基于K-shell及动力学模型的SPN关键节点识别算法  

Identification algorithm of key nodes in SPN based on K-shell and dynamical model

作　　者：王云花 邓翠艳 姚旭清 WANG Yunhua;DENG Cuiyan;YAO Xuqing(Jinzhong University,Shanxi 030619,China;China mobile communications Group Shanxi Co,Ltd,Shanxi 030032,China)

机构地区：[1]晋中信息学院,山西晋中030800 [2]中国移动通信集团山西有限公司,山西太原030032

出　　处：《通信与信息技术》2024年第6期48-53,共6页Communication & Information Technology

基　　金：2022年度山西省高等学校科技创新项目“无线传感器网络节点定位方法的研究”(项目编号:2022L663)。

摘　　要：SPN网络采用切片分组组网策略进行网络传输,由于其高效的业务承载能力,SPN网络已成为各种切片业务综合性传输承载网络。在通信网络中,SPN网络具备小世界复杂网络特性,针对该复杂网络特性提出了一种基于K-shell及动力学模型的SPN关键节点识别算法。本文针对通信运营商SPN网络拓扑特性及动力学特征,通过构建复杂网络模型,抽取网络拓扑及动力学关键特征实现SPN网络关键节点识别。首先构建了SPN网络的拓扑图,基于最短路径算法构建网络故障传播模型。其次运用复杂网络分析方法,从网络拓扑及故障模型中提取能够表征节点重要性的特征([CB]、[CD]、[CC]、SoCF、FoP),并通过计算节点的故障信息与拓扑信息特征之间的Kendall相关系数,量化网络级联故障与拓扑结构之间的相关性。最后在相关性分析的基础上,使用HDBSCAN聚类算法对两类特征进行聚类分析。实验结果表明,在SPN网络关键节点识别中,可以采用动力学特征(SoCF和[CC])或(FoP和[CB])对网络关键节点进行有效识别。The SPN network adopts the slicing and grouping networking strategy for transmission.Due to its efficient network ca-pacity,the SPN network has become a comprehensive transmission network for various slicing services.Considering the SPN network has the small-world characteristics in the communication network.in response to this,an algorithm for identifying key nodes in the SPN based on the K-shell and dynamical model algorithm is proposed.Based on the structure of the SPN network,the goal is con-structing a complex network model to extract key topological and dynamical features for characterizing whether a network node is a key node.Firstly,this paper constructs the topology graph of the SPN network and builds a network fault propagation model based on the shortest path algorithm.Secondly,using complex network analysis methods,features that can characterize the importance of nodes(CB、CD、CC、SoCF、FoP)are extracted from the network topology and fault model,and the Kendall correlation coefficient between the fault information and topological information features of the nodes is calculated to quantify the correlation between network cascading failures and topological structures.Finally,based on the correlation analysis,the HDBSCAN clustering algorithm is used for cluster analysis of two types of features.The experimental results show that in the identification of key nodes in the SPN network,dynamical features(SoCF and CC)and(FoP and CB)can be used to identify key network nodes.

关 键 词：SPN网络 复杂网络 K-SHELL 关键特征提取 Kendall系数 HDBSCAN 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]