Vehicles are becoming connected entities, and with the advent of online gaming, on demand streaming and assisted driving services, are expected to turn into data hubs with abundant computing needs. In this article, we show the value of estimating vehicular mobility as 5G users move across radio cells, and of using such estimates in combination with an online algorithm that assesses when and where the computing services (virtual machines, VM) that are run on the mobile edge nodes are to be migrated to ensure service continuity at the vehicles. This problem is tackled via a Lyapunov-based approach, which is here solved in closed form, leading to a low-complexity and distributed algorithm, whose performance is numerically assessed in a real-life scenario, featuring thousands of vehicles and densely deployed 5G base stations. Our numerical results demonstrate a reduction of more than 50% in the energy expenditure with respect to previous strategies (full migration). Also, our scheme self-adapts to meet any given risk target, which is posed as an optimization constraint and represents the probability that the computing service is interrupted during a handover. Through it, we can effectively control the trade-off between seamless computation and energy consumption when migrating VMs.

中文翻译：

---

车联网的移动感知和MEC服务的动态迁移

车辆已成为连接的实体，随着在线游戏的出现，按需流媒体和辅助驾驶服务有望变成具有大量计算需求的数据中心。在本文中，我们展示了估算5G用户在无线电小区之间移动时的车辆机动性的价值，以及将此类估算与在线算法结合使用的价值，该算法可以评估在何时何地在何处运行计算服务（虚拟机，VM）移动边缘节点将被迁移以确保车辆的服务连续性。通过基于Lyapunov的方法解决了此问题，在此方法以封闭的形式解决，从而导致了低复杂度和分布式算法，其性能是在现实生活中以数字方式评估的，其中包括成千上万辆汽车和密集部署的5G基础站。我们的数值结果表明，与以前的策略（完全迁移）相比，能源消耗减少了50％以上。此外，我们的计划可自适应满足任何给定的条件风险目标，作为优化约束，它表示切换期间计算服务被中断的概率。通过它，我们可以有效地控制迁移虚拟机时在无缝计算和能耗之间的权衡。