科技通报
科技通報
과기통보
BULLETIN OF SCIENCE AND TECHNOLOGY
2015年
6期
190-192
,共3页
鱼群算法%流量预测%特征优选
魚群算法%流量預測%特徵優選
어군산법%류량예측%특정우선
fish algorithm%traffic prediction%feature selection
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。
對內容中心網絡的域間流量鑑測是未來互聯網架構和網絡安全鑑護的重要內容,通過域間流量鑑測,防止網絡擁堵和不穩,提供網絡性能,同時縮減網絡冗餘流量。傳統方法中,對域間流量的預測採用雙譜分析的方法進行特徵提取,實現對跼域網流量的預測,算法受到短時緩遲流量的非線性特徵的影響,性能不好。提齣一種基于魚群信息鏈特徵優選的域間流量預測算法,進行網絡流量信號模型分析,構建流量鑑護數據樣本驅動空間,進行魚群信息鏈特徵優選繫統模型與數據預處理,得到域間流量為一階空間權矩陣,域間網絡流量信息數據錶示為一箇方陣,實現基于魚群信息鏈特徵優選的域間流量預測算法的改進。實驗得齣,採用該算法,在較低的信譟比下,仍然具有較高的預測精度,域間流量估計誤差較低,性能較優。
대내용중심망락적역간류량감측시미래호련망가구화망락안전감호적중요내용,통과역간류량감측,방지망락옹도화불은,제공망락성능,동시축감망락용여류량。전통방법중,대역간류량적예측채용쌍보분석적방법진행특정제취,실현대국역망류량적예측,산법수도단시완충류량적비선성특정적영향,성능불호。제출일충기우어군신식련특정우선적역간류량예측산법,진행망락류량신호모형분석,구건류량감호수거양본구동공간,진행어군신식련특정우선계통모형여수거예처리,득도역간류량위일계공간권구진,역간망락류량신식수거표시위일개방진,실현기우어군신식련특정우선적역간류량예측산법적개진。실험득출,채용해산법,재교저적신조비하,잉연구유교고적예측정도,역간류량고계오차교저,성능교우。
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.