Computation Rate Maximization in Active RIS-Assisted Hybrid FDMA-NOMA MEC Systems: A Deep Reinforcement Learning Approach

This letter investigates an active RIS-assisted mobile edge computing system where IoT devices (IDs) offload compute-intensive tasks to a base station under limited energy budget. IDs communicate in pairs leveraging hybrid frequency-division and nonorthogonal multiple-access schemes. To maximize the sum computation rate of IDs, we jointly optimize the energy allocation for offloading and local computing, RIS phase shifts, and receive beamforming. To tackle the non-convex problem, we apply a deep reinforcement learning approach based on the proximal policy optimization algorithm, ensuring stable and efficient training. The simulation results highlight the superiority of the proposed approach when compared to benchmark methods.