Efficient generation of macromodels via the loewner matrix approach for the stochastic analysis of high-speed passive distributed networks

Stochastic distributed networks can be characterized in the frequency-domain by augmented multiport Y-parameter sampled data based on a stochastic Galerkin's formulation of the network equations. In this paper, a Loewner Matrix approach towards generating time-domain macromodels from the augmented multiport data is proposed. The key benefit of this work is that the superior scaling of the computational complexity of the Loewner Matrix approach with respect to number of network ports is utilized to generate the macromodel much more efficiently than the traditional Vector Fitting approach. The advantage of the proposed approach is validated by a numerical example.