Deep learning‐assisted reconfigurable intelligent surface for enhancing 6G mobile networks

Abstract

AbstractThis article presents the role of Reconfigurable Intelligent Surface (RIS) using the Multi‐Input Multi‐Output (MIMO) technique as the enabling technology to boost the achievable data rate for mobile networks of the sixth generation (6G). The RIS has been adopted to mitigate the interference at the Cell Edge User (CEU) placed where two adjacent cells are separated. That is by reflecting the incident interference signal in the CEU direction out of phase with the basic interference signals coming from the interfering BS towards the CEU. This article adopts an efficient solution for designing the RIS redirecting (reflection) matrix with trivial training overhead using Deep Learning (DL) technology. Whereas a few of the reflecting elements in RIS are chosen to be active (attached to the baseband), whilst the majority are chosen to be passive, in which the active element's channels are known and used as medium indicators and indicate further the positions of the transmitter and receiver. This article illustrated the efficacy of the adopted framework with DL, by changing the training parameters regarding the data rates that can be achieved, the Spectral Energy Efficiency (SEE), and the Satisfaction Rate (SR). As a result, the proposed model improves the achievable data rate by a near average of 97% above the reference model. Furthermore, it enhances the achievable data rate above the baseline model that assumes no RIS used by an average of 115%. Therefore, The DL method demonstrated that the proposed model is promising for enhancing 6G Mobile Networks.