Abstract: In recent years, vehicle edge networks have made great progress in integrating mobile edge computing technologies into vehicular networks, however, in real-time road traffic,emergency tasks represented by autonomous driving are usually concurrent with streaming application data, which brings additional task offloading energy consumption and offloading delay. Considering the characteristics of emergency tasks and general tasks, firstly, a task dynamic offloading framework based on non-orthogonal multiple access technique and Lyapunov theory is constructed, which takes the system stability as the premise and controls the transmission power and wireless interface of the vehicle in real time. Secondly, for the energy efficiency optimization problem of general tasks and the execution delay optimization problem of emergency tasks, based on the exact potential game theory. A joint power and channel allocation algorithm compatible with dual task types is proposed to obtain the optimal pure strategy Nash equilibrium solution. Finally, the numerical results verify that the proposed scheme has significant advantages over other benchmark schemes in terms of system stability and energy cost reduction.

Key words: vehicular edge networks, emergency tasks, Lyapunov theory, exact potential games

摘要： 近些年来，车辆边缘网络将移动边缘计算技术融合进车联网取得了极大的进展，然而，在实时道路交通中，以自动驾驶为代表的紧急任务通常与流媒体应用数据并发，从而带来额外的任务卸载能耗与卸载时延。综合考虑紧急任务与普通任务的特性，构建了一个基于非正交多址技术和李雅普诺夫理论的任务动态卸载框架，该框架以系统稳定为前提，对车辆的传输功率与无线接口进行实时控制；针对普通任务的能量成本优化问题以及紧急任务的执行时延优化问题，在精确势博弈理论的基础上提出了兼容双任务类型的联合功率与信道分配算法来获得最优的纯策略纳什均衡解。数值结果验证了所提方案相较于其他基准方案在系统稳定以及减小系统能量成本等方面具有显著的优势。

关键词: 车辆边缘网络, 紧急任务, 李雅普诺夫理论, 精确势博弈