VEC中基于计算资源动态变化的服务迁移策略  

Service Migration Strategy Based on Dynamic Changes in Computing Resources in VEC

作　　者：范艳芳[1] 宋志文 蔡英[1] 陈若愚[1] FAN Yan-fang;SONG Zhi-wen;CAI Ying;CHEN Ruo-yu(Beijing Information Science&Technology University,Beijing 100101,China)

机构地区：[1]北京信息科技大学,北京100101

出　　处：《计算机仿真》2025年第2期134-139,共6页Computer Simulation

基　　金：国家自然科学基金(61672106);北京市自然科学基金(L192023);促进高校内涵发展一面向边缘计算的创新科研平台建设项目(2020KYNH105);北京信息科技大学‘勤信人才’培育计划(QXTCP C202111)。

摘　　要：在车载边缘计算中,通过虚拟化技术将应用程序及其依赖项封装为服务实体,并且随着车辆的移动而不断迁移服务实体,极大提升了车辆的服务质量。然而,如果迁移到计算资源占用率过高的边缘服务器,就出现因服务需求得不到满足,而导致服务质量下降和系统总开销上升的情况。因此,将边缘服务器的计算资源占用率和车辆的计算资源需求纳入迁移策略,以最小化系统总开销。并提出了基于计算资源动态变化的服务迁移策略。将服务迁移问题描述为一个马尔可夫决策过程,并提出了一种基于深度强化学习的迁移算法进行求解。仿真结果表明,所提策略可以有效降低计算开销来提升服务质量。与其它策略相比,系统总开销减少了10%以上。In Vehicular Edge Computing,the application program and its dependencies are encapsulated as service entities by virtualization technology,and the service entities are constantly migrated with the movement of vehicles,which greatly improves the quality of service.However,if the service entity is migrated to an edge server with high computing resource usage,the quality of service deteriorates and the total system cost increases due to unmet service requirements.Therefore,this paper takes the computing resource usage rate of the edge server and the computing resource demand of the vehicle into the migration strategy to minimize the total system cost.A service migration strategy based on dynamic change of computing resources is proposed.This paper describes the service migration problem as a Markov decision process and proposes a migration algorithm based on deep reinforcement learning to solve it.The simulation results show that the proposed strategy can effectively reduce the computing cost and improve the service quality.The total system cost is reduced by more than 10%compared to other strategies.

关 键 词：车载边缘计算 服务迁移 深度强化学习 

分 类 号：TN929.5[电子电信—通信与信息系统]