【作者】 谭谨；

【导师】 张端明；

【作者基本信息】 华中科技大学 ， 凝聚态物理， 2007， 硕士

【摘要】 鲁棒性研究是复杂网络中的一个重要的研究领域。近年来,网络的攻击与出错的研究得到了一个普遍的结论:即无标度网络对随机故障具有较高的容错性,但对于蓄意攻击非常脆弱。以往研究中的攻击方式都是完全摧毁所攻击的目标,然而在现实网络中,攻击往往不能完全摧毁网络的节点或者边,通常只是造成节点或者边的部分失效。本文针对这种情况提出了一个普遍的攻击策略模型,即一般性打击策略。在这个模型中,利用对边的加权造成网络效率的降低,攻击的强度大小由一个连续加权系数来表示,其范围为1到无穷大。1表示无攻击,无穷大等同于删去点或者边。本文利用计算机模拟研究了应用该打击策略,在连续性最大化伤害方式下,网络效率的演化规律。研究结果表明:对点的一般性打击,无论在无标度网络或随机网络上,当加权系数很小时,攻击两点就可以达到删除单点造成的网络效率伤害;对边的一般性打击下,无标度网络比随机网络更脆弱。本文所提出的一般性打击对网络造成的伤害主要用网络效率来衡量,网络传输效率是一些网络的重点关注对象,如英特网、万维网等现实网络,所以本文的工作有助于帮助人们维护网络效率,预测网络在遭受一般性攻击时,将可能对网络效率造成的最大伤害,并有助于在网络的维护中及时发现出现小故障却造成大问题的节点或边。更多还原

【Abstract】 Robustness is an important research field of complex networks. Recently, error and attack tolerance of complex network research generally concluded that scale-free networks display a topological robustness against random node failures but fragile to intentional attacks. The most popular attack method is complete destruction of a selected target, while in the fact, the most part of attack can not achieve so ideal effect but cause partial damage of nodes or edges.We established a generic hit strategy model to represent the situation above. In the model, we add the edge’s weight to attenuate network efficiency, and defined a continuous weight coefficientωto represent attack intensity, andωincrease continuously from 1 to infinity where 1 represents no attack and infinity represents complete destructive attack.We numerically studied the model under successive maximal damage method. The result shows that: with generic hit on the node, either on scale-free networks or random networks, when weight coefficient is very small, hit on two nodes could achieve the same damage with the elimination of a single selected node; while with generic hit on the edge, scale-free networks is much more frangible than random network.The damage causing by the generic hit was monitored by the network efficiency and the most attention part of network usually is the network transmission efficiency, such as Internet, WWW etc. Our studies are be propitious to help the maintenance of network, moreover, we can predict that the most potential and largest damage causing by the generic hit. Rely on this ,we can find out which node or edge having partial malfunction but causing a huge damage to the whole network.更多还原