基于Q-learning算法的vEPC虚拟网络功能部署方法

发布时间：2018-02-11 09:23

本文关键词： G 虚拟网络功能服务功能链部署 Q-learning　出处：《通信学报》2017年08期 　论文类型：期刊论文

【摘要】：针对虚拟化演进分组核心网(v EPC)环境下,现有虚拟网络功能(VNF)部署方法无法在优化时延的同时保证服务链部署收益的问题,提出一种改进的基于Q-learning算法的v EPC虚拟网络功能部署方法。在传统0-1规划模型的基础上,采用马尔可夫决策过程建立了v EPC服务链部署的空间—时间优化模型,并设计了改进的Q-learning算法求解。该方法同时考虑了空间维度下的EPC服务链虚拟映射和时间维度下的VNF生命周期管理,实现了VNF部署的收益—时延多目标优化。仿真结果表明,与其他VNF部署方法相比,该方法在降低网络时延的同时提高了VNF部署的收益和请求接受率。
[Abstract]:Under the environment of virtualization evolution packet core network (VEPC), the existing methods of virtual network function deployment can not guarantee the benefit of service chain deployment while optimizing delay. In this paper, an improved method for deploying the functions of v EPC virtual network based on Q-learning algorithm is proposed. Based on the traditional 0-1 programming model, the space-time optimization model of v EPC service chain deployment is established by using Markov decision process. An improved Q-learning algorithm is designed, which considers the virtual mapping of EPC service chain in spatial dimension and VNF lifecycle management in time dimension, and realizes the multi-objective optimization of pay-delay in VNF deployment. Compared with other VNF deployment methods, this method not only reduces the network delay, but also improves the VNF deployment revenue and request acceptance rate.
【作者单位】：国家数字交换系统工程技术研究中心;移动互联网安全技术国家工程实验室;
【基金】：国家高技术研究发展计划(“863”计划)基金资助项目(No.2015AA01A706) 国家自然科学基金资助项目(No.61521003) 科技部支撑计划基金资助项目(No.2014BAH30B01)~~
【分类号】：TP393.0

【相似文献】