Trust management ix 9th ifip wg 11 11 international conference ifiptm 2015 hamburg germany may 26 28

Page 1

International Conference IFIPTM 2015

Hamburg Germany May 26 28 2015

Proceedings 1st Edition Christian Damsgaard Jensen

Visit to download the full and correct content document: https://textbookfull.com/product/trust-management-ix-9th-ifip-wg-11-11-international-c onference-ifiptm-2015-hamburg-germany-may-26-28-2015-proceedings-1st-edition-c hristian-damsgaard-jensen/

Trust
WG 11
Management IX 9th IFIP
11

More products digital (pdf, epub, mobi) instant download maybe you interests ...

ICT Systems Security and Privacy Protection 30th IFIP TC 11 International Conference SEC 2015 Hamburg Germany May 26 28 2015 Proceedings 1st Edition Hannes Federrath

https://textbookfull.com/product/ict-systems-security-andprivacy-protection-30th-ifip-tc-11-international-conferencesec-2015-hamburg-germany-may-26-28-2015-proceedings-1st-editionhannes-federrath/

Trust Management XIII 13th IFIP WG 11 11 International Conference IFIPTM 2019 Copenhagen Denmark July 17 19 2019 Proceedings Weizhi Meng

https://textbookfull.com/product/trust-management-xiii-13th-ifipwg-11-11-international-conference-ifiptm-2019-copenhagen-denmarkjuly-17-19-2019-proceedings-weizhi-meng/

Computer and Computing Technologies in Agriculture IX 9th IFIP WG 5 14 International Conference CCTA 2015 Beijing China September 27 30 2015 Revised Selected Papers Part I 1st Edition Daoliang Li

https://textbookfull.com/product/computer-and-computingtechnologies-in-agriculture-ix-9th-ifip-wg-5-14-internationalconference-ccta-2015-beijing-china-september-27-30-2015-revisedselected-papers-part-i-1st-edition-daoliang-li/

Environmental Software Systems Infrastructures Services and Applications 11th IFIP WG 5 11 International Symposium ISESS 2015 Melbourne VIC Australia March 25 27 2015 Proceedings 1st Edition Ralf Denzer

https://textbookfull.com/product/environmental-software-systemsinfrastructures-services-and-applications-11th-ifipwg-5-11-international-symposium-isess-2015-melbourne-vicaustralia-march-25-27-2015-proceedings-1st-edition-ralf-denz/

Open

Source Systems Adoption and Impact 11th IFIP WG 2

13 International Conference OSS 2015 Florence Italy May 16 17 2015 Proceedings 1st Edition Ernesto Damiani

https://textbookfull.com/product/open-source-systems-adoptionand-impact-11th-ifip-wg-2-13-international-conferenceoss-2015-florence-italy-may-16-17-2015-proceedings-1st-editionernesto-damiani/

Health Information Science 4th International Conference

HIS 2015 Melbourne Australia May 28 30 2015 Proceedings 1st Edition Xiaoxia Yin

https://textbookfull.com/product/health-information-science-4thinternational-conference-his-2015-melbourne-australiamay-28-30-2015-proceedings-1st-edition-xiaoxia-yin/

Codes Cryptology and Information Security First

International Conference C2SI 2015 Rabat Morocco May 26 28 2015 Proceedings In Honor of Thierry Berger 1st Edition Said El Hajji

https://textbookfull.com/product/codes-cryptology-andinformation-security-first-internationalconference-c2si-2015-rabat-morocco-may-26-28-2015-proceedings-inhonor-of-thierry-berger-1st-edition-said-el-hajji/

Algorithms and Complexity 9th International Conference CIAC 2015 Paris France May 20 22 2015 Proceedings 1st Edition Vangelis Th. Paschos

https://textbookfull.com/product/algorithms-and-complexity-9thinternational-conference-ciac-2015-paris-francemay-20-22-2015-proceedings-1st-edition-vangelis-th-paschos/

Data

Driven Process Discovery and Analysis 5th IFIP WG

2 6 International Symposium SIMPDA 2015 Vienna Austria December 9 11 2015 Revised Selected Papers 1st Edition Paolo Ceravolo

https://textbookfull.com/product/data-driven-process-discoveryand-analysis-5th-ifip-wg-2-6-international-symposiumsimpda-2015-vienna-austria-december-9-11-2015-revised-selectedpapers-1st-edition-paolo-ceravolo/

Christian Damsgaard Jensen Stephen Marsh

Theo Dimitrakos Yuko Murayama (Eds.)

Trust Management IX

9th IFIP WG 11.11 International Conference, IFIPTM 2015 Hamburg, Germany, May 26–28, 2015 Proceedings

123
IFIP AICT 454

IFIPAdvancesinInformation andCommunicationTechnology

Editor-in-Chief

KaiRannenberg,GoetheUniversity,Frankfurt,Germany

EditorialBoard

FoundationofComputerScience

JacquesSakarovitch,TélécomParisTech,France

Software:TheoryandPractice

MichaelGoedicke,UniversityofDuisburg-Essen,Germany Education

ArthurTatnall,VictoriaUniversity,Melbourne,Australia

InformationTechnologyApplications

ErichJ.Neuhold,UniversityofVienna,Austria CommunicationSystems

AikoPras,UniversityofTwente,Enschede,TheNetherlands SystemModelingandOptimization

FrediTröltzsch,TUBerlin,Germany InformationSystems

JanPries-Heje,RoskildeUniversity,Denmark ICTandSociety

DianeWhitehouse,TheCastlegateConsultancy,Malton,UK ComputerSystemsTechnology

RicardoReis,FederalUniversityofRioGrandedoSul,PortoAlegre,Brazil SecurityandPrivacyProtectioninInformationProcessingSystems

YukoMurayama,IwatePrefecturalUniversity,Japan

Arti ficialIntelligence

TharamDillon,CurtinUniversity,Bentley,Australia

Human-ComputerInteraction

JanGulliksen,KTHRoyalInstituteofTechnology,Stockholm,Sweden

EntertainmentComputing

MatthiasRauterberg,EindhovenUniversityofTechnology,TheNetherlands

454

IFIP – TheInternationalFederationforInformationProcessing

IFIPwasfoundedin1960undertheauspicesofUNESCO,followingtheFirstWorld ComputerCongressheldinParisthepreviousyear.Anumbrellaorganizationfor societiesworkingininformationprocessing,IFIP’saimistwo-fold:tosupportinformationprocessingwithinitsmembercountriesandtoencouragetechnologytransferto developingnations.Asitsmissionstatementclearlystates,

IFIP’smissionistobetheleading,trulyinternational,apoliticalorganizationwhich encouragesandassistsinthedevelopment,exploitationandapplicationofinformationtechnologyforthebenefitofallpeople.

IFIPisanon-profitmakingorganization,runalmostsolelyby2500volunteers.It operatesthroughanumberoftechnicalcommittees,whichorganizeeventsandpublications.IFIP’seventsrangefromaninternationalcongresstolocalseminars,butthe mostimportantare:

• TheIFIPWorldComputerCongress,heldeverysecondyear;

• Openconferences;

• Workingconferences.

The flagshipeventistheIFIPWorldComputerCongress,atwhichbothinvitedand contributedpapersarepresented.Contributedpapersarerigorouslyrefereedandthe rejectionrateishigh.

AswiththeCongress,participationintheopenconferencesisopentoalland papersmaybeinvitedorsubmitted.Again,submittedpapersarestringentlyrefereed.

Theworkingconferencesarestructureddifferently.Theyareusuallyrunbya workinggroupandattendanceissmallandbyinvitationonly.Theirpurposeisto createanatmosphereconducivetoinnovationanddevelopment.Refereeingisalso rigorousandpapersaresubjectedtoextensivegroupdiscussion.

PublicationsarisingfromIFIPeventsvary.ThepaperspresentedattheIFIPWorld ComputerCongressandatopenconferencesarepublishedasconferenceproceedings, whiletheresultsoftheworkingconferencesareoftenpublishedascollectionsof selectedandeditedpapers.

Anynationalsocietywhoseprimaryactivityisaboutinformationprocessingmay applytobecomeafullmemberofIFIP,althoughfullmembershipisrestrictedtoone societypercountry.FullmembersareentitledtovoteattheannualGeneralAssembly, Nationalsocietiespreferringalesscommittedinvolvementmayapplyforassociateor correspondingmembership.Associatemembersenjoythesamebenefitsasfullmembers,butwithoutvotingrights.CorrespondingmembersarenotrepresentedinIFIP bodies.Affiliatedmembershipisopentonon-nationalsocieties,andindividualand honorarymembershipschemesarealsooffered.

Moreinformationaboutthisseriesathttp://www.springer.com/series/6102

ChristianDamsgaardJensen • StephenMarsh

TheoDimitrakos • YukoMurayama(Eds.)

9thIFIPWG11.11

InternationalConference,IFIPTM2015

Hamburg,Germany,May26–28,2015

Proceedings

Trust ManagementIX
123

Editors

ChristianDamsgaardJensen TechnicalUniversityofDenmark Lyngby

Denmark

StephenMarsh UniversityofOntario Oshawa,ON

Canada

TheoDimitrakos BTResearch&Innovation

Ipswich

UK

YukoMurayama IwatePrefecturalUniversity

Takizawa

Japan

ISSN1868-4238

ISSN1868-422X(electronic)

IFIPAdvancesinInformationandCommunicationTechnology

ISBN978-3-319-18490-6ISBN978-3-319-18491-3(eBook) DOI10.1007/978-3-319-18491-3

LibraryofCongressControlNumber:2015937744

SpringerChamHeidelbergNewYorkDordrechtLondon © IFIPInternationalFederationforInformationProcessing2015 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartofthe materialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation, broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodologynow knownorhereafterdeveloped.

Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse.

Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbookare believedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsortheeditors giveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforanyerrorsor omissionsthatmayhavebeenmade.

Printedonacid-freepaper

SpringerInternationalPublishingAGSwitzerlandispartofSpringerScience+BusinessMedia (www.springer.com)

Preface

DearReader

WelcometotheIFIPTM2015Proceedings!

Thisvolumecontainstheproceedingsofthe9thIFIPWorkingGroup11.11 InternationalConferenceonTrustManagement.TheconferencewasheldinHamburg, Germany,May26–28,2015.

IFIPTMisatrulyglobalconference,spanningresearch,development,policy,and practicefortheincreasinglyimportantareasoftrustmanagementandcomputational trust.Giventhebreadthofapplicationoftheseareas,andtruetoourhistorical underpinningsestablishedatthe firstIFIPTMconferencein2007,IFIPTM2015 focusedonseveralareas,includingtrustandreputationandmodelsthereof,therelationshipbetweentrustandsecurity,socio-technicalaspectsoftrust,reputation,and privacy,trustinthecloud,andbehavioralmodelsoftrust.

Theconferencereceived28submissionsfromawidevarietyofcountries,including France,Germany,TheNetherlands,UK,Algeria,Norway,Singapore,Greece,Denmark,China,Japan,Malaysia,Luxembourg,Romania,China,USA,Australia,and Canada.Everysubmissionwassubjectedtoathoroughpeerreviewprocess,withat leastthreeandmostoftenfourreviewsperpaper.Followingtheseweacceptedeight longand fiveshortpapers(anacceptancerateforlongpapersof32%).Inaddition, sinceIFIPTMwascolocatedwiththeIFIPSECconference,wesolicitedtwopapers fromSECthatweremoresuitablefortheTrustManagementarea,eachofwhichwas alsoreviewedbyIFIPTMProgramCommitteemembers.Theresultingprogramis broadandwehopestimulatingfortheattendeesandyourself.

IFIPTMalsohostseveryyeartheWilliamWinsboroughCommemorativeAddress inmemoriamofouresteemedcolleagueProf.WilliamWinsborough.Theawardis giventoanindividualwhohassignifi cantlycontributedtotheareasofcomputational trustandtrustmanagement.In2015,theWorkingGroupwaspleasedtohostProf. EhudGudesofBen-GurionUniversityoftheNegev,whokeynotedtheconferenceand providedanextendedabstractwhichcanbefoundintheseproceedings.

Inadditiontopapersandkeynoteaddress,IFIPTMhostedatutorialonidentityand accessmanagementbyProf.AudunJøsangoftheUniversityofOslo,aspecialsession onDataProtection,Privacy,andTransparencyorganizedbyDr.RehabAlnemrfrom HPLabsandDr.CarmenFernández-GagofromUniversityofMálagaandkeynotedby MaritHansen,DeputyChiefofUnabhängigesLandeszentrumfürDatenschutz,Germany.Finally,theconferencehostedaspecialsessiononTrustedCloudEcosystems organizedandchairedbyDr.TheoDimitrakosofBT,fromwhichpapersanda messagefromDr.Dimitrakosareincludedintheseproceedings.

Conferencesaremultiheadedbeasts,andassuchrequireateamofdedicatedpeople totamethem.ToourProgramCommitteeandassociatedreviewers,whodelivered thoughtful,insightfulandverymuchontimereviews,ourthanks.Thisyearwehave beenluckytoworkwithtrulyprofessionalandhelpfulWorkshop,tutorial,Posterand

Demonstration,Publicity,andLiaisonChairs.SinceIFIPTMiscolocatedwithIFIP SEC,thetaskoflocalorganizationandregistrationfellontheIFIPSECteam,notably Dr.DominikHerrmannoftheUniversityofHamburg,towhom,specialthanksfor puttingupwithourfrailties.ThanksalsototheUniversityofHamburgforproviding thefacilities.

Noconferencewouldsucceedwithoutauthors.Toallofthosewhosubmitted,our thanksandcongratulationsforbeingpartofagrowing,important,andvibrantresearch area.Therearemany,manyconferencesforwhichtrustislistedaseitherakeyoran associatedareaofinterest,andwearekeenlyawareoftheapplicabilityoftrustand trustmanagementtoagreatmanyaspectsofcomputersecurity,HumanComputer Interaction,privacy,thesocialsciences,andbeyond.Wecontinuetotrytobuild IFIPTMasacross-disciplinaryconferenceofchoice,andappreciateyoursupport.

Formoreinformationontheworkinggroup,pleasevisit http://www.i fiptm.org/.

Wehopeyouenjoytheconferenceandtheproceedings.

March2015 StephenMarsh ChristianDamsgaardJensen

VIPreface

IFIPTrustManagementIX

9thIFIPWG11.11InternationalConference onTrustManagement,2015 Hamburg,Germany

May26–28,2015

GeneralChairs

TheoDimitrakosSecurityResearchCentre,BTGroupCTO andUniversityofKent,UK

YukoMurayamaIwatePrefecturalUniversity,Japan

ProgramChairs

ChristianDamsgaardJensenTechnicalUniversityofDenmark,Denmark

StephenMarsh UniversityofOntarioInstituteofTechnology, Canada

WorkshopandTutorialChairs

SheikhMahbubHabibTechnischeUniversitätDarmstadt,Germany

Jan-PhilippSteghöferGöteborgUniversity,Sweden

PosterandDemonstrationChairs

DhirenPatelNITSurat,India

AudunJøsangUniversityofOslo,Norway

PanelandSpecialSessionChairs

Jean-MarcSeigneurUniversityofGeneva,Switzerland MasakatsuNishigakiShizuokaUniversity,Japan

PublicityChairs

TimMullerNanyangTechnologicalUniversity,Singapore AnirbanBasuKDDIR&DLaboratories,Japan

GraduateSymposiumChairs

NuritGal-OzSapirAcademicCollege,Israel JieZhangNanyangTechnologicalUniversity,Singapore

LocalOrganizationChair

DominikHerrmannUniversityofHamburg,Germany

ProgramCommittee

RehabAlnemr

HPLabsBristol,UK

ManHoAu HongKongPolytechnicUniversity,HongKong

AnirbanBasu KDDIR&DLaboratories,Japan

ElisaBertino PurdueUniversity,USA

PamelaBriggs NorthumbriaUniversity,UK

DavidChadwick UniversityofKent,UK

PiotrCofta

LynneCoventry NorthumbriaUniversity,UK

FrédéricCuppens TELECOMBretagne,France

TheoDimitrakos SecurityResearchCentre,BTGroupCTO andUniversityofKent,UK

NatashaDwyer VictoriaUniversity,Australia

BabakEsfandiari CarletonUniversity,Canada

RinoFalcone InstituteofCognitiveSciencesandTechnologies, Italy

HuiFang

NanyangTechnologicalUniversity,Singapore

CarmenFernández-GagoUniversityofMálaga,Spain

JosepFerrer UniversitatdelesIllesBalears,Spain

SimoneFischer-HübnerKarlstadUniversity,Sweden

SaraForesti Università degliStudidiMilano,Italy

NuritGal-Oz SapirAcademicCollege,Israel

DieterGollmann HamburgUniversityofTechnology,Germany

StefanosGritzalis UniversityoftheAegean,Greece

EhudGudes Ben-GurionUniversityoftheNegev,Israel

SheikhMahbubHabibCASED/TechnischeUniversitätDarmstadt, Germany

OmarHasan UniversityofLyon,France

PeterHerrmann NTNUTrondheim,Norway

XinyiHuang FujianNormalUniversity,China

RoslanIsmail UniversitiTenagaNasional,Malaysia

ValerieIssarny Inria,France

ChristianDamsgaardJensenTechnicalUniversityofDenmark,Denmark

AudunJøsang UniversityofOslo,Norway

YuecelKarabulut VMware,USA

TracyAnnKosa UniversityofOntarioInstituteofTechnology, Canada

CostasLambrinoudakisUniversityofPiraeus,Greece

GabrieleLenzini

SnT/UniversityofLuxembourg,Luxembourg

JosephLiu MonashUniversity,Australia

VIIIIFIPTrustManagementIX

YangLiu NanyangTechnologicalUniversity,Singapore

JavierLopez UniversityofMálaga,Spain

StephenMarsh UniversityofOntarioInstituteofTechnology, Canada

FabioMartinelli IIT-CNR,Italy

SjoukeMauw UniversityofLuxembourg,Luxembourg

WeizhiMeng InstituteforInfocommResearch(I2R),Singapore

MaxMühlhäuser TechnischeUniversitätDarmstadt,Germany

TimMuller NanyangTechnologicalUniversity,Singapore

YukoMurayama IwatePrefecturalUniversity,Japan

WeeKeongNg NanyangTechnologicalUniversity,Singapore

MasakatsuNishigakiShizuokaUniversity,Japan

ZeinabNoorian UniversityofSaskatchewan,Canada

DhirenPatel

NITSurat,India

GüntherPernul UniversitätRegensburg,Germany

SiniRuohomaa UniversityofHelsinki,Finland

PierangelaSamaratiUniversità degliStudidiMilano,Italy

Jean-MarcSeigneurUniversityofGeneva,Switzerland

MuratSensoy ÖzyeğinUniversity,Turkey

KetilStølen SINTEF,Norway

TimStorer UniversityofGlasgow,UK

MaheshTripunitaraTheUniversityofWaterloo,Canada

ClaireVishik IntelCorporation,UK

IanWakeman UniversityofSussex,UK

ShouhuaiXu UniversityofTexasatSanAntonio,USA

JieZhang NanyangTechnologicalUniversity,Singapore

JianyingZhou InstituteforInfocommResearch(I2R),Singapore

ExternalReviewers

NaipengDongNationalUniversityofSingapore,Singapore

IdaMariaHaugstveitSINTEF,Norway

RaviJhawarUniversityofLuxembourg,Luxembourg

SpyrosKokolakisUniversityoftheAegean,Greece

FranciscoMoyanoUniversityofMálaga,Spain

AidaOmerovicSINTEF,Norway

RubenRios UniversityofMálaga,Spain

AggelikiTsohouIonianUniversity,Greece

DongxiaWangNanyangTechnologicalUniversity,Singapore

YangZhangUniversityofLuxembourg,Luxembourg

IFIPTrustManagementIXIX

Contents

WinsboroughAwardInvitedPaper

Reputation-fromSocialPerceptiontoInternetSecurity...............3 EhudGudes

FullPapers

MathematicalModellingofTrustIssuesinFederatedIdentityManagement ....13 Md.SadekFerdous,GethinNorman,AudunJøsang,andRonPoet

SimpleandPracticalIntegrityModelsforBinariesandFiles............30 YongzhengWuandRolandH.C.Yap

EnablingNAME-BasedSecurityandTrust.........................47 NikosFotiouandGeorgeC.Polyzos

TrustDrivenStrategiesforPrivacybyDesign......................60 ThibaudAntignacandDanielLeMétayer

LightweightPracticalPrivateOne-WayAnonymousMessaging..........76 AnirbanBasu,JuanCamiloCorena,JaideepVaidya,JonCrowcroft, ShinsakuKiyomoto,StephenMarsh,YungShinVanDerSype, andToruNakamura

Privacy-PreservingReputationMechanism:AUsableSolutionHandling NegativeRatings...........................................92 PaulLajoie-Mazenc,EmmanuelleAnceaume,GillesGuette, ThomasSirvent,andValérieVietTriemTong

ObscuringProvenanceConfidentialInformationviaGraphTransformation...109 JamalHussein,LucMoreau,andVladimiroSassone

SocialNetworkCultureNeedstheLensofCriticalTrustResearch........126 NatashaDwyerandStephenMarsh

PredictingQualityofCrowdsourcedAnnotationsUsingGraphKernels.....134 ArchanaNottamkandath,JasperOosterman,DavideCeolin, GerbenKlaasDirkdeVries,andWanFokkink

AnArchitectureforTrustworthyOpenDataServices..................149 AndrewWong,VickyLiu,WilliamCaelli,andTonySahama

ShortPapers

1,2,Pause:LetsStartbyMeaningfullyNavigatingtheCurrentOnline AuthenticationSolutionsSpace.................................165 IjlalLoutfiandAudunJøsang

DataConfidentialityinCloudStorageProtocolBasedonSecret SharingScheme:ABruteForceAttackEvaluation...................177 AlexandruButoi,MirceaMoca,andNicolaeTomai

TheDetailofTrustedMessages:RetweetsinaContextofHealth andFitness...............................................185 NatashaDwyerandStephenMarsh

ReusableDefenseComponentsforOnlineReputationSystems...........195 JohannesSänger,ChristianRichthammer,ArturRösch, andGüntherPernul

ContinuousContext-AwareDeviceComfortEvaluationMethod..........203 JingjingGuo,ChristianDamsgaardJensen,andJianfengMa

SpecialSession:TowardTrustedCloudEcosystems

Foreword:TowardsTrustedCloudEcosystems......................215 TheoDimitrakos

ACloudOrchestratorforDeployingPublicServicesontheCloud – TheCase ofSTRATEGICProject............................................217 PanagiotisGouvas,KonstantinosKalaboukas,GiannisLedakis, TheoDimitrakos,JoshuaDaniel,GéryDucatel, andNuriaRodriguezDominguez

IntegratingSecurityServicesinCloudServiceStores.................226 JoshuaDaniel,FadiEl-Moussa,GéryDucatel,PramodPawar, AliSajjad,RobertRowlingson,andTheoDimitrakos

BuildinganEco-SystemofTrustedServicesviaUserControl andTransparencyonPersonalData..............................240 MicheleVescovi,CorradoMoiso,MattiaPasolli,LorenzoCordin, andFabrizioAntonelli

Security-as-a-ServiceinMulti-cloudandFederatedCloudEnvironments....251 PramodS.Pawar,AliSajjad,TheoDimitrakos,andDavidW.Chadwick

TheRoleofSLAsinBuildingaTrustedCloudforEurope.............262 AnaJuanFerrerandEnricPagesiMontanera

AuthorIndex ............................................277 XIIContents
WinsboroughAwardInvitedPaper

Reputation-fromSocialPerception toInternetSecurity

Ben-GurionUniversity,84105Beer-Sheva,Israel

ehud@cs.bgu.ac.il

Abstract. Reputationisaconceptthatweuseinmanyaspectsofour sociallifeandaspartofourdecisionmakingprocess.Weusereputation inourinteractionwithpeopleorcompanieswedonotknowandweuseit whenwebuymerchandizeorreservearoominahotel.However,reputationplaysalsoanimportantroleintheinternetsocietyandenablesusto establishtrustwhichisessentialforinteractioninthevirtualworld.ReputationhasseveralimportantaspectssuchasAggregation,Identityand Transitivitywhichmakeitapplicableincompletelydifferentdomains. Inthispresentationweshowtheuseoftheseaspectsinseveraldifferent domainsanddemonstrateitwithourownpreviousandcurrentresearch onreputation.

Agoodnameismoredesirablethangreatriches; tobeesteemedisbetterthansilverorgold. Proverbs22:1

1Introduction

Reputationisakeyconceptinoursociallife.Manyofourdaytodaydecisionssuchaswhichbooktobuyorwhichphysiciantoconsultwitharebased onTrust.Thistrustisbasedeitheronourowndirectexperienceorwhensuch directexperienceislacking,onotherpeople(whoseopinionwevalue)direct experience.Howeverwhennosuchdirectorindirectexperienceisavailablewe tendtorelyonanaggregatedopinionofalargesetofpeopleoracommunity whichismanifestedasReputation.Reputationplaysalsoamajorroleinvirtualcommunitiesandsocialnetworks.Attemptstotarnishreputationinsocial networkshavecausedmuchdamagetopeopleinrecentyears(severalcasesof suicidehavebeenreportedasaresultoftarnishedreputation).Somaintaininga goodonlinereputationbecomesacriticalissueforbothpeopleandbusinesses. Theexistenceofeasilyaccessiblevirtualcommunitiesmakesitbothpossibleand legitimatetocommunicatewithtotalstrangers.Suchinteractionhowevermust bebasedontrustwhichisusuallybasedonpersonalexperience.Whensuch experienceisnotreadilyavailable,oneoftenreliesonreputation.Thus,computingreputationtocaptureacommunity’sviewpointisanimportantchallenge. Reputationhasbecomeakeycomponentofseveralcommercialsystemssuch asE-bay[3].Also,quiteafewmodelsfortrustandreputationweredeveloped.

c IFIPInternationalFederationforInformationProcessing2015 C.D.Jensenetal.(Eds.):IFIPTM2015,IFIPAICT454,pp.3–10,2015. DOI:10.1007/978-3-319-18491-3 1

Differentmodelsusedifferentconceptualframeworksincludingsimpleaverage ofratings,bayesiansystems,beliefmodels[11]whichenabletherepresentation ofuncertaintyinrating,flowmodelsinwhichtheconceptoftransitivetrustis centralsuchasEigen-trust[13]andPage-rank[16]andgroup-basedmodelssuch astheKnotmodel[7].Inthispresentationwediscussthreeimportantaspectsof reputationandshowhowtheyareusedindifferentdomains.Whilethefirsttwo domainswediscussinvolvereputationofreal-lifeusers,thethirddomaindeals withabstractentities,internetdomains,yetcomputingandusingreputationin thisdomainissimilartoitsuseinthesocialdomain.

Thefirstaspectwedealwithistheuseofreputationaspartofan Identity Inthesocialdomains,reputationisanimportantpartofapersonidentity,and theidentityofapersondeterminesitspermittedactions.Anexpertprogrammer maygainmoreaccessrightstoanopensourcecodemanagedbysomecompany, asherreputationincreases.Suchrightsmaybereviewormodifycodeatdifferentlevels.OurfirstdomainthenistheAuthorizationdomainandtheuse ofreputationforfine-grainedaccesscontrol.InSect. 2 wepresentsomemodels whichusereputationaspartofauseridentityandconsideritinmakingaccess controldecisions.

Thesecondaspectweexamineis Aggregation.Mostreputationcomputationalmodelsusesomeformofaggregationofratingstocomputethereputation[12].However,suchaggregationisusuallydonewithinasinglecommunity.In real-life,usersmaybeactiveinseveralcommunitiesandtoprotecttheirprivacy, usersmayusedifferentidentitiesindifferentcommunities.Amajorshortcomings isthatusereffortstogainagoodreputationinonecommunityarenotutilized inothercommunitiestheyareactivein.Anothershortcomingistheinabilityof onecommunitytolearnaboutthedishonestbehaviorofsomememberasidentifiedbyothercommunities.Thustheneedarisestoaggregatereputationfrom multiplecommunities.WedevelopedtheCross-CommunityReputation(CCR) modelforthesharingofreputationknowledgeacrossvirtualcommunities[5, 6, 9]. TheCCRmodelisaimedatleveragingreputationdatafrommultiplecommunitiestoobtainmoreaccuratereputation.Itenablesnewvirtualcommunities torapidlymaturebyimportingreputationdatafromrelatedcommunities.The useofAggregationintheCCRmodelisdiscussedinSect. 3

Thethirdaspectwediscussis Transitivity,animportantpropertyoftrust whichhasimplicationsonthecomputationofreputation.Itenablesustocomputereputationnotonlyfromourownexperienceorourfriendsexperience butalsofromour“friendsoffriends”experience,etc.Severalflowmodelsfor computingreputationwhilepracticingthetransitivityproperty,havebeenpublished,includingEigen-trust[13]andPage-rank[16].Ouruniquecontribution hereisintransferringtheseideastothecomputationofInternetdomainsreputation.Today’sinternetworldisfullofthreatsandmalware.Hackersoftenuse variousdomainstospreadandcontroltheirmalware.Thedetectionofthesemisbehavingdomainsisdifficultsincethereisnotimetocollectandanalyzetraffic datainreal-time,thustheiridentificationaheadoftimeisveryimportant.We usetheterm domainreputation toexpressameasureofourbeliefthatadomain

4E.Gudes

isbenignormalicious.ComputingdomainreputationbyusingtheTransitivity propertyandaFlowalgorithmwasinvestigatedbyus[15]andwillbediscussed inSect. 3.

2Identity-ReputationandAccessControl

Conventionalaccesscontrolmodelslikerolebasedaccesscontrolaresuitable forregulatingaccesstoresourcesbyknownusers.However,thesemodelshave oftenfoundtobeinadequateforopenanddecentralizedmulti-centricsystems wheretheuserpopulationisdynamicandtheidentityofallusersarenotknown inadvance.Forsuchsystems,theremustbe,inadditiontouserauthentication,sometrustmeasureassociatedwiththeuser.Suchtrustmeasurecanbe representedbytheuserreputationasoneattributeofitsidentity.Chakraborty andRay[2]presentedTrustBAC,atrustbasedaccesscontrolmodel.Itextends theconventionalrolebasedaccesscontrolmodelwiththenotionoftrustlevels. Usersareassignedtotrustlevelsinsteadofrolesbasedonanumberoffactors likeusercredentials,userbehaviorhistory,userrecommendationetc.Trustlevelsareassignedtoroleswhichareassignedtopermissionsasinrolebasedaccess control.InTrustbac,whenthereputationofauserdecreasesbecauseofpast actions,itsassignmenttotheoriginalrolemaynotbevalidanymoreandanew rolewithlesspermissionsisassignedbythesystem.Anexampleofsuchscenario inthedigitallibrarydomainisgivenin[2].Theswitchingofrolesmaynotbe desirableinallcases.Inamedicaldomainforexample,aphysicianwithless reputationmaynotloseitsroleas“doctor”butmayloseinsteadsomeofher permissions.Thisdynamicassignmentofpermissionsforthesamerole,basedon theuserreputationmaybemuchmoreflexibleandcanpreventtheproliferationoftoomanyroles.In[14]wedefinethisdynamicmodelformallyandshow adetailedexampleofitsoperationinthesoftwaredevelopmentdomain.The mainobservationofthisisthatwhenoneconsidersreputationaspartofthe useridentity,onecansupportmuchmoreflexiblerole-basedmodelswithout theneedtoincreasesignificantlythenumberofrolesinthesystem.

3AggregationandCrossCommunityReputation

Inthissectionwebrieflydescribethewayreputationisaggregatedfromseveral communitiesusingtheCCRmodel[5, 9].TheCCRmodeldefinesthemajorstages requiredtoaggregatethereputationofacommunitymemberwiththereputation ofthatmemberinothercommunities.Thefirststagedeterminestheconfidence onecommunityhasinanotherasapreconditionforreceivingreputationinformationfromthelatter.Thesecondstageinvolvestheconversionofreputationvalues fromthedomainvaluesofonecommunitytothoseoftheother.Inthethirdstage, amatchingprocedureiscarriedoutbetweenthesetsofattributesusedbytheparticipatingcommunitiestodescribereputation.Asanexample,supposethereare twosportcommunitiesinwhichacommentatorisactive,oneforBasketball,the

Reputation-fromSocialPerceptiontoInternetSecurity5

otherforFootball.AssumethatBobacommentatorlikestoimport(andaggregate)hisreputationfromthefootballcommunityintothebasketballcommunity. Thefirststageconsidersthegeneralconfidencethatbasketballcommunitymembershaveforreputationcomputedinthefootballcommunity.Thesecondstage considersthestatisticaldistributionofreputationvaluesinthetwocommunities andapplytherequiredtransformation(e.g.,averygoodratinginonecommunitymayonlybeconsidered“good”intheother).Thethirdstagemapsthespecificattributesthatareusedtocomputethereputationinthetwocommunities (e.g.,theattribute“predictionaccuracy”inthefootballcommunitymaybepartiallymappedtotheattribute“generalreliability”inthebasketballcommunity). Adetailedmathematicalmodelwhichexplainstheprocessofthemappingand aggregationofCCR,isdescribedin[5].TheCCRmodelwasimplementedasthe TRICsoftware.TRICisconcernedprimarilywithaggregatingdifferentreputationmechanismsacrosscommunitiesandwithprotectinguserrightstoprivacy andcontroloverdataduringthisaggregation.TheCCRcomputationprocess[5] beginswhena requestingcommunity thatwishestoreceiveCCRdataregarding oneofitsusers,sendsarequesttorelevant respondingcommunities.Communities thathavereputationdataoftheuserandarewillingtosharetheinformationreply withtherelevantreputationdata.Thereceiveddataisaggregatedandassembled intoanobjectcontainingtheCCRdataoftheuserinthecontextoftherequesting community.ThisprocessisillustratedinFig. 1.

Fig.1. RequestforCCRscenario:(1):ArequestingcommunitysendsTRICarequest fortheCCRofacommunitymember;(2):TRICcompilesarequestand(3)submitsit toallpotentialrespondingcommunities;(4):Respondingcommunitiessubmitareputationobjectofthememberatsubject;(5):TRICprocessesallreputationobjectsand compilesaCCRobject;(6):TRICsendstheCCRobjecttotherequestingcommunity

Oneoftheimportantgoalsassociatedwithsharingreputationbetweencommunitiesisdealingwithprivacy.WithintheCCRmodel,weidentifiedthree majorprivacyconcernsthatarenotpresentorthatarelesssignificantinsingle

6E.Gudes

communitydomains.FirstUnlinkabilityisaprimaryconcernraisedbytheCCR model.Althoughweaimtocomputeauser’sCCRfromseveralcommunities,we providethemeanstodosowithoutcompromisingtheuser’sanonymityineach communityandwhileupholdingtherequirementofunlinkabilitybetweenthe communities.Controllingthedisseminationofreputationinformationisanother privacyrequirement.Wepresentapolicy-basedapproachthatenablesboththe usersandthecommunitiestohavecontroloverthedisseminationofreputationdata.Thethirdprivacyissueweaddressisthetradeoffbetweenprivacy andtrust.WesuggestthetransparencymeasureforevaluatingCCRobjects. Toattainahightransparencyrank,membersareencouragedtodisclosetheir reputation-relatedinformationwheneveritisclearthatdisclosingtheirinformationispreferableandmorevaluabletothemthanthepotentialimpairmentof theirprivacy.TheissueofPrivacywithintheCCRmodelisdiscussedin[8].

4TransitivityandComputingDomainsReputation

Aswasdiscussedearlier,computingdomainreputationandidentifyingsuspiciousdomainsisaveryimportantprobleminInternetsecuritytoday.Ourapproachtotheproblem[15]usesagraphofdomainsandIPswhichisconstructed frommappinginformationavailableinDNSlogrecords.TheDomainNameService(DNS)mapsdomainnamestoIPaddressesandprovidesanessentialservice toapplicationsontheinternet.ManybotnetsuseaDNSservicetolocatetheir nextCommandandControl(C&C)site.Therefore,DNSlogshavebeenusedby severalresearcherstodetectsuspiciousdomainsandfiltertheirtrafficifnecessary.Wetakethefamousexpression TellmewhoyourfriendsareandIwilltell youwhoyouare,motivatingmanysocialtrustmodels,intotheinternetdomains world.Thusadomainthatisrelatedtomaliciousdomainsismorelikelytobe maliciousaswell.ThisTransitivitypropertymotivatestheuseofaFlowalgorithm.AlthoughDNSdatawasusedbyseveralresearchersbeforetocompute domainreputation(see[1]),in[15]wepresentanewapproachbyapplyingaflow algorithmontheDNSgraphtoobtainthereputationofdomainsandidentify potentiallymaliciousones.Computingreputationfordomainsraisesseveralnew difficulties:

–Ratinginformationifexists,issparseandusuallybinary,adomainislabeled either“white”or“black”.

–Staticsourceslikeblacklistsandwhitelistsareoftennotup-to-date.

–Thereisnoexplicitconceptoftrustbetweendomainswhichmakesitdifficult toapplyafloworatransitivetrustalgorithm.

–Reputationofdomainsisdynamicandchangesveryfast.

Thesedifficultiesmaketheselectionofanadequatecomputationalmodelfor computingdomainreputationachallengingtask.Ourapproachisbasedona flowalgorithm,commonlyusedforcomputingtrustinsocialnetworksandvirtual communities.Wearemainlyinspiredbytwomodels:theEigentrustmodel[4] whichcomputestrustandreputationbytransitiveiterationthroughchainsof

Reputation-fromSocialPerceptiontoInternetSecurity7

trustingusersandthemodelbyGuhaetal.[10]whichcombinestheflowoftrust anddistrust.Themotivationforusingaflowalgorithmistheassumptionthat IPsanddomainswhichareneighborsofmalwaregeneratingIPsanddomains,are morelikelytobecomemalwaregeneratingaswell.Weconstructagraphwhich reflectsthetopologyofdomainsandIPsandtheirmappingsandrelationships anduseaflowmodeltopropagatetheknowledgereceivedintheformofblack list,tolabeldomainsinthegraphasmaliciousorsuspecteddomains.Although wedonotclaimthateverydomain(orIP)connectedtoamaliciousdomainin ourgraphismalicious,ourresearchhypothesisisthatsuchdomains(IPs)have ahigherprobabilitytobecomemalicious.Ourpreliminaryexperimentalresults supportthishypothesis.

ThemaininputtotheflowalgorithmistheDomains/IPsgraph.Thisgraphis builtfromthefollowingsources:(1)A-records:adatabaseofsuccessfulmappings betweenIPsanddomains,collectedfromalargeISPoverseveralmonths.These mappingbasicallyconstructtheedgesbetweenDomainsandIPs.(2)Whois:a queryandresponseprotocolthatiswidelyusedforqueryingdatabasesthatstore theregisteredusersorassignersofanInternetresource.Thisdatabasegroups IPswhichhavesimilarcharacteristicsandisthereforethebaseforIPtoIPedges. InadditionthereareDomaintoDomainedgeswhicharerelatedtosimilarity betweendomainnames.(3)Feed-framework:alistofmaliciousdomainswhich iscollectedoverthesameperiodoftimeasthecollectedA-records.Thislist isusedastheinitial“malicious”domainsset.(4)Alexa:Alexadatabaseranks websitesbasedonacombinedmeasureofpageviewsanduniquesiteusers.The initial“benign”domainsisderivedfromthislist.(5)VirustTotal:awebsitethat providesfreecheckingofdomainsforvirusesandothermalware.Weuseitto testourresultsaswillbedescribedbelow.Themostdifficultpartinconstructing theDomain/IPgraphisassigningtheweightontheedges,sincetheweightis proportionaltotheamountofflowontheedge.Wetestedseveralmethodsto assignweightswhichconsidertopologiesofthegraphandotherfactors,see[15]. OncetheDNSgraphisbuiltandthesetsof“benign”and“malicious”domains areextracted,thealgorithmcanbeperformed.Theentireprocessisdepicted inFig. 2

TheflowalgorithmmodelstheideathateveryIPanddomaindistributetheir reputationtoIPsordomainsconnectedtothem.Thisisdoneiterativelyand thereputationineachiterationisaddedtothetotalreputationofadomainor IP,withsomeattenuationfactor.Theattenuationfactorisameanstoreduce theamountofreputationonevertexcangainfromavertexthatisnotdirectly connectedtoitbytransitivity.Theflowalgorithmisexecutedseparatelyto propagategoodreputationandbadreputationandthenthetworeputation valuesarecombinedinseveralmannersresultingwithseveralvariationsofthe algorithm(seedetailsin[15].)

Theimportantcontributionofthesealgorithmsistheirabilitytocorrectly predictfuturemaliciousdomains.Althoughnotallmaliciousedomainsareidentified,asignificantamountisdiscovered.Inoneoftheexperimentsweused DNSlogsovera3monthsperiodfromwhichalargeDomain-IPgraphwas

8E.Gudes

Fig.2. Theprocessforcomputingthescore:(1)Createthegraphandassignweights representedasmatrix;(2)Createtheinitialvectorusedforpropagation;(3)Combine thematrixandthevectortoexecutetheflowalgorithm;(4)Getthefinalscores.

constructedwithnearlyonemillionnodes,andtheflowalgorithmwasapplied toit.Theresultswerethatoutofthetop1000highlysuspecteddomains,30% werefoundtobeknownmalicious(usingVirusTotal),whileinarandomsetof 1000domainsonly0.9%wereknownasmalicious.

5Conclusions

Reputationisakeyconceptinmakingdecisionsinoursociallife.Inthispaper wehavediscussedthreekeyaspectsofreputation:Identity,Aggregationand Transitivitywhichareimportantwhenmigratingtheconceptofreputationfrom onedomaintoanother.Thiswasshownbybrieflyreviewingseveralresearch papersofours.Themainconclusionisthatreputationplaysamajorroleina widerangeofdomainsbesidethesocialarenadomain.

References

1.Antonakakis,M.,Perdisc,R.,Dagon,D.,Lee,W.,Feamster,N.:Buildinga dynamicreputationmodelforDNS.In:USENIXSecuritySymposium,pp.273–290 (2010)

2.Chakraborty,S.,Ray,I.:TrustBAC:integratingtrustrelationshipsintotheRBAC modelforaccesscontrolinopensystems.In:Proceedingsofthe11thACMsymposiumonAccessControlModelsandTechnologies(SACMAT2006),pp.49–58. ACM,NewYork(2006)

3.Dellarocas,C.:Analyzingtheeconomicefficiencyofebay-likeonlinereputation reportingmechanisms.In:ACMConferenceonElectronicCommerce,pp.171–179 (2001)

4.Kamvar,S.D.,Schlosser,M.T.,Garcia-Molina,H.:Theeigentrustalgorithmfor reputationmanagementinP2Pnetworks.In:WWW,pp.640–651(2003)

5.Gal-Oz,N.,Grinshpoun,T.,Gudes,E.:Sharingreputationacrossvirtualcommunities.J.Theor.Appl.Electr.Commer.Res. 5(2),1–25(2010)

Reputation-fromSocialPerceptiontoInternetSecurity9 Graphconstruction IPData Domains Arecords Algorithm Initial Good Domains InitialBad Domains Vector Final 1 23 3 4

6.Gal-Oz,N.,Grinshpoun,T.,Gudes,E.,Meisels,A.:Cross-communityreputation: policiesandalternatives.In:ProceedingsoftheInternationalConferenceonWeb BasedCommunities(IADIS-WBC2008)(2008)

7.Gal-Oz,N.,Gudes,E.,Hendler,D.:Arobustandknot-awaretrust-basedreputationmodel.In:Proceedingsofthe2ndJointiTrustandPSTConferenceson Privacy,TrustManagementandSecurity(IFIPTM2008),Trondheim,Norway, June2008,pp.167–182(2008)

8.Gal-Oz,N.,Grinshpoun,T.,Gudes,E.:Privacyissueswithsharingreputation acrossvirtualcommunities.In:Proceedingsofthe2011InternationalWorkshop onPrivacyandAnonymityinInformationSociety,PAIS2011,Uppsala,Sweden, p.3.March2011

9.Grinshpoun,T.,Gal-Oz,N.,Meisels,A.,Gudes,E.:CCR:amodelforsharingreputationknowledgeacrossvirtualcommunities.In:Proceedingsofthe IEEE/WIC/ACMInternationalConferenceonWebIntelligenceandIntelligent AgentTechnology(WI2009),pp.34–41.IEEE(2009)

10.Guha,R.,Kumar,R.,Raghavan,P.,Tomkins,A.:Propagationoftrustanddistrus. In:WWW,pp.403–412(2004)

11.Jøsang,A.,Ismail,R.:Thebetareputationsystem.In:Proceedingsofthe15th BledElectronicCommerceConference,vol.160,pp.17–19(2002)

12.Jøsang,A.,Ismail,R.,Boyd,C.:Asurveyoftrustandreputationsystemsfor onlineserviceprovision.Decis.SupportSyst. 43(2),618–644(2007)

13.Kamvar,S.,Schlosser,M.,Garcia-Molina,H.:TheeigentrustalgorithmforreputationmanagementinP2Pnetworks.In:Proceedingsofthe12thInternational ConferenceonWorldWideWeb(WWW2003),pp.640–651.ACM(2003)

14.Lavi,T.,Gudes,E.:AdynamicreputationbasedRBACmodel.Report,TheOpen UniversityRaananaIsrael(2015)

15.Mishsky,I.,Gal-Oz,N.,Gudes,E.:Aflowbaseddomainreputationmodel.Report, Ben-GurionUniversity,Beer-Sheva,Israel(2015)

16.Parreira,J.X.,Donato,D.,Michel,S.,Weikum,G.:Efficientanddecentralized pagerankapproximationinapeer-to-peerwebsearchnetwork.In:Proceedingsof the32ndInternationalConferenceonVeryLargeDataBases,pp.415–426(2006)

10E.Gudes

FullPapers

MathematicalModellingofTrustIssues inFederatedIdentityManagement

Md.SadekFerdous1(B) ,GethinNorman1 ,AudunJøsang2 , andRonPoet1

1 SchoolofComputingScience,UniversityofGlasgow, GlasgowG128QQ,Scotland {sadek.ferdous,gethin.norman,ron.poet}@glasgow.ac.uk

2 DepartmentofInformatics,UniversityofOslo,0316Oslo,Norway josang@mn.uio.no

Abstract. Withtheabsenceofphysicalevidence,theconceptoftrust playsacrucialroleintheproliferationandpopularisationofonlineservices.Infact,trustistheinherentqualitythatbindstogetherallinvolved entitiesandprovidestheunderlyingconfidencethatallowsthemtointeractinanonlinesetting.TheconceptofFederatedIdentityManagement (FIM)hasbeenintroducedwiththeaimofallowinguserstoaccessonline servicesinasecureandprivacy-friendlywayandhasgainedconsiderablepopularitiesinrecentyears.Beingatechnologytargetedforonline services,FIMisalsoboundbyasetoftrustrequirements.Eventhough therehavebeennumerousstudiesonthemathematicalrepresentation, modellingandanalysisoftrustissuesinonlineservices,acomprehensivestudyfocusingonthemathematicalmodellingandanalysisoftrust issuesinFIMisstillabsent.Inthispaperweaimtoaddressthisissueby presentingamathematicalframeworktomodeltrustissuesinFIM.We showhowourframeworkcanhelptorepresentcomplextrustissuesin aconvenientwayandhowitcanbeusedtoanalyseandcalculatetrust amongdifferententitiesqualitativelyaswellasquantitatively.

Keywords: Trust · FederatedIdentityManagement · Mathematical modelling

1Introduction

Unlikethebrickandmortarworld,thephysicalevidenceandvisualcuesthat canbeusedtoestablishtrustandgainconfidencearelargelyabsentinonline services.Despitethis,thepopularityofonlineserviceshasgrownexponentially inthelastdecadeorso.Theconceptoftrustplayedacrucialroleinpopularisingonlineservices.Infact,trustistheinherentqualitythatbindstogether allinvolvedentitiesandprovidestheunderlyingconfidencethatallowsthemto interactinanonlineservice.Themathematicalmodellingandanalysisofdifferenttrustrequirementsinonlineservicesareaboundandisawellestablished researcharea.Suchamodelhelpstoexpressandtoreasonwithtrustissuesina c IFIPInternationalFederationforInformationProcessing2015 C.D.Jensenetal.(Eds.):IFIPTM2015,IFIPAICT454,pp.13–29,2015. DOI:10.1007/978-3-319-18491-3 2

formalwaywhichcanultimatelyhelptocreatenovelwaysfordeterminingtrust amonginvolvedentities.

TheconceptofFederatedIdentityManagement(FIM)hasbeenintroduced toeasetheburdenofmanagingdifferentonlineidentitiesandtoallowusers toaccessonlineservicesinasecureandprivacy-friendlyway[1].FIMoffers anarrayofadvantagestodifferentstakeholdersandhasgainedconsiderable popularitiesinrecentyears.Beingatechnologytargetedfortheonlinesetting, FIMisalsoboundbyasetoftrustrequirements.Surprisingly,themathematical representation,modellingandanalysisofdifferenttrustrequirementsofFIM havereceivedlittleattentionsofar.Theaimofthispaperistofillthisgap.

Here,wepresentacomprehensivemathematicalframeworkconsideringdifferenttrustaspectstargetedforFIM.Indoingso,weshowhowourframework canformallyexpresstrustinFIMandhowsuchexpressionscanbeusedto analyseandevaluatetrustqualitativelyandquantitatively.Themaincontributionsofthepaperare:

1.Inspiredbythenotationoftrustpresentedin[14],wepresentanotationto expresstrustbetweendifferententitiesinFIM.

2.Weusethisnotationtodevelopthefirstmathematicalframeworktomodel, analyseandderivetrustindifferenttypesofidentityfederations.

3.WeexploretrusttransformationsresultingfrominteractionsinFIM.

4.Finally,wepresentasimplemethodtoevaluatetrustquantitativelyinFIM.

Thepaperisstructuredasfollows.Section 2 providesabriefintroductiontoFIM andtherequiredtrustissuesinthissetting.Section 3 introducesthenotation andtheinteractionmodelthatwillbeusedinourframework.Thetrustissues indifferenttypesofidentityfederationsaremodelledinSects. 4 and 5.Weshow howtrusttransformationsoccurwithindifferentfederationsusingourframework inSect. 6 andhowtrustcanbecalculatedquantitativelyinSect. 7.Section 8 discussestherelatedworkandfinallySect. 9 concludesthepaper.

2Background

Inthissection,weprovideabriefintroductiontoFIM,todifferentaspectsof trustingeneralandtotrustissuesinFIMspecifically.

FederatedIdentityManagement. IdentityManagementconsistsoftechnologiesandpoliciesforrepresentingandrecognisingentitiesusingdigitalidentifiers withinaspecificcontext[7].Asystemthatisusedformanagingtheidentity ofusersiscalledanIdentityManagementSystem(IMS).EachIMSincludes thefollowingtypesofparties: ServiceProviders(SPs) or RelyingParties(RPs) -entitiesthatprovidesservicestousersorotherSPs, Identity Providers(IdPs) -entitiesthatprovidesidentitiestouserstoenablethem toreceiveservicesfromSPsand Clients/Users -entitiesthatreceiveservices fromSPs.AmongdifferentIMS,theFederatedIdentityManagement(FIM)has gainedmuchattentionandpopularity.

14M.S.Ferdousetal.

TheFederatedIdentityManagementisbasedontheconceptofIdentityFederation.AfederationwithrespecttoIdentityManagementisabusinessmodel inwhichagroupoftwoormoretrustedpartieslegallybindthemselveswith abusinessandtechnicalcontract[1, 17].Itallowsausertoaccessrestricted resourcesseamlesslyandsecurelyfromotherpartnersresidingindifferentIdentityDomains.Anidentitydomainisthevirtualboundary,contextorenvironmentinwhichanidentityofauserisvalid[17].SingleSignOn(SSO)isthe capabilitythatallowsuserstologintoonesystemandthenaccessotherrelated butautonomoussystemswithoutfurtherlogins.Italleviatestheneedtologin everytimeauserneedstoaccessthoserelatedsystems.Agoodexampleisthe GoogleSingleSignOnservicewhichallowsuserstologinaGoogleservice,e.g., Gmail,andthenallowsthemtoaccessotherGoogleservicessuchasCalendar, Documents,YouTube,Blogsandsoon.

(a)Type1.(b)Type2.

Fig.1. Federatedidentitydomain.

AfederatedidentitydomaincanbeformedbyoneIdPinanidentitydomain andanumberofSPswitheachSPresidinginaseparateidentitydomain(Type 1inFig. 1(a)).Severalfederatedidentitydomainscanbecombinedtoformalarger federatedidentitydomainwhereeachsmallerfederateddomainisofType1(Type 2inFig. 1(b)).AType2federationallowsanIdPofaType1federationtodelegatetheauthenticationtasktoanotherIdPinadifferentType1federation.To enablethis,bothIdPsneedtoactasbothIdPsandSPs.Theissueoftrustisa fundamentalconceptinFIMasdifferentautonomousbodiesneedtotrusteach otherinsidethefederation.Suchpartiesinsideafederationaresaidtoformthe so-calledCircleofTrust(CoT).

Afederationcanbeoftwotypesdependingonhowitiscreated.Thetraditionalfederation,alsocalleda StaticFederation,iswherethefederationis createdattheadminlevelandisboundwithalegalcontractusingaspecified setofadministrativeprocedures.Ontheotherhand,ina DynamicFederation anyuser,notonlyadministrators,cancreatethefederationinadynamicfashion withoutadministrativeinterventionoralegallybindingcontract[3].

Trust. Theconceptoftrustandtrustmanagementinthesettingofonline servicesisawidelystudiedtopicandhasbeendefinedinnumerousways.For thepurposeofthispaper,weusethefollowingdefinitiontakenfrom[11]which wasoriginallyinspiredby[13].

MathematicalModellingofTrustIssuesinFIM15

“Trustistheextenttowhichonepartyiswillingtodependonsomethingor somebodyinagivensituationwithafeelingofrelativesecurity,eventhough negativeconsequencesarepossible.”

Thedefinitiongivesadirectionalrelationshipbetweentwoentities:thefirstis regardedasthe Trustor andthesecondthe Trustee.Thetrustorandtrusteecan beanyentity,however,inthescopeofthispaper,onlythoseinvolvedinFIM willbeconsidered(i.e.users,IdPsandSPs).Thepairwisetrustrelationswe considerareuser-IdP,user-SP,IdP-SPandIdP-IdPwhichisinlinewithcurrent IMSsettingandtherelationshipsthatoccurinsideafederation.

Trustcanbeoftwotypes:DirectTrust(DT )andIndirectTrust(IT )[12]. Directtrustsignifiesthatthereexistsatrustrelationshipbetweentheentities basedonfirsthandexperienceandevidence.Ontheotherhand,indirecttrust, alsoknownasTransitiveTrust,isatrustrelationshipbetweentwoentitiesbased onreferralfromoneormoreintermediatethirdparties.

Everytrustrelationshiphasascopethatsignifiesthespecificpurposeor contextintowhichthattrustrelationshipisvalid.Thetruststrength(alsoknown asthetrustdegree)signifiestheleveloftrustatrustorhasoveratrustee[14]. Thetypeandvalueusedtodefinetheleveloftrustwillvarydependingonthe trustscopesaswell.Trustcanbedefinedas MutualTrust onlyifthereisa bi-directionaltrustrelationshipwiththesametrusttype,scopeandstrength betweenthecorrespondingentities.Insuchcase,bothentitiescanactasthe trustorandthetrustee.Trustoftenexhibitsthetransitivityproperty[11]:ifan entity A trustsanotherentity B and B trustanotherentity C,atrustrelation canbederivedbetween A and C.Toderivesuchatransitivetrustrelation,the trustscopemustbesame.Thetrusttransformationistheprocesswhenatrust relationshipbetweentwoentitieschangesduetothechangeoftruststrength whilethetrusttyperemainsthesame.Suchatransformationoccursnormally fortworeasons:(i)whenthetrustisderivedfollowingthetransitivityproperty and(ii)whenoneentityinteractswithanotherentitytoperformacertainaction whichultimatelytriggersthechangeinthetruststrength.Thetransformation canbepositive,meaningthenewtruststrengthishigherthanwhatwasbefore, orcanbenegative,meaningthenewtruststrengthislowerthanwhatwas before.

Atrustwithasinglescopecanbedefinedasatomictrust.Compoundtrust canbedefinedasthecombinedtrustofseveraldifferentatomictrustswherethe trustor,trusteeandthetrustdirectionandstrengthbetweenthemremainthe same.Thecompoundtrustwillalsohavethesametrustdirectionandstrength.

TrustIssuesinIdentityManagement.

Theissueoftrustisafundamental conceptinFIMasdifferentparticipatingorganisationsneedtotrusteachother insidethefederationatasufficientleveltoallowthemtoexchangeandtrust userinformation.Wewillconsidersuchtrustissuesusingtwoseparateinstances.

Thefirst,called HighLevel trust,istheabstractleveloftrustthatisassumed betweenfederatedentities(IdPsandSPs)inafederation.Thisleveloftrustis commonintheexistingliteratureonFIM.Forexample,itiscommontoexpress thattwoentitiestrusteachotheriftheybelongtothesameCoT.Insuchan

16M.S.Ferdousetal.

expression,thetrustistreatedatanabstractlevelandisusedmostlytosignify theirarchitecturalrelationinsideafederation.

Thesecond,called Fine-grained trust,isadetailedexpressionoftrustincludingthescopebetweenentities(includingusers)inafederation.Theexpression may(optionally)includeatrusttypeorstrength.Inspiredbytherequirements outlinedin[8, 12],theauthorsin[2]haveoutlinedasetoffine-grainedtrust requirementsinthetraditionalfederationwhichareapplicableforbothType1 andType2federations.Wewillusetheirrequirementstorepresentfine-grained trustsinSect. 4

Trustinadynamicfederationismodelledusingthreeclassesofentities[3]: FullyTrusted entitiesareIdPsandSPsinthetraditionalSAML(Security AssertionMarkupLanguage)federationwhichhavealegalcontractbetween them[18]; Semi-trusted entitiesareSPsinadynamicfederationthathave beenaddeddynamicallytoanIdPinsidethefederationunder someconditions withoutacontractandtowhomanyuseroftheIdPhasagreedtoreleasea subsetofherattributesand Untrusted entitiesareIdPsandSPsinadynamic federationwhichhavebeenaddeddynamicallyunder someconditions without acontract.Adetaileddiscussionoftheseclassescanbefoundin[3].

3Notation

Inthissectionwewillintroducethenotationthatwillbeusedtobuildupthe model.Weuse E todenotethesetofentities,with U thesetofusers, SP the setofserviceprovidersand IDP thesetofidentityproviders.Sinceeachuser, SPandIdPisalsoanentity,wehave E = U ∪ IDP ∪ SP .Inaddition, F denotes thesetoffederationsandwillusesubscriptfrom F todefinethecontextsof entities(i.e.thefederationinwhichtheybelong).Forexample, Ef willbeused todenotethesetsofentitiesinafederation f .Weuse T todenotethesetoftrust types.Asexplainedabove,weconsidertwotypesoftrust:directtrust(denoted by DT )andindirecttrust(denotedby IT ).Therefore, T = {DT , IT }.

Weuse S forthesetoftrustscopes.Differenttrustscopescanbedefined dependingonthetrustrequirements.Weconsiderthefollowingtrustscopesfor FIMbasedonthefine-grainedtrustrequirementsof[2]:

– REG istrustintheimplementationoftheregistrationprocess;

– STO istrustinsecureattributestorage;

– AUTHN istrustintheimplementationoftheauthenticationmechanism;

– AP istrustinallowingtheuseofanonymousorpseudonymousidentifiers;

– CONSENT istrustinthereleaseofonlythoseattributesconsentedto;

– ABU isthetrustthatanentitywillnotabuseattributesreleasedtoit;

– CARE isthetrustanentityhandlesherattributeswithadequatecare;

– HON isthetrustthatanentityprovidesattributevalueshonestly;

– ACDA isthetrustthatanentityadherestotheagreedpoliciesandprocedures duringaccesscontrolanddelegatedaccess;

– SRV isthetrustinserviceprovisioning;

– MIN -ATT isthetrustthatanentityrequestsonlyminimalattributes;

MathematicalModellingofTrustIssuesinFIM17

– REL isthetrustinanentitycorrectlyreleasingattributes; – ND isthetrustinanentityadheringtothenon-disclosureofattributes; – FED istrustbetweenfederatedentities.

WeconsiderthefollowingtypesoftruststrengthsinFIM.

SubjectiveTrust. ThisdefinesthesubjectivetrustausermayhaveinIdPs andSPsinafederationandwillbedenotedwith conf .Itcanhavedifferent levels,however,wehaveoptedforthreelevels: LOW(L),MED(M),HIGH(H). LevelofAssurance(LoA). Thisdefinesthetruststrengthbetweenfederated IdPsandSPsandisusedduringserviceprovisioning.ItisbasedontheNIST LoAguidanceof1to4whereLevel1canbeusedtomodelthelowesttrustand Level4thehighest[15].Itwillbedenotedas loa withvaluesfrom1to4. FederationTrust. ThelasttypeconcernsthetruststrengthbetweenfederatedIdPsandSPswithrespecttotheirarchitecturalrelations.Itisdenoted with fed -trust andcantakefourdifferentvalues: UNTRUSTED(UT), SEMITRUSTED(ST), RESTRICTED-TRUSTED(RT) and FULLY-TRUSTED(FT). Thelowesttruststrength UT meansatrustordoesnottrustatrusteeatalland isassociatedbetweenentitiesfederatedinadynamicfashionorbetweenentities inatransitivetrustinstaticfederations(seebelow).Thestrength ST means atrustortrustsatrusteeuptoacertainlevel.Anexampleisthetruststrength betweenadynamicallyfederatedIdPandanSPandthefactthattheIdPmay notwantreleasesensitiveattributestotheSPastherearenoformalagreement betweenthem.Thestrength RT ishigherthan ST,butlowerthan FT.Sucha strengthisexhibitedwhenthetrustrelationshipbetweenatrustorandtrusteeis derivedusingtransitivityandthetrustormaynotfullytrustthetrusteeasthere arenoformalagreementsbetweenthem.Thestrength FT signifiesthehighest strengthandisexhibitedwhenthetrustorandtrusteearepartofatraditional federation.Thefederationtruststrengthsareranked:

UT < ST < RT < FT

Toindicateanentity e1 ∈ Ef (thetrustor)has t ∈ T trustoveranentity e2 ∈ Ef (thetrustee)inafederation f ∈F withatrustscopeof s ∈ S andthetrust strengthof v ,wewillusethefollowingnotation,inspiredby[14]:

e1 t : s → v e2

where v representsthetruststrength(either conf , loa or fed -trust ).Toexpress thesametrust t betweentwoentities e1 and e2 withsametruststrength v ina numberofdifferentscopes, s1 ,...,sn ,weextendthenotationto:

e1 t : {s1 ,...,sn } → v e2

Ifthereexistsamutualtrust(t)betweentwoentitiesinthesametrustscope (s)withthesametruststrength(v ),weusethenotation:

e1 t : s ←−−−−→ v e2

18M.S.Ferdousetal.

3.1InteractionModel

Toenableaprotocolflowinafederation,eachentityinteractswithanother entityinordertoperformanactionatanotherentity.Auserinteractingwith anIdPtoauthenticateherselfbyprovidinganidentifier(e.g.username)and acredential(e.g.password)isexampleofaninteraction.Interactionbetween entitiestoperformanactioncancausethetrustbetweentheinvolvedentities totransform.Theinteractionmodelconsistsoftheactionsthatanentitycan performatanotherentityinafederation.Suchinteractionsmustbecarriedout usingacommunicationchannel.Wewillusethenotation CHANNEL todefine thesetofchannels.Twotypesofchannelswillbeconsidered:securechannels, denoted SC ,modelsecureHTTPSconnectionswhereasunsecuredchannels, denoted UC ,modelunsecuredHTTPconnections.

Todenoteaninteractionthatrepresentsanentity e1 performsaction a at entity e2 usingcommunicationchannel c,wewillusethefollowingnotation: c(e1 a e2 ).Therecouldbemanyinteractionsinafederation,however,tothe scopeofthispaper,werestrictattentiontothefollowinginteractions:

– c(u RG idp )representinguser u registeringatIdP idp throughchannel c;

c(u A idp )representinguser u authenticatingherselfatIdP idp through channel c;

c(idp AP u )representingIdP idp allowinguser u touseanonymousor pseudonymousidentifiersthroughchannel c;

c(idp C u )representingIdP idp providinguser u withtheopportunityto provideconsentforreleasingselectedattributesthroughchannel c;

c(idp RL sp )representingIdP idp releasinguser u’sselectedattributestothe SP sp throughchannel c.

4TrustModellinginTraditional(Static)Federations

Inthissection,wemodeltrustbetweendifferententitiesintraditionalfederations.Wewillconsiderfirsthighleveltrustandthenfine-grainedtrust.

4.1HighLevelTrustModelling

WecanexpressthehighleveltrustinaType1federation f ∈F betweenan IdP idp ∈ IDP f andanSP sp ∈ SP f by:

idp DT : FED ← → FT sp

Thissignifiesthat idp and sp haveamutualdirecttrustinthescopeofthe federation.SinceitisaType1federation,theentitiestrusteachotherfully, hencethetruststrengthisfullytrusted(FT ).

MathematicalModellingofTrustIssuesinFIM19

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.