ITU-T Workshop on Security Trends of Biometrics Technology Standardization 14 May 2002 Naohisa Komatsu Waseda University, Japan Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 1 Authentication process user user authentication terminal authentication terminal cryptosystem network system Knowledge-based : Threat of forgetting e.g. password Possession-based : Threat of loss e.g. card Individual characteristics : No threat of forgetting or loss e.g. fingerprint, voice, handwriting Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 2 Parameters for User Authentication knowledge password ...... threat of forgetting possessions key,ID card ...... Individual characteristics physiological behavioral characteristics characteristics fingerprint,face handwriting,voice hand,eye ...... keystroke ...... threat of loss change through time passing ? stored data = input data stored data → personal features a. =? input data → personal features ? b. stored data = input data Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 3 Characteristics of Biometrics (OMRON Corp.) Distance between system number and user high 10-4% 0.01% 0.1% 1% 0 ideal 0.5m Iris Retina Each biometrics has its own merits or Pattern of vein demerits. There are no ideal biometrics. 0.01m 0 Fingerprint Face 0 0 Hand geometry 0 Finger geometry Signature 0.1m 1~3m Voice low low acceptability high Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 4 The Standardization of Biometrics Technologies Why standardization is necessary? Data format (CBEFF) and Application Program Interface (BioAPI) Security requirements (X9.84) Accuracy test (Best Practice) Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 5 Needs of Standards To accelerate fair competition by clarifying vulnerability and countermeasures. To reduce the cost of system development Accuracy test Standards for applying biometrics Application program interface Data format For effective development through common framework for biometric system. Common Criteria Privacy guideline Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 6 Goal of BioAPI & CBEFF Data interoperabity:CBEFF Program interoperability:BioAPI Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 7 Purpose of BioAPI Purpose Interoperability and development cost reduction of biometric authentication systems. Providing a high-level generic biometric authentication model Authentication/Identification, Server/Client Scope Any form of biometric technology Enrollment,authentication,identification,database interface Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 8 History of BioAPI NIST merged HA-API, BAPI and BioAPI. 1997 1998 1999 2000 2001~ I/O Software joined BioAPI consortium BAPI 1.0 Former BioAPI DRAFT HA-API HA-API 1.0 2.0 BioAPI BioAPI Specification Ver.1.0 Specification Ver.1.1 BioAPI Reference Implementation Ver1.0 Beta BioAPI Reference Implementation Ver1.1 NIST:National Institute of Standards and Technologies HA-API: Human Authentication API Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 9 Outline of BioAPI Structure Application API API:Application Program Interface Middleware mediates between API & SPI ・Reference implementation for windows is available BioAPI Framework SPI SPI SPI SPI:Service Provider Interface BSP BSP BSP BSP:Biometrics Service Provider Biometric function provided by technology vendors Device Device Device Biometric Device fingerprint scanner, camera, etc... Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 10 Outline of CBEFF Purpose Interoperability between different systems. Accommodation to any biometric technology. History Sponsor Developing organization CBEFF Technical Development Team Cooperating with NIST ITL,Biometrics Consortium BioAPI Consortium,X9.F4 Working Group,IBIA,TeleTrustT Publication NISTIR6529 “Common Biometric Exchange File Format” (NIST, January 3, 2001) http://www.nist.gov/cbeff NIST ITL:Information Technology Laboratory IBIA:International Biometric Industrial Association Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 11 CBEFF Data Structure Includes three blocks SBH (Standard Biometric Header) Header of CBEFF file BSMB(Biometric Specific Memory Block) Contains the biometric data Vendors can place any biometric data directly into this block Biometric information, template, original header, etc... SB(Signature Block) Contains signature or MAC for integrity Optional SBH BSMB SB (Standard Biometric Header) (Biometric Specific Memory Block) (Signature Block) Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 12 Relation Among Standardizations The standardizations are progressing to convergence on BioAPI and CBEFF BioAPI Specification ver.1.1(2001/3) “CBEFF” NISTIR6529(2001/1) BAPI BioAPI merged into HA-API adoption ANSI X9.84 adoption CBEFF ISO7816-11 ANSI X9.84: Operating requirements for biometrics considering adoption authentication systems for the financial industry Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 13 Outline of X9.84 Approved in March, 2001 by committee on Financial Services, X9 and subcommittee on Information Security, X9F. A standard of biometric data management and security for financial biometric system. X9.84 specifies Security requirements of enrollment, verification/identification, storage, termination,etc... Template format compatible with CBEFF Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 14 Requirements of X9.84 Common requirements of data management in enrollment, verification/identification, storage, termination, etc... To maintain the integrity of biometric data and verification results To mutually authenticate between sender and receiver component of biometric data and verification results. To ensure the confidentiality of the biometric data Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 15 Initial Enrollment for Example Enrollment Model Data Collection Signal Processiong Storage Matching Mechanism and procedure shall be in place to Authorization to perform the enrollment process Authentication of the enrollee Maintain integrity and authenticity of templates Meet level 2 physical security requirement in a controlled environment and level 3 in an uncontrolled environment. Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 16 Outline of “Best Practice” Purpose Scope To provide the best method for the accuracy test of biometric system in real world Any biomerics and application Features Experimental evaluation Three test methods depending on the aim of evaluation Definition of experimental condition Technology evaluation: algorithm Scenario evaluation: Specific system assumed by an evaluator Operational evaluation: Running system How to select subjects, to collect biometric data, to match them… Representation of performance ROC curve for accuracy Failure to enroll and acquire for usability Detailed report for repeatability Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 17 The Standardization of Biometrics Technologies in Japan Position of INSTAC/AIM/JBAA Activities of JBAA Operating Requirements Decision Guideline Vulnerability of Biometrics Technologies Biometrics and PKI Biometrics and Privacy Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 18 Standardization Activities in Japan 1996 1997 1998 1999 2000 ECOM WG6 △V0.5 △Evaluation criteria for biometrics authentication V1.0 2001 2002以降 Accuracy Test JIS-TR JIS-TR△ △ IPA Project △ Accuracy Test Guideline △ ORD Guideline 1994 CC V1.0 JBAA BDPP,X9.84 1995 BS7799 ECOM:Electronic Commerce Promotion Council of Japan JBAA:Japan Biometric Authentication Association IPA:Information-technology Promotion Agency,Japan CC:Common Criteria INSTAC:Information Technology Research and Standardization Center BS7799:British Standard7799 JIS:Japanese Industrial Standard BDPP:Biometric Devices Protection Profile ORD: Operating Requirements Decision Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 19 Position of INSTAC,AIM,JBAA ASIA Committee Bio WG Taiwan Bio WG Malaysia BEAM: Biometrics EnAbled Mobile Commerce BEAM Consortium SIngapore Asia Biometrics Joint Meeting Korea Biometrics Association(KBA) JBAA AIMJ ISO/IEC SC17/SC27 INSTAC/JSA (International) Standardization Biometrics Consortium Biometrics Working Group EU/USA Committee INSTAC/JSA : Information Technology Research and Standardization Center / Japanese Standards Association AIM : Automatic Identification Manufactures Association, Japan JBAA: Japan Biometric Authentication Association Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 20 How to Expand Biometrics Market? - Based on different methods and data for accuracy test - Publication of the best results (1) Accuracy?(Technology) (2) Cost effectiveness?(Business) (3) User acceptability?(Social) - Few examples - Indefinite requirements for security, convenience, etc. Standardization of accuracy test is important. IPA/Hitachi Project ’99 (16 companies) Image processing → Security technology Solution for market creation is necessary. IPA: Information-Technology Promotion Agency,Japan Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 21 Outline of Complete Activities (1) ECOM Personal Authentication WG April/1996~March/1998 Examine the scheme of the test and evaluation and the personal authentication model using biometrics (www.ecom.or.jp) (2) IPA/Hitachi National Project January/1999~December/1999 Examine the standards scheme of accuracy test and operation requirements sponsored by IPA(MITI) (www.sdl.hitachi.co.jp/ipa_biotest/ipa/english.htm) ECOM: Electronic Commerce Promotion Council of Japan MITI: The ministry of International Trade and Industry IPA: Information-technologies Promotion Agency of Japan Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 22 Outline of Current Activities Standardization activities are done in two organizations (1) Biometrics WG of INSTAC/JSA “Make a Standardization of test and evaluation of biometrics device and system in Electronic Commerce application” (2) Biometrics WG of AIM “Enlighten the biometrics technology and research the biometrics market” (3) Biometrics WG of JBAA Discussion about implementation of “Biometric Authentication Authority” which provides network type biometric identification Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 23 Biometrics WG of INSTAC/JSA Purpose The standardization of the biometrics authentication technologies is done in Electric Commerce application Members of WG Chairman : N. Komatsu(Waseda Univ.) Members : METI, Animo*, Casio, Fujitsu, Hitachi*, KDDI, Matsushita, MELCO, NEC, NTT-data* , OKI, Sony, Toshiba, etc. * : Working Group leader Contents of activities (1) Draft Japanese Industrial Standards of Test and Evaluation (Physical and behavioral characteristics) (2)Liaison with ISO/IEC JTC1/SC17 Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 24 Purpose of The Project guide for Design Guidelines for requirements decision Requirements decision Vendors Evaluation method Proposal Users Guidelines for accuracy test Valuation basis Creation of real-based biometric market Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 25 Policy of Accuracy Test (1) Objective evaluation for multiform products, various evaluators “viewpoints” and individual tests (2) Common basis with Europe and America ・Japanese accuracy test could be accepted ・Refer to proposals of NBTC about mathematical basis (3) Focus on fingerprint based authentication systems Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 26 Classification of Functional Structures Authentication system Verification Device Fingerprint capture Fingerprint Fingerprint Matcher Distance Parameters Decision function Verification Device Fingerprint capture Fingerprint Matcher Output distance Fingerprint Matcher Image enhancement Feature extraction Templates Matching functions Output distance Authentication system Verification Device Decision function Output result Result (True/False) Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 27 Definition of Test Items Fingerprint Collection Verification Accuracy Calculation Defined Items Guidelines for accuracy tests Accuracy Test Software Test process start ・Collection environment ・Number of fingers ・Number of fingerprints ・Experimental subjects ・Training to input finger ・Combination of genuine ・Combination of imposter ・Calculation methods ・Result description ・ROC Curve ・Availability Rate Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 28 Results Description FMR & FNMR as ROC curve Calculated at each threshold or parameters Described by a logarithmic ROC curve Availability rate Rate of persons who can use the product Frequency Imposter hg (t) Genuine hi (t) Threshold Th FMR FNMR Distance (t) Distance Distribution False Match Rate(FMR) 0.1% 0.01% 0 0.01% False Non-Match Rate(FNMR) 0.1% ROC Curve Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 29 Comparison with Best Practice Biometrics Target Application Evaluation Method Results Description Accuracy Test Guideline Fingerprint Verification only (1 to 1) Best Practice All biometrics technologies All applications ・Verification ・(Positive) Identification ・Negative Identification Depend on a object of evaluation For algorithms Technical For devices Scenario For systems Operational ・ROC(FNMR,FMR) curve ・ROC(FNMR,FMR) curve ・Accuracy Test Guideline ・Failure to acquire, Failure to enroll ・Test Specification ・Test Specification ・BinningError vs Penetration curve Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 30 Comparison with Best Practice Item Number of Biometric Information Best Practice Accuracy Test Guideline expected ・As many as possible with Decided accuracy ・No strict criterion ・There must be enough time Requirements interval between collecting for Fingerprint enrollment data and test data. Collection ・The length of interval is not specified. Fingerprint data that failed to enroll or verify are not used Availability Rate for accuracy evaluation. ・The time interval must be longer than general time of healing of that body part. (2 to 3 weeks for fingerprints) ・Fingerprint data that failed to enroll or verify are not used for accuracy evaluation. ・Failure to Enroll Rate ・Failure to Acquire Rate Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 31 Outline of the ORD Guideline (1) The guidelines provide the methods to decide the requirements for application (2) Investigation of applications using authentication 6 fields (Finance, Public, Medical, PD/Retail, Housing,etc.) Hearing from 50 users (3) Application model leads requirements on usability (4) Risk analysis leads requirements on security ORD: Operating Requirements Decision Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 32 Security Level Classification Usability Safety Level Criterion (H) (M) ・Very Highl Risk ・Relation to Social Safety Example ・Area Control in Nuclear Applications Power Plant ・Area Control in Mint Bureau ・Access Control for Arms ・Area Control in Smart Card Issuer ・Access Control of CA’s Private Key FAR(:ex) Expression FRR 0.00006% 1 ( Population ) ( PCOR) (L) ・High Risk ・Relation to Social Trust ・Low Risk ・No necessary Security ・Area Control in Bank ・Immigration ・Access Control of Smart Card ・Debit/Credit Card ・Remote Banking ・Medical chart ・ATM ・DB in Enterprise ・PC Log in ・Entrance of Apartment ・Attendance of Office ・User Tracking ・Observing 1%~0.01% ( Permisible FAR) ( Member) ( PCOR) FAR takes first priority About 1% FRR takes first priority Functional Requirements Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 33 Model Classification for Biometric Applications ①Access Control Safety Real Space Authenticate Protected Space (Physical or Electronic) Value ②Flow Control Real Space Electronic Space Authenticate e-Doc ・Signature ・Seal Flow of Sanction ③Tracking Real Space Electronic Record(Log) ID Place Time Authenticate Usability Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 34 Procedure of ORD Start Model Classification Functional Requirements Specification Threats Analysis Occurrence Rate Value Evaluation Value Risk Analysis Security Level Classification Usability Requirements Adjustment Requirements Safety Requirements C Report B of device: Evaluation A Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 35 Japan Biometric Authentication Association Object To create the fair biometric market, JBAA (1) Investigates problems of standardization of biometric authentication (2) Proposes activities for standardization, promote projects for common framework by academic, business, and governmental circles Agenda (1) Interoperability ・Investigation of standardization of data format and API ・Clarifying PKI model (2) Performance ・Investigation of standardization for accuracy test ・Investigation of privacy and other compliance (3) Assurance ・Investigation of security standardization and protection profiles ・Investigation of operating requirements decision guideline and proposal of a draft. ・Clarifying policy for examination of vulnerability Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 36 Activities of Technology WG of JBAA Purpose - System integrators/users can select appropriate biometric devices Term - Sep 2000 - now Members - Hitachi, Omron, Oki, Mitsubishi, Computer Associates, Japan Telecom, NEC, Secure Generation, Sharp, Secom, Cyber Sign, Waseda university etc. Outline of project (1) Research of biometric technologies (2) Discussion about problems of service/business model (3) Comprehensive discussion about common technical problems for standardization (4) Experiment to verify above problems Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 37 Privacy Consideration Biometric information can be easily stolen and forged → Vulnerability Biometrics is the ultimate privacy data Privacy protection in X9.84 is based on HIPAA (Healthcare Insurance Portability and Accountability Act). Act on the prohibition of unauthorized access to computer systems (Feb, 2000) bans dishonest acquisition and use of identification code. Biometrics fall under this category. We should make a privacy guideline for biometric authentication systems and lead system integrator and operator to a better understanding of privacy. Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 38 Vulnerability Consideration What’s “Vulnerability” for information systems? Vulnerability for Biometric authentication systems is… Characteristic of the system causes the system not to perform the requirement designed. The characteristics causes impersonation The characteristics causes the impediment of the system availability For secure biometric authentication systems To define all of the vulnerability on the system To define the risk of every vulnerable characteristic To define the countermeasure of the vulnerable characteristics Need for definition of the vulnerability of biometric systems Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 39 Examples of The Vulnerability for Biometric Authentication Systems Biometrics specific vulnerability False Acceptance rate Artificial biometric object Hill-climbing attack etc… Common vulnerability for information system Forgery or alternation of Template data of users Matching software Result of the matching Electronic Biometric data etc… Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 40 Comparison of PKI/Biometrics Models Client Model Basic model Server Model Authentication Authentication Basic model server Model server Model Templates are stored in Client Client BCA BCA Verified in Client Client Application Authentication server Digital authentication in Application Authentication server Application Authentication server ・Confidentiality in client Security Requirements ・Consistency with PKI ・Integrity of biometric info. ・ Consistency with PKI Suitable model should be selected according to various system requirements Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 41 Server Model (Basic Model) PKI based authentication →Biometrics authentication User Certificate Secret key Biometric Data Input biometrics & Sign (1)Challenge code (2)User signature, Certificate & biometrics BCA Certificate Verification of Signature & Biometrics CRL User Template BCA (3)Service Client Terminal CA CA Certificate Application Server Template Database Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 42 Client Model( Basic Model ) Biometrics authentication → PKI based authentication User Certificate Biometric Data Secret key Biometric Verification & Sign User Template Client Terminal CA (1) Challenge code (2) User signature & Certificate CA Certificate CRL Signature Verification (3) Service Application Server BCA Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 43 Biometric Authentication on Mobile Phones BCA - Server Model - Certificate Mobile phone Biometric Certificate Database CA ③ Biometric Base Station Carrier Net IP Network ①Challenge code ② Signature,Certificate,fingerprint data Certificate Application Server CA Certificate ④Service Biometric Data Signature Verification Certificate Secret key Sign Biometric Verification UIM Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 44 Biometric Authentication Authority (Japan Telecom) Biometric Authentication Authority Biometric DB Verification Engine Biometric Authentication Server ⑤ Verification ④ Authentication Request ⑥ Result EC Sites ⑥ Result Biometric Data ② Authentication Request Client ③ Acquisition of Biometric Data PKI Certificate Authority ① Access Biometrics HandFingerprint geometry Iris ・・・ Face Voce Signature Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 45 Biometrics Campus User Authentication COOP User Authentication Issuing Machines for Certificates Safety Box (Mitsubishi Corp.) User Authentication Vending Machine Parking Lot Holder Authentication Copy Machine ID Card(IC) Intra-CampusLAN Workflow System for office workers Lecture Room Authorization Attendance Labs/ Computer Rooms Library/ Dormitory Access Control Servers Library System Authentication Authentication Servers Access Control for Intra-campus LAN Home Campus On-line Registration/ On-line inquiry Internet Mobile Campus Cell. Phone User Authentication Theater/ Restaurant/ Salon Student Discount On-line Attendance Ticketing Center User Authentication Application for Parking Lot On-line Registration for Certificates/ Student Discount Satellite Campus Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 46 Study Items ■Standardization ・Operating requirement decision guideline ・Privacy consideration ・Vulnerability consideration ・Authentication model (PKI+biometrics) ・etc. ■Utilizing merits of biometrics ・Authentication without user’s consciousness ・Authentication with user’s feelings Copyright(C) 2002, All rights reserved. SDL, Hitachi, Ltd. and Waseda University. 47