摘要
采用当前方法进行光纤网络流量异常监测过程中,特征选择法无法全面描述流量异常特征监测的不足,存在监测效果较差的问题。为此,提出一种基于改进特征选择法的异常流量监测方法。首先采用分光方式对光纤网络流量进行分析,获取光纤网络流量时间序列,并描述用于流量异常监测的多时间序列之间的相互关系,然后利用改进特征选择法对网络出口流量进行特征提取。利用聚类算法选择网络流量异常最优类数和聚类中心,来对网络流量异常现象进行过滤,从而实现网络异常流量特征抽取、特征选择改进算法和网络流量异常监测的研发,从而提高光纤网络流量异常现象监测的准确度。仿真实验结果证明,通过这种方法,能有效地对网络流量异常现象进行监测,且算法简单,能够满足网络流量异常监测的应用需求,实用价值较高。
In the process of anomaly monitoring of optical network traffic by using the current method,feature selection method can not fully describe the lack of monitoring of traffic anomaly characteristics,and there is a poor monitoring effect.To this end,an abnormal flow monitoring method based on improved feature selection method is proposed.First,the traffic of optical fiber network was analyzed by splitting the optical mode,the optical network traffic time series was gotten,and the relationship between multiple time series was described used for traffic anomaly monitoring,and then the improved feature selection method was used to extract the characteristics of network traffic flow.The network traffic anomaly optimal class number and clustering center was selected using clustering algorithm to filter the abnormal network traffic,so as to realize the network traffic anomaly feature extraction,feature selection algorithm and improved research of network traffic anomaly monitoring,so as to improve the fiber network traffic anomaly monitoring accuracy.Simulation results show that this method can effectively monitor the abnormal phenomenon of network traffic,and the algorithm is simple,which can meet the application needs of network traffic anomaly monitoring,and has high practical value.
作者
任春雷
刘世民
朱继阳
曹秀峰
REN Chun-lei;LIU Shi-min;ZHU Ji-yang;CAO Xiu-feng(Information&Telecommunication Branch,State Grid East Inner Mongolia Electric Power Company Ltd,Hohhot 010021,China;School of Computer Science and Technology,Wuhan University of Technology,Wuhan 430000,China)
出处
《科学技术与工程》
北大核心
2018年第23期50-54,共5页
Science Technology and Engineering
关键词
改进特征选择法
光纤网络
流量异常
监测
安全
improved feature selection
optical network
traffic anomaly
monitoring
security