Investigation of Antenna Topology Optimization Using Genetic Algorithms

Abstract

This paper presents a new initialization method and optimum parameterization for genetic algorithms (GAs) in pixelated antenna design. The mathematical model of pixelated antenna design is first illustrated. In order to solve this mathematical model, GAs are widely used as the optimization algorithm. In general, GAs are initialized by a randomized population, which may lead to unbalanced exploration and slower convergence. To overcome such a limitation, this paper presents orthogonal arrays serving as the initialization mechanism to generate a fairly-distributed population; as a result, the efficiency is greatly enhanced. In addition, the optimal parameterization is clarified, and the associated convergence history exhibits robust performances as compared to conventional approaches. Hence, the design cycle of pixelated antennas is reduced significantly by the proposed technique.