基于模糊-隐马尔可夫模型的复合式攻击预测方法研究

《基于模糊-隐马尔可夫模型的复合式攻击预测方法研究》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、万方数据击场景的识别和攻击意图序列的预测三项功能，将成为主动防御的重要组成部分，并应用于主动防御实践中。通过仿真实验可以看出：经过训练的隐马尔可夫模型比未经过训练的隐马尔可夫模型在对复合式攻击行为的识别和预测效果更佳，而且在当前四类主流的复合式攻击预测方法中，本文提出的基于模糊一隐马尔科夫模型的复合式攻击预测方法在报警信息预处理、报警信息关联等方面的综合性能最优。关键词：报警信息关联规则模糊评价法隐马尔可夫模型复合式攻击万方数据AbstractAsweentertheinformationage，therequirementsof

2、securityofinformationtransmission，informationstorageandinformationprocessingareofamuchhigherstandard．Thenetworksecuritynotonlyrelatestothesafetyofthecountry,thedevelopmentofeconomyandtheprogressofthescienceandtechnology,butalsorelatestothevitalinterestsofeveryone．Netw

3、orkisadouble-edgedsword；itnotonlyacceleratestheinformatizationofsociety,butalsobringsahugechallengefortheinformationsecurityissues．Inrecentyears，therateofnetworksecuritycrimesriseseveryyear．Particularly,withtheadventofonlinebanking，mobilebanking，e—commerceandotheronli

4、neservices，andthenetworksecurityissuescausedbytheconstructionofavarietyofprivatenetworks，thefollowingnetworksecurityproblemsbecomehotissuesaswell．Inthisstage，thetraditionalpassivedefenseCan’tadapttothedynamicchangesofnetworksecurity．Afterapplyingofthedefenseindepthoft

5、hemilitaryfieldtothesecurityareas，activedefenseemerges．Inthispaper,throughthestudyoftheexistedmulti—stepaRackspredictionmethods，theimprovedApriorialgorithmandfuzzyevaluationareappliedtothehiddenmarkovmodel，andtheapproachtoforecastingmulti-stepattackbasedonfuzzyhiddenm

6、arkovmodelisproposed．AfterthesemanticanalysisofrawalertsandfeaturesanalysispossessedbyaRacks，therawalertsarefusedintosuperalertaccordingtotherulesfirstly．TherealIntentionsofattackersarehiddenandCan’tbeobservedbyobserversdirectly．Buttherawalertscanbeobserveddirectly．In

7、thisregard，hiddenmarkovmodelisappliedtomulti‘stepattackspredictionmethods，inwhichthealertsareasobservationlayerandtherealintentionsareashiddenlayenThentheattackscenariowhichthealertsbelongtoisrecognizedbytheForwardalgorithmofHMMandthenextpossibleattacksequenceisforeca

8、stedbyViterbialgorithm．Finally,theexistedhiddenmarkovmodelistrainedbyBaum-WelchalgorithmofHMMandwegetanewfuzzyhiddenmarkovmo