Antenna Design Using the Efficient Global Optimization (EGO) Algorithm

Abstract : In this report, we discuss antenna design optimization using the efficient global optimization (EGO) algorithm. The first design is a parasitic super directive array where we compare EGO with a classic optimization technique called Nelder-Mead and then with GA optimization. The second antenna design optimization problem is a wideband antenna element backed by both a metamaterial ground plane and a conducting ground plane. The antenna element over a conducting ground plane was fabricated and we present a comparison of test results with predicted results. The third antenna design is an infinite, periodic array of fragmented patch elements. We also investigate two potential areas of improvement to the EGO algorithm. The first is in the "endgame" where we consider techniques to make the algorithm more accurate in the final estimate of the location of the global minimum. The last area involves the selection of initial data sets with the goal of more efficient algorithms.