Lightning simulator. Supplementary knowledge: Smart metasurface, also called "reconfigurable intelligent surface" or "intelligent reflective surface", in English is RIS (Reconfigurable Intelligence Surface), or IRS (Intelligent Reflection Surface). Metasurfaces composed of subwavelength resonators are capable of controlling the wavefront of electromagnetic waves. Due to its ultra-compact structure, it has important uses in many fields, including imaging, detection and information processing. However, most metasurfaces that have been proposed so far are static, meaning that once fabricated, they will have a fixed functionality.
Encoded metasurfaces are one of the most promising candidates for enabling dynamic adjustment. In the coded metasurface, each subwavelength cell contains an electrically or optically tunable diode that can switch between two states with high contrast reflection amplitude and phase, resulting in digital signals that can be reconfigured in real time. 2D outline. This means we can tune these diodes to tune the response of the metasurface device.
Recently, some scientists have combined codable metasurfaces and deep learning to design smart devices that can adapt to the environment. A conceptual diagram of the device is shown in Figure 1. The system consists of three main parts: metasurface, sensor, and control system. The sensor senses the external environment and feeds this information to the trained deep neural network model. The network can quickly output results, and the controller adjusts the metasurface diode based on these results, ultimately achieving the purpose of adapting to the environment. Figure 1c demonstrates maintaining a fixed direction of the reflection spectrum by adjusting the tilt direction of the metasurface. Because the deep neural network model can quickly output results, the device can achieve real-time control.