Surrogate-assisted multi-objective optimization of compact microwave couplers

Abstract

This work presents a rigorous methodology for expedited simulation-driven multi-objective design of microwave couplers with compact footprints. The proposed approach is a viable alternative for computationally expensive population-based metaheuristics and exploits a surrogate-assisted point-by-point Pareto set determination scheme that utilizes – for the sake of computational efficiency – space-mapping-corrected equivalent circuit models. The technique is showcased using a complex design example of a compact rat-race coupler, for which a set of nine alternative design solutions is efficiently identified. The latter design solutions illustrate the best possible trade-offs between conflicting design objectives for the structure at hand, that is, its operational bandwidth and the layout area. The overall design cost corresponds to approximately 20 high-fidelity electromagnetic simulations of the miniaturized coupler. Several selected trade-off designs have been manufactured and measured for the purpose of method validation.