A parametric study of genetic algorithms for pixelated antenna design

To reduce the design cycle of antenna development, pixelated antenna design becomes a very important approach. The success of pixelated antenna design relies on understanding the effect of genetic operators and learning thoroughly how to set the level of parameters. The aim of this paper is to present a numerical study for designing genetic algorithms (GAs) that perform well and robust on pixelated-antenna-design problems. We show that for a GA using orthogonal arrays as initialization condition we can enhance the capability of exploring the solution domain. On the other hand, the important guiding features to generate a successful design, such as the “sweet spot” of GA parameters and population sizing, are illustrated in this paper.