The ever increasing amount of connected mobile devices and users in the Internet of vehicles and the fact bicycles, electric scooters and users' smartphones are also connected poses a real challenge in terms of ensuring quality of service. The constantly changing topology due to the high‐mobility of devices and users greatly impacts its stability and the connectivity of devices. Furthermore, to ensure safety of people in the case of autonomous vehicles it is of paramount importance to ensure excellent reliability. This is why we have developed a solution that is based on machine learning to classify devices according to their mobility profile and that uses a scoring system to select the best candidates to act as mobile relays amongst devices with a suitable mobility profile. The scoring system allows to find critical locations in terms of user density. This solution does not require a dedicated infrastructure such as road side units. Simulations results will show the proposed solution increases the packet delivery ratio by up to 6%, reduces the energy consumption by up to 30% and increases the efficiency of bandwidth usage without sacrificing the end delay of users and devices compared with the state‐of‐the‐art.

中文翻译：

---

根据车辆互联网中移动性类型和本地化网络指标的分类来选择中继器

车辆互联网以及自行车，电动踏板车和用户的智能手机也已连接的互联移动设备和用户的数量不断增加，这对确保服务质量构成了真正的挑战。由于设备和用户的高移动性，拓扑结构不断变化，极大地影响了其稳定性和设备的连接性。此外，为了确保自动驾驶汽车的人员安全，至关重要的是确保出色的可靠性。这就是我们开发基于机器学习的解决方案的原因，该解决方案基于设备的移动性概况对设备进行分类，并使用评分系统选择最佳候选者，以在具有合适的移动性概况的设备之间充当移动中继。计分系统允许根据用户密度查找关键位置。该解决方案不需要专用的基础设施，例如路边单元。仿真结果表明，与状态相比，该解决方案可将数据包的传输率提高多达6％，将能耗降低多达30％，并提高了带宽使用效率，而不会牺牲用户和设备的最终延迟。艺术。