Diffusion LMS Networks In The Presence Of Noisy Nodes: A Convergence Rate And MSD Analysis
Amanda De Paula, Cristiano Panazio

In this article, we provide a detailed analysis of the diffusion least mean square algorithm in the specific context where each node in the network presents different SNR values. We show, through an eigenvalue analysis, the condition that the algorithm step-size values should obey in order to provide a given convergence rate. Under this condition, it is shown how to set the algorithm step-size in each node that lead to the minimum mean square deviation. We also provide simulation results that corroborate our theoretical analysis.