When MetaVerse Meets Computing Power Networking: An Energy-Efficient Framework For Service Placement

The Metaverse, as the next-generation Internet, disrupts the physical world to create an immersive virtual digital realm. Its implementation relies on key enablers, including virtual reality, AI, blockchain, and 6G. However, these technologies involve computationally intensive operations with high energy consumption, posing a challenge to the sustainability of the Metaverse. A promising approach is to leverage ubiquitous and heterogeneous computing power, encompassing end-edge-cloud devices, to optimize resource allocation for computing and networking. Unfortunately, existing edge-cloud architectures fall short in meeting the diverse resource demands of Metaverse applications. In this article, we present an energy-efficient framework that delivers Metaverse services meeting quality-of-service metrics. The framework is based on computing power networking, a network-driven distributed computing environment enabling the free flow of isolated computing power. We introduce a typical Metaverse scenario involving video analytics and formulate the problem of optimal service placement, which we solve using the deep reinforcement learning algorithm, proximal policy optimization. Experimental results demonstrate the superior energy efficiency of our proposed framework. Finally, we discuss future directions for building a sustainable Metaverse, emphasizing the need for ongoing advancements in energy efficiency.