Machine Learning Based Antenna Design

Abstract

The communication era is evolving exponentially with new technologies emerging progressively, to satisfy ubiquitous high data rate transfer. In this context, antenna design has become critical, since efficient communication system requires appropriately designed antenna serving its purpose. An antenna design strategy based on machine learning that accomplishes directional communication using patch antenna is presented here. Genetic Algorithm (GA) is popularly employed for solving limited and unbounded optimization issues that is based on natural selection, which is the primary driver of biological evolution, where the population of individual solutions are repeatedly transformed into newer versions, in search for optimal solutions. NSGA-II (Non-Dominated Sorting Genetic Algorithm-II) is an optimization technique that enables to optimize multiple objectives without being dominated by any one solution. The algorithm is configured to maximize gain & directivity and minimize aperture. The simulation results confirm that suggested antenna design is suitable for high gain applications where miniaturization is of priority.