基于粒子滤波的网络安全态势预测方法研究.pdf

第 34 卷 第 1 期 计算机应用与软件 VoL 34 No . 1 2017年 1 月 Computer Applications and Software Jan . 2017 基于粒子滤波的网络安全态势预测方法研究 汤 永 利 李 伟 杰 于 金 霞 闫 玺 玺 (河南理工大学计算机科学与技术学院河南焦作454000) 摘 要 为了使网络管理由被动防御转为主动防御，需对网络安全态势进行预测。针对已有态势预测方法存 在的准确性不高、需要样本训练问题，提出基于粒子滤波的网络安全态势预测方法。该方法利用带权粒子集逼近 系统的后验概率密度函数，通过重要性采样、权值更新、状态估计等近似积分操作来实现非线性状态预测。实验 结果表明，该方法不仅能体现网络安全状态的非线性，也预测了网络安全态势值。对比其他态势预测模型，该方 法准确性更高，且适应于复杂网络环境。 关键词 网 络 安 全 态 势 预 测 粒 子 滤 波 中图分类号 TP393.08 文献标识码 A D O I 10. 3969/j. issn. 1000-386x. 2017.01.053： RESEARCH ON A PREDICTION METHOD OF NETWORK SECURITY SITUATION BASED ON PARTICLE FILTER Tang Yongli Li Weijie Y u Jinxia Yan Xixi {School o f Computer Science and Technology,Henan Polytechnic University ,Jiaozuo 454000,Henan,China) Abstract In order to change the network management from passive defense to active defense,it is necessary to predict network security situation. To solve the problems of the low accuracy and samples training in existing forecast methods, the prediction method of network security situation based on particle filter is proposed. This method uses the weighted particles to approximate the posteriori probability density and realizes nonlinear situation prediction by the approximate integral operations include importance sampling, weight updating, state estimation. Experimental results show that this method can not only reflect the nonlinear of network security situation, but also forecast the value of network security situation. Comparing with the other situation forecasting models,this method is more accurate and can be adapted to the complex network environment. Keywords Network security Situation prediction Pa