Fingerprints On Mobile Devices: Abusing and Leaking

Similar documents
The Android Developers Guide to 3 rd -Party SDK Assessment and Security

How we keep harmful apps out of Google Play and keep your Android device safe

What someone said about junk hacking

Do Containers fully 'contain' security issues? A closer look at Docker and Warden. By Farshad Abasi,

FIDO Trust Requirements

Exploiting Trustzone on Android

McAfee Vulnerability Manager 7.5.1

2-FACTOR AUTHENTICATION WITH OPENLDAP, OATH-HOTP AND YUBIKEY. Axel Hoffmann

NetIQ Advanced Authentication Framework

Threat Model for Mobile Applications Security & Privacy

How Secure is Authentication?

PAKITI Patching Status System

Comprehensive Security for Internet-of-Things Devices With ARM TrustZone

State of Minnesota. Office of Enterprise Technology (OET) Enterprise Vulnerability Management Security Standard

How to remove Encrypted File guide. How to manually remove Encrypted File

Developing Secure Software in the Age of Advanced Persistent Threats

Device-Centric Authentication and WebCrypto

Scalable Authentication

Excellence Doesn t Need a Certificate. Be an. Believe in You AMIGOSEC Consulting Private Limited

PFP Technology White Paper

SSOScan: Automated Testing of Web Applications for Single Sign-On Vulnerabilities. Yuchen Zhou and David Evans Presented by Yishan

How to Install Applications (APK Files) on Your Android Phone

<COMPANY> PR11 - Log Review Procedure. Document Reference Date 30th September 2014 Document Status. Final Version 3.

Security Technology for Smartphones

Security in Android apps

Android Security 2015 Year In Review

Seven Things To Consider When Evaluating Privileged Account Security Solutions

Windows Phone 7 Internals and Exploitability

Software that provides secure access to technology, everywhere.

Bomgar Corporation. Bomgar Application Security Assessment Summary January 26, This document is the property of Bomgar Corporation.

Information Security for Modern Enterprises

Change Management. Why Change Management? CHAPTER

Penetration testing. A step beyond missing patches and weak passwords

Research on Situation and Key Issues of Smart Mobile Terminal Security

BYPASSING THE ios GATEKEEPER

Data Masking Best Practices

Hong Kong Information Security Outlook 2015 香 港 資 訊 保 安 展 望

CDM Software Asset Management (SWAM) Capability

Tutorial on Smartphone Security

M-Shield mobile security technology

Securing your Virtual Datacenter. Part 1: Preventing, Mitigating Privilege Escalation

CITY OF BOULDER *** POLICIES AND PROCEDURES

Unknown threats in Sweden. Study publication August 27, 2014

Port Scanning and Vulnerability Assessment. ECE4893 Internetwork Security Georgia Institute of Technology

Developing National Frameworks & Engaging the Private Sector

Security and Cloud Computing

Applying Cryptography as a Service to Mobile Applications

Configuring on Mobile Devices

The Mobile Problem. Alex Bobotek Co-Chairman, M 3 AAWG October 2012 New Delhi, India

Secunia Vulnerability Intelligence Manager (VIM) 4.0

Module II. Internet Security. Chapter 7. Intrusion Detection. Web Security: Theory & Applications. School of Software, Sun Yat-sen University

Threat Modeling. Frank Piessens ) KATHOLIEKE UNIVERSITEIT LEUVEN

Analysis of Secure Key Storage Solutions on Android

FIDO Security Reference

Top 8 Steps for Effective Mobile Security

University of California, Riverside Computing and Communications. IS3 Local Campus Overview Departmental Planning Template

CYBER TRENDS & INDUSTRY PENETRATION TESTING. Technology Risk Supervision Division Monetary Authority of Singapore

Advanced ANDROID & ios Hands-on Exploitation

e-government Agency Delivering Secure, Public-Oriented e-government Facilities in Africa A Holistic Approach

Now and Tomorrow. IEEE/AICCSA Conference November 2014 Malike Bouaoud Cyber Security Advisor

An Information Security and Privacy Perspective for Procurement Services Projects

Orchestrated Security Network. Automated, Event Driven Network Security. Ralph Wanders Consulting Systems Engineer

Introduction to SquareTrade Security Services

Topic 5a Operating System Fundamentals

SSOScan: Automated Testing of Web Applications for Single Sign-On vulnerabilities

IBM Security Intelligence Strategy

HOW OBSERVEIT ADDRESSES 7 OF THE SANS 20 CRITICAL SECURITY CONTROLS

Securing the Internet of Things: Mapping Attack Surface Areas Using the OWASP IoT Top 10

How To Test A Control System With A Network Security Tool Like Nesus

State of Security. Top Five Critical Issues Affecting Servers. Decisive Security Intelligence You Can Use. Read Our Predictions for 2013 and Beyond

NOK NOK LABS AUTHENTICATION & OTT SERVICES

A Study of Android Application Security

Product Roadmap Symantec Endpoint Protection Suzanne Konvicka & Paul Murgatroyd

Network and Host-based Vulnerability Assessment

Running head: USING NESSUS AND NMAP TOOLS 1

A!Team!Cymru!EIS!Report:!Growing!Exploitation!of!Small! OfCice!Routers!Creating!Serious!Risks!

QUIRE: : Lightweight Provenance for Smart Phone Operating Systems

Transcription:

Fingerprints On Mobile Devices: Abusing and Leaking I. Introduction YulongZhang,ZhaofengChen,HuiXue,andTaoWei FireEyeLabs Fingerprintscannersarebecomingmoreandmorepopularonthemodernmobiledeviceslike HTCOneMax,HuaweiMate7,iPhone5s/6/6+,andSamsungGalaxyS5.Basedonarecent projection[1],50%ofsmartphoneshipmentswillhaveafingerprintsensorby2019.those fingerprintscannersaremorethanjustagimmicktounlockyourphonexxtheycanalsoletyou convenientlyauthorizemoneytransactionswithaswipeofyourfinger.withtheintroductionof ApplePay,mobilepaymentisgoingtobeaprimarydriverforfingerprintsensors.Mostof ApplescompetitorshavereactedquicklytomatchApplesmobilepaymentleadinthemarket. However,howsecurethosefingerprintframeworksaredesignedremainsthecustomers biggestconcern.inthetraditionalpasswordxbasedauthsystems,victimscaneasilyreplacethe stolenpasswordswithanewone.butfingerprintslastforalifexxonceleaked,theyareleaked fortherestofyourlife.moreover,fingerprintsareusuallyassociatedwitheverycitizen sidentity, immigrationrecord,etc.itwouldbeahazardiftheattackercanremotelyharvestfingerprintsin alargescale. Previouslytherehavebeensomeworksfocusingonfingerprintspoofingattack[2][3].These worksdemonstratethatfingerprintscanbestolenfrompolishedsurfaces(e.g.smartphone screens)orfromawavinghandphoto,andcanbespoofedusingelectricallyconductive materials.wecategorizethistypeofattackasopticalattacks.inthispaper,wewillratherfocus onthesystemattacksofmobilefingerprintauthframework.toourknowledge,wearethefirst todiscussthistypeofthreats. Wewillanalyzethemobilefingerprintauthenticationandauthorizationframeworks,anddiscuss severalsecuritypitfallsofthecurrentdesigns,including: ConfusedAuthorizationAttack Unsecurefingerprintdatastorage Trustedfingerprintsensorsexposedtotheuntrustedworld BackdoorofpreXembeddingfingerprints Thispaperisstructuredasfollows:wewillintroducethebackgroundoffingerprintauth frameworkinsectionii,anddescribedetailedvulnerabilitiesinsectioniii.furtherdiscussions willbeprovidedinsectioniv.weconcludethepaperinsectionv.

II. Background 1. Original Mobile Fingerprint Auth Framework Figure1:Originalmobilefingerprintauthframework Figure1illustratesthenormaldesignofmobilefingerprintauthentication/authorization framework.anexampledevicewiththisdesignishtconemax.inthisdesign,thekernel interactsdirectlywiththefingerprintsensor,andvendorfingerprintlibraries(nativecode)and fingerprintservices(javacode)providefunctionalwrappingofcommonfingerprintauth operations.thehighlevelauthlogicisimplementedinspecificappsbyinvokingvendor exposedapis. ThereisanobvioussecurityweaknesswiththisdesignXXthefingerprintsarejustassecureas thekernel.iftheattackerrootsthedevice,he/shecanstealthefingerprintdata.unfortunately therearequiteafewpublicknownkernelvulnerabilitiesthatcanbeexploitedtorootthe majorityofandroiddevices,likeframaroot[4],towelroot[5],andpingpongroot[6],etc.thus thevendorsarenowmovingtoamoresecureddesignwiththehelpofarmtrustzone[7]. 2. Mobile Fingerprint Auth Framework with TrustZone

Figure2:TrustZoneisolationonmobiledevices Figure3:TheimprovedmobilefingerprintauthframeworkwithTrustZoneisolation Figure2showsthegeneralarchitectureformobileplatformsequippedwithTrustZone. TrustZoneisolatesthenormalworld(thenormaluser/kernelmode)andthesecureworld (TrustedExecutionEnvironments,orTEE)bycreatingadditionaloperatingmodes,knownas thesecuremodeandthemonitormode.thesecuremodehasthesamecapabilitieswiththe normalworldwhileoperatinginaseparatememoryspace.thetrustzonemonitoractsasa virtualgatekeepercontrollingmigrationbetweenthetwoworlds.thenormalworldcanissue requeststothesecureworldviasecuremonitorcalls(smc). Sincethenormalworldcannottouchthesecureworld smemory,fingerprintscanbewell protectedintrustzonefrombeingaccessedbynormalworldattackers.figure3depictsthe enhancedfingerprintauthframeworkutilizingtrustzone sprotection.inthisdesign,the fingerprintsensordriver,fingerprintrecognitionlogic,andthefingerprintdataareallisolatedin thesecureworld,sothefingerprintauthframeworkremainssecureevenifthenormalworld kerneliscompromised.

Althoughtheenhanceddesignhasbeenmuchsecurerthantheoriginaldesign,inthenext sectionwewillrevealthattherearestillmanyseveresecuritypitfalls. III. Vulnerabilities 1. Confused Authorization Attack Thefirstvulnerabilityisthatallthefingerprintframeworksarepronetothe Confused AuthorizationAttack,whichhaslongbeenoverlooked.Authorizationgrantsaccessrightsto resources,whileauthenticationverifieswhoyouare.securitysystemsoftenmistakenlytreat authorizationasauthentication,orfailtoprovidecontextprooffortheauthorizationobjects. Withoutpropercontextproof,theattackercanmisleadthevictimtoauthorizeamalicious transactionbydisguisingitasanauthenticationoranothertransaction.forexample,asshown infigure4,theattackercaneasilyfakealockscreentofoolthevictimtothinkthathe/sheis swipingfingertounlockthedevice,butthefingerprintisactuallyusedtoauthorizeamoney transferinthebackground. CurrentlytheFIDOAllianceisdevelopingthespecificationofsecureauthenticationand authorizationprotocolsforthemobileecosystem.asfidodescribesinthespecification[12]: Basically(if(a(FIDO(UAF(Authenticator(has(a(transaction(confirmation(display(capability,(FIDO( UAF(architecture(makes(sure(that(the(system(supports(What(You(See(is(What(You(Sign(mode( (WYSIWYS).(A(number(of(different(use(cases(can(derive(from(this(capability(EE(mainly(related(to( authorization(of(transactions((send(money,(perform(a(context(specific(privileged(action,( confirmation(of(email/address,(etc). FIDO sspecificationrequiresthat the(transaction(confirmation(display(component(implementing( WYSIWYS(needs(to(be(trusted.However,theoriginalfingerprintauthframeworkhasno reliablewaytoprovidetheauthorizationcontextproof.theframeworkwithtrustzonecanbe improvedtoachievethisgoal(thetrustletmodulesintrustzonecanbemodifiedtoprovidethe contextproof),butsofar(june2015)wehaven tseenanymajorvendorthatimplementedthis feature.

Figure4:Illustrationofusingconfusedauthorizationattacktoauthorizemoneytransfer 1 2. Fingerprint Data Storage Vulnerability Thesecondissueisthatnotallthevendorsstorethefingerprintssecurely.Whilesomevendors claimedthattheystoreuser sfingerprintsencryptedinasystempartition,theyputusers fingerprintsinplaintextandinaworldxreadableplacebymistake.oneexampleishtcone MaxXXthefingerprintissavedas/data/dbgraw.bmpwith0666permission(worldXreadable) 2. Anyunprivilegedprocessesorappscanstealuser sfingerprintsbyreadingthisfile.other vendorsstorefingerprintsintrustzoneorsecureenclave,buttherearestillknown vulnerabilitiesforattackerstoleveragetopeekintothesecretworld[8][9]. Tomakethesituationevenworse,eachtimethefingerprintsensorisusedforauthoperation, theauthframeworkwillrefreshthatfingerprintbitmaptoreflectthelatestwipedfinger.sothe attackercansitinthebackgroundandcollectthefingerprintimageofeveryswipeofthevictim. Notethatthereissomespecialtyoftheformatoftherawfingerprintbitmapimage.Normallythe sizeofeachbitmaprowisroundeduptoamultipleof4bytesbypadding.butasamplerowof therawfingerprintbitmaplookslike: 01 FE 02 00 09 09 14 20 60 50 70 70 70 40 60 50 70 70 70 70 70 70 70 70 60 70 60 70 70 80 70 80 80 70 70 70 70 80 90 A0 A0 A0 A0 A0 A0 B0 A0 80 [...] B0 B0 B0 A0 B0 A0 A0 A0 A0 A0 A0 A0 90 A0 A0 90 90 90 90 90 90 90 80 80 70 B0 70 1 Thisisageneralissueforfingerprintbasedpaymentformobile.Thedeviceshownisjustforillustration andisnotnecessarilyvulnerable. 2 HTChaspatchedthisvulnerabilityperournotification

Weobservedthatalltherowsstartwith0xFE01.Thisisprobablyusedtomarkthebeginningof eachfingerprintimageline.andeachlinehas187bytesofdata(including0xfe01),whichis not4xbytealigned.ifoneopenstherawfingerprintbitmapimagedirectly,itwilllooklikefigure 5(a).Afterappropriatelypadding,itwillbecorrectlylooklikeFigure5(b).Thefingerprintbitmap imageisblurryinthebottomandhasmoredetailsintheupperpart.also,onemaynoticethat theimageisvisiblyseparatedintotwohorizontalparts,andtheleftpartisalmosttwicethewidth oftherightpart.thisisthedesignnatureofcertainsensorchips.wejustpresenttheimageas itis. (a)(b) Figure5:FingerprintbitmapobtainedfromHTCOneMax.Boththerawbitmap(a)andthereX alignedversion(b)areshown.weonlypastedasmallportionoftheimagestoprotectthe fingerprintowner sanonymity. 3. Fingerprint Sensor Exposure Vulnerability EveniftheprotectionoffingerprintdatainTrustZoneisindeedtrustworthy,itonlymeansthat thefingerprintspreviouslyregisteredonthedevicesaresecured.wefoundthatthefingerprint sensoritselfinmanydevicesisstillexposedtotheattackers.althoughthearmarchitecture enablesisolatingcriticalperipheralsfrombeingaccessedoutsidetrustzone(e.g.by programmingthetrustzoneprotectioncontroller),mostvendorsfailtoutilizethisfeatureto protectfingerprintsensors 3.Tothebestofourknowledge,wearethefirstoftheworldtoput forwardthefingerprintsensorspyingthreats. 3 Asofwriting,wehaveconfirmedthisvulnerabilityonHTCOneMax,SamsungGalaxyS5,etc.All vendorshaveprovidedpatchesperournotification.

Figure6:Hownormalworldissupposedtointeractwiththefingerprintsensor(TrustZone shouldalwaysactasthemediator) Figure7:Malwareinthenormalworldcandirectlyreadthefingerprintsensor AsshowninFigure6,insteadofdirectlycommunicatingwiththefingerprintsensor,allthe normalworldcomponentsaresupposedtoinvokethetrustzonefingerprintforsensor operations.however,mostvendorsfailtolockdownthesensor(frombeingaccessedbythe normalworldprograms)whentheprocessorswitchedbackfromthesecureworld.withoutthe properlockxdown,theattackerfromnormalworldcandirectlyreadthefingerprintsensor(shown infigure7).notethatattackerscandothisstealthilyinthebackgroundandtheycankeep readingthefingerprintsoneverytouchofthevictim sfingers.thisalsoindicatesthatattackers withremotecodeexecutionexploits(e.g.[11])canremotelyharvesteveryone sfingerprintsina largescale,withoutbeingnoticed. Giventhatthefingerprintsensorisexposedtothenormalworld,theonlyprotectionisthe accesspermissionenforcedbythenormalworldkernel.unfortunately,onsomephones,the sensorisonlyguardedby system privilegeinsteadof root.thisdecreasestheattack difficultysincetherearesomeeasierxtoxexploitvulnerabilities(e.g.cvex2015x1474)togainthe system privilegecomparedtothe root privilege.

Afterobtainingenoughprivilegetoread/writethesensor,theremainingmatterishowto configurethesensorintothestatetoactivelyreadfingerprints.usuallythefingerprintsensor s IOoperationsareopenXsourcedandmayincludethefollowing: IOCTL_POWER_ON IOCTL_POWER_OFF IOCTL_DEVICE_RESET IOCTL_SET_CLK IOCTL_CHECK_DRDY IOCTL_SET_DRDY_INT IOCTL_REGISTER_DRDY_SIGNAL IOCTL_SET_USER_DATA IOCTL_GET_USER_DATA IOCTL_DEVICE_SUSPEND IOCTL_STREAM_READ_START IOCTL_STREAM_READ_STOP IOCTL_RW_SPI_MESSAGE IOCTL_GET_FREQ_TABLE IOCTL_DISABLE_SPI_CLOCK IOCTL_SET_SPI_CONFIGURATION IOCTL_RESET_SPI_CONFIGURATION IOCTL_GET_SENSOR_ORIENT Notethatforcertainimplementations,althoughthesensorgotexposedtothenormalworld,the normalworldkernelhasstrippedoutcertainioctlhandlers.malwarecanutilizesomekernel exploittodynamicallyenablesuchfunctionalities(e.g.throughcodeinjectionorrop).malware withrootprivilegecanalsoflashacustomizedkernelwithsuchioctlhandlersenabledintothe Android sbootpartition. MalwarecanuseIOCTL_REGISTER_DRDY_SIGNAL toregisterastheeventsignallistenerofthe sensordevice.thenitcansequentiallyinvokeioctl_power_on,ioctl_set_clk, IOCTL_DEVICE_RESET,IOCTL_SET_DRDY_INT,IOCTL_CHECK_DRDY, IOCTL_STREAM_READ_START,etc.toinitializethedevicetotheactivelyreadingfingerprintsstate. Oncethesensorisintheactivereadingstate,malwarecankeepstealingusers fingerprints silentlyinthebackground.forsomedevices,thefingerprintsensorisintegratedontothehome button,whichmeansthattheattackercanstealusers fingerprintsoneverytouchofthehome button. 4. Pre-embedded Fingerprint Backdoor Moreover,theattackerscanstealthilyembedprefabricatedfingerprintsinthedevicesasan authorizationbackdoor,beforeprovidinganewdevicetothevictim.therootcauseofthis vulnerabilityisthattheuidisplayingthenumberofregisteredfingerprintsisaseparate

component(inthenormalworld,withouttrustzone sprotection)fromtheactualfingerprintauth frameworkinthesecureworld.attackerscandeceivetheusertobelievethatthereareonlyn fingerprintregisteredonthedevicebutthereareactuallymorethann.suchextraprex embeddedfingerprintcanbeusedtobypasstheauthframeworklikeabackdoor. ItisusuallytheSettingsappthatdisplaystheregisteredfingerprintnumbertotheusers,sothe attackerneedstomodifythesettingsapp.forexample,onmanydevices,onecanmodifythe enrolledfingerprintnummethodof!the!class com/android/settings/fingerprint/fingerprintsettingsinsecsettings.apk,by changingthereturnvalueofgetenrolledfingerstobenem,wherenistheactualregistered fingerprintnumberandmisthenumberoffingerprintsprexembeddedbytheattacker. NotethatSettingsisasystemapp,theattackerneedstosigningthemodifiedSettingsappwith thesameprivatekeythatsignstheothersystemapps,whichisthephonevendor sprivatekey anddifficultforattackerstoobtain.theattackercouldalsoextracttheromandresigningall thesystemapps(includingthereplacedsettingsapp)usingtheattacker sownprivatekey. Iftheattackerhasrootprivilege,anotheralternativeistodirectlydisablethesystemsignature checking.mostofandroiddevicesenforcethesystemsignaturecheckingbasedonthe comparesignaturesmethodintheclasscom/android/server/pm/packagemanagerservice implementedin/system/framework/services.jar.itwillreturnzeroifsignaturematch,and nonxzerootherwise.therefore,onecanmodifythismethodtoalwaysreturnzero,sothatthe systemsignaturecheckingwillalwayssuccess. IV. Discussion 1. Suggestions to mobile users and vendors Toavoidbeingattackedbymalwareorbeingexploitedforremotecodeexecution,wesuggest normaluserstochoosemobiledevicevendorswithtimelypatching/upgradingtothelatest version(e.g.androidlollipop),andalwayskeepyourdeviceuptodate.also,itisalwaysa goodpracticetoinstallpopularappsfromreliablesources.enterprise/governmentusersshould seekforprofessionalservicestogetprotectionsagainstadvancedtargetedattacks. Mobiledevicevendorsshouldimprovethesecuritydesignofthefingerprintauthframeworkwith improvedrecognitionalgorithmagainstfakefingerprintattacks,andbetterprotectionofboth fingerprintdataandthescanningsensor.moreover,vendorsshouldfigureouthowto differentiateauthorizationwithauthenticationandprovidecontextproof.theexistingfingerprint authstandardshouldbefurtherimprovedtoprovidemoredetailedandsecuredguidelinesfor

developerstofollow.finally,givenasecuritystandard,vendorsstillneedprofessionalsecurity vetting/auditstoenforcesecureimplementations. 2. Suggestions to the overall fingerprint auth ecosystem Actuallyallthefourvulnerabilities/attacksdescribedinthispaperarecommonlyapplicabletoall thefingerprintbasedauthentication/authorizationplatforms.forexample,manyhighxend laptopsequipfingerprintscannerstoauthenticateandauthorizeuserlogin.sincethefingerprint driverisakernelmodule,itisonlyassecureasthekernel.attackerswithkernelexploitcan stealfingerprintdataorcollectfingerprintfromthesensorinthebackground. Forexternalfingerprintscannersusedforidentityrecognition(e.g.inthecustomhouse, immigrationoffice,andthedmv),dooraccesscontrol,ormoneytransactioninbanks,the situationissimilar.sowesuggestthatthefingerprintauthframeworkforallplatformsshould alsobeimprovedtobetterprotectfingerprintdataandsensor(andprovidedefenseofanyother attacksdescribedinthispaperifapplicable). V. Conclusion Inthistalk,werevealedsomesevereissueswiththecurrentAndroidfingerprintframeworksthat havelongbeenneglectedbyvendorsandusers.weprovidedinxdepthsecurityanalysisofthe popularmobilefingerprintauthentication/authorizationframeworks,anddiscussedthesecurity problemsofexistingdesigns,including(1)theconfusedauthorizationattackthatenables malwaretobypasspayauthorizationsprotectedbyfingerprints,(2)insecurefingerprintdata storage,(3)fingerprintsensorexposedtotheuntrustedworld,and(4)prexembeddedfingerprint backdoor.wealsoprovidedsuggestionsforvendorsanduserstobettersecurethefingerprints. References [1]http://www.marketresearch.com/land/product.asp?productid=8918844&progid=87404 [2]https://srlabs.de/spoofingXfingerprints/ [3]http://www.ccc.de/en/updates/2014/ursel [4]http://forum.xdaXdevelopers.com/apps/framaroot/rootXframarootXoneXclickXapkXtoXrootXt2130276 [5]https://towelroot.com [6]http://forum.xdaXdevelopers.com/galaxyXs6/general/rootXpingpongrootXs6XrootXtoolXt3103016 [7]http://www.arm.com/products/processors/technologies/trustzone/index.php [8]https://www.blackhat.com/usX14/briefings.html#reflectionsXonXtrustingXtrustzone [9]https://www.blackhat.com/usX15/briefings.html#attackingXyourXtrustedXcoreXexploitingXtrustzoneXonX android [10] http://infocenter.arm.com/help/topic/com.arm.doc.dto0015a/dto0015_primecell_infrastructure_amba3_tz pc_bp147_to.pdf

[11]https://www.blackhat.com/docs/usX14/materials/usX14XWeiXSidewinderXTargetedXAttackXAgainstX AndroidXInXTheXGoldenXAgeXOfXAdXLibs.pdf [12]https://fidoalliance.org/wpXcontent/uploads/html/fidoXuafXoverviewXv1.0XpsX20141208.html