Rational modeling of spectral data using orthonormal vector fitting

Abstract

Vector fitting (VF) is an iterative rational macromodeling technique by B. Gustavsen and A. Semlyen (1999) that became quite popular over the last years, due to its simplicity and availability. Although the VF method provides accurate broadband macromodels, the numerical stability of the algorithm is not always optimal. In this paper, the orthonormal vector fitting (OVF) algorithm is introduced, which reduces the numerical sensitivity of the model parameterization to the choice of starting poles significantly, and limits the number of required iterations.