Computational Offloading for MEC Networks with Energy Harvesting: A Hierarchical Multi-Agent Reinforcement Learning Approach

Multi-access edge computing (MEC) is a novel paradigm that leverages nearby MEC servers to augment the computational capabilities of users with limited resources. In this paper, we investigate offloading problem in multi-user multi-server systems energy harvesting, aiming minimize both system latency and consumption by optimizing task offload location selection ratio.We propose hierarchical strategy based on multi-agent reinforcement learning (MARL). The proposed decomposes into two sub-problems: high-level low-level ratio problem. complexity reduced decoupling. To address these sub-problems, framework proximal policy optimization (MAPPO), where each agent generates actions its observed private state avoid action space explosion due increasing number user devices. Simulation results show HDMAPPO outperforms other baseline algorithms terms average latency, consumption, discard rate.