The Internet of Vehicles connects all vehicles and shares dynamic vehicular data via wireless communications to effectively control vehicles and improve traffic efficiency. However, due to vehicular movement, vehicular data sharing based on conventional cloud computing can hardly realize real-time and dynamic updates. To address these challenges, artificial intelligence (AI)-empowered mobile/multi-access edge computing (MEC) has been regarded as a promising technique for intelligently supporting various vehicular services and applications in proximity of vehicles. In this article, we investigate the issue of collaborative data sharing in vehicular edge networks (VENs) with the deployment of AI-empowered MEC servers. Furthermore, we present a specific mode for collaborative data sharing. Then we propose a novel collaborative data sharing scheme with deep Q-network and federated learning to ensure efficient and secure data sharing in the VEN. Evaluation results demonstrate the effectiveness of the proposed scheme on reducing latency of vehicular data sharing. Finally, we discuss several open issues and future challenges of the AI-empowered VEN.

中文翻译：

---

联邦学习赋能的车辆边缘网络协作数据共享


车联网将所有车辆连接起来，通过无线通信共享车辆动态数据，有效控制车辆，提高交通效率。然而，由于车辆的运动，基于传统云计算的车辆数据共享很难实现实时动态更新。为了应对这些挑战，人工智能（AI）支持的移动/多路访问边缘计算（MEC）被认为是一种有前途的技术，可以智能地支持车辆附近的各种车辆服务和应用。在本文中，我们研究了通过部署人工智能支持的 MEC 服务器来实现车辆边缘网络 (VEN) 中的协作数据共享问题。此外，我们提出了一种协作数据共享的特定模式。然后，我们提出了一种采用深度 Q 网络和联邦学习的新型协作数据共享方案，以确保 VEN 中高效、安全的数据共享。评估结果证明了所提出的方案在减少车辆数据共享延迟方面的有效性。最后，我们讨论了人工智能驱动的 VEN 的几个悬而未决的问题和未来的挑战。