Green orchestra: joint spatiotemporal task scheduling and hybrid energy coordination in computing power networks

Recent advancements in information and communication technologies necessitate powerful computing power and network capabilities. Fortunately, Computing Power Networks (CPNs) have emerged to seamlessly integrate distributed computing resources via network orchestration, enabling high-throughput and on-demand computing services. However, CPNs consume substantial energy and generate significant carbon emissions when processing massive data. What’s worse, the interplay between CPN nodes and network paths, and spatiotemporal variations in renewable energy complicate energy-efficient task scheduling. To address these issues, we design a joint spatiotemporal task scheduling and hybrid energy coordination mechanism to efficiently manage and allocate computing, network, and energy resources, achieving green orchestra in CPNs. Firstly, we design a novel green CPN framework that synergizes computing resources, network resources, and enhanced energy coordination systems. Then, we propose a spatiotemporal task scheduling scheme with a triple selection of CPN nodes, routing paths, and forwarding time. The scheme can optimize energy consumption and carbon emissions while ensuring delay constraints and load balancing. Lastly, we formulate the problem as a Markov Decision Process (MDP) and design a customized Deep Reinforcement Learning (DRL) approach to solve it. Simulation results validate our scheme outperforms the benchmark schemes in learning efficiency, energy savings, carbon reduction, and renewable energy utilization efficiency.