Iterative Loewner Matrix Macromodeling using CUR Decomposition for Noisy Frequency Responses

Abstract

This paper presents an efficient macromodeling technique for modeling distributed circuits characterized by noisy frequency-domain data. The proposed method is based on an iterative Loewner matrix (LM) approach. Using the LM approximation of previous iterations, state space matrices of the system are made to be more accurate. This approach is shown to minimize the biasing effect of the noisy data resulting in more accurate poles and residues while reducing the computation time by taking advantage of CUR decomposition instead of using the usual singular value decomposition (SVD) decomposition. A numerical example is presented to illustrate the efficiency of the proposed method.