Steady-state behavior and design of the Gaussian KLMS algorithm

The Kernel Least Mean Square (KLMS) algorithm is a popular algorithm in nonlinear adaptive filtering due to its simplicity and robustness. In kernel adaptive filters, the statistics of the input to the linear filter depends on the parameters of the kernel employed. A Gaussian KLMS has two design parameters; the step size and the kernel bandwidth. Thus, its design requires analytical models for the algorithm behavior as a function of these two parameters. This paper studies the steady-state behavior and the stability limits of the Gaussian KLMS algorithm for Gaussian inputs. Design guidelines for the choice of the step size and the kernel bandwidth are then proposed based on the analysis results. A design example is presented which validates the theoretical analysis and illustrates its application.

[1]  T. Kailath,et al.  SOME USEFUL PROBABILITY DISTRIBUTIONS , 1965 .

[2]  G. Wahba,et al.  Some results on Tchebycheffian spline functions , 1971 .

[3]  Thomas Kailath,et al.  RKHS approach to detection and estimation problems-V: Parameter estimation , 1973, IEEE Trans. Inf. Theory.

[4]  Paulo Sergio Ramirez,et al.  Fundamentals of Adaptive Filtering , 2002 .

[5]  Y. Engel.  Kernel Recursive Least Squares , 2004 .

[6]  Paul Honeine,et al.  On-line Nonlinear Sparse Approximation of Functions , 2007, 2007 IEEE International Symposium on Information Theory.

[7]  Weifeng Liu,et al.  The Kernel Least-Mean-Square Algorithm , 2008, IEEE Transactions on Signal Processing.

[8]  Paul Honeine,et al.  Online Prediction of Time Series Data With Kernels , 2009, IEEE Transactions on Signal Processing.

[9]  Weifeng Liu,et al.  Kernel Adaptive Filtering , 2010 .

[10]  Cédric Richard,et al.  Stochastic Behavior Analysis of the Gaussian Kernel Least-Mean-Square Algorithm , 2012, IEEE Transactions on Signal Processing.