【摘要】 为了在满足用户服务体验的同时，提高移动设备的能耗利用率，对移动边缘计算(Mobile Edge Computing, MEC)中的任务卸载问题进行研究。考虑整体智能家居等专用MEC应用场景，引入物理机分簇机制，提出一种基于单用户和任务分类的MEC任务卸载策略。结合任务的处理需求，将任务分为低速、中速和高速三类，搭建带有异构边缘的MEC系统架构，构建具有单输入流与多类顾客的排队模型，评估任务的平均处理时延和移动设备的能耗水平等性能指标。进行数值实验，分析不同系统参数对各性能指标变化带来的影响。权衡不同性能指标，构建系统成本函数，应用卡罗需-库恩-塔克(Karush-Kuhn-Tucker, KKT)条件方法，对任务卸载策略进行联合优化，从而实现系统成本的最小化。更多还原

【Abstract】 In order to meet the service experience of users and improve the energy efficiency of mobile device, a task offloading problem in Mobile Edge Computing(MEC) is studied. By considering private MEC scenarios, such as the multifunctional smart home, and introducing the mechanism of physical machine clustering, a task offloading strategy with single-user and tasks classification is proposed. To meet the processing requirements of different tasks, the tasks are classified into low-speed tasks, medium-speed tasks and high-speed tasks. The architecture of a MEC system with heterogeneous edge is built, and a queueing model with single-arrival flow and multiple-customers is established. The performance measures in terms of the average execution delay of tasks and the energy consumption level of mobile device are evaluated. The system cost function is constructed to trade off different performance measures, and the Karush-Kuhn-Tucker(KKT) condition method is applied to optimize the task offloading strategy with the minimum system cost.更多还原