Programmable transmission metasurface scattering control under obstacles based on deep learning

Abstract

The emergence of 5G represents a pivotal step in merging mobile communication networks with the Industrial Internet of Things. Despite the numerous advantages of 5G, the presence of unknown obstacles can adversely affect user signals. Although mitigating signal pressures can be achieved by increasing base station density, it often involves bulky equipment and high costs. To address this, we propose a deep learning-based method for controlling tunable transmissive metasurfaces and validate their scattering control capabilities in the presence of obstacles. By constructing a network model to analyze the mapping relationship between metasurface arrays and far-field scattering, rapid control of scattering characteristics is achieved. AI-driven high-performance tunable metasurfaces exhibit vast potential applications in intelligent communication, offering a universal solution for intelligent control in complex signal environments.