Diffusion Constrained Least Mean M-estimate Algorithm for Adaptive Networks

Abstract

Distributed adaptive networks are widely used in many fields. Most of the existing distributed adaptive algorithms are designed to solve the problem of network optimization under unconstrained conditions. However, in actual situations, there exist some network optimization problem under constrained conditions need to be solved, and considering that the distributed network is usually interfered by impulsive noise, a novel diffusion algorithm called diffusion constrained least mean M-estimate (D-CLMM) is proposed by using the modified Huber (MH) function, which can provide robust learning ability when network is disturbed by impulsive interference. Finally, the performance of the proposed algorithm is verified under different non-Gaussian noise environments. Simulation results show that the D-CLMM algorithm performs better than the diffusion-constrained least mean square algorithm (D-CLMS) based on mean square error (MSE) criterion.