基于拓扑感知节点排序的虚拟网络嵌入

Virtual Network Embedding Based on Topology-aware Node Sorting

摘要:

在共享底层上嵌入多个虚拟网络(VN)是云计算平台和大规模可切片网络测试平台的一个挑战性问题。本文利用马尔可夫随机游走模型,根据网络节点的资源和拓扑属性对其进行排序,这种新的拓扑感知节点排序方法可反映节点的相对重要性。利用节点排序设计了两种VN嵌入算法：RW-MaxMatch和RW-BFS。仿真实验表明：与现有的嵌入算法相比,拓扑感知节点排序具有较好的资源度量,并且所提出的基于RW的算法增加了长期平均收益和接受率。

Abstract:

Embedding multiple virtual networks (VNs) on the shared bottom is a challenge for cloud computing platforms and large-scale sliceable network testing platforms. In this paper, Markov random walk model is used to rank nodes according to their resources and topological attributes. This new topology-aware node ranking method can reflect the relative importance of nodes. Two VN embedding algorithms, RW-MaxMatch and RW-BFS, are designed by using node sorting. The simulation results show that compared with the existing embedding algorithms, the ranking of topology-aware nodes has better resource metrics, and the proposed RW-based algorithm increases the long-term average revenue and acceptance rate.