A parametric approach to adaptive IIR filtering : finite precision implementation and stability aspects

Abstract

It is well known that a digital filter can be characterized with different parametrizations and realized with different structures. The main objective of this thesis is two-fold. The first one is to investigate the effects of parametrizations and structures on the performance of adaptive IIR filters, particularly on the stability aspect, and hence to derive some new adaptive algorithms. The second one is to examine how a well-designed adaptive algorithm can be degraded when implemented with finite precision consideration.