基于线性优化的移动边缘计算中任务迁移算法的设计与实现  

Design and Implementation of Task Offloading Algorithm in Mobile Edge Com⁃puting Based on Linear Optimization

作　　者：沈欣 叶宇皓 司迎利 SHEN Xin;YE Yuhao;SI Yingli(The First Air Force Armament Department Representative's Office in Luoyang,Luoyang 471009,China;Xi’an Jiaotong University,Xi’an 710049,China;China Airborne Missile Academy,Luoyang 471009,China;National Key Laboratory of Air-based Information Perception and Fusion,Luoyang 471009,China)

机构地区：[1]空装驻洛阳地区第一军事代表室,河南洛阳471009 [2]西安交通大学,陕西西安710049 [3]中国空空导弹研究院,河南洛阳471009 [4]空基信息感知与融合全国重点实验室,河南洛阳471009

出　　处：《郑州航空工业管理学院学报》2024年第3期95-103,共9页Journal of Zhengzhou University of Aeronautics

基　　金：航空科学基金项目(2023Z004012001)。

摘　　要：移动边缘计算(MEC)是一种可以提高移动端计算速度和安全性的重要技术。用户将移动端任务卸载到附近边缘设备上,以达到减少移动端负载及计算耗能等的效果,如何设计移动端任务在边缘设备中的迁移算法是提高MEC效率的关键问题。文章提出了目前在边缘计算中两种比较普遍的任务迁移算法:整数线性规划(ILP)和贪心启发式算法。这两种迁移解决方案除考虑位置、带宽、用户、迁移等方面带来的能量损耗外,还考虑了由于车辆的机动性所带来的损耗,并使用马尔科夫链预测车辆位置,将位置预测和任务迁移相融合。实验结果表明,首先,ILP算法与贪心启发式算法的耗能近乎相同,但在同样耗能的情况下,贪心启发式算法的求解延迟远远小于ILP算法,从而确立了贪心算法在迁移方案中的优势;其次,通过任务刷新频率的选择,可进一步降低贪心启发式算法的耗能,优化后的算法比初始算法耗能降低了约20%。Mobile Edge Computing(MEC)is an important technology that can improve the computational speed and security of mobile devices.Users offload mobile tasks to nearby edge devices,reducing the load on the mobile devic⁃es and decreasing computational energy consumption.The task migration algorithm in edge devices is a key issue for enhancing task processing efficiency.This paper proposes two commonly used task migration algorithms in edge com⁃puting:Integer Linear Programming(ILP)and Greedy Heuristic algorithm.Both migration solutions consider location loss,bandwidth loss,user loss,and migration loss.Additionally,the solutions for task migration need to consider not only parameters related to energy consumption but also vehicle mobility.This paper utilizes a Markov chain to pre⁃dict vehicle locations and integrates location prediction with task migration.The experimental results show that ILP and Greedy Heuristic algorithm have nearly the same energy consumption.However,under the same energy consump⁃tion,the Greedy Heuristic algorithm exhibits significantly lower time delay compared to the ILP algorithm,establish⁃ing the advantage of the Greedy algorithm in the migration scheme.Furthermore,by selecting the task refresh fre⁃quency,the energy consumption of the Greedy Heuristic algorithm can be further reduced.The optimized algorithm reduces energy consumption by approximately 20%compared to the initial algorithm.

关 键 词：ILP 贪心算法 马尔可夫链 刷新频率 

分 类 号：TJ760[兵器科学与技术—武器系统与运用工程]