Rapid multi-objective optimization of a MIMO antenna for UWB applications

Rapid multi-objective design optimization of ultrawideband (UWB) multiple-input-multiple-output (MIMO) antenna is presented. Our approach allows for finding the best possible design trade-offs (a so-called Pareto set) with respect to the two objectives: minimization of the reflection coefficient and reduction of the antenna footprint. Computational efficiency is obtained by using variable-fidelity electromagnetic simulation models, auxiliary response surface approximations, as well as initial design space reduction. The optimization process is a five-step procedure: (i) design space reduction, (ii) gathering the coarse-discretization EM-simulation data, (iii) construction of the response surface approximation (RSA) model of the samples, (iv) identification of the initial Pareto set by multi-objective evolutionary optimization of the RSA model, and (v) design refinement, where the selected designs are improved to elevate the Pareto set to high-fidelity EM model level.