面向移动边缘计算中多应用服务的虚拟机部署算法

Virtual Machine Placement Algorithm for Supporting Multiple Applications to Mobile Edge Computing

移动边缘计算(MEC)通过在用户近端以虚拟机(VM)形式部署应用服务,能有效降低服务响应延迟并减少核心网络数据流量。然而,当前MEC中虚拟机部署的大多数研究尚未具体考虑用户对多种应用服务的需求。因此,该文针对MEC中多应用服务的虚拟机部署问题,提出两种启发式算法,即基于适应度的启发式部署算法(FHPA)和基于分治的启发式部署算法(DCBHPA),通过在边缘网络中配置支持多种应用服务的虚拟机来最大限度地减少网络中的数据流量。FHPA和DCBHPA分别基于边缘服务器的网络连接特征和用户对应用请求的差异性,定义了不同的适应度计算模型。在此基础上,通过子问题划分机制实现VM配置。仿真结果表明,相比于基准算法,所提算法能更好地控制系统数据流量,有效地提高边缘网络服务资源的利用率。

In Mobile Edge Computing(MEC)environment,deploying application services in the form of Virtual Machines(VM)at the edge of the network can effectively reduce the service response delay and reduce the data traffic of core network.There have been many solutions to the problem of optimal allocation of edge network resources,but few studies consider the optimal deployment of VM that provide users with multiple application services to mobile edge networks.To this end,two heuristic algorithms are proposed,Fitness-based Heuristic Placement Algorithm(FHPA)and Divide-and-Conquer Based Heuristic Placement Algorithm(DCBHPA).By distributing VMs that support multiple application services to the MEC network,these two algorithms aim to minimize the data traffic in MEC architecture.Besides,FHPA and DCBHPA define respectively different fitness computing models based on the network connection characteristics of edge servers,as well as the differences in users’application requests.Thus,VM configuration can be realized through the sub-problem division mechanism.Compared with the baseline algorithms,the simulation results show that the proposed algorithms can better control the system data traffic and improve effectively the utility of edge network service resources.