Diffusion Variable Step Size Algorithm Based on Maximum Correntropy Criterion

Abstract

As a new kind of method using localized similarity measure by kernel function, maximum correntropy criterion has the better performance in the environment of non-Gaussian noise. And in order to accelerate the convergence speed of the algorithm, this paper introduces a variable step size method inspired by maximum correntropy criterion, proposes a variable step size distributed algorithm based on maximum correlation criterion (VSS-DMCC). Simulation results show that our algorithm has smaller steady-state error and faster convergence speed than the minimum mean square variable step size distributed algorithm (VSS-DLMS) and maximum correntropy criterion distributed algorithm (DMCC).