Antenna design using surrogate models and adjoint sensitivity

Abstract

Design closure aided by electromagnetic (EM) simulations is often a necessary step in the antenna design process. It is also challenging because of the fact that accurate EM simulations are computationally expensive. In this paper, we discuss simulation-driven antenna optimization using surrogate models and adjoint sensitivity. We demonstrate that incorporating sensitivity information into the surrogate model improves the robustness of the optimization process and can also be utilized to reduce its computational cost.