分布式认知网络多域并行入侵实时预警仿真

Simulation of Real-Time Early Warning of Multi-Domain Parallel Invasion in Distributed Cognitive Network

对分布式认知网络中的多域并行入侵进行预警在提高分布式认知网络安全性能方面具有重要意义。由于多域并行入侵者的攻击行为具有较强的随机性,采用当前入侵预警方法预测入侵意图难,无法阻止分布式认知网络攻击行为,存在预警反应时间长,入侵攻击对抗性差等问题,提出一种基于三层攻击图的分布式认知网络多域并行入侵实时预警方法。通过对分布式认知网络多域并行入侵者攻入主机后的攻击数据进行分析,建立了分布式认知网络三层攻击图,通过对多域并行入侵意图的概率分析来定量攻击图,采用隐马尔科夫模型设计分布式认知网络多域并行入侵攻击行为预测模型,以多域并行攻击行为预测模型为核心,构建主动入侵实时预警策略。实验结果表明,该方法能够迅速发现一些未知入侵攻击,入侵攻击对抗性强。

A real - time prediction method for multi - domain parallel intrusion in distributed cognitive network based on three - layer attack graph is presented. Through analyzing the attack data that multi - domain parallel invad- ers in distributed cognitive network attacked host computer, we establish three - layer attack graph in distributed cog- nitive network. Through the probabilistic analysis of multi - domain parallel intrusion intention, the attack graph is quantified. The hidden Markov model is used to design prediction model of multi - domain parallel intrusion attack in distributed cognitive network. Taking prediction model of multi - domain parallel attack behavior as core, we present the early warning strategy of active intrusion. Simulation results show that this method can quickly find some unknown intrusion attacks.