Optimizing Energy Consumption through Scheduling in Low-resource Edge Clusters using Multi-agent PPO

With the growing demand for computing resources, data centers must optimize energy consumption while maintaining performance. This paper focuses on optimizing job scheduling in low-resource edge clusters using Multi-Agent Proximal Policy Optimization (MAPPO). Cloud computing offers scalability and flexibility but faces challenges in energy efficiency due to the high consumption of traditional data centers. By leveraging low-resource computational clusters at the edge, we aim at reducing energy costs while meeting performance needs. A MAPPO-based scheduling policy is proposed to dynamically allocate jobs between the cloud and machines in a low-power cluster, balancing energy efficiency and scalability. The policy was designed for real-world deployment, ensuring fast decision-making and effective resource management. We evaluate the model's effectiveness in minimizing energy usage.