가중치 함수 기반 적응적 조합 계수를 갖는 Diffusion LMS 알고리즘

Abstract

In this paper, we propose a new combination of diffusion LMS coefficients, whereby combination coefficients control the amount of obtained information from each node. Specifically, the adaptive combination coefficients obtained by using a decreasing and positive weighting function are utilized to improve better convergence. In various noisy environments, the weighting function assigns more proper values, proportional to SNR of nodes, to combination coefficients. The computer simulation results show that the proposed algorithm yields better performance rather than other algorithms for system identification.