Reliable low-cost co-kriging modeling of microwave devices

A reliable methodology for accurate modeling of microwave devices is presented. Our approach exploits co-kriging which utilizes low- and high-fidelity EM simulation data and combines them into a single surrogate model. Densely sampled low-fidelity data determines a trend function which is further corrected by sparsely sampled high-fidelity simulations. Low-fidelity EM data is also enhanced by using a frequency scaling. With our method, accurate models can be obtained at a fraction of cost required by conventional approximation models that are exclusively based on high-fidelity simulations. Two cases of microstrip bandpass filters are considered. Comparisons with conventional approximation models as well as application examples are also given.