Steady-state and stability analyses of diffusion sign-error LMS algorithm

摘要

Diffusion sign-error LMS is a useful algorithm that allows agents in an adaptive network to perform in a robust manner under impulsive noise. In the literature, the mean stability condition and mean-square transient performance of this algorithm were analyzed with a reasonable impulsive noise model. However, it is important to analyze its mean-square stability condition and second-order steady-state performance for both theoretical understanding and practical implementation of the algorithm. This paper complements previous works by achieving this analysis with the same noise model. Simulations are performed to validate the theoretical findings.