Using model-based parameter estimation to increase the efficiency of computing electromagnetic transfer functions

Two main ideas are introduced: (1) the use of model-based parameter estimation based on rational function approximations, which reduces the number of frequencies at which solutions or samples are required; (2) a sampling approach that uses frequency derivatives of the response and a novel analytical technique based on differentiating the moment method impedance equation, which provides derivative information in a time proportional to N/sup 2/ in contrast with the N/sup 3/ dependence in solving the original problem. Antenna input admittances are modeled using frequency samples and derivatives. The rational function model is shown to offer a large advantage over polynomial interpolation of a frequency response. Application of the frequency-derivative approach is demonstrated for problems having well-defined resonances such as a dipole antenna, and for more challenging problems having narrow resonances. >