Diffusion Spline Adaptive Filtering with Adaptive Step-size Normalised Least Mean Square Algorithm

Abstract

Diffusion adaptation on spline adaptive filtering (SAF)is presented with combine-then-adapt (CTA) strategy for each node. SAF consists of an adaptive linear filtering and a spline interpolation function. Normalised least mean square algorithm is furnished in the adaptive linear filtering part. An adaptive averaging step-size mechanism is applied for both tap-weight vector of linear and nonlinear filtering parts to provide the fast convergence with low computation complexity. Statistical results testify that the proposed diffusion algorithm is able to provide promising and competitive results to the conventional diffusion strategy algorithm.