复杂网络的攻击策略研究

发布时间：2018-05-08 14:59

本文选题：复杂网络 + 攻击策略　；参考：《青岛理工大学》2015年硕士论文

【摘要】：随着科技的发展,我们的生活中到处充满着复杂系统,复杂网络逐渐成为科学家们研究的热点。复杂网络给我们的生活带来了很大的便利,但是复杂网络无时无刻不受到各种攻击,包括随机攻击和恶意攻击。因此,复杂网络的抗毁性研究变得尤为重要,其中复杂网攻击策略是抗毁性研究的关键。本文首先分析了网络的抗毁性和经典攻击策略的攻击效率,结合网络的抗毁性和拓扑结构特性,提出了联合型攻击策略,包括IDB策略和RDB策略。并提出了基于熵的攻击策略,包括IE(Information entropy)、CE(Cross entropy)和LE(Local entropy)。最后,结合攻击效率和攻击代价来衡量新提出的攻击策略的攻击效果。本文主要研究工作如下:(1)网络模型的建立与分析。为了全面分析不同网络模型的抗毁性,本文选取了三种标准的网络模型(ER随机网络、WS小世界网络和BA无标度网络)和三种现实网络模型。(2)网络抗毁性分析。指数网络都具有很强的鲁棒性,在静态攻击下至少移除80%的节点网络才濒临崩溃,即使在最有效的攻击策略RBC的攻击下也要移除40%的节点。BA无标度网络在受到攻击时比较脆弱,静态攻击下只需移除40%的节点网络就陷入瘫痪,而在动态攻击下仅仅移除20%的节点网络就瘫痪。三种real-world网络无论面临随机攻击还是恶意的攻击都非常易毁,这是由于real-world网络不规则,当网络中少数的hub节点受到攻击时网络的拓扑结构就遭到严重破坏。(3)攻击策略的攻击效果分析。本文以首先分析了经典攻击策略的攻击效率,以基于DC(Degree centrality)和BC(Benweenness centrality)的攻击策略为参考衡量新提出的攻击策略的攻击效率。在WS小世界网络中,IDB的攻击效率与IDC相比提高了20%,RDB较RDC提高了30%;在The power-grid网络中,IDB的攻击效果是最好的,攻击效率远大于其它攻击方式。在ER随机网络和BA无标度网络中,ICE的攻击效果是最好的;其中,BA无标度网络中,ICE策略的攻击效率与IBC相比提高了25%。总之,本文所提出的攻击策略IDB、RDB和基于熵的攻击策略,在不同的网络模型中与经典攻击策略相比具有很大优势。
[Abstract]:With the development of science and technology, our life is full of complex systems. Complex networks bring great convenience to our life, but complex networks are always subject to various attacks, including random attacks and malicious attacks. Therefore, the research on the invulnerability of complex networks becomes more and more important, and the attack strategy of complex networks is the key to the research of survivability. In this paper, firstly, the invulnerability of the network and the attack efficiency of the classical attack strategy are analyzed. Combined with the invulnerability and topology characteristics of the network, a joint attack strategy, including IDB strategy and RDB strategy, is proposed. An Entropy based attack strategy is proposed, including IE(Information entropyr cross entropyy and LE(Local entropyy. Finally, the attack effect of the proposed attack strategy is evaluated by combining the attack efficiency and attack cost. The main work of this paper is as follows: 1) Establishment and analysis of network model. In order to analyze the invulnerability of different network models, three kinds of standard network models, ER random network / WS small-world network and BA scale-free network, and three real network models, respectively, are selected in this paper. Exponential networks have strong robustness, at least 80% of the nodes removed under static attack are on the verge of collapse. Even under the most effective attack strategy RBC attack, 40% of the node. BA scale-free network is vulnerable when it is attacked. Under static attack, only 40% of the node network is paralyzed, while only 20% of the node network is paralyzed under dynamic attack. The three kinds of real-world networks are vulnerable to random or malicious attacks, because real-world networks are irregular. When a few hub nodes in the network are attacked, the topology of the network is severely damaged. In this paper, the attack efficiency of the classical attack strategy is analyzed firstly, and the attack efficiency of the new attack strategy is evaluated by reference of the attack strategy based on DC(Degree centrality and BC(Benweenness centrality. Compared with IDC, the attack efficiency of IDC in WS small world network is 30% higher than that of RDC. In The power-grid network, the attack efficiency of IDB is the best, and the attack efficiency is much higher than that of other attack methods. In ER random network and BA scale-free network, the attack effect of ICE is the best, and the attack efficiency of ICE strategy in BA scale-free network is 25% higher than that of IBC. In a word, the attack strategy proposed in this paper is IDBN RDB and Entropy based attack strategy, which has a great advantage over the classical attack strategy in different network models.
【学位授予单位】：青岛理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.08

【参考文献】