Abstract: Epidemic attacks such as worms and viruses in email networks are serious network security problems. Designing effective immunization is a very important way to address these problems. The Node Betweenness（NB） strategy is the state of the art. An immunization strategy is designed based on graph partitioning methods. In all of the simulations in synthetic networks and the Enron email networks, the new immunization strategy is consistently better than the NB strategy. In some simulation scenarios, there is almost a 50% improvement than the NB strategy.

Key words: epidemic attacks, network immunization methods, graph partitioning, email networks

摘要： 蠕虫、病毒等在电子邮件网络中的传播型攻击是非常严重的网络安全问题, 设计有效的网络免疫方法对于解决这个问题是非常重要的。基于节点介数（Node Betweenness，NB）的方法是目前最好的方法。利用图分割（graph partitioning）方法设计了一种网络免疫方法, 在生成的电子邮件网络拓扑模型以及Enron电子邮件网络真实拓扑数据上的仿真表明，所设计的方法比NB方法更有效。在某些仿真场景下，所设计的免疫方法能够比NB方法达到50%的改进。

关键词: 传播型攻击, 网络免疫方法, 图分割, 电子邮件网络