Data-Driven Parameterized Model Order Reduction Using z-Domain Multivariate Orthonormal Vector Fitting Technique

TL;DR: In this paper, a robust multivariate extension of the z-domain Orthonormal Vector Fitting technique is presented, which provides accurate and compact rational parametric macromodels based on numerical electromagnetic simulations or measurements in either frequency-domain or time-domain.

Abstract: Efficient real-time design space exploration, design optimization and sensitivity analysis call for Parameterized Model Order Reduction (PMOR) techniques to take into account several design parameters, such as geometrical layout or substrate characteristics, in addition to time or frequency. This chapter presents a robust multivariate extension of the z-domain Orthonormal Vector Fitting technique. The new method provides accurate and compact rational parametric macromodels based on numerical electromagnetic simulations or measurements in either frequency-domain or time-domain. The technique can be seen as a data-driven PMOR method.