Explainable Planar Multiband Antenna Designer with Wasserstein Generative Adversarial Network

In the antenna design process, predicting return loss via electromagnetic (EM) simulation is crucial for understanding the antenna's behavior. As EM simulation can be time-intensive, its role is often confined to simulating the response, and not extended to proposing alternative designs. To address this, we propose a novel artificial intelligence (AI) framework for antenna design. It comprises a regressor for accurate and fast response prediction, a generative designer for proposing a vast number of new designs that meet users' requirements, and an explainer for analyzing the impact of design parameters. Application of the proposed AI framework to the design of planar multiband antennas has demonstrated its accuracy and capability.