A robust state space model encompassing early-time transients for electromagnetic signal extrapolation

This paper presents a control systems methodology, based on the state space method (SSM), to derive rational transfer function models from time-domain simulated data. Unlike existing methods such as Prony´s and pencil of functions, which discard a significant part of the early-time transient response due to instability in model extrapolation, the proposed SSM produces causal, stable, low-order parametric representations for the entire signal using a sequence that includes early-time transients. SSM is applied to model time-domain signatures of highly resonant circuits, such as notch filters and dielectric resonators, and it is shown that excellent model corroboration can be achieved using modest resources.