Intelligent and efficient Metaverse rendering and caching in UAV-aided vehicular edge computing

The extensive application of the Metaverse in the Internet of Vehicles (IoV) has provided broader application scenarios and innovative opportunities for intelligent vehicle travel. The implementation of the Metaverse, which necessitates low latency, high precision, and swift feedback and interaction, can be effectively addressed by harnessing unmanned aerial vehicle (UAV)-assisted IoV technology. However, the actual wireless communication environment of UAV-assisted IoV networks, characterized by variability and complexity amidst numerous uncertain and uncontrollable interference factors, underscores the urgent need for research on the efficient communication and computing within the Metaverse. In this work, we investigate an efficient rendering scheme for Metaverse applications in UAV-aided edge computing networks, where multiple UAVs perform various Metaverse applications for vehicles with the help of a ground base station. Considering image quality and frame refresh rate as key metrics, we formulate a joint system utility optimization problem to minimize response time and energy consumption. To provide stable and high-quality vehicular Metaverse services, we develop an intelligent rendering and caching method for intelligent vehicular Metaverse, where a diffusion probabilistic model-based Metaverse frame rendering algorithm and a deep learning-based Metaverse frame caching algorithm are jointly designed. The proposed method can achieve optimal resource allocation results with low time complexity by fully exploring the benefits of a double auction model between vehicles and UAVs and a social model between different vehicles. Based on real-world datasets, we conduct extensive simulation experiments. Numerical results indicate that the proposed algorithm can improve resource utilization and reduce Metaverse frame rendering time and system energy consumption significantly.