Noise Constrained Diffusion Least Mean Squares over adaptive networks

Abstract

This paper presents the design of a new diffusion algorithm over adaptive networks. The algorithm assumes knowledge of variance of additive noise. The design is based on the Noise-Constrained Least-Mean Squares (LMS) Algorithm and the new algorithm becomes a type of variable step-size algorithm for which the step-size variation rule results directly from the constraint. The design of the Noise-Constrained Diffusion LMS algorithm has been included. Simulation results show that the new algorithm outperforms the existing Diffusion LMS algorithm as well as its Incremental counterpart.