摘要
针对传统的光纤网络离群恶意数据检测方法存在的检测准确性低、检测时间较长的问题,提出一种基于LODCD算法的离群恶意数据自动检测方法,为了解析离群恶意数据信号,构建了离群数据传输模型,采用经验模态分解算法对传输过程中的离群恶意数据进行解析,建立了离群恶意数据信号分析模型;将离群恶意数据分析模型划分成多个窄带信号,得到离群恶意数据包络特征,通过计算恶意数据包络特征在各个通信信道的干扰信号比提取出离群恶意数据特征;根据离群恶意数据特征计算恶意数据之间的距离,通过引用数据信息熵概念得到恶意数据离群属性,利用局部链接离群因子度量备选数据库中恶意数据的离群程度,得到离群恶意数据检测结果。通过仿真实验结果表明,所提自动检测方法较传统检测方法相比,检测准确率更高,检测耗时更少。
Aiming at the low detection accuracy and long detection time of traditional fiber network outlier malicious data detection methods,an automatic detection method of outlier malicious data based on LODCD algorithm is proposed.In order to resolve outlier malicious data signals,a new method is constructed.The outlier data transmission model uses the empirical modal decomposition algorithm to analyze the outlier malicious data in the transmission process,and establishes an outlier malicious data signal analysis model.The outlier malicious data analysis model is divided into multiple narrowband signals to obtain the separation.The group malicious data envelope feature extracts the outlier malicious data characteristics by calculating the interference signal ratio of the malicious data envelope feature in each communication channel;when calculating the distance between the malicious data according to the outlier malicious data feature,by referring to the data information entropy The concept obtains the outlier attribute of malicious data,and uses the local link outlier to measure the outlier degree of malicious data in the candidate database,and obtain the outlier malicious data detection result.The simulation results show that the proposed automatic detection method has higher detection accuracy and less time-consuming detection than the traditional detection method.
作者
刘征
孙汉卿
LIU Zheng;SUN Hanqing(School of information engineering Henan university of animal husbandry and economcy,Zhengzhou 450044,China)
出处
《激光杂志》
北大核心
2019年第8期111-114,共4页
Laser Journal
基金
国家自然科学基金项目(No.61572444)
关键词
光纤网络
离群恶意数据
数据检测
LODCD算法
离群数据
optical network
outlier malicious data
data detection
LODCD algorithm
outlier data characteristics
