`
`133
`
`where S is the sample feature vector of components Si (i=1 to f) and Ti is
`the true (reference) feature mean value of variance Vi.
`This distance measure requires only five to ten true signatures to
`establish initial values for user behaviour statistics at enrolment.
`The 'best' feature set selected as standard by the above method contained
`features: 1 , 2, 3, 6, 11, 12, 13, 14, 16, 20, 22, 25, 26, 27, 28, 29, 30, 32,
`33, 40, 41, 42, 43 and 44. 24 features in all.
`
`REFERENCES
`
`[1] Masuyama, H. (1985) 'Properties of personal identification systems using
`question-answer techniques' Trans lnt Electronics and Communications
`Engineering J69D/4, 613-620.
`[2] Kniessler, H. (1985) Identification of individuals with computer graphics SRI
`Report, SRI International Inc.
`[3] Harmon, L. D., Khan, M. K., Lasch, R. and Ramig, P. F. {1981) 'Machine
`identification of human faces' Pattern Recognition 13(2) 97-110.
`(4] Siemens, A. G. (19�2) Fingerprint personal identification system DE 3036
`912. 13 May 1982.
`[5] Asai Koh "{1980) Device for extracting a density as one of a number of pattern
`. . . GB 2050026. 3L Dec 1980.
`features extracted for each feature point
`Nippon Electric Co Ltd.
`[6] Sparrow, M. K. and Sparrow, P. J. (1985) A Topological approach to the
`Matching of Single Fingerprints US Dept of Commerce, NBS. Oct. 1985.
`[7] Isenor, D. K. and Zaky, S. G. {1986) 'Fingerprint identification using graph
`matching' Pattern Recognition 19{2) 113-122.
`[8] Sibany Mfg. Corp. (1966) Recognising fingerprints GB 1150511. 29 April
`1966.
`[9] Stellar Systems Inc. (1983) Personnel identification system using characteristic
`data US Au8335163. 31 March 1983.
`[10] Fowler, R. C., Ruby, K., Sartor, F. F. and Sartor, T. F. (1985) Fingerprint
`Imaging Apparatus US 4537484. 27 Aug 1985. ldentix Inc, Palo Alto, Ca.
`[11] Fingermatrix Inc. (1984) Fast action fingerprint check for access control US
`EP125-532. 21 Nov 1984.
`[12] IBM Corp. (1979) Palm print identifier GB 1535467.
`[13] Palmguard Inc. (1982) Image recognition system for recognising human palm.
`US 4186378. 20 Jan 1982.
`[14] Stellar Systems Inc. (1984) Personnel Identification Devices Using Hand
`Measurement Techniques April, 1984.
`[15] Mitsubishi Denki K. K. {1985) Individual identification apparatus based on
`finger length comparison JP EP132665A. 13 Feb 1985.
`(16] Zuccarelli, H. {1983) 'Ears hear by making sounds' New Scientist 100 BPC.
`(17] EYE-D Developments IT Ltd (1984) Ocular fundus reflectivity pattern identi
`fication apparatus US EP126549A. 28 Nov 1984.
`[18] Rice, J. (1986) Method of, and Apparatus for the Identification of Individuals
`UK 8509389 86.
`
`Page 151 of 201
`
`UNITED SERVICES AUTOMOBILE ASSOCIATION
`Exhibit 1003
`
`
`
`134
`
`Integrated Circuit Cards
`
`[19] Abberton, E. and Fourcin, A. J. (1978) 'Intonation and speaker identificatimi'
`·
`.
`Language & Speech 21(4) 305.
`[20] Bolt, R. H., Cooper, F. S., David Jr, E. E. and Denes, P. B. {1973)
`'Speaker identification by speech spectrograms: some further observations'
`JASA 54(2) 531-537, Acoustic Society of America.
`[21] Holden, A. D. C. and Cheung, J. Y. (1977) The role of idiosyncracies in
`linguistic stress cues, and accurate formant analysis, in Speaker Identification
`Carnahan Conference 1977, University of Kentucky, 33-37.
`(22] McGlone, R. E. and Hollien, H. (1976) 'Partial analysis of acoustic signal of
`stressed and unstressed speech' Carnahan Conference, University of Kentucky,
`19
`[23] Davis, R. L., Sinnamon, J. T. and Cox, D. L. (1982) Voice verification
`upgrade. Report Texas Instruments Inc.
`[24] Rosenberg, A. E. (1976) 'Automatic 'speaker verification: a review' Proc
`IEEE 64(4) 475--'487.
`[25] DeGeorge, M. (1981) 'Experiments in automatic speaker verification'
`Carnahan Conference 1981, University of Kentucky, 103-110.
`[26] Bunge, E., Hofker, U., Hohne, H. D. and Kriener, B. (1977) 'Report about
`speaker-recognition investigations with the AUROS system' Frequenz 31(12)
`382.
`[27] Kuhn, M. H. and Geppert, R. (1980) 'A low cost speaker verification device'
`Carnahan Conf 1980, University of Kentucky, 57-61.
`[28] Ney, H. and Gierloff, R. (1982) 'Speaker recognition using a weighting
`technique' IEEE Int. Conf on Acoustics, Speech and Signal Processing
`I
`1645-1648
`[29] IBM Corp. (1967) Speech recognition GB 1179029. 19 April 1967.
`[30] Perkin Elmer Corp. (1970) Speech recognition GB 1289202. 21 April 1970.
`[31] NCR Corp (1977) Verification GB 1532944. 21 Feb 1977.
`[32] Kashyap, R. L. 'Speaker recognition from an unknown utterance and speaker
`speech interaction' Tr Acou. Speech & Signal Processing ASSP 24(6) 451.
`[33] Tokyo Shibaura Denki (1983) Individual verification apparatus based on
`speech recognition JP EP-86-064. 27 Jan 1983.
`[34] Reitboeck,-H. J. (1977). 'Speaker identification over telephone transmission
`channels� Carnahan 1977, University of Kentucky, 237-238.
`(35) Shridhar, M., Baraniecki, M. and Mohankrishnan, N. (1982) 'A unified
`approach to speaker verification with noisy speech input' Speech Communi-
`,
`cation 103-112.
`[36] Kiyohiro, S. (1985) Text-Independent Speaker Recognition Experiments using .
`.
`Codebook Computer Science Dept C-M U, 9 April 1985.
`[37) Moore, R. K. 'Systems for isolated and connected word recognition' NATO
`AS/ 16 74-143 Springer-V�rlag.
`[38] Nagel, R. N. and Rosenfeld, A. (1977) 'Computer detection of freehand
`forgeries' Trans on Computers C 26(9) 895.
`[39] Rediffusion Computers Ltd (1983) 'Rediffusion introduces low-cost signature
`verifier' Financial Times 24 Feb 1983, 19.
`[40] Chainer, T. J., Scranton, R. A. and Worthington T. K. {1985) Data input
`pen for signature verification US 4513437. 23 April 1985.
`[41] Crane, H. D. and Ostrem, J. S. {1983) 'Automatic signat�re verification
`
`1
`i
`
`Page 152 of 201
`
`
`
`Automated Personal Identification Methods
`
`135
`
`Using a three-axis force-sensitive pen' Trans on Sys., Man & Cybernet SMC-
`13{3) 329-337.
`[42] Quest Automation Ltd (1979) Transducer pad for electrographies GB 1539755.
`[43] Greenaway, D. L. (1978) Apparatus and Method for Producing an Electrical
`Signal Responsive to Handwriting Characteristics US 4122435. 24 Oct 1978.
`[44] Darringer, J. A., Denil, N. J. and Evangelisti, C. J. (1975) 'Speed pen' IBM
`Tech Disclosure Bulletin 18(7) 2374-5.
`(45] Radice, P. F. (1980) Personal Verification. Device US 4234868. 18 Nov 1980.
`[46] Worthington, T. K., Chainer T. J., Williford, J. D. and Gundersen, S. C.
`(1985) 'IBM dynamic signature verification' Computer Security, Elsevier
`Science Publishers BV, 129-54.
`[47] Zimmermann, K. P. and Varady, M. J. 'Handwriter, identification from one
`bit quantized pressure patterns' Pattern Recognition 18[1] 63-72.
`[48] Hale, W. J. and Paganini, B. J. (1980) 'An automatic personal verification
`system based on signature writing habits' Carnahan Conf. on Crime Counter,
`University of Kentucky, 121-125.
`[49] Stuckert, P. E. (1979) 'Magnetic pen and tablet' IBM Technical Disclosure
`Bul 22(3) 1245-1251.
`[50] de Bruyne, P. (1984) 'An ultrasonic radar graphic input tablet' Scienta
`Electrica 1-26.
`[5i] Lew, J. S. (1980). 'Optimal accelerometer layouts for data recovery in
`signature verification' IBM Journal of R & D 24(4) 496-511.
`[52] . Herbst, N. M. and. Liu, C. N. (1977) 'Automatic signature verification based
`on accelerometry' IBM Jour of R & D 245.
`[53] Liu, C. N., Herbst, N. M. and Anthony, N: J. (1979) 'Automatic signature
`verification: system description and field test results' TR Sys., Man & Cyber-
`·
`netics SNC 9(1) 35.
`[,54] de Bruyne, P. (1977) 'Developments in signature verification' I.nt Conf on
`Crime Countermeasure, Oxford Univ.
`[55) Watson, R. S. and Pobgee, P . .1. (1980) 'A computer to check signatures'
`Visible Language 13(3).
`[56] Fox, P. F. (1982) 'A practical method of personal identification by signature
`validation' IACSS Conference, Zurich, SWitzerland.
`[57] Umphress, D. and Williams, G. (1985) 'Identity verification through keyboard
`characteristics' Int J Man-Machine Studies 263-273.
`[58] Haberman, W. and Fejfar, A. (1976) 'Automatic identification of personnel
`through, speaker and signature verification - system design and testing'
`Carnahan Conference 1976, University of Kentucky, 23-30.
`
`Page 153 of 201
`
`
`
`Chapter 8
`Cryptography and the Smart
`Card
`D. W. DAVIES
`(Data Security Consultant)
`
`Cryptology is the key technology for secure systems ..
`
`8.1 Introduction
`
`The close relationship between smart card and cryptogtaphic techniques
`can be looked at from two directions. The smart card can be used as a
`component of a cryptographic system to improve its convenience or level
`of security. From the other viewpoint, the smart card itself is the main
`component of the system and cryptography is called upon to help it with
`its task. In this chapter we shall mainly adopt the second viewpoint,
`which is centred on smart card applic�tions but first let us look at the
`smart card as an adjunct to cryptography.
`The confidentiality of data on a communication line can be protected
`by enciphering it. Encipherment is a transformation which makes the
`transformed data seem meaningless to an outsider, yet which allows an
`inverse transformation, for those authorised to receive the information,
`which turns it back into its clear text. To separate the authorised readers
`from others, the authorised readers hold a secret value called a crypto
`graphic key without which decipherment is impossible. In the usual form
`of cryptography, this secret key is used as a parameter for both the
`encipherment and the decipherment functions.
`When cryptography is used to protect data travelling some distance,
`before it can go into operation a secret key must be established at both
`the sending and the receiving end. Conveying the key from one place to
`the other entails a risk of losing it to an opponent. A smart card can be
`used to store a key for secure transport. The use of this key can be
`authorised by means of a password, known only to authorised users, and
`the smart card itself can take part in the complex process of key manage
`ment. Some of the techniques are described later in the. chapter.
`Sometimes, cryptography is used to encipher information not for com
`munications purposes but to protect it while it is stored locally. It might
`be difficult to protect the local store from illicit access or information
`
`.,
`f
`
`Page 154 of 201
`
`
`
`Cryptography and the Smart Card
`
`137
`
`stored on a removable medium might be stolen or copied. When crypto
`
`graphy is used for stored data, the keys are not transported but their
`
`security is very important because they_ can unlock all the protection
`
`
`provided. Most computers are physically insecure, so a smart card can be
`used to hold the keys and the card taken away by its owner and stored in
`
`
`a safe place. Here also, a password can be used to unlock the secret key
`from the card.
`A related problem of cryptography is the protection
`of the cryptographic
`
`mechanism itself. Not only must the key be protected but also the place
`
`
`
`where the cryptographic transformations take place. Smart cards can help
`engine' of
`in this problem by becoming, themselves, the 'cryptographic
`
`the system. If they have enough processing power for the purpose, they
`
`
`
`can hold all the protective mechanism of a secure system, particularly at
`the terminal end where the processing demands are less severe. The
`
`
`computer itself, which might be an intelligent terminal, is physically
`
`insecure and any part of its store or process is open to tapping or
`
`
`'bugging'. To counter this, cryptographic methods are used and the keys,
`
`together with the cryptographic transformations, are contained entirely in
`
`smart cards. When these are removed, the system is locked up and the
`
`information it contains is safe against illegal access.
`These are· examples of the close relationship
`between smart cards and
`
`
`cryptography, seen from the side of the cryptographer who regards the
`
`
`smart card as an additional tool. Our viewpoint in the rest of this chapter
`
`is to think of the smart card as a main component of the system and see
`
`how cryptographic techniques are used for its purpose.
`
`8.2 PROTECTION F�OM PASSIVE AND ACTIVE ATTACKS
`
`Cryptographic techniques can be used in a large number of ways and for
`
`
`many different purposes. The basic purpose is to protect a system against
`
`
`misuse by impostors or unauthorised people. The first stage in protecting
`a system is to analyse the threats to the system and the risks they entail.
`
`We shall consider only those threats that are amenable to cryptographic
`protection and, as a first step we divide these into passive and active
`attacks.·
`A passive attack attempts to read information without changing it.
`Examples are the tapping of a telephone line, stealing or copying a
`diskette,
`observing a password by looking at the keyboard while it is
`
`
`entered or picking up stray electromagnetic radiation from which a mean
`ingful signal can be reconstructed. Generally speaking, these attacks are
`
`
`not difficult to carry out and in a widespread communication network it is
`impossible to prevent them. The tapping or bugging of voice conver
`sations-is a 'highly developed art which can be applied (with a few
`
`Page 155 of 201
`
`
`
`138
`
`Integrated Circuit Cards
`
`changes) to the collecting of digital data from communication lines, I/0
`channels, processors, stores, keyboards or any other part of the infor
`mation system. In these. circumstances, to preserve the confidentiality of
`information it must be transformed by encipherment and then transformed
`back into clear form for processing, printing or display. Wherever the
`information is in clear form there must be other ways to protect it, such
`as not allowing unauthorised persons into offices where information is
`displayed or printed.
`On the other hand, an active attack is one which seeks to alter the
`information, perhaps to falsify a transaction, prevent a debit to a bank
`account from reaching it or even destroy an entire file. Generally speaking,
`these active attacks are much more difficult to carry out and they require
`more skill and sometimes more luck to achieve their purpose. When they
`do succeed, the consequences are often more serious than those of a
`passive attack.
`It might seem that encipherment was all that is needed to prevent an
`active attack on data but this is not so. Some of the more extreme active
`attacks such as placing a bomb in the computer room are not amenable to
`cryptographic protection but, these aside, there is a much wider range of
`possibjlities against which protection is needed. Suppose, for example,
`that an important file is stored in enciphered form and updated from time
`to time, while still being enciphered. If we are not careful, an attacker
`simply replaces today's file by an older, one which will pass as genuine
`because it is properly enciphered. Whole transactions might be knocked
`out on a · communications line without detection. Some encipherment
`methods allow an attacker to change individual bits of a message or file
`without knowing the cryptographic key. Thus there is more to protection
`against an active attack than encipherment alone and the methods used
`have in the past been known as 'authentication' because they ensure that
`the information remains authentic. This term is discouraged by the recent
`usage of the International Standards Organisation and we speak therefore
`of data integrity meaning that the data takes the values it was intended,
`not those altered or substituted by an impostor.
`In an extended communication system it may not be possible to prevent
`the changing of information but we can at least hope to detect w�en the
`information loses its integrity. Thus we shall normally be speaking about
`integrity verification rather than the prev�ntion of an active attack.
`One form of active attack is the masquerade which means an unauthorised
`user pretending to the system that he is an authorised one. It goes further
`than this because any part of the system can be subject to a masquerade.
`A well known example is the program which asks the user to present his
`password. If the program is an impostor, the passwords can be given
`away to an enemy. Protection against masquerading should therefore be
`
`included in every interaction within the system where it is physically
`
`I I I i
`1
`I
`I
`
`I I
`
`\
`
`f. I
`
`i
`
`:; I
`I
`I j'
`
`Page 156 of 201
`
`
`
`Cryptography and the Smart Card
`
`139
`
`possible for an intruder to slip in. In the work of the International
`Standards Organ�ation this is known as peer entity authentication since
`the two communicating parts of the system are entities at the same level,
`i.e. peer entities. The particular case which we meet most frequently is the
`authentication of a user to the system. Masquerading as an authorised
`user is the commonest type of active attack on teleprocessing systems,
`typified by the 'hacker.' who mainly has to contend with rather simple
`password systems. Authentication of users can take much more secure
`forms, using smart cards, which will be described later in section 8.5. As
`in the present use of 'automatic teller machines' (cash dispensers), a
`token which the user holds together with a password which he remembers
`adds a lot to the security. A smart card, which is very difficult to forge,
`makes an ideal token for this purpose.
`Masquerading forms a bridge between passive and active attacks. By
`tapping the line to find the password or guessing the password or trying a
`lot of passwords the attacker obtains the means to enter the system in the
`guise of a genuine user. This gives him the possibility of both a passive
`and an active attack, the active attack requiring no more skill than normal ....
`use of the terminal, if a public network is used for access.
`In the next three sections of this chapter we shall describe the use of
`cryptography against passive attacks, preservation of data integrity and
`user authentication.
`·
`
`8.3 CRYPTOGRAPHY
`
`Cryptography is an ancient art which, like many others, is changing
`rapidly with the development of information processing technology. We
`are concerned only with ciphers, which use a cryptographic algorithm and
`not with codes which are based on large, arbitrary code books. Figure 8.1
`shows the nature of a cipher which employs two algorithms E and D to
`encipher and decipher information respectively. Encipherment operates
`on the plaintext using a cryptographic key (k) as parameter. Decipherment
`
`Key k
`
`Key k
`
`Plain text
`X
`
`E
`
`Cipher text
`y = Ek(x)
`
`D
`
`Plain text
`X
`
`Encipherment
`Decipherment
`Fig. 8.1 Notation for encipherment and decipherment.
`
`Page 157 of 201
`
`
`
`140
`
`Integrated Circuit Cards
`
`uses the same key to operate on the ciphertext y and restore the plaintext
`x. The symbols x and y can either represent single blocks of data on
`which the transformation takes place or streams of data, for example, a
`stream of ASCII characters passing over a communication link. This dis- ·
`tinction corresponds to two different types of cipher algorithm, a block
`cipher and a stream cipher. Later we shall give examples of both.
`
`8.3.1 Attacks on a cipher
`The cipher derives its strength, in part, from the key. If the secrecy
`depended on the algorithm alone, when this algorithm became known to
`an enemy, either by chance or by hard cryptographic work, it would be
`necessary to go back to the drawing board and design a new algorithm.
`All modern ciphers make use of a key so that, even when the algorithm is
`known, decipherment is not possible unless the particular key employed
`is known.
`The possible number of different keys is important, like the number of
`'differs' of the keys you use in a lock. If this is too small, a processor can
`be set to try all the keys and see which one generates a plausible plaintext.
`The total number of different key values is known as the size of the key
`space.
`If nothing were known about the pl�intext and it could therefore be
`any random set of digits, the most elaborate and fastest machine for
`searching through the keys would still tell .us nothing. We must either
`have some knowledge of redundancy in the plaintext or, better still, some
`examples of actual plaintext with their ciphertext equivalents. In many
`cases, probable words or phrases from the plaintext can be guessed when
`their place in the text is unknown. With any of these clues about the
`plaintext an attack can be mounted, by searching through all the keys, in
`principle. A very large keyspace prevents the breaking of a cipher by
`simple key searching but it does not ensure that other, cleverer methods
`will not succeed. Only long experience with the practical breaking of
`ciphers can enable the strength of a cipher to be evaluated. Ultimately,
`there can be no guarantee because the whole range of possible methods
`of attack can never be enumerated.
`
`8.3.2 The Data Encryption Standard
`
`The US Government employs 'Federal Information Processing Standards'
`for its own use. In 1973, the US National Bureau of Standards announced
`that it was contemplating a standard for data encryption and asked for
`proposals. The response was disappointing but after a second call in 1974
`
`·i : .
`
`I j
`
`Page 158 of 201
`
`
`
`Cryptography and the Smart Card
`
`141
`
`,..
`
`a proposal from IBM was seen to have promise and, after some changes,
`it was published in 1975 as a draft. Eventually this was adopted in 1976 as
`the Data Encryption Standard defined by FIPS Publication 46.
`The Data Encryption Standard or DES had a much wider influence
`than was originally intended. The US standards body ANSI made it a US
`standard and it was widely adopted, particularly in banking and financial
`services. Because of restrictions placed on the export of chips for the
`DES algorithms from the USA, its use outside the USA has largely been
`confined to banking and financial services. It became a de facto standard
`within this community.
`The DES is a block cipher with plaintext and ciphertext blocks of 64
`bits and it employs a key of 64 bits, but has 8 parity bits so that its
`keyspace depends only on 56 bits.
`The structure of the algorithm is described in detail in reference (1]. It
`uses a combination of bit permutations and substitutions. By substitution
`we mean that an input field is used as an address to look up a table and
`produce an output field. The DES employs eight substitutions, each with
`a 6 bit input and 4 bit output. These are known as the 'S boxes'.
`From the outset, the strength of the DES has been controversial. The
`size of its keyspace was criticised on the basis that a complete search
`through its keyspace. could be carried out in less than one day by a
`machine containing one million devices, each able to test one million keys
`. per second. The cost of such a machine was argued about but believed at
`the time to be in the region of $10M. The structure of the S boxes was
`clearly not random but the criteria for choosing these tables has never
`been published and the work done to develop the DES is classified. Many
`felt that this was unsatisfactory for a standard.
`The whole idea of a published algorithm was a new one. Though the
`secrecy of the algorithm is not assumed to be essential to the security of a
`cipher, no cipher had been published, studied and discussed so widely
`and in such detail:
`Studies in many places revealed interesting properties of the DES,
`including the existence of four weak keys and twelve semi-weak keys
`which give the cipher special properties. Interesting facets of the design of
`the DES continue to be discovered. However, none of these has shown
`any significant weakness of the algorithm beyond that which is implied by
`the size of its keyspace.
`As information technology improved, the cost of a key searching
`machine decreased until, by about 1985 it became obvious that a searching
`machine would be within the bounds of expenditure by a large corporation
`or a large criminal organisation, assuming that the searching time was
`allowed to extend to about one month. This is · a real threat when the
`same key is used for longer than this time. Multiple encryption using the
`DES with at least two different keys has been adopted by the financial
`
`Page 159 of 201
`
`
`
`142
`
`Integrated Circuit Cards
`
`community to overcome this problem. This is particularly relevant in key
`management for a large system where a master key remains in place for
`some time.
`By 1988 the US had decided not to renew their endorsement of the
`DES for federal purposes. When this was announced in .1986 it caused
`consternation among the financial users but the US Government stated
`· that it would not discourage use of the DES for financial transaction
`purposes.
`At the present time, adoption of the DES or any other cipher, as an
`international standard has been stopped by a resolution by the ISO
`Council so there is no role for the international standards body in the
`development of a replacement. This leaves the financial community in an
`unsatisfactory position, particularly where the use of the DES for safe
`guarding large payments is concerned. For small payments, with a good
`key management system, the DES is usually considered adequate.
`The algorithm was designed for implementation solely in hardware on a
`special chip, according to the US Federal standard. The ANSI standard
`removed this condition and many systems now employ the DES im
`plemented in software on a microprocessor. Because of its origins and the
`original intention to use hardware, the DES is not a convenient algonthm
`for software implementation and wa� outside the ability of the early smart
`cards. Recent developments in smart cards have made the DES more
`feasible so that it is now possible to, implement the DES in software
`within the. same smart cards. Where this is the algorithm required by
`financial transactions, incorporation of the cipher in a smart card is a
`useful contribution to security, enabling the card to store its own secret
`key and use it internally.
`A replacement for the DES will probably be a block cipher with the
`same size of plaintext and ciphertext blocks in order to ease the change
`from DES to the new cipher. The key size would have to be larger, at
`least 64 bits. A replacement for DES should be designed to be easy to
`·
`implement in hardware and software.
`
`8.3.3. Methods of using a block cipher
`
`There is a limited number of applications for a cipher which can only
`handle 64 bits. A longer message could be broken into blocks of 64 bits,
`for example blocks of eight ASCII characters, and each block in tum could
`be enciphered. This simple method of enciphering a longer stream is not
`recommended for two reasons. Firstly, an enemy could extract pieces of
`the message made of a number of blocks and reuse them or rearrange
`them in messages of his own construction. Secondly, though eight characters
`may have a large number of potential values, in some contexts there
`
`Page 160 of 201
`
`
`
`Cryptography and the Smart Card
`
`143
`
`64 bit store
`
`k
`
`E
`
`I I I
`I I I I I
`
`Cipher
`
`Plain
`
`Plai n
`
`Fig. 8.2 Cipher block chaining.
`would be blocks which repeated quite frequently and could be recognised
`in. ciphertext, so that methods used for breaking codebo6ks could be
`applied. For example, in English text the words 'of the', wHh the three
`spaces, when it occurs on the eight character boundary will probably be
`the most common block and can thus be recognised. In computer output,
`sequences such as all zeros or all spaces will easily be detected.
`To overcome these problems a method of using the DES was introduced
`known as cipher block chaining ( CBC) and this can be applied to any
`block cipher. It is illustrated in Figure 8.2 which shows the ciphertext
`produced by the sender added, modulo 2, to the next plaintext block.
`This chains each block to its neighbour, preventing .the code book analysis
`method and the separation and reassembly of blocks of text. When applying
`this method, the block which is modulo 2 added to the first plaintext
`block is called the initialising variable or IV. In many systems, a new IV is
`sent when each new key is distributed and the IV is kept as secret as the
`key.
`Decipherment presents no problems, since the ciphertext can be stored
`and modulo 2 added to the plaintext of the next block. There is no
`synchronising problem in the sense that, whatever happens to modify or
`momentarily interrupt the ciphertext, provided the boundaries of the 64
`btt block can be recognised at the receiver, the system will recover. For
`example, a single error in ciphertext results in completely random output
`for the plaintext block into which it is transformed. The same error goes
`into the 64 bit store and emerges in the following block, so two blocks of
`text are affected. After this, the error does not propagate any more. Even
`this amount of error extension can be troublesome. It can interact badly
`with error correction schemes and it greatly increases the average error
`rate on the line.
`The CBC mode of operation is most useful where the information
`already has some imposed block structure, as in formatted messages.
`When these messages have been assembled in a store, the CBC encipher
`ment can be applied to the whole stored message and then the result
`transmitted or stored, as required.
`
`Page 161 of 201
`
`
`
`144
`
`Integrated Circuit Cards
`64 bit shift register
`--
`
`64 bit shift register
`-
`
`k
`
`E
`
`64
`
`Select
`left bit
`1
`
`1
`
`1
`
`1
`
`E
`
`k
`
`64
`
`Select
`left bit
`
`1
`
`1
`
`Plain
`
`Plain
`
`Cipher
`Fig. 8.3 One-bit cipher feedback.
`This mode of operation is not convenient when the natural unit of
`
`
`
`
`
`communication is smaller than a block. For example, much. communication
`
`takes place in the form of 8 bit units or less, for example 8 bit ASCII
`
`characters or 5 bit telegraph codes. For these, the use of a 64 bit block
`
`
`
`
`
`cipher would mean that characters were held up, waiting for transmission,
`
`
`
`until the block was filled. Interactive communication between two operators
`
`
`were would require that, after a short pause, the residual characters
`
`
`
`
`
`padded out and transmitted. To satisfy these requirements, another method
`
`of using a block cipher was introduced which is cipher feedback (CFB).
`
`This is illustrated in Figure 8.3. At the top of the figure is a 64 bit shift
`
`
`
`register into which the ciphertext is introduced, bit by bit, from the right
`
`
`hand end. Thus it contains, at each end of the communication line, a
`
`64 bits of the ciphertext. At each end of the
`record of the last
`line, this
`
`
`
`
`block is enciphered and from the result only the left hand bit is extracted.
`
`This bit stream is added into the plaintext to produce the ciphertext.
`
`Effectively, it is a random bit stream generated by feeding back the
`is in this case used only in
`
`
`ciphertext itself. Note that the cipher algorithm
`
`
`
`
`the encryption mode; in fact a reversible algorithm is not really needed.
`
`
`Because this works on each bit as it arrives, it is completely transparent
`
`
`to the procedure used by the communicating parties, whether they are
`
`
`
`people or machines. It is normally used within the OSI architecture at the
`
`
`
`
`
`physical layer because it treats only the bit stream, without reference to
`
`
`its structure, and passes on each bit as it arrives.
`
`
`The entire DES algorithm calculation must be made again for each bit,
`
`this method will be slow:
`so if there is a limited speed of processing,
`
`As with the CBC mode, the shift register must be loaded before the
`
`
`
`process begins. In this case the initialising variable (IV) can be sent in
`
`
`
`clear over the line in a preamble. It can be in clear because the block is
`
`Page 162 of 201
`
`
`
`Cryptography and the Smart Card
`
`145
`
`Key
`
`E
`
`Key
`
`E
`
`64
`
`64
`
`Select
`left j bits
`
`Select
`left j bits
`
`· Plain text
`
`Cipher text
`
`Plain text
`
`Fig. 8.4 Output feedback.
`
`enciphered before it is ever used. Another possibility is to send no IV but
`to transmit random data for the first 64 bits, which will ensure that the
`two registers are synchronised.
`For this method, there is no synchronising because there is no 64 bit or
`other boundary to be observed. If an error occurs on the line or, for
`example, a bit is missed or an extra bit inserted, this causes an immediate
`error in the delivered plaintext and also affects the shift register during
`the next 64 bits. Consequently the error extension property of thi