Deep-Learning Estimation of Complex Reverberant Wave Fields with a Programmable Metasurface

Abstract

Electromagnetic environments are becoming increasingly complex and congested, creating a growing challenge for systems that rely on electromagnetic waves for communication, sensing, or imaging, particularly in reverberating environments. The use of programmable metasurfaces provides a potential means of directing waves to optimize wireless channels on demand, ensuring reliable operation and protecting sensitive electronic components. Here we introduce a technique that combines a deep-learning network with a binary programmable metasurface to shape waves in complex reverberant electromagnetic environments, in particular ones where there is no direct line of sight. We apply this technique for wavefront reconstruction and control, and accurately determine metasurface configurations based on measured system scattering responses in a chaotic microwave cavity. The state of the metasurface that realizes desired electromagnetic wave field distribution properties was successfully determined even in cases previously unseen by the deep-learning algorithm. Our technique is enabled by the reverberant nature of the cavity, and is effective with a metasurface that covers only approximately 1.5% of the total cavity surface area.