MODEL-BASED PARAMETER ESTIMATION IN ELECTROMAGNETICS: III--Applications to EM Integral Equations

Abstract

Problem solving in electromagnetics, whether by analysis, measurement or computation, involves not only activities specific to these particular categories, but also some concepts that are common to all. Fields and sources are sampled as a function of time, frequency, space, angle, etc. and boundary conditions are satisfied through mathematical imposition or experimental conditions. The source samples, usually the unknowns in a problem, are found numerically or analytically by requiring them to satisfy both the appropriate form of Maxwell's Equations as relationships between them, together with the applicable boundary conditions. Alternatively, source samples may be measured under prescribed experimental conditions. These sampled relationships can be interpreted from the viewpoint of signal and information processing, and are mathematically similar to various kinds of filtering operations. It is this similarity that is discussed here in the context of modelbased parameter estimation, where the dependence of electromagnetic fields and sources that produce them are both regarded as generalized signals. MBPE substitutes the requirement of obtaining all samples of desired quantities (physical observables such as impedance, gain, RCS, etc. or numerical observables such as impedance-matrix coefficients, geometrical-diffraction coefficients, etc.) from first-principles models (FPMs) or from measured data (MD) by instead using a reduced-order, physically-based approximation, a fitting model (FM), to interpolate between, or extrapolate from, FPM or MD samples. When used for electromagnetic observables, MBPE can reduce the number of samples that are required to represent responses of interest, thus increasing the efficiency of obtaining them. When used in connection with the FPM itself, MBPE can decrease the computational cost of its implementation. Some specific possibilities for improving FPM efficiency are surveyed, specifically in terms of using FMs to simplify frequency and spatial variations associated with FPMs. Examples of MBPE applications are included here as well as speculative possibilities for their further development in improving FPM performance. [Vol. 10, No. 3 (1995), Special Issue on Advances in the Application of Method of Moments to Electromagnetic Radiation and Scattering Problems, pp 9-29]