A comparison of full pattern and response feature based modeling of antenna radiation patterns

This paper compares the interpolation based modeling of antenna radiation pattern response features when the models are built directly on the response feature of interest, or on an expansion coefficient describing the full radiation pattern. It is shown that by interpolating the expansion coefficients of an orthogonal set of basis functions for the full radiation pattern, the expected rational response in frequency (and some geometric parameters) can be exploited to construct accurate interpolation models using very few training data. In many cases it is not possible to predict the functional form of the response features, and therefore interpolation directly on these data requires more training samples than for the expansion coefficient case.