Enhancing the Incident Angle Band in Carpet Cloaking using Deep Neural Networks

Abstract

Metasurface-based carpet cloaks although are very thin and have a good performance, are very sensitive to the incident angles and work only for a very narrow range of incident angles. In this paper, we have addressed this issue using PSO algorithm in combination with a deep neural network. A passive single-layer metasurface capable of making objects under it invisible for a wide range of incident angles, 0 to 70 degrees, is designed and optimized using PSO algorithm in combination with a deep neural network. The cloak has been designed and evaluated using full-wave numerical simulations, and the results show a good performance in the mentioned incident angle band.