Given a topology, the choice of the adaptation algorithm crucially affects the performance of an adaptive filtering system. A fast affine projection (FAP) algorithm can potentially converge almost as fast as the RLS and have a low complexity comparable to that of the LMS. However, the existing FAP algorithms have certain shortcomings.

This talk will start with an echo cancellation system as an example of adaptive filtering applicatioin, and briefly review the existing adaptation algorithms - their pros and cons. Special attention will be paid to the existing FAP algorithms, with emphasis on the problems they have.

The essence of the talk is to present two stable and robust FAP algorithms that have just been invented. The first one can provide an exact solution at a moderate complexity - in contrast with the fact that existing stable FAP algorithms are all approximate. The second FAP algorithm, based on a novel reciprocating recursion concept, gives a very good approximate solution at a complexity significantly lower than that of any existing FAP algorithm.

The proposed FAP algorithms are expected to be very practical in telecom and other applications of adaptive filtering.

Dr. Heping Ding joined Acoustics and Signal Processing Group, IMS, National Research Council, Canada, in 2001, where he is holding the position of Senior Research Officer. He has also served in Bell Northern Research and Nortel Networks, Canada, as a Senior Audio Signal Processing Specialist. Dr. Ding is an Adjunct Professor with the School of Information Technology and Engineering (SITE) at the University of Ottawa, Canada, and an Adjunct Professor at the School of Graduate Studies, Ryerson University, Canada. In addition, he has been a part-time professor at SITE, University of Ottawa, to teach a graduate level course on digital signal processing. So far, Dr. Heping Ding holds 9 issued patents and more than 37 open and confidential publications in fields of subjective audio, telephony, adaptive filtering, echo control, non-stationary signal processing, noise reduction, audio coding, and electronic equipment. In addition, he has received more than 13 awards from industry, universities, and science and technology communities. Dr. Ding is a Senior Member of IEEE.