On low-cost space mapping optimization of antenna structures

In this paper, a low-cost formulation of space mapping (SM) for design optimization of antenna structures is investigated. The major challenge is that low-fidelity (or coarse) antenna models are normally obtained from coarse-discretization EM simulations so that their evaluation cost cannot be neglected. Consequently, the SM algorithm has to aim at reducing both the number of fine and coarse model evaluations throughout the optimization process. In our approach, a fast parameter extraction procedure is proposed along with the trust-region-based surrogate optimization routine. As demonstrated using a microstrip PIFA antenna, this allows for maintaining overall low optimization cost even with relatively expensive coarse model.