Crosstalk statistics via collocation method

A probabilistic model for the evaluation of transmission lines crosstalk is proposed. The geometrical parameters are assumed to be unknown and the exact solution is decomposed into two functions, one depending solely on the random parameters and the other on the frequency. The stochastic collocation method is used to estimate the crosstalk statistical moments. The results are obtained from a limited number of carefully-chosen values of the random geometrical parameters. The estimated statistical moments are then used to build the probability density function of the crosstalk parameters. A Monte Carlo validation demonstrates the accuracy and efficiency of the advocated method.