Reduced-cost data-driven modeling of antenna structures

Abstract

Fast and accurate replacement models (surrogates) are indispensable in the design of microwave and antenna structures. Conventional data-driven modeling of antennas is usually prohibitively expensive in terms of data acquisition due to highly nonlinear responses of typical responses (e.g., reflection coefficient versus frequency). In this paper, a technique for reduced-cost modeling is proposed. The technique is based on performance-driven constraining of the surrogate model region of validity. Our approach permits dramatic reduction of the sampled portion of the design space without formally reducing geometry parameter ranges. It is demonstrated using a dielectric resonator antenna and favorably compared to conventional modeling using kriging interpolation.