Dynamic and Real-Time Frame Rendering for Edge Computing-Enabled Metaverse Systems

Metaverse is an immersive, seamless, interactive, and comprehensive virtual world, as well as a replication, extension, and transcendence of the real world. The constantly improving image resolution and enriched scene details of Meta-verse applications have increased complexity and difficulty in rendering Metaverse panoramic frames. However, due to factors such as device computing, communication, and battery power, existing Metaverse scene rendering methods are unable to meet the high-quality rendering task requirements of users, resulting in image lag, high rendering latency, and reduced user experience quality. In this paper, through the collaborative use of the computation resources provided by edge servers and terminal devices, we propose a multi-terminal collaborative adaptive meta-universe panoramic frame rendering method, in which the gradient provided by model-based meta boundary rendering problem is used to generate integer rendering decisions using a Deep Reinforcement learning-based Frame Rendering (DRFR) framework. The meta-universe rendering task is decomposed and deployed to different computing platforms for execution. Comprehensive simulation results demonstrate that the DRFR model can reduce the Metaverse frame rendering time and improve user experience.