Energy-Efficient Hybrid Precoding Schemes for RIS-Assisted Millimeter-wave Massive MIMO

Reconfigurable intelligent surfaces (RIS)-assisted multiuser millimeter-wave (mmWave) wireless transmitters are envisioned to be an attractive paradigm for the future 6G wireless systems that can achieve a transmission in a cost-effective, energy-efficient and intelligent way. In this paper, RIS is introduced as a promising technology to assist the machine learning (ML) inspired energy-efficient hybrid precoding scheme that requires low-cost and energy-efficient switches and inverters, and employs the idea of cross entropy (CE) optimization. More specifically, adaptive CE (ACE) algorithm is adopted as an attractive technique to find the corresponding optimal precoding weights. Extensive simulations have been conducted to evaluate the achievable sum-rate and the energy efficiency (EE) of the proposed RIS-assisted ACE-based hybrid precoding scheme. In addition, the performance of the proposed hybrid precoder is compared to the performance of conventional hybrid precoding schemes. The results demonstrate that the proposed hybrid precoding architecture outperforms the conventional schemes when the number of reflecting elements is higher than the number of users, and can potentially offer high EE.