| 基于SDN和NFV融合的网络切片资源分配优化算法 |
| Optimization algorithm for network slice resource allocation based on SDN and NFV |
| |
| DOI： |
| 中文关键词: 软件定义网络；网络功能虚拟化；5G；网络切片；对偶分解 |
| 英文关键词:software defined network (SDN); network function virtualization (NFV); 5G; network slicing; dual decomposition |
| 基金项目: |
| 作者 | 单位 | | 朱晓荣 | 南京邮电大学 通信与信息工程学院，江苏 南京 210003 | | 高 健 | 南京邮电大学 通信与信息工程学院，江苏 南京 210003 |
|
| 摘要点击次数: 2682 |
| 全文下载次数: 2010 |
| 中文摘要: |
| 软件定义网络(Software Defined Network,SDN)和网络功能虚拟化(Network Function Virtualization,NFV)已经成为新一代网络体系结构的新范式。SDN和NFV可以有效地提高部署和管理服务功能链(Service Function Chains，SFCs)的灵活性。文中融合SDN和NFV，并应用到网络切片的资源优化中，为未来的5G三大业务场景实现定制服务的有效部署，提出了一种基于SDN和NFV的网络切片资源优化算法。首先针对5G uRLLC，eMBB，mMTC三大应用场景的不同业务需求，将底层的物理节点按照功能类型划分为三个虚拟子网的节点集合。然后根据不同应用场景的不同服务功能链的要求，分别对每一类的网络切片进行建模，形成混合整数线性规划的基于SDN和NFV的网络切片的数学模型，并提出了基于拉格朗日对偶分解的算法，将网络切片的数学模型转换为节点和链路的子问题，再对分解的子问题进行映射方案的求解。仿真结果表明，与以往算法相比，文中提出的算法在资源利用率、接受率、平均执行时间方面具有更好性能。 |
| 英文摘要: |
| Software defined network (SDN) and network function virtualization (NFV) have become a new paradigm for the new generation of network architecture. SDN and NFV can effectively improve the deployment and the management of service function chains(SFCs).The integration of SDN and NFV is applied to the resource optimization of network slicing, to realize the effective deployment of customized services for the three major business scenarios of 5G in the future. a network slicing resource optimization algorithm is proposed based on SDN and NFV. Firstly, according to the different business requirements of the 5G uRLLC, eMBB and mMTC, the underlying physical nodes are divided into three virtual subnet node sets according to function types. Then, according to the requirements of different service function chains in different application scenarios, a mixed integer linear programming mathematical model for each type of network slice is formed based on SDN and NFV. Finally, an algorithm based on Lagrangian dual decomposition is used to convert the mathematical model of network slicing into sub-problems of nodes and links. The mapping scheme for the decomposed sub-problems is solved. Compared with previous algorithms, simulation results show that the algorithm has better performance on resource utilization, acceptance rate and average execution time. |
| 查看全文 查看/发表评论 下载PDF阅读器 |