/
Defending computer system with decoys Defending computer system with decoys

Defending computer system with decoys - PDF document

lois-ondreau
lois-ondreau . @lois-ondreau
Follow
389 views
Uploaded On 2017-04-04

Defending computer system with decoys - PPT Presentation

2RelatedWorkTheuseofdeceptivetechniquessuchasdisinformativepropagandatothwartonesenemieshasplayedapartinmilitarycon ictsinceantiquityNoonehassummarizedtheimportanceofdisinformationinthecontextofco ID: 335485

2RelatedWorkTheuseofdeceptivetechniques suchasdisinformativepropaganda tothwartone'senemieshasplayedapartinmilitarycon ictsinceantiquity.Noonehassummarizedtheimportanceofdisinformationinthecontextofco

Share:

Link:

Embed:

Download Presentation from below link

Download Pdf The PPT/PDF document "Defending computer system with decoys" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.


Presentation Transcript

ofcomputers.1IntroductionOrganizationsacrosstheglobearebecomingincreasinglyawareoftheimportanceofsecuringtheircomputersystems.Asaconsequence,worldwidesalesofsecuritysoftwareroseby7.5%in2011[5].Governmentagenciesareparticularlyconsciousoftheneedtodefendtheircomputinginfrastructure.Thisisexempli edbythefactthattheUnitedStatesgovernmentincreasedfundingforcybersecurityresearchby35%from2011to2012[15].Attentivenesstosecuritypracticeshasalsorisenattheindividuallevel,as90%ofAmericanadultsnowbelievethatasafeInternetiscriticaltotheU.S.economy[16].Yetinspiteoftheheightenedscrutinythatsecuritypracticeshavebeenunder,computercrimescontinueto ourish.ArecentstudybythePonemonInstitutefoundthatthenumberofcyberattackshasmorethandoubledsince2010[14].Sincetheseattacksarealsobecomingmorecomplexanddiculttoanticipate,anaveragecompanycancurrentlyexpecttobethevictimof1.8successfulo ensivee ortspermonth[14].Thevastmajorityofexistingcomputersecuritymeasuresfocusoncontrollingaccesstokeepmaliciousactorsout.Otherapproachesattempttoeliminatesystemvulnerabilitiesoratleastpreventtheirexploitation.Theaforementionedtrendsinsecuritystatisticsleavelittledoubtthatthesetechniquesarenotcapableofsucientlysecuringtoday'scomputernetworks.Overtimesuchsecuritysolutionswillinvariablyfail,allowingadversariestoillicitlyaccesssystemcredentials,data,and nancialresources.Furthermore,traditionalsecuritytechniqueso ernodefenseagainst\insiders"whoinitiallyholdlegitimatecredentialsbutlaterchoosetogorogue.Thispaperproposestheuseofdecoysasanewparadigmforaddressingcomputersecurityissuesthatexistingdefensesarenotcapableofdetecting.Decoysareconstructswhichcontaindatathatappearsvaluablebutisinfactspurious.Sinceauthenticuserswillhaveanatural2 2RelatedWorkTheuseofdeceptivetechniques,suchasdisinformativepropaganda,tothwartone'senemieshasplayedapartinmilitarycon ictsinceantiquity.NoonehassummarizedtheimportanceofdisinformationinthecontextofcombatmoreconciselythanSunTzu,whowrotethat\allwarfareisbasedondeception"intheArtofWar[17].AwellknownexampleofdeceptioninamilitarycontextisOperationBodyguard,whichwasanAlliedplanusedduringWorldWarIItodistractGermanforcesfromtheinvasionofNormandy[8].Althoughdeceptionisanancientconcept,ithasonlyrecentlybeenappliedtotheprocessofsecuringcomputersystems.Cli Stollwasthe rstpersonknowntoutilizemisdirectioninordertosecureanetworkofcomputers.Stollestablishedaspurioussetofcomputingresourcesinordertocatchhackerswhowereattemptingtoex ltrateinformationfromLawrenceBerkeleyNationalLaboratory[6].Computerswhoseprimaryfunctionistoattracttheattentionofmaliciousactorsareoftencalled\honeypots."Entirenetworksofsuchspuriousmachinesareknownas\honeynets."Thesesystemsareusuallyconstructedinawaysuchthattheyappearasthoughtheyareanunassumingcomponentofalargernetworkarchitecture.Inreality,however,theyfailtocontainanyusefuldataandarecordonedo fromnetworkresourceswhichareactuallyvalued.Honeypotsandhoneynetscanbequitee ectivewhenusedtodetectexternalthreats.Theirapplicabilitytowardsdefendingagainstattacksoriginatingfromwithinanorganizationislimited,though.Thisisduetothefactthatthisclassofadversariestypicallyalreadyhavetheknowledgethatisrequiredtoaccesstheportionofanetworkwherelegitimatedataresides.Furthermore,honeypotso ernoutilityafterasuccessfulattackhasalreadyoccurred.Spitznerextendedtheconceptofhoneypotstothedomainofinsiderthreatdetectionbyinventingtheconceptof\honeytokens"[12].Honeytokensaredeceptivesecurityconstructsthatworkatamuch nergranularitythanhoneypotsorhoneynetworks.Theyareindividual4 bankcredentialsismorelikelytocapturetheirmaliciousactivitythanonewhosecontentconcernsmedicalinformation.Similarly,certaingenresofdecoysmaybemoreapplicabletospeci ccorporateenvironments.WehaveincludedadetaileddiscussionofdecoyusagescenariosinSection4.Inordertodesigndecoysthatarease ectiveaspossible,itisalsobene cialtoanalyzetheminamoregeneralsensebyconsideringcharacteristicsthatareindependentofapar-ticularcontext.AsinitiallyexploredbyBowenet.alin[1],severalabstractpropertiesexistthatde nehowadecoyshouldoperateunderidealcircumstances.Someoftheseattributesconcerntherelationshipbetweenadversariesanddecoydata,whileotherspertaintotheinteractionsbetweenlegitimateusersanddeceptivematerial.A\perfectlybelievabledecoy"wouldpreciselyconformtoalloftheseguidelines,thoughpracticalrestrictionspreventthisfromoccurringinmostsituations.Althoughthereexistssomeoverlapbetweenthesetraits,itisalsoworthnotingthattheyarenotcompletelyorthogonal.Forexample,believabilityanddi erentiabilityareincontentiontosomeextent.3.1BelievabilityOneofadecoy'sprimaryfunctionsistobebelievable.Uponinspection,adecoyshouldappearauthenticandtrustworthy.Intheabsenceofanyadditionalinformation,itshouldbeimpossibletodiscernaspuriousdecoyfromauthenticdata.Forexample,adecoytaxdocumentshouldcontainallofthesame eldsasonethatisactuallyinuse,andeachofits eldsshouldbepopulatedwithrealisticvalues.Believabilitycanbeformalizedviathefollowingthoughtexperiment.Considerapoolof les,someofwhichcontainrealdataandsomeofwhicharefabricateddecoys.Selectadecoy leandrealpieceofdatafromthispool,andpresentittoanadversary.Theselecteddecoycanbeconsideredperfectlybelievableifthisattackerhasanequalprobabilityofselectingthedecoyandthelegitimatedocument.Thischaracteristicisofcriticalimportancetoexternallyobservablefeaturesofdecoys.6 locationssuchasauser'sdesktop.Italsodemonstratesthatitishelpfultoplacedocumentsinhightrac lesystemlocations,includingworkingfolderswhere lesthatareaccessedonaday-to-daybasisarestored.Filesystemsearchesarealsouseractionsthatmayresultinthepresenceofdecoys.Conspicuousdecoysshouldthereforebeeasilylocatedbysearchqueries.Thisrelatesconspicuousnessbacktoenticingness,however,asthesearchtermsthatanadversaryemployeeswillbeheavilydependentontheirunderlyingmotivation.3.4DetectabilityTheaforementioneddecoypropertiesallconcerntherelationshipbetweendecoydocumentsandapotentialattacker.Detectability,ontheotherhand,describestheabilityofdecoystonotifytheirownerwhentheyhavebeenaccessed.Anidealdecoysystemwouldissueanalerteachandeverytimeadecoyisaccessed,buttechnicalchallenges,includingnetworkavailabilityandvariabilitybetweensoftwareplatforms,meanthatthismaynotalwaysbepossibleinpractice.Deployingmultipleoverlappingdecoymonitorsthatoperateatdi erentsystemlevelscanhelpmitigatethepossibilityofanattackeraccessingadecoywhileremainingunde-tected.Featuresofthedecoydocumentsthemselvescanbeleveragedtoequipthemwithembeddedalertcode.Monitoringsoftwarecanbeplacedintheoperatingsystemtodetectpredeterminedtokensplacedwithindecoyswhentheyareopened.Further,operatingsys-temauditingcanbeenabledtorecorddecoyinteractions.Inordertocheckfordocumentex ltration,softwarecanbeplacedonnetworkequipmenttocheckforsuchtokensaswell.Finally,thecontentofdecoydocumentscanalsoserveasanalertsystem.Forexample,credentialsforspuriousaccountscanbeplacedwithinadecoy.Sincethereisnoreasonthatalegitimateuserwouldeveraccesstheseaccounts,anyactivitytheyexhibitwouldsendastrongsignalofmaliciousintent.Itisparticularlycriticalthatdecoyaccesseventsaredetectablewhileanattackistakingplace.Continuingtomonitorthisinformationallowsforcon dentiallyviolationstobe8 3.6StealthWhileitisclearlydesirablethateverydecoyaccesseventbeperceptibletotheownersofasystem,caremustbetakenlestthealarmsthataccomplishthisarousesuspicion.Anovertmechanismforissuingalertbeaconswouldprovideadversarieswithanobvioussignalthatanelementcontainsatrap,whichcompletelyviolatesthepropertyofdecoyvariability.Themessagesthataretransmittedbydecoysmustthereforebeassubtleandcovertaspossible.Raisinganalertthatdecoycontenthasbeenaccessednecessarilyinvolvestakingsomeaction,however.Evenifprecautionsaretaken,thereisalwaysthepossibilitythatthisactwillbeperceptibletoamaliciousactor.Itisthereforealsodesirabletotriggerbeaconeventsasearlyaspossibletopreventtheirinterception.Forexample,alertsfor lebaseddecoysshouldberaisedassoonastheyareaccessedandpriortoanycontentbeingdisplayed,iffeasible.Thiswouldeliminatethepossibilityofadecoybeingrecognizedanddiscardedbeforethedecoysystemhasanopportunitytodetectthatishasbeenaccessed.3.7Non-interferenceThispropertyisthe rsttodescribehowdecoysshouldcoexistwithlegitimateuserswhoarenotmasqueradingwithassumedcredentials.Anoptimalmasqueraderdetectionnetworkwouldnota ectthehabitsoftypicalusersinanyway.Byinsertingdecoymaterialintoanoperatingenvironment,however,weintroducethepossibilitythatthisdatawillconfuseusersorotherwisehindertheirabilitytocompletetheireverydaytasks.Itisthereforedesirablefordecoystodemonstratethepropertyofnon-interferencebynotobstructingthebehaviorofnormalusers.Ifa lesystemispopulatedwithdecoydocumentsthatserveasintrusionsensors,forexample,theprobabilitythatthe lesystem'sprimaryownerisabletoaccessaparticularstandarddocumentshouldremainthesameasitwaspriortotheintroductionofthedecoycontent.Similarly,introducingdecoyapplicationstoamobiledevice'soperatingsystemshouldnotimpactauser'sabilitytoaccessrealapplicationsastheynormallywould.10 anoptimalleveloffunctionalityandafterwhichtheirecacybeginstodiminish.Itisthusdesirablefordecoydeploymentsystemstoincludeamechanismbywhichdecoyscanbeupdated,potentiallyextendingtheirshelflifeinde nitely.4DecoyUsageScenariosThissectionintroducesseveralbroadsituationsinwhichsecuritycanbebolsteredbyde-ployingadecoydefensenetwork.Italsoattemptstodiscusssomeofthechallengesthatmustbemetinorderfordecoystobeusede ectivelyineachenvironment.4.1HostDecoysThemostcommonusagescenariofordecoysistoplacethemonaterminalthatiswithinalocalcomputernetwork.Thisistheoperatingenvironmentforwhichdecoydocumentswereoriginallydesigned,andassuchtheyrequirelittlemodi cationtobeutilizedinthismanner.Therearestillsomedeploymentquestionsthatmustbeconsidered,however.Forexample,systemadministratorsmustdeterminewhethertheywillpushdecoydocumentsouttoclientsystemsorrequireuserstopulldecoysfromadistributionsourcethemselves.Theformerplaceslessofaburdenonindividualusers,butmaytemporarilyleadtoanincreaseinfalsepositivesasusersbecomeacquaintedwiththenewdecoysthathavebeenplacedintheirworkspace.Experimentsfrom[13]showhowdecoyaccessisa ectedbyadocument's lesystemlocationandthenumberofdeployeddecoys.4.2NetworkandBehavioralDecoysInane orttodetectsilentattackersastheyeavesdropontransmissionsbetweencomputersystems,wehavealsodevelopeddecoysthatoperateonanetworklevel.Ratherthanhostbaseddata les,thesedecoysconsistofbogusdata owsthatareinjectedintoanetwork.Thistracmustappearlegitimateintermsofprotocolspeci cationsbutalsocontainsbait12 sensitivecorporateororganizationaldata.Furthermore,gainingaccesstoanorganization'sinfrastructureonthecloudisachievedthroughprovidingausernameandpassword.Suchalightweightauthenticationmechanismpresentsanincreaseintheproblemofmasqueraderinsiderthreats.Accordingto[18],observingdeviationsfromtypicaluserbehaviorcanbeusedintandemwithdecoydocumentsinacloudenvironment.Ifabnormaldataaccessisnoticed,thecloudcanreturndecoyinformationthatlookslegitimatetotheattackerwhotriggeredtheunusualusagepattern.Intheeventofafalsepositive,anauthorizeduserwouldbeabletorecognizeanyfalseinformationreturnedbythecloudandthencorrectlyrespondtoaseriesofauthenticationchallengestoprovehisorherlegitimacy.Thecloudwouldpreventanyunauthorizeddisclosureofinformationbycontinuouslyreturningfalsedatatoadversaries.Alsousefulinadistributedenvironmentsuchasthecloudistheconceptofcomputationaldecoys.Oftenasinglecomponentofacloudwillberesponsibleforacertaintask.Ifthiscomponentiscompromised,thedistributedsystemallowsmigrationoftheuncorruptedcomponentselsewhere.Thecompromisedcomponentscanthenactuallyusecomputationaldeceptiontoreturnfalsedata,thusdivertinganattacker'sattention[11].4.4SoftwareDecoysCompaniesareconstantlyfacedwiththechallengeofpreventingemployeesfromstealingproprietarysoftware.InMayof2012,forexample,acomputerprogrammerwasfoundguiltyofillegallycopyingsoftwarefromtheFederalReserveBankofNewYork[2].Softwareisperhapsthemostvaluableassetforanorganizationandisconsequentlyahighlypro tabletargetforinsiders.Toguardagainsttheunauthorizedex ltrationofproprietarysoftware,wehavedevisedsoftwaredecoysthatlooklikelegitimatesourcecodebuthavebeaconsthattriggerwhenthecodeiscompiledorexecuted[19].InadditiontothepropertiesofdecoysdescribedinSection3,softwaredecoysshouldadheretoseveraladditionalrequirements.Thecodemustbecompilableandexecutablein14 Aseconduseofdecoysinamobilesetting,whichissimilartotheirusageinacloudenvironment,isforremoteaccesstosensitivedocuments.Peopleareincreasinglyusingmo-bilephonestoaccessprivateinformation.Wearecurrentlyworkingonamobileapplicationthatinterfaceswithauser'sdocumentsremotelyanddetectsunauthorizedaccess.Whenuploadingdocumentstotheremoteserver,ausercreatesaauthenticationgesturethatislaterusedtoretrieveadocument'scontent.Ifauserprovidesanincorrectgesturewhenattemptingtoaccessadocument,illegitimatedataisreturned.5DecoyGenerationHavingestablishedwhatqualitiesdecoysshouldpossessandwheretheyareintendedtobeused,wenowturnourattentiontohowtheyshouldbecreated.Userscouldcertainlycraftdecoymaterialbyhand.Forinstance,uponcompletinganinvoice,ausercouldcreateasecondaryfakeinvoicethatmirrorstheformattingortheauthenticversionbutcontainsbogusinformation.Suchdecoyswouldbehighqualityintermsoftheirbelievabilitybecausetheywouldcloselymirrorrealdata.Theywouldalsobeverydi erentiable;sinceuserswouldhavecreatedthedocumentsthemselves,theywouldeasilybeabletorecognizetheirphonycontent.Theprocessofmanuallyintroducingdecoycontenttoasystemisverytedious,however,sinceeachtimenewinformationissavedonthesystemanequalamountofspuriousmaterialwouldneedtobecreatedaswell.Userswouldalsoberesponsibleforcheckingaccesseventsforthese les.Needlesstosay,manualdecoycreationwouldscaleverypoorlytoalargeorganizationwithmanycomputersandusers.Making,managing,andmonitoringdecoysisthusanontrivialproblem.Asanalternativetoperformingthesestepsmanually,wesuggestusingasystemthatdoessowithminimaluserinvolvement.ThisispreciselythepurposeoftheDecoyDocumentDistributor(D3)System[1].D3isatoolforgeneratingandmonitoringdecoyswhichcanbeaccessedby16 throughoutthecomputersonhisorhernetworkinordertodefendthemfrominsiderthreats.Tomanuallyplacedecoysina lesystem,thesystemadministratorwould rsthavetocollectfeedbackfromusersregardingwhichlocationsintheir lesystemtheywouldliketoplacedecoys.Afteraggregatingthisdata,heorshewouldthenhavetorequestabatchofdecoysandthengothroughthepainstakingprocessofcopyingthemtotheirdestinationdirectoriesonebyone.Suchaprocessdoesnotscalewelltoanenvironmentwithalargenumberofcomputersandusers.Incontrast,theDDTrequiresneitherknowledgeregarding lesystemspeci csnorindividual leplacement.Itreducesthetaskofdecoydocumentmanagementtothesimplestepsofspecifyinghowmanydecoysaredesiredandinwhichportionofa lesystemtoplacethem.Ourdistributionapplicationisthereforecapableofreducingthetimerequiredtoestablishasystemofinsiderthreatsensorsbyatleastanorderofmagnitudewhileretainingallofthesecuritybene tsofmanualdecoyusage.TheDDThastwomainobjectives.The rstistoautomaticallydeterminelocationsinacomputer's lesystemthataremostlikelytobeaccessedbyamaliciousinsider.Thesecondistoplacedecoydocumentsintheseselectedlocations,eitherdirectlyalongwithexistingdocumentsorinaseparatefolder.TheDDTallowsausertoselectasourcedirectoryofdecoydocumentsthatshouldbedistributedonthetargetmachine.TheusercanchooseanexistingfoldercontainingdecoysorcreateanewsetbyaccessingtheFOGwebsite[7]throughtheDDTandspecifyinganumberofdecoystobegenerated.Thisapproachenables exibilityinthetypesofdocumentsthataredeployedasdecoys.Oncetheuserchoosesasourcedirectoryofdecoydocuments,heorshethenspeci esadestinationdirectory.Thisdestinationdirectoryisusedasarootfromwhichtargetlocationsareselected.Enablingtheusertospecifytherootfromwhichtargetdirectoriesareidenti edallowsmorefreedomindecoyplacement.Forexample,ifaparticularownerofamachineonlyusesdirectorieswithin\C:nSal,"heorshemaywanttoconsiderplacingdecoyssolelyindirectoriesbranchingfromthisroot.18 inwithcurrentlyexistingdocuments,orinseparatefoldersthatcontaindecoysonly.6.3NamingofDecoyDocumentsandFoldersAswithlocation,thenamesofdocumentsdirectlyin uencehowenticingtheylooktoad-versaries.Foldernamescanalsoimpactauser'sdecisiontoaccessalocationinthe lesystem.Whencreatingnewdirectoriesinwhichdecoyswillbeplaced,theDDTcreatesfourenticingly-namednewfoldersinthespeci eddestinationdirectoryandevenlydispersesthedecoysoverthesefolders.Inane orttoincreasethevariabilityofdecoys,theDDTusesthreemethodsfornamingdecoydocumentswhenattemptingtoblendthemwithexistingdocumentsinafolder,ran-domlyselectingamongtheseapproaches.Theprimeobjectiveofthenamingschemeistocreate lenamesthatblendinwithexistinglegitimatedocumentssothattheydonotlookovertlysuspicious.Atthesametime,thedecoynamesshouldluremalevolentusersintoopeningthedocuments.The rstnamingmethodselectsanexisting leinthetargetdirectoryandappendseither\- nal"or\-updated"totheendofthe lename.Thelogicbehindthisschemeisthatthemostrecentlymodi edversionsof lesmayseemmoreocialthanolderversions.Notethatifeitheroftheterms\ nal"or\updated"arealreadyusedinthe lename,theDDTselectsanothernamingmethodforthisdecoy.ThesecondnamingmethodemployedbytheDDTappendsadatestringtotheendofarandomlyselectedexisting lenameinthetargetdirectory.Appendingadatestringmakesadocumentappearasifithasbeenmarkedasamoreauthentic,ocialversion.TheDDTusestheformatmmddyyforthedatestrings.Forexample,adecoydocumentmaybenamedcompany employees-010412.pdfandplacedinthetargetfolderalongsidealegitimatedocumentnamedcompany employees.pdf.Blending lenamesbyuseofdelimitersisthe nalnamingapproach.TheDDTcalcu-latesthedelimiterusedmostofteninthetargetdirectoryandmodi esadecoy's lename20 date.Ifanadversarydecidedtosortafolder'scontentsbydate,thedecoydocumentswouldthenappearatthetopofthesortedlist,makingthemconspicuousandlikelytobemoreattractivetargets.Anamingconventionthatappendsadatestringtothedecoy's lenameshouldintuitivelysetthe le'screationdatetothatusedintheselecteddatestring.TheDDT rstdeterminesanappropriatedateforadecoydocumentby ndingthemediancreationdateofexisting lesinthedecoy'stargetdirectory.Thedecoy'screationdateisthensetwithina48-hourwindowofthemedian.Thedatestringappendedtothe lenameisobtainedfromtheproposedcreationdate.Whenplacingdecoydocumentsinaseparatenewfolder,theDDThasnoexistingdocu-mentdatesonwhichtobasea le'stargetdate.Therefore,themodifydatesofthedecoysaredeterminedbythedatethatthedecoysaregeneratedbytheFOGsystem.Acreationdateisthensettouptotwodaysbeforetheexistingmodifydate.TheDDTsubsequentlyaltersthemodifyandcreationdatesofthenewfolderinwhichthedecoysareplaced.Thecreationdateofthefolderissettothecreationdateoftheoldestdecoy leplacedinside;themodifydateissettothatofthemostrecentlymodi eddocument.Weconsideredthisdatingapproachthemostpracticalforthecaseinwhichnewdecoyfoldersarecreated,sinceusershavedirectcontroloverthedatesappliedtothedecoydocuments.Inourfuturework,theDDTwillincludeafeatureto\refresh"thedatesofdecoyssothattheyremainamongthemostrecentdocumentsinthe lesystem.7ConclusionsTosummarize,thispaperintroducedanovelsecurityparadigmwhichwerefertoasdecoytechnology.Decoysrepresentadrasticdeparturefromexistingsecuritysolutionsinseveralimportantways.Byplacingcontentthatisspuriousyetbelievableandenticinginthepathofpotentialadversaries,decoyscanserveasapotentlastlineofdefenseagainstattacks22 Networks,2009.[2]B.Katz.ChineseManPleadsGuiltytoNYFedCyberTheft.Availableathttp://www.reuters.com/article/2012/05/29/usa-crime-fed-idUSL1E8GTBG120120529,2012.[3]B.Bowen,P.Prabhu,V.Kemerlis,S.Sidiroglou,A.Keromytis,andS.Stolfo.Botswindler:Tamperresistantinjectionofbelievabledecoysinvm-basedhostsforcrimewaredetection.InRecentAdvancesinIntrusionDetection,page118137,2010.[4]B.M.Bowen,V.P.Kemerlis,P.Prabhu,A.D.Keromytis,andS.J.Stolfo.Automatingtheinjectionofbelievabledecoystodetectsnooping.InProceedingsofthethirdACMconferenceonWirelessnetworksecurity,page8186,2010.[5]C.PetteyandR.vanderMeulen.GartnerSaysSecuritySoftwareMarketGrew7.5Percentin2011.Availableathttp://www.gartner.com/it/page.jsp?id=1996415,2012.[6]C.Stoll.TheCuckoo'sEgg,1989.[7]ColumbiaUniversityIntrusionDetectionSystemsLab.FOGComputing.Availableathttp://ids.cs.columbia.edu/FOG/,2012.[8]J.Rubin.Deception:Theother'D'inD-Day.Availableathttp://www.msnbc.msn.com/id/5139053/ns/msnbc_tv-the_abrams_report/t/deception-other-d-d-day,2004.[9]J.VorisandN.BoggsandS.Stolfo.LostinTranslation:ImprovingDecoyDocumentsviaAutomatedTranslation.InWorkshoponResearchforInsiderThreat,2012.[10]J.YuillandM.ZappeandD.DenningandF.Feer.Honey les:DeceptiveFilesforIntrusionDetection.InWorkshoponInformationAssurance,2004.[11]A.D.Keromytis,R.Geambasu,S.Sethumadhavan,S.J.Stolfo,J.Yang,A.Benameur,M.Dacier,M.Elder,D.Kienzle,andA.Stavrou.TheMEERKATScloudsecurity24