摘要
提出一种改进证据理论(dempster-shafer evidence theory,DSET)的物联网智能终端网络安全评估方法.通过k-means方法得到不同识别框架的聚类中心,作为数据驱动曲线的中线值,改进了数据驱动法的信息源基本概率分配(basic probability allocation,BPA),特别是避免了焦元支持率为1 的冲突证据生成.进一步为了分配证据间的冲突,基于所有证据源计算关键证据,引入相似度权重冲突分配规则进行融合,从而得到评估结果.为了验证本文提出的方法,搭建视频终端实验网络,采用设备应用控制器(device application controller,DAC)设备探查设备运行资产,以视频终端网络要素层安全评估为例,比较了不同方法得到的要素层运行资产BPA在融合评估中的结果,以及针对冲突证据融合时不同融合规则得到的融合结果.实验结果表明,相比其他改进方法,本文提出的改进DSET方法能在避免生成冲突极大的证据同时合理分配冲突,并且保证了较高的物联网终端网络安全评估正确率.
Abstract
An improved DSET(dempster-shafer evidence theory)network security assessment method of Internet of Things intelligent terminal is proposed in this paper.The cluster centers of different frames of discernment are obtained through the k-means method as the median value of the data-driven curve,which improves the BPA(basic probability allocation)accuracy of the data-driven method,especially to avoid the generation of conflict evidence with a focal element support rate of 1.To distribute the conflict between evidences,key evidence is calculated based on all evidences'sources,and similarity weight conflict allocation rules are addressed.In order to verify the method proposed in this paper,a video terminal network is built,and DAC(device application controller)equipment is used to explore equipment operating assets information.Taking the element layer security assessment of video terminal network as an example,fusion results of the elements BPA obtained by different methods and the fusion results of fusing conflict evidences obtained by different fusion rules were compared.The experimental results show that the improved DSET method proposed can greatly avoid generating conflict evidences and distribute conflict reasonably,thus ensuring a high accuracy rate of Internet of things terminal network security assessment.
基金项目
上海市"科技创新行动计划"高新技术领域项目(18511105800)
国家电网科技项目(H2020-179)
维普
万方数据