Abstract:

The computation offloading technology in Mobile Edge Computing（MEC） offloads the computing tasks of terminal devices to the edge of the network to solve the problems such as extension, large energy consumption, and high load in the cloud computing center. The concept of MEC, the current mainstream network architecture and deployment solutions of MEC are introduced. From the aspect of offloading decision-making, a detailed study of the offloading technology of computing-intensive applications in the MEC environment is carried out. This paper analyzes and compares the four computation offloading schemes from the optimization goals of minimizing delay, minimizing energy consumption, balancing delay and energy consumption, and maximizing revenue, and summarizes their key. By analyzing the development trend of MEC offloading technology in 5G environment, the IIoT-MEC network deployment architecture supporting 5G is introduced. On this basis, a lightweight task offloading strategy based on deep reinforcement learning and a MEC offloading strategy based on D2D collaboration are analyzed. The core challenges of offloading decision-making, interfere with the management, mobility management are summarized, which are currently faced by computation offloading technologies in MEC.

Key words: Mobile Edge Computing（MEC）, computation offloading, MEC deployment scenario, offloading schemes, 5G

摘要：

移动边缘计算（Mobile Edge Computing，MEC）中的计算卸载技术通过将终端设备的计算任务卸载到网络边缘处，以解决云计算中心时延长、能耗大和负载高等问题。介绍了MEC的概念、目前主流的MEC网络架构和部署方案。从卸载决策方面对MEC环境下计算密集型应用的卸载技术进行了详细研究，从最小化时延、最小化能耗、权衡时延和能耗及最大化收益为优化目标的4种计算卸载方案进行了分析和对比，并总结出各自的关键研究点。通过分析5G环境下MEC卸载技术的发展趋势，介绍了支持5G的IIoT-MEC网络部署架构，在此基础上分析了基于深度强化学习的轻量级任务卸载策略和基于D2D协作的MEC卸载策略。总结和归纳了目前MEC中计算卸载技术所面临的卸载决策、干扰管理、移动性管理等方面的核心挑战。

关键词: 移动边缘计算, 计算卸载, MEC部署方案, 卸载策略, 5G