Evaluation of vector fitting for compact interconnect model representation

In this work, we present a study on the utilization of vector fitting as a compression technique for storing passive interconnect model data for multiport systems. A canonical transmission line case and three different modeling examples from the literature are explored, where the error is compared as a function of the number of poles used to extract the rational approximation. The compression ratio for data storage with respect to conventional S-Parameter models is discussed as well. It will be shown that the compression ratio is highly dependent of the electrical length of the configuration and the type of response; however, ratios over 5 could be obtained for all the cases when using a number of poles that guarantees a maximum mean relative error below 1%.