Efficient Resource Allocation for Building the Metaverse with UAVs: A Quantum Collective Reinforcement Learning Approach
Yuhang Wang, Ying He, F. Richard Yu, Bin Song, Victor C. M. Leung
Abstract
The Metaverse offers a highly immersive virtual world where users interact with others and objects in real time through their avatars. The integration of Internet of Things (loT) devices plays a pivotal role in achieving seamless synchronization between these two realms. In this article, we explore the utilization of unmanned aerial vehicles (UAVs) for image capture and data transmission to base stations, which then render and optimize images to construct the Metaverse. Considering the limited resources, when the communication or storage resources of base stations are insufficient, UAV tasks can be offloaded to the Web3 cloud servers for storage and processing, so as to avoid wasting the information collected by UAVs. This continuous interaction among the Web3 cloud servers, UAVs, and base stations enables the real world and Metaverse updating simultaneously. To enhance the effectiveness of our approach, we introduce a quantum collective reinforcement learning method, which empowers UAVs with fast training capabilities. This collective learning enables UAVs to autonomously adapt and swiftly respond to novel scenarios, thereby ensuring seamless integration with new environments. Extensive simulations validate the effectiveness of our proposed method in achieving efficient offloading and enhancing overall system performance within the Metaverse and Web3 ecosystem.