Macromodeling of microwave structures based on noisy frequency-domain data

Frequency-domain macromodeling tools are of paramount importance for the design and study of microwave structures. In this paper, a reliable identification method, called orthormal vector fitting method, is combined with a relaxation constraint. This leads to more accurate fitting models, in a limited amount of computation time, especially if the spectral data is contaminated with noise. Its performance is analyzed on a measurement-based example and compared to other existing approaches in the field.