摘要
针对恶劣网络通信环境导致的数据采集和信息交互异常问题,提出了一种基于多维时序数据子序列的相关性检测方案.该方案针对相同时间段内的多维资源状态时序数据进行相关性计算.在获得其相关性矩阵后,先与已训练好的正常检测模式比对,若结果被判定为异常,再与已训练好的异常检测模式比对,若结果仍被判定为异常,则最终认定结果为异常.通过这种双重检测,尽可能避免对正常数据的误判.仿真结果显示:该方案在确保检测实时性的同时提高了检测的准确率.
To solve the problem of abnormal data in the process of data collection and information interaction caused by bad network communication environment, a correlation detection scheme based on sub-sequences of multidimensional temporal data was proposed. The scheme calculated the correlation of multi-dimensional resource state temporal data in the same period. After obtaining the correlation matrix, it was firstly compared with the trained normal detection mode. If the result was judged as the abnormal one, it was then compared with the trained abnormal detection mode. If the result was still judged as the abnormal one, it was finally identified as the abnormal one. Through this dual detection, as far as possible to avoid misinterpretation of normal data. Simulation results show that the proposed scheme improves the detection accuracy while ensuring real-time performance.
作者
刘尧
邓晓衡
桂劲松
刘斌
付琨
LIU Yao;DENG Xiaoheng;GUI Jinsong;LIU Bin;FU Kun(School of Computer Science and Engineering,Central South University,Changsha 410083,China;Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100094,China)
出处
《湖南科技大学学报（自然科学版）》
CAS
北大核心
2021年第3期88-96,共9页
Journal of Hunan University of Science And Technology：Natural Science Edition
基金
全军共用信息系统装备预研专用技术项目资助(315105202)。
关键词
弱链路
对等模式
相关性检测
资源信息
共享
weak-link
peer mode
correlation detection
resource information
sharing
作者简介
通信作者:桂劲松,E-mail:jsgui06@163.com。
