A Design Methodology for the Gaussian KLMS Algorithm

International audienceThe Gaussian kernel least-mean-square (Gaussian KLMS) algorithm has been studied under different implementation conditions. Though analytical models that predict its behavior are available, methodologies for determining the algorithm parameter values to satisfy given design criteria is still missing from the literature. In this paper we propose, test, and validate a methodology for the design of the Gaussian KLMS algorithm. Designing the algorithm consists in selecting adequate values for its free parameters from available theoretical performance models. These parameters comprise the filter length, the adaptive step-size, and the kernel bandwidth. The objective is to achieve specific design objectives, e.g., fast convergence time, good steadystate performance and/or reduced computational load. These goals are quantified in terms of different performance measures. Particularly, the time to convergence, the residual mean-squarederror (MSE), and the filter order