端到端網(wǎng)絡(luò)流量的混合估計(jì)方法-論文.pdf

《端到端網(wǎng)絡(luò)流量的混合估計(jì)方法-論文.pdf》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫(kù)。

1、第43卷第7期光子學(xué)報(bào)Vo1．43No．72014年7月ACTAPHOToNICASINICAJuly2014doi：10．3788／gzxb20144307．0706023端到端網(wǎng)絡(luò)流量的混合估計(jì)方法蔣定德，趙祖耀，許宏偉，王興偉(東北大學(xué)信息科學(xué)與工程學(xué)院，沈陽(yáng)110819)摘要：利用主成分分析法獲取端到端網(wǎng)絡(luò)流量的主要特征分量并獲得其初始估計(jì)結(jié)果．為克服其初值敏感性將估計(jì)結(jié)果作為遺傳算法的初始值、鏈路流量估計(jì)偏差作為遺傳算法的適應(yīng)度函數(shù)，通過構(gòu)建合適的交叉和變異概率函數(shù)來控制遺傳算法的交叉

2、和變異過程．采用合適的約束迭代函數(shù)，利用遺傳算法通過迭代尋優(yōu)獲得端到端流量的估計(jì)結(jié)果，仿真結(jié)果表明所提出的方法是可行的．關(guān)鍵詞：端到端流量；流量估計(jì)；主成分分析；迭代過程；流量建模中圖分類號(hào)：TP393文獻(xiàn)標(biāo)識(shí)碼：A文章編號(hào)：1004—4213(2014)07—0706023—6MixedEstimationApproachtoEnd-to—EndNetworkTrafficJIANGDing—de，ZHAOZu—yao，XUHong—wei，WANGXing—wei(College0，form

3、ationScienceandEngineering，NortheasternUniversity，Shenyang110819，China)Abstract：Principalcomponentanalysiswasexploitedtoextracttheprincipalfeaturesofend-to-endnetworktrafficandtoattaintheinitialestiamtionresults．Thisresultsaretakenasthepriorvalueofge

4、neticalgorithmtoovercomeitssensitivenesstothepriorvalue．Theestiamtionbiasesoflinktrafficisregardedasthefitnessfunctionofgeneticalgorithm．Thecrossoverandmutationprobabilityfunctionsarebuilttocontro1itscorossoverandmutationprocesses．Theappropriateitera

5、tivefuntionwithcontraintsisbuilt．Thegeneticalgorithmisusedtoattaintheend-to—endtrafficestimationresultsintheiterativeway．Simulationresultsshowthattheproposedmethodisfeasible．Keywords：End—to—endtraffic；Trafficestimation；Principalcomponentanalysis；Iter

6、ativeprocess；TrafficmodelingoCISCodes：060．4256；060．4251；060．4258；060．1155networkmanagementandtrafficengineering．Dueto0Introductionthedifficultyindirectlyattainingthem，end-to-endwiththerapiddevelopmentofinformationtrafficestimationhasreceivedmoreatten

7、tionfromtechnologies，communicationnetworksarebecomingresearchersandoperatorsaroundtheworldE．moreandmorecomplexandtheirscalesareTheestimationmethodsinRefs．F7—9]isbasedonincreasingl_l_2]．Thenetworktrafficpresentsallkindsofstatistica1theoriestobeputforw

8、ard．Theyusedthenewfeatures。suchasspatio—temporalcorrelationstatisticalmodelstosetupnetworks"source-destinationfeatures，self-similaritynature，heavy-tailednodeflowmodelandexploitedthestatisticaltheorytodistributionandsoon[]．End_to—endtrafficshowstheest