基于鱼群信息链特征优选的域间流量监测  

Inter Domain Traffic Monitoring Based on Fish Information Chain Feature Selection

作　　者：徐燕文[1] 董峰[2] 

机构地区：[1]河南中医学院信息技术学院,郑州450046 [2]黄河科技学院现代教育技术中心,郑州450063

出　　处：《科技通报》2015年第6期190-192,共3页Bulletin of Science and Technology

基　　金：2014年度河南省科技厅科技攻关重点项目(142102210100);2014年度郑州市科技局科技攻关重点项目(20140662)

摘　　要：对内容中心网络的域间流量监测是未来互联网架构和网络安全监护的重要内容,通过域间流量监测,防止网络拥堵和不稳,提供网络性能,同时缩减网络冗余流量。传统方法中,对域间流量的预测采用双谱分析的方法进行特征提取,实现对局域网流量的预测,算法受到短时缓冲流量的非线性特征的影响,性能不好。提出一种基于鱼群信息链特征优选的域间流量预测算法,进行网络流量信号模型分析,构建流量监护数据样本驱动空间,进行鱼群信息链特征优选系统模型与数据预处理,得到域间流量为一阶空间权矩阵,域间网络流量信息数据表示为一个方阵,实现基于鱼群信息链特征优选的域间流量预测算法的改进。实验得出,采用该算法,在较低的信噪比下,仍然具有较高的预测精度,域间流量估计误差较低,性能较优。The contents of the central network inter domain traffic monitoring is an important content of the future Internet architecture and network security monitoring, through the inter domain traffic monitoring, to prevent network congestion and instability, provides the network performance, at the same time reduce redundant network traffic, in the traditional method, to predict flow inter domain using bispectrum analysis method feature extraction, to achieve the prediction of LAN traffic, affect the algorithm by non-linear characteristics of short-term buffer flow, performance is not good. A prediction al?gorithm flow fish information chain feature selection based inter domain is proposed, analysis of network traffic signal mod?el, construction flow monitoring data samples of chain drive space, characteristics of fish information preferred system mod?el and data preprocessing, get the inter domain traffic to a first-order spatial weight matrix, the inter domain network traffic information data representation for a square matrix, to achieve improved prediction algorithm flow fish information chain feature selection based on the inter domain. Experiments, using the algorithm in the low SNR, the prediction accuracy is still higher, inter domain traffic estimation error is low, it has a good performance.

关 键 词：鱼群算法 流量预测 特征优选 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]