摘要
边缘计算的提出,为减少骨干网络的数据流量、优化物联网应用的响应延迟提供了可能。与云相比,边缘节点的计算和存储能力十分有限。为了打破边缘节点的硬件资源瓶颈,缩短应用程序完成时间,文中提出了一种基于多边缘节点协作的新型应用卸载算法(MNC)。该算法采用细粒度级别的应用程序分区方法,分析方法级组件间的依赖关系,利用网络中的可用边缘节点并行处理应用程序组件。在真实的物联网场景中,对所提出的算法与MCOP分区卸载算法、云端执行分别进行了对比,结果表明在边缘节点数量适宜时,MNC算法下应用程序总完成时间更少,能够满足大部分物联网场景的时延要求。
The proposal of edge computing provides a possibility to reduce the data traffic of the backbone network and optimize the response delay of Internet of Things ( IoT) applications.The computing power and the storage capacity of edge nodes are very weak compared with the cloud.To break the hardware resource bottleneck of the edge nodes and reduce the application completion time,a new application offloading algorithm based on multi-edge node collaboration is proposed.The algorithm uses a fine-grained application partitioning method to analyze the dependencies between method-level components and utilize the available edge nodes in the network to parallelly process application components.In the real IoT scenario, the proposed algorithm is compared with the MCOP partition offloading algorithm and the cloud execution, respectively.The results show that when the number of edge nodes is appropriate,the total execution time of the application on the MNC algorithm is less,meeting the delay requirements of most IoT scenarios.
作者
吴正坤
江凌云
穆晏如
WU Zhengkun;JIANG Lingyun;MU Yanru(College of Telecommunications & Information Engineering,Nanjing University of Posts and Telecommunication,Nanjing 210003,China)
出处
《南京邮电大学学报（自然科学版）》
北大核心
2019年第4期96-102,共7页
Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition
基金
国家自然科学基金(6127123)资助项目
关键词
物联网
边缘计算
多网关协同
应用卸载
Internet of Things( IoT)
edge computing
multi-gateway collaboration
application offloading
