大数据环境下的网络流量非线性预测建模  

Nonlinear Prediction Modeling of Network Traffic in Big Data Environment

作　　者：郭海蓉 GUO Hairong(Modern Education Technology Center,Chengdu Medical College,Chengdu 610500)

机构地区：[1]成都医学院现代教育技术中心

出　　处：《微型电脑应用》2019年第8期149-151,共3页Microcomputer Applications

摘　　要：为了改善大规模网络流量预测结果,建立了大数据环境下的网络流量非线性预测模型。首先采集大规模的网络流量历史数据,然后根据云计算技术的Map/Reduce处理模式对其进行细分,得到多个数据量相对较小的子训练样本集合,并引入数据挖掘技术中的状态回声网络对子训练样本集合的网络流量进行预测,最后对子训练样本集合的网络流量预测结果进行融合,并与当前经典网络流量预测模型进行了对照实验,模型的网络流量预测精度超过90%,网络流量训练时间得到了大幅度缩短,网络流量整体性能要明显优于当前经典网络流量预测模型,对比实验验证了本模型用于当前网络流量建模与预测的优越性。In order to improve the prediction result of large scale network traffic,a non-linear prediction model of network traffic based on large scale data environment is designed.Firstly,we collect historical data of network traffic,then subdivide it according to Map/Reduce processing mode of cloud computing technology,and get a set of sub-training samples with relatively small amount of data.Then we introduce the state echo network of data mining technology to predict the network traffic of sub-training samples.Finally,we advance the prediction results of network traffic of sub-training samples.We compare it with the current classical network traffic prediction model.The network traffic prediction accuracy of this model exceeds 90%,the network traffic training time has been greatly shortened,the overall performance of network traffic is significantly better than the current classical network traffic prediction.The comparative experiments verify the superiority of this model in the current network traffic modeling and prediction.

关 键 词：单处理模式 云计算平台 回声状态网络 训练时间 网络流量预测精度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]