Model Order Reduction of Parameterized State-Space Systems with Sequential Sampling

Abstract—Parameterized reduced order models are important for the design and analysis of microwave structures and systems. Quite often, a large set of models (nodes) with respect to a design parameter variation are uniformly chosen in the parameter design space, which are subjected to model order reduction algorithms and interpolated into a multidimensional model. In order to preserve passivity in the parameterization step, positive interpolation operators are frequently used. This paper demonstrates the importance of sequential sampling for selecting the nodes and building the parameterized models. It is shown that sequential sampling algorithms can significantly reduce the model evaluation cost. The present approach is validated by means of a microstrip example. I