全局分配策略在级联故障中的建模与研究

Modeling and Research of Global Allocation Policy in Cascading Failures

现实情况中,遭到攻击的网络由于结构和功能发生改变,网络全局负载随之发生变化。基于此,文中级联故障模型舍弃传统的基于节点邻域或局域的负载局部分配策略,采用全局分配策略,定义正常节点的分配负载与距故障节点的最短路径长度成反比关系,采用"步长"刻画故障扩散快慢。仿真无标度网络、小世界网络和随机网络,网络节点的平均故障结果表明无标度网络的抵制故障能力较强;在全局分配策略较弱时,不同网络在同一容忍系数下取到不同步长最值;网络冗余度较低时,后续故障节点更倾向选择上一迭代中故障节点的非邻居节点。

In reality, the global loads of network usually alter when the structure and function of network change by a list of attacks. Based on this, the paper defines extra loads of the normal node inversely proportional to the distance of failure node and presents a new indicator named ＂diffusion step＂ to failure diffusion process, which uses the global allocation policy instead of neighbor or local feature of node. By simulating scale-free network, small-world network and random network, the average results of network show that the scale-free network has strong ability to resist failures. When the influence of global allocation policy is weak, the different networks achieve respective maximum of diffusion steps in the same tolerance coefficient. The following failed nodes prefer to select the non-neighbor nodes of the previous iteration of the failed nodes when the network redundancy is low.