Fast multi-objective antenna design through variable-fidelity EM simulations

A technique for fast multi-objective antenna optimization is introduced. A kriging interpolation surrogate constructed from sampled coarse-mesh EM simulations is utilized by multi-objective evolutionary algorithm (MOEA) to obtain the initial Pareto front approximation. The surrogate is defined in a subset of the original design space, determined by means of independently optimized individual objectives. Response correction techniques are subsequently used to refine the designs selected from the initial Pareto set. Our approach is demonstrated using a planar UWB dipole antenna.