Artificial intelligence for RIS-aided wireless communications

The recently proposed Reconfigurable Intelligent Surface (RIS) can reconstruct the wireless channels between the transceivers, thus it is regarded as a promising technology for future 6G wireless networks to enlarge their coverage and improve the capacity. However, RISs also impose some new challenges, such as an unaffordable overhead for channel estimation and high complexity for real-time beam-forming. Fortunately, the impressive success of Artificial Intelligence (AI) in various fields has inspired its application in RIS-aided communications to address these challenges. In this paper, two pairs of dominant methodologies of using AI for RIS-aided wireless communications are discussed. The first one is the AI-based algorithm design, which is illustrated by some examples of typical transmission techniques. The second one is the AI-based architecture design, which breaks the classical block-based design rule of wireless communications in the past few decades. The interplay between AI and RIS is also highlighted. Finally, key challenges and future opportunities in this emerging area are pointed out. We expect that this paper will stimulate more promising AI-based investigations for RIS-aided wireless communications.