摘要
采用空时网格编码方法辨识异常光纤通信流量时,低于大数据背景中噪声干扰的能力差,辨识结果误差高、稳定性差。深入研究大数据背景的光纤通信流量异常辨识方法,构建基于LSSVM的光纤通信网络流量挖掘模型,挖掘大数据中光纤通信流量,采用小波变换对挖掘出的光纤通信网络流量实行去噪后,依据光纤通信流量的相对熵通过训练与辨识两个过程辨识异常流量。实验结果表明,所提方法的辨识值与实际值相同,可有效辨识大数据背景的光纤通信异常流量,且鲁棒性高达99%以上、辨识准确率均值超过99%、辨识耗时少,具有较高的鲁棒性、准确率和辨识效率。
When using time grid coding method to identify abnormal optical fiber communication flow, the ability of noise interference is lower than that of big data background , and the error of identification results is high and the sta bility is poor. The optical fiber communication traffic anomaly identification method based on big data background was studied in depth , and an optical fiber communication network traffic mining model based on LSSVM was built. The op tical fiber traffic in big data was excavated, and the optical fiber traffic was denoised bywavelet transform. According to the relative entropy of optical fiber traffic, the abnormal traffic is identified by training and identification. The experi mental results show that the identification value of the proposed method is the same as the actual value, and it can ef fectively identify the abnormal traffic of optical fiber communication underbig data background, and the robustness is as high as 99%, the average identification accuracy rate is over 99%, and the identification time is smaller. It has high robustness, accuracy and identification efficiency.
作者
马宗梅
张睿萍
MA Zongmei;ZHANG Ruiping(Department of Computer Science and Technology, Zhongyuan University of Technology, Zhengzhou 450007 , China)
出处
《激光杂志》
北大核心
2019年第7期75-78,共4页
Laser Journal
基金
河南省科技厅科技攻关项目(No.172102210594)
关键词
大数据
光纤通信
流量
异常
辨识
小波变换
big data
optical fiber communication
traffic
anomaly
identification
wavelet transform