基于变分图自动编码器与K均值聚类的虚拟网络嵌入算法应用
Application of Virtual Network Embedding Algorithm Based on Variational Graph Autoencoder and K-means Clustering
姚丽敏1
作者信息
摘要
将虚拟网络映射到物理网络是网络功能虚拟化中一项重要的任务.为了有效地分配虚拟网络请求,需要将虚拟网络嵌入到物理网络拓扑中.然而,由于虚拟网络的复杂性和物理网络的限制,这一任务变得非常具有挑战性.鉴于此,研究在现有虚拟网络嵌入算法(Virtual Network Embedding,VNE)模型基础上进行改进,融入了变分图自动编码器(Variational Graph Auto-Encoders,VGAE),提出了一种新型虚拟网络嵌入算法模型.通过编码器对虚拟网络的嵌入特征进行提取,随后利用K-means聚类算法对所得到的嵌入特征进行分类,最终得到合适的嵌入分配方法.实验结果表明,该新模型相较于其他同类型的嵌入算法性能表现最佳,稳定性最好,其平均嵌入请求接受率为60％,长期平均CPU资源利用率最高达97％.综上所述,研究提出的新型虚拟网络嵌入算法在资源利用率和嵌入质量方面表现出色,能够有效应对复杂的网络环境和大规模的虚拟网络请求.
Abstract
Mapping virtual networks to physical networks is an important task in network functional virtualization.In order to effectively allocate virtual network requests,it is necessary to embed the virtual network into the physical network topology.However,due to the complexity of virtual networks and the limitations of physical networks,this task has become very challenging.In view of this,the study improves the existing Virtual Network Embedding(VNE)model by incorporating Variational Graph Auto Encoders(VGAE),and a new virtual network embedding algorithm model is proposed.The embedded features of the virtual network is extracted through an encoder.And then used the K-means clustering algorithm,the obtained embedded features is classfied,ultimately a suitable embedding allocation method is obtained.The experimental results show that the new model performs the best and has the best stability compared to other embedding algorithms of the same type.Its average embedding request acceptance rate is 60％,and the long-term average CPU resource utilization rate is as high as 97％.In summary,the new virtual network embedding algorithm proposed in the study performs well in resource utilization and embedding quality,and can effectively cope with complex network environments and large-scale virtual network requests.
关键词
虚拟网络/变分图自动编码器/K-means/嵌入算法/特征分配Key words
Virtual network/Variational diagram autoencoder/K-means/Embedding algorithm/Feature allocation引用本文复制引用
基金项目
2021年度滁州城市职业学院专项重点项目(2021bmzx01)
出版年
2024
NETL
NSTL
万方数据