基于节点重要度的复杂网络可靠性研究

Reliability of complex network based on importance of nodes

为研究复杂网络的节点重要度及其可靠性,在随机网络、小世界网络和无标度网络模型下,理论分析了度值中心性、半局部中心性、介数中心性和PageRank算法在节点重要度评估上的准确性。通过仿真模拟方法,对网络理论模型和实际复杂网络分别进行研究,采用逐步移除节点和重复计算的方式,考察网络最大连通子图和节点移除比例的关系,同时对数值仿真结果进行合理分析。研究结果表明,基于网络全局信息的介数中心性和PageRank算法对节点重要度评估的效果更好,同时说明网络的可靠性与网络拓扑结构有关,实际复杂网络一般同时具备多种拓扑特性。

To evaluate the significance of nodes and reliability of the complex network, considering the degree centrality, semi-local centrality, betweenness centrality and PageRank algorithm in random network, small-world network and scale-free network, this paper used the simulation to remove gradually the nodes in the theoretical network models and real complex networks, and analysed the relation between the largest connected component and the proportion of the removed nodes. The results show that the betweenness centrality and PageRank algorithm which are based on the whole situation are more effective. And the reliability of network is related to its topology structure, the real complex networks contain many topology features.