Multi-objective EM-based design optimization of compact branch-line coupler

Piotr Kurgan,Slawomir Koziel,Qingsha S. Cheng

DOI: https://doi.org/10.1109/NEMO.2017.7964238

2017-01-01

Abstract:This work addresses a problem of multi-objective design optimization of a computationally expensive compact branch-line coupler. Circuit miniaturization is achieved here primarily by using intricate slow-wave structures instead of conventional transmission lines. The presented approach exploits a point-by-point Pareto set exploration with consecutive trade-off designs found by applying adjusted design specifications and executing a surrogate-based optimization routine with a low-fidelity model of the slow-wave structure composed of duplicated electromagnetic simulation data blocks of its constitutive element. The optimization engine is a trust-region-embedded gradient search with Jacobian estimation. A set of six Pareto-optimal designs representing trade-offs between the coupler layout area and its bandwidth is obtained at the cost corresponding to less than forty high-fidelity simulations of the entire coupler.