网络流量异常检测的小波分析方法

《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