欢迎来到天天文库
浏览记录
ID:36565589
大小:4.38 MB
页数:115页
时间:2019-05-12
《网络异常流检测技术研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、VlⅢ764分类号UDC编号:04110614舻幽建喾博士学位论文网络异常流检测技术研究学位申请人导师4:t名厦职称陆伟古书顺争教授通信与信息系统ABSTRACTInrecentyears,denial-of-service(DOS)attackpresentsmostseriousthreatstothestabilityoftheInternet.DoSattackgeneratesalargevolumeoftrafficloadstoexhaustnetworkresourcesandpreventlegitimateusersfr
2、omaccessinganInteractservice.Nowadays,newstrategieshavebeenappliedtoDoSattacksinordertoescapedetection.Attackmachinescalldecreasetheirattackratestomimiclegitimateusersorsendhighnarrowspikestolaunchapulsingbaseddenialofserviceattack.However,mostcurrentresearchesCalldetectth
3、eoccurrenceofDoSattack,butfailtoexplicitlydistinguishlegitimateusersandattackmachines.TllispaperpresentsahiddenMarkovmodel(}玎Ⅷ旧basedmethodtodistinguishlegitimatetrafficandattacktraffic.111emethodfirstusesalargenumberoflegitimatelaserrequestsequencestotrainanHMMandthenusest
4、hismodeltocheckeachincomingrequestsequences.Incomingtrafewhosesequence’Slikelihoodfallsintounreasonablerangeswouldbeclassifiedaspotentialattackmachineandshouldbecontrolled.WefurtheruseanHMM—basedclusteringmethodtoclusterthelegitimateusers’accesspattemsandgetadedicatedmodel
5、foreachcluster.TllispaperalsopresentsallHSMM(HiddenSemi-MarkovModel)baseddetectionmethodtodetectpulsingbaseddenialofserviceattack.nlemethodCannotonlydetectaggregativestreamcontainingattackflowsbutalsoidentifytheattackflows.111emethodUSeSCBF(CountingBloomFilter)torecordthet
6、ra:陆contheboUlenecklink,andHSMMtoidentifytheexistenceofpulsingbaseddenialofserviceattack.Iftheattackexists,themethodlocatesandfilterstheattackflowbasedonthetra伍crecord.Finally,inconsideringthatanDoSattackmaycausethedegradeofnetworkperformance,wepresentanewschemetomeasurean
7、destimatepacketlossrate,one-wayqueuingdelayanddelayjitterofthenetworkperformancemetrics,whichisvalidatedindiffiemtscenarios.Keywords:Anomalydetection,HiddenMarkovModel,DenialofserviceattackIV本人郑重声明:原创性声明所呈交的学位论文,是本人在导师的指导下,独立进行研究工作所取得的成果。除文中已经注明引用的内容外,本论文不包含任何其他个人或集体已经发表或撰
8、写过的作品成果。对本文的研究作出重要贡献的个人和集体,均已在文中以明确方式标明。本人完全意识到本声明的法律结果由本人承担。学位论文作者签名:日期:7o明砗f2月3日学位论文使用授
此文档下载收益归作者所有