Some applications of modelbased parameter estimation in computational electromagnetics
Abstract
All of electromagnetics, whether measurement analysis, or computation, may be regarded as activities of information acquisition, processing, and presentation. It is relevant to mention the informationintensive nature of electromagnetics because this property can provide a useful, unifying perspective from which might be developed efficiency improvements in these various areas. If, for example, redundant information can be reduced or avoided in numerical calculations, the result should be a possibly substantial decrease in the amount of computer time that is needed for computer modeling. The use of a generalized signalprocessing approach called modelbased parameter estimation (MBPE) for making various improvements in either the efficiency of electromagnetic computer modeling or the efficiency of representing electromagnetic observables is considered. One is to increase the efficiency of the basic model computation itself, a typical example being that of replacing the rigorous Green's function for the antennainterface problem by a simpler, more easily computable, approximation. Another is to make more efficient use of the results that are computed, for example, by reducing the number of samples needed to construct a transfer function over some frequency band. A third arises because the electromagnetic fields/sources in various transformedpair domains, such as exhibited by the far fields and locations of a linearsource array, are described by exponential and pole series from which physically relevant parameters can be extracted, which are useful for such purposes as data compression and physical interpretation. The discussion deals primarily with MBPE based on exponential and polesseries models which yield generalized waveform and spectraldomain response functions of various transformpair variables. It is demonstrated how sampling such observables in terms of the appropriate variable or derivatives thereof leads to a data matrix from which the model parameters can be computed. Various kinds of data and models are used to illustrate the breadth of potential applications, with an emphasis on estimating frequency responses from frequencysampled data. Some concluding remarks are addressed to use of MBPE for the Green'sfunction application.
 Publication:

In AGARD
 Pub Date:
 September 1989
 Bibcode:
 1989madu.agarR....M
 Keywords:

 Antenna Design;
 Computerized Simulation;
 Electromagnetism;
 Estimating;
 Mathematical Models;
 Signal Processing;
 Data Compression;
 Data Processing;
 Far Fields;
 Green'S Functions;
 Position (Location);
 Transfer Functions;
 Communications and Radar