Rapid dimension scaling of dual-band antennas using variable-fidelity EM models and inverse surrogates

Computationally, efficient scaling of geometry parameters that allows finding the structure dimensions for various operating frequencies is important to expedite the antenna design process. Unfortunately, the scaling problem is challenging, particularly for dual- and multi-band antennas because certain geometry parameters may influence several bands. Also, the relationship between the optimum dimensions and operating frequencies is normally nonlinear and cannot be accurately described using analytical models. In this work, we address the scaling problem for dual-band antennas so that optimum dimensions ensuring that the antenna operates at the two frequencies of choice can be obtained at minimum cost. The presented approach exploits inverse surrogate models where geometry parameters of the structure are represented through design surfaces. The latter are identified using a set of reference designs obtained for coarse-discretization electromagnetic (EM) simulation models and further corrected to be applicable at the high-fidelity EM model level. The proposed method is demonstrated using two examples of dual-band planar antennas. Simulation results are validated using experimental data.