基于隐马尔可夫的网络实时风险评估  

Network real-time risk assessment based on hidden Markov models

作　　者：马煜[1] 刘建华[1] 尚星[2] 田东平[1] 

机构地区：[1]西安邮电学院信息安全研究中心,陕西西安710061 [2]西安电子科技大学通信工程学院,陕西西安710071

出　　处：《计算机工程与设计》2009年第11期2656-2659,共4页Computer Engineering and Design

基　　金：2007年度通信软科学计划基金项目(信息产业部2007R100)

摘　　要：在基于隐马尔可夫模型的网络安全实时风险评估中,状态转移概率矩阵的确定是关键一步,目前基本上都是依据经验给出,具有很大主观性,不能客观反映网络安全的风险状况。为此,引入了攻击难度系数的概念,通过对数据集的统计学习,给出了状态转移概率矩阵。此外,通过对威胁进行分类,根据各类威胁的影响,给出了相应的权重。实验结果表明,该方法使得网络安全实时风险评估更加客观,为网络安全的风险管理提供了决策支持。In the field of network risk assessment based on hidden Markov models, the state transition matrix are usually derived from expert experience. It leads to much subjectives to the result ofthe assessment which can not objectively reflect the real risk ofthe network. Therefore, a conception of attack difficulty coefficient is introduced. Through analyzing and statistical learning the dataset, the state transition matrix is derived. Taxonomy of those threats are done in the dataset, then weights all of them according to their influence. The experiment indicates that the result of the assessment using the method proposed is much more objective while providing better support for network risk management.

关 键 词：网络安全风险评估 网络攻击 实时风险 威胁分类 隐马尔可夫模型 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]