A Variable Step-Size Proportionate Affine Projection Algorithm for Identification of Sparse Impulse Response. (24th September 2009)

• Record Type: Journal Article
• Authors: Liu, Ligang; Fukumoto, Masahiro; Saiki, Sachio; Zhang, Shiyong
• Other Names: Bermudez Jose Carlos Academic Editor.
• Is Part Of: EURASIP journal on advances in signal processing. Volume 2009(2009)
• Publication Date: 2009-09-24
• Subjects: Signal processing -- Periodicals; Traitement du signal; Signal processing; Periodicals; 621.3822
• DOI: 10.1155/2009/150914
• Languages: English
• ISSNs: 1687-6172

Abstract

Abstract : Proportionate adaptive algorithms have been proposed recently to accelerate convergence for the identification of sparse impulse response. When the excitation signal is colored, especially the speech, the convergence performance of proportionate NLMS algorithms demonstrate slow convergence speed. The proportionate affine projection algorithm (PAPA) is expected to solve this problem by using more information in the input signals. However, its steady-state performance is limited by the constant step-size parameter. In this article we propose a variable step-size PAPA by canceling the a posteriori estimation error. This can result in high convergence speed using a large step size when the identification error is large, and can then considerably decrease the steady-state misalignment using a small step size after the adaptive filter has converged. Simulation results show that the proposed approach can greatly improve the steady-state misalignment without sacrificing the fast convergence of PAPA.