Expedited microstrip antenna array design through surrogate-based optimization

In this work, a technique for expedited design of microstrip antenna arrays using surrogate-based optimization is presented. The proposed methodology exploits an antenna array model which superposes EM-simulated responses (both radiation and reflection) of the array with only one radiator active at a time. Low optimization cost is maintained by constructing the model from coarse-discretization EM-simulation data. In the process, the antenna array model is iteratively corrected and re-optimized to yield an improved set of excitation amplitudes and/or phases. Our approach permits simultaneous handling of the radiation pattern and the reflection coefficients, i.e., accounting for design specifications imposed on both reflection and radiation responses. It is demonstrated using a planar 10 by 10 microstrip array with the optimized design obtained at the cost corresponding to a few evaluations the high-fidelity EM model.