Multitask Diffusion Adaptive Tuning Affine Projection Algorithm for Distributed Model

Abstract

This paper presents combine-then-adapt diffusion model on affine projection adaptive algorithm with adaptive tuning mechanism. An adaptive affine projection algorithm is suggested in the distributed network. Diffusion strategy involves with the combination stage and adaptation stage. By adding the extra-stage in the combination stage, the combination between nodes and inside nodes are built with the global cost function for multitask model to eliminate unknown data. As dealt with fast convergence, the adaptive tuning parameters are assigned in terms of step-size value. Experimental results reveal that this proposed algorithm is able to contribute the approving results for system identifications.