Energy Efficiency of RIS-Assisted NOMA-Based MEC Networks in the Finite Blocklength Regime

In this paper, we investigate a reconfigurable intelligent surface (RIS)-assisted mobile edge computing (MEC) network aiming to maximize the energy efficiency in the finite blocklength (FBL) regime under both coding length and maximum decoding error rate constraints. We first analyze the single user equipment (UE) case and propose a three-step alternating optimization algorithm to solve the problem. Extending the system model, we subsequently investigate a network with multiple UEs, in which non-orthogonal multiple access (NOMA) transmission is adopted. In this more general setting, we also conduct a convergence analysis. Furthermore, we introduce a UE-grouping scheme for hybrid NOMA-TDMA transmission and develop a dynamic CPU frequency allocation algorithm at the mobile edge computing (MEC) server. Numerical results show that the proposed algorithms solve the problem efficiently. Via numerical results, we also identify the impact of various parameters (e.g., coding blocklength, the number of RIS elements, computational resources, number of UEs) on the energy efficiency. Furthermore, with the numerical results, we verify the validity of UE grouping method and demonstrate that the proposed dynamic CPU frequency allocation can enhance the performance substantially.