Vehicular Metaverses
Vehicular metaverses integrate digital twins of vehicles (Vehicle Twins or VTs) with immersive virtual services, aiming to revolutionize the automotive industry by enhancing user experience and efficiency. Current research heavily focuses on optimizing VT migration between edge servers (including UAVs) to ensure seamless service despite vehicle mobility and resource constraints, employing techniques like reinforcement learning (including diffusion-based and multi-agent approaches) and auction-based resource allocation mechanisms. These efforts address challenges in efficient resource management, security (including defense against adversarial attacks), and trust, ultimately impacting the design and implementation of robust and secure vehicular metaverse platforms.
Papers
Diffusion-based Reinforcement Learning for Dynamic UAV-assisted Vehicle Twins Migration in Vehicular Metaverses
Yongju Tong, Jiawen Kang, Junlong Chen, Minrui Xu, Gaolei Li, Weiting Zhang, Xincheng Yan
Multi-attribute Auction-based Resource Allocation for Twins Migration in Vehicular Metaverses: A GPT-based DRL Approach
Yongju Tong, Junlong Chen, Minrui Xu, Jiawen Kang, Zehui Xiong, Dusit Niyato, Chau Yuen, Zhu Han