基于自相似特性的網(wǎng)絡(luò)業(yè)務(wù)流的研究.pdf

1、江南大學(xué)碩士學(xué)位論文基于自相似特性的網(wǎng)絡(luò)業(yè)務(wù)流的研究姓名：毛藍(lán)申請學(xué)位級別：碩士專業(yè)：計算機應(yīng)用技術(shù)指導(dǎo)教師：劉淵20070301江南人學(xué)傾I學(xué)位論史AbstractTitle：ResearchontheselfsimilarityofnetworktrafileSubject：ComputerApplicationsGraduateStudent：MaoLanTufor：LiuYuanSincetheseminalstudyofLel

2、and，Taqqu，Willinger，andWilsonwhopointedouttheexistenceofscalingbehaviorinthenetworktraffic，thesocalledselfsimilarity，thereisnowampleevidencethatscale—invafiantburstinessisanubiquitousphenomenoninawiderangeofgeneralizedda

3、tatypes，fromlocal—areaandwide—areanetworkstoIPandATMprotocolstackstocopperandfiberoptictransmissionmediaHurstparameterisaveryimportantparametertoevaluatenetworkselfsimilarityInthispaper，anewmethodtoestimatetheHurstparame

4、teroftheincrementprocessinnetworktrafficaprocessthatisassumedIobeselfsimilarispresentedTheconfidenceintervalsareobtainedfortheestimatesusingthenewmethod1、isnewmethodisthenapplied“topseudorandomdataandtorealtrafficdataWec

5、omparetheperformanceofthenewmethodtothatofthewidelyusedwaveletmethod，anddemonstratethattheformerjsmuchfasterandprodnoesmuchsmallerconfidencejntervalsoftheHurstparameterestimateAndthentheestimationbasedonHurstparameterisu

6、sedtodetectDoSattack，researchedontheaffectofHurstparameterchangebroughtbyDoSattackByanalyzingthe1998DARPAIntrusionDetectionEvaluationdatasetitisverifythatthismethodcandetectDoSattackandismorereliableontherecognitionofall

7、kindsofDoSattackthananyothermethodbasedonmeasureprecisionTheexistenceofselfsimilarityshowsthatPoissonorMarkovprocesscannotaccuratelydescribetherealnetworktrafficInthispaper，thetrafficpredictiononselfsimilarityisresearche

8、dLeastMeanKurtosis(LMK)basedonQPSO，whichcanobtainsignalerrorratiolessthanLMK，isproposedtopredicttheselfsimilarIrafficThesimulationresultswiththerealtrafiletracesshowtheaccuracyefficiencyofthemodelKeywords：selfsimilarity，