Blockchain-based pseudonym management for vehicle twin migrations in vehicular edge metaverse Jiawen Kang, Xiaofeng Luo, Jiangtian Nie, Tianhao Wu, Haibo Zhou, Yonghua Wang, Dusit Niyato, Shiwen Mao, Shengli Xie Abstract—Driven by advances in metaverse and edge computing, vehicular edge metaverses are expected to disrupt current intelligent transportation systems. Vehicle Twins (VTs) deployed in edge servers provide metaverse services to improve driving safety and on-board satisfaction. To maintain uninterrupted metaverse experiences, VTs must migrate among edge servers following vehicle movements. This raises privacy concerns during dynamic communications. Pseudonyms, as temporary identifiers, can be used by VMUs and VTs to achieve anonymous communications. However, existing pseudonym management methods fall short in meeting the extensive pseudonym demands in vehicular edge metaverses, diminishing privacy preservation performance. To address these concerns, we present a cross-metaverse empowered dual pseudonym management framework. We utilize cross-chain technology to enhance pseudonym management efficiency and data security. We propose a metric to assess privacy levels and employ Multi-Agent Deep Reinforcement Learning (MADRL) to obtain an optimal pseudonym generating strategy. Numerical results demonstrate that our proposed schemes are high-efficiency and cost-effective, showcasing their promising applications in vehicular edge metaverses. Index Terms—Vehicular metaverse, cross-chain, twin migration, pseudonym management, deep reinforcement learning. ### I. INTRODUCTION The fast evolution of IoT systems has paved the way for the novel paradigm of metaverse, considered a creative application of B5G networks to meet people's growing demands for hyper spatiotemporal and surreal digital services. By integrating edge computing and intelligent transportation systems, metaverses can transition into a distinct paradigm called vehicular edge metaverses. Functioning as surreal realms that merge virtual spaces with the physical space at the network edge, vehicular edge metaverses can offer remarkable metaverse services such as Augmented Reality (AR) navigation to Vehicular Metaverse Users (VMUs) with lower latency and higher fidelity. These services can significantly increase VMUs' driving safety and on-board satisfaction throughout their journey. Vehicle Twins (VTs) as specific AI agents are one of the core components of delivering metaverse services, covering the entire life cycles of the vehicle and VMUs in vehicular edge metaverses. Embedded with versatile multimodal Large Language Models (LLMs), the VTs can process multimodal sensory inputs from their VMUs and the vehicle to enhance environmental perception and understanding. The VTs in virtual spaces are capable of continuously updating themselves through interacting with other VTs and their associated VMUs, thereby providing customized services back to the VMUs in the physical space. Owing to the inherent limitations in computing and storage resources of vehicles, the memory and computation-intensive tasks of maintaining VTs and runningBlockchain-based pseudonym management for vehicle twin migrations in vehicular edge metaverse Jiawen Kang, Xiaofeng Luo, Jiangtian Nie, Tianhao Wu, Haibo Zhou, Yonghua Wang, Dusit Niyato, Shiwen Mao, Shengli Xie Abstract—Driven by advances in metaverse and edge computing, vehicular edge metaverses are expected to disrupt current intelligent transportation systems. Vehicle Twins (VTs) deployed in edge servers provide metaverse services to improve driving safety and on-board satisfaction. To maintain uninterrupted metaverse experiences, VTs must migrate among edge servers following vehicle movements. This raises privacy concerns during dynamic communications. Pseudonyms, as temporary identifiers, can be used by VMUs and VTs to achieve anonymous communications. However, existing pseudonym management methods fall short in meeting the extensive pseudonym demands in vehicular edge metaverses, diminishing privacy preservation performance. To address these concerns, we present a cross-metaverse empowered dual pseudonym management framework. We utilize cross-chain technology to enhance pseudonym management efficiency and data security. We propose a metric to assess privacy levels and employ Multi-Agent Deep Reinforcement Learning (MADRL) to obtain an optimal pseudonym generating strategy. Numerical results demonstrate that our proposed schemes are high-efficiency and cost-effective, showcasing their promising applications in vehicular edge metaverses. Index Terms—Vehicular metaverse, cross-chain, twin migration, pseudonym management, deep reinforcement learning. ### I. INTRODUCTION The fast evolution of IoT systems has paved the way for the novel paradigm of metaverse, considered a creative application of B5G networks to meet people's growing demands for hyper spatiotemporal and surreal digital services. By integrating edge computing and intelligent transportation systems, metaverses can transition into a distinct paradigm called vehicular edge metaverses. Functioning as surreal realms that merge virtual spaces with the physical space at the network edge, vehicular edge metaverses can offer remarkable metaverse services such as Augmented Reality (AR) navigation to Vehicular Metaverse Users (VMUs) with lower latency and higher fidelity. These services can significantly increase VMUs' driving safety and on-board satisfaction throughout their journey. Vehicle Twins (VTs) as specific AI agents are one of the core components of delivering metaverse services, covering the entire life cycles of the vehicle and VMUs in vehicular edge metaverses. Embedded with versatile multimodal Large Language Models (LLMs), the VTs can process multimodal sensory inputs from their VMUs and the vehicle to enhance environmental perception and understanding. The VTs in virtual spaces are capable of continuously updating themselves through interacting with other VTs and their associated VMUs, thereby providing customized services back to the VMUs in the physical space. Owing to the inherent limitations in computing and storage resources of vehicles, the memory and computation-intensive tasks of maintaining VTs and running