A Study on Machine Learning-based Approaches for Reconfigurable Intelligent Surface

Next-generation wireless communication needs to support an increase in mobile users, the rapid expansion of cellular data traffic, and various applications. For this next-generation wireless communication, reconfigurable intelligent surfaces (RIS) have recently received much attention. RIS can autonomously perform improvements in radio transmission coverage and capacity. Therefore, RIS has emerged as a promising solution for sixth-generation communication infrastructure. To further maximize these advantages of RIS, many machine learning (ML) techniques can be applied. In this paper, we provide a thorough review of the ML algorithms used in RIS. To better understand these two technologies, a brief overview of RIS, a summary of ML methods with RIS architecture, and a comparison of the available methodologies are provided. In addition, the importance of open research topics is emphasized to provide useful directions for further research.