摘要
移动边缘计算(Mobile Edge Computing,MEC)将云计算服务从云端下沉到网络边缘,能够满足第五代移动通信(5G)新业务超低时延的需求,被认为是未来5G通信的关键技术之一。针对5G应用场景下的多用户多边缘服务器的MEC网络架构和系统模型,设计了一种基于强化学习的多节点计算资源分配方案。该方案中用户设备将计算任务发送到家庭基站,家庭基站基于过往经验独立决策选择哪个MEC节点上传卸载的计算任务,在不引入额外信令开销的同时,实现了MEC服务器端的负载均衡。仿真实验表明,提出的方法在负载均衡上优于其他对比方法,尤其是在高任务强度时。
MEC(Mobile Edge Computing)pushes cloud service from central cloud to network edges,satisfying low latency requirements of new applications for the fifth generation mobile communication(5G).It is considered as one of the critical technologies of future 5G communication.According to the characteristics of MEC network,this paper designs a new MEC-double layer network architecture and describes the MEC model consisting of multiple users and multiple edge servers under this network architecture.Then a multinode computing resource allocation algorithm based on reinforcement learning is proposed.User equipment sends computing tasks to the femto basestation.Femto basestation using this algorithm makes independent decision based on past experience to choose a MEC node to upload the computing task with the intention of achieving load balancing among the MEC servers.No additional signaling overhead is introduced.Finally,simulation experiments indicate that the proposed method is superior to other comparison methods in load balancing,especially when task intensity is high.
作者
余萌迪
唐俊华
李建华
YU Meng-di;TANG Jun-hua;LI Jian-hua(School of Cyber Security,Shanghai Jiaotong University,Shanghai 200240,China;Shanghai Key Laboratory of Integrated Administration Technologies for Information Security,Shanghai 200240,China)
出处
《通信技术》
2019年第12期2920-2925,共6页
Communications Technology
基金
国家自然科学基金重点项目（No.61831007,No.61431008)~~
关键词
移动边缘计算
MEC
计算资源分配
负载均衡
mobile edge computing
MEC
computing resource allocation
load balancing
