欢迎来到天天文库
浏览记录
ID:36523476
大小:2.00 MB
页数:80页
时间:2019-05-11
《网络流量异常检测的小波分析方法》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、电子科技大学硕士学位论文网络流量异常检测的小波分析方法姓名:高军申请学位级别:硕士专业:通信与信息系统指导教师:胡光岷20060601摘要关键词:异常检测,多尺度,小波包,滑动时窗,无穷可分层叠ABSTRACTABSTRACTNetworktratncanomalyreferstothestatusthattrafficbehaviorsdepartfromthenormalbehaviors.Manyreasons,suchasthemisuseofnetworkequipments,networkoperationsanoma
2、ly,flashcrowd,networkintrusionandsoonwillCaUSenetworkanomaly.ThecharacteristicofanomalytrafficsisthatitbreaksoutwithoutanyomenandCandestroynetworksandcomputersinashorttime(Forinstance,theoutburstoftrafficbehaviorcausedbyspecificattackprogramsorwormburst).Therefore,tod
3、etectanomalyrapidlyandaccuratelyisoneofthepreconditionofensuringtheefficientnetworkoperationanddetectionofanomaloustraffichasbecometheattractiveandvaluablesubjectinthepresentacademicandindustrialcircles.Thispaperintendstoresearchthewaveletpacketanalysismethodwhichcand
4、etecttheanomalyaccuratelyandreliably,andachievethereal—timeperformance;aimattheproblemoftrafficanomalydetection,researchthefastwavelet(paeket)transformalgorithmbasedonslidingwindow;aSthetrafficanomalymakeadifferencetothenetworktrafficIDCcharacteristic,researchandimple
5、mentthedetectionbasedontrafficIDCcharacteristic.(1)Atfirst,thispapersummarizesthenetworktrafficanomalydetectionmethods.Analyzingtheresearchactualitybothhereandabroad,wecategorizethenetworktrafficanomalydetectionmethods,pointoutthedeficiencyofthepresentresearches,aimat
6、theproblems,proposedsomenewresearchclewsandpossibletechnologylines.(2)r111ispaperpresentssomebasictheoryaboutwaveletanalysis,andproposesafastwaveletQacket)algorithmbasedonslidingwindow.Bystoringthepartofwaveletpacketcoefficientsinmemoryinadvance,Weavoidrepeatingcalcul
7、ationandgainafastwaveletalgorithmatthecostofincreasingsomestoragespace.nleexperimentsprovedthatthefastalgorithmcansufficientlysupportOurnetworktrafficanomalydetectionmechanism.(3)Aimingatthedeficiencyofpresentmethodsofnetworktrafficanomaly,thispaperproposesanewscale—a
8、daptivedetectionmechanism.Bymeansofwaveletpacketdecomposition,Ourmethodhasthesanledetectiveabilitytomiddleandhighfrequencyas
此文档下载收益归作者所有