An improved variable tap-length algorithm for structure adaptation

This paper investigates the variable tap-length algorithm for structure adaptation. Among existing algorithms, the Segmented Filter (SF) and Gradient Descent (GD) algorithms are of interest as both can track the tap-length variations quickly. In this paper, we first compare the SF and GD algorithms and show that each has advantages/disadvanges relative to the other. Then we propose an improved variable tap-length algorithm which not only has better performance, but also has less complexity, than existing algorithms. The proposed algorithm has great significance in both theory and applications.