Model-based parameter estimation in electromagnetics. III. Applications to EM integral equations

For pt.II see ibid., vol.40, no.2, p.51-66 (1998). This article has considered the rationale and illustrated the application of model-based parameter estimation (MBPE) to achieve reduced-order representations of electromagnetic observables via fitting models, the model-based part of MBPE, that derive from the physics of EM fields. The parameter-estimation part of MBPE is the process of obtaining numerical values for the coefficients of the fitting model by matching or fitting it to sampled values of the EM observable of interest. Although a wider range of fitting models are feasible, attention here is focused on what are termed waveform-domain models, comprised of exponential series, and spectral-domain models, comprised of pole series. These kinds of fitting models are shown to provide natural basis functions for many kinds of EM observables, whether these observables are based on experimental measurement or numerical computation.