adaptation capability is achieved with little added computational complexity, and the coefficients consist of a fixed part and an adjustable part. In this paper, we modify the basic algorithm to support linear constraints on the filter transfer function. This is accomplished by setting constraints of the type C~T w = a on the fixed part of the transfer function, and C~T w = 0 in the adaptive extension. The algorithm is described, the stability condition is given, and examples of operation are shown.