Network Working Group                                        N Doraswamy
Internet Draft                                            [Bay Networks]
                                                               P Metzger
                                                             W A Simpson
expires in six months                                          July 1997

                      The ESP Triple DES Transform

Status of this Memo

   Follows draft-simpson-esp-des3-x-01.txt.

   This document is an Internet-Draft.  Internet Drafts are working doc-
   uments of the Internet Engineering Task Force (IETF), its Areas, and
   its Working Groups.  Note that other groups may also distribute work-
   ing documents as Internet Drafts.

   Internet Drafts are draft documents valid for a maximum of six
   months, and may be updated, replaced, or obsoleted by other documents
   at any time.  It is not appropriate to use Internet Drafts as refer-
   ence material, or to cite them other than as a ``working draft'' or
   ``work in progress.''

   To learn the current status of any Internet-Draft, please check the
   ``1id-abstracts.txt'' listing contained in the internet-drafts Shadow
   Directories on: (Africa) (Europe) (US East Coast) (US West Coast) (Pacific Rim)

   Distribution of this memo is unlimited.


   This document describes the "Triple" DES-EDE3-CBC block cipher trans-
   form interface used with the IP Encapsulating Security Payload (ESP).
   It provides compatible migration from RFC-1851.

Doraswamy, Metzger & S... expires in six months                 [Page i]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

1.  Introduction

   The Encapsulating Security Payload (ESP) [RFC-1827x] provides confi-
   dentiality for IP datagrams by encrypting the payload data to be pro-
   tected.  This specification describes the ESP use of a variant of the
   Cipher Block Chaining (CBC) mode of the US Data Encryption Standard
   (DES) algorithm [FIPS-46, FIPS-46-1, FIPS-74, FIPS-81].

   This variant, colloquially known as "Triple DES", processes each
   block three times, each time with a different key [Tuchman79].

   For an explanation of the use of CBC mode with this cipher, see [RFC-

   For more explanation and implementation information for Triple DES,
   see [Schneier95].

   This document assumes that the reader is familiar with the related
   document "Security Architecture for the Internet Protocol"
   [RFC-1825x], that defines the overall security plan for IP, and pro-
   vides important background for this specification.

   In this document, the key words "MAY", "MUST", "recommended",
   "required", and "SHOULD", are to be interpreted as described in

1.1.  Availability

   There were a number of US patents (see [Schneier95] for listing).
   All patents have expired.  Several freely available implementations
   have been published world-wide.

1.2.  Performance

   As this specification requires "outer" chaining, it is not possible
   to provide parallel computation for the same data stream.  Triple DES
   is approximately 2.5 times slower than "single" DES (rather than 3
   times), because inner permutations may be removed.

   Phil Karn has tuned DES-EDE3-CBC software to achieve 6.22 Mbps with a
   133 MHz Pentium.  Other DES speed estimates may be found at
   [Schneier95, page 279].  Your milage may vary.

Doraswamy, Metzger & S... expires in six months                 [Page 1]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

2.  Description
2.1.  Block Size

   The US Data Encryption Standard (DES) algorithm operates on blocks of
   64-bits (8 bytes).  This often requires padding before encrypting,
   and subsequent removal of padding after decrypting.

   The output is the same number of bytes that are input.  This facili-
   tates in-place encryption and decryption.

2.2.  Mode

                      P1             P2             Pi
                      |              |              |
               IV->->(X)    +>->->->(X)    +>->->->(X)
                      v     ^        v     ^        v
                   +-----+  ^     +-----+  ^     +-----+
               k1->|  E  |  ^ k1->|  E  |  ^ k1->|  E  |
                   +-----+  ^     +-----+  ^     +-----+
                      |     ^        |     ^        |
                      v     ^        v     ^        v
                   +-----+  ^     +-----+  ^     +-----+
               k2->|  D  |  ^ k2->|  D  |  ^ k2->|  D  |
                   +-----+  ^     +-----+  ^     +-----+
                      |     ^        |     ^        |
                      v     ^        v     ^        v
                   +-----+  ^     +-----+  ^     +-----+
               k3->|  E  |  ^ k3->|  E  |  ^ k3->|  E  |
                   +-----+  ^     +-----+  ^     +-----+
                      |     ^        |     ^        |
                      +>->->+        +>->->+        +>->->
                      |              |              |
                      C1             C2             Ci

   The DES-EDE3-CBC algorithm is a simple variant of the DES-CBC algo-
   rithm [RFC-wwww, RFC-1829x].  The "outer" chaining technique is used.

   In DES-EDE3-CBC, an Initialization Vector (IV) is XOR'd with the
   first 64-bit (8 byte) plaintext block (P1).  The keyed DES function
   is iterated three times, an encryption (Ek1) followed by a decryption
   (Dk2) followed by an encryption (Ek3), and generates the ciphertext
   (C1) for the block.  Each iteration uses an independant key: k1, k2
   and k3.

   For successive blocks, the previous ciphertext block is XOR'd with
   the current plaintext (Pi).  The keyed DES-EDE3 encryption function
   generates the ciphertext (Ci) for that block.

Doraswamy, Metzger & S... expires in six months                 [Page 2]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

   To decrypt, the order of the functions is reversed: decrypt with k3,
   encrypt with k2, decrypt with k1, and XOR the previous ciphertext

   Note that when all three keys (k1, k2 and k3) are the same, DES-
   EDE3-CBC is equivalent to DES-CBC.  This property allows the DES-EDE3
   hardware implementations to operate in DES mode without modification.

2.3.  Interaction with Authentication

   There is no known interaction of DES with any currently specified
   Authenticator algorithm.  Never-the-less, any Authenticator MUST use
   a separate and independently generated key.

3.  Initialization Vector

   DES-EDE3-CBC requires an Initialization Vector (IV) that is 64-bits
   (8 bytes) in length [RFC-wwww].

   By default, the 64-bit IV is generated from the 32-bit ESP Sequence
   Number field followed by (concatenated with) the bit-wise complement
   of the same 32-bit value:

      SN || -SN

   Alternative IV generation techniques MAY be specified when dynami-
   cally configured via a key management protocol.

   Security Notes:

      Using the Sequence Number provides an easy method for preventing
      IV repetition, and is sufficiently robust for practical use with
      the DES algorithm.  But, when used alone, cryptanalysis might be
      aided by the rare serendipitous occurrence when the Sequence Num-
      ber increments in exactly the same fashion as a corresponding bit
      position in the first block.

      No commonly used IP (Next Header) Protocols exhibit this property.
      Never-the-less, inclusion of the bit-wise complement ensures that
      Sequence Number bit changes are reflected twice in the IV.

Doraswamy, Metzger & S... expires in six months                 [Page 3]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

4.  Keys

   The secret DES-EDE3 key shared between the communicating parties is
   effectively 168-bits long.  This key consists of three independent
   56-bit quantities used by the DES algorithm.  Each of the three
   56-bit sub-keys is stored as a 64-bit (8 byte) quantity, with the
   least significant bit of each byte used as a parity bit.

4.1.  Weak Keys

   DES has 64 known weak keys, including so-called semi-weak keys and
   possibly-weak keys [Schneier95, pp 280-282].  The likelihood of pick-
   ing one at random is negligible.

   For DES-EDE3, there is no known need to reject weak or complementa-
   tion keys.  Any weakness is obviated by the other keys.

   However, since checking for weak keys is quite easy, the configura-
   tion mechanisms are expected to incorporate the test.

4.2.  Manual Key Management

   When configured manually, three independently generated keys are
   required, in the order used for encryption, and 64-bits (8 bytes) are
   configured for each individual key.

   Keys with incorrect parity SHOULD be rejected by the configuration
   utility, ensuring that the keys have been correctly configured.

   Each key is examined sequentially, in the order used for encryption.
   A key that is identical to a previous key MAY be rejected.  The 64
   known weak DES keys [RFC-1829x] SHOULD be rejected.

4.3.  Automated Key Management

   When configured via a Security Association management protocol, three
   independently generated keys are required, in the order used for
   encryption, and 64-bits (8 bytes) are returned for each individual

   The key manager MAY be required to generate the correct parity.
   Alternatively, the least significant bit of each key byte is ignored,
   or locally set to parity by the DES implementation.

   Each key is examined sequentially, in the order used for encryption.

Doraswamy, Metzger & S... expires in six months                 [Page 4]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

   A key that is identical to a previous key MUST be rejected.  The 64
   known weak DES keys [RFC-1829x] MUST be rejected.

4.4.  Refresh Rate

   To prevent differential and linear cryptanalysis of collisions [RFC-
   wwww], no more than 2**32 plaintext blocks SHOULD be encrypted with
   the same key.  Depending on the average size of the datagrams, the
   key SHOULD be changed at least as frequently as 2**30 datagrams.

5.  ESP Alterations
5.1.  ESP Sequence Number

   The Sequence Number is a 32-bit (4 byte) unsigned counter.  This
   field protects against replay attacks, and may also be used for syn-
   chronization by stream or block-chaining ciphers.

   When configured manually, the first value sent SHOULD be a random
   number.  The limited anti-replay security of the sequence of data-
   grams depends upon the unpredictability of the values.

   When configured via an automated Security Association management pro-
   tocol, the first value sent is 1, unless otherwise negotiated.

   Thereafter, the value is monotonically increased for each datagram
   sent.  A replacement SPI SHOULD be established before the value
   repeats.  That is, no more than 2**32 datagrams SHOULD be sent with
   any single key.

5.2.  ESP Padding

   The Padding field may be zero or more bytes in length.

   Prior to encryption, this field is filled with a series of integer
   values to align the Pad Length and Payload Type fields at the end of
   a 64-bit (8 byte) block boundary (measured from the beginning of the
   Transform Data).

   By default, each byte contains the index of the byte.  For example,
   three pad bytes would contain the values 1, 2, 3.

   After decryption, this field MAY be examined for a valid series of
   integer values.  Verification of the sequence of values is at the
   discretion of the receiver.

Doraswamy, Metzger & S... expires in six months                 [Page 5]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

Operational Considerations

   The specification provides only a few manually configurable parame-

      Manually configured SPIs are limited in range to aid operations.
      Automated SPIs are pseudo-randomly distributed throughout the
      remaining 2**32 values.

      Default: 0 (none).  Range: 256 to 65,535.

   SPI LifeTime (SPILT)
      Manually configured LifeTimes are generally measured in days.
      Automated LifeTimes are specified in seconds.

      Default: 32 days (2,764,800 seconds).  Maximum: 182 days
      (15,724,800 seconds).

   Replay Window
      Long term replay prevention requires automated configuration.
      Also, some earlier implementations used pseudo-random values.
      This check must only be used with those peers that have imple-
      mented this feature.

      Default: 0 (checking off).  Range: 32 to 256.

   Pad Values
      New implementations use verifiable values.  However, some earlier
      implementations used pseudo-random values.  This check must only
      be used with those peers that have implemented this feature.

      Also, some operations desire additional padding to inhibit traffic

      Default: 0 (checking off).  Range: 7 to 255.

      Three 56-bit keys are configured as a 192-bit quantity, with par-
      ity included as appropriate.

   Each party configures a list of known SPIs and symmetric secret-keys.

   In addition, each party configures local policy that determines what
   access (if any) is granted to the holder of a particular SPI.  For
   example, a party might allow FTP, but prohibit Telnet.  Such consid-
   erations are outside the scope of this document.

Doraswamy, Metzger & S... expires in six months                 [Page 6]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

Security Considerations

   Users need to understand that the quality of the security provided by
   this specification depends completely on the strength of the Triple
   DES algorithm, the correctness of that algorithm's implementation,
   the security of the Security Association management mechanism and its
   implementation, the strength of the key [CN94], and upon the correct-
   ness of the implementations in all of the participating nodes.

   The padding bytes have a predictable value.  They provide a small
   measure of tamper detection on their own block and the previous block
   in CBC mode.  This makes it somewhat harder to perform splicing
   attacks, and avoids a possible covert channel.  This small amount of
   known plaintext does not create any problems for modern ciphers.

   It was originally thought that DES might be a group, but it has been
   demonstrated that it is not [CW92].  Since DES is not a group, compo-
   sition of multiple rounds of DES is not equivalent to simply using
   DES with a different key.

   Triple DES with independent keys is not, as naively might be
   expected, as difficult to break by brute force as a cryptosystem with
   three times the keylength.  A space/time tradeoff has been shown
   which can brute-force break triple block encryptions in the time
   naively expected for double encryption [MH81].

   However, "double" DES (DES-EE2) can be broken with a meet-in-the-
   middle attack, without significantly more complexity than breaking
   DES requires [ibid].  DES-EDE3 with three independant keys is actu-
   ally needed to provide significantly more security than "single" DES.

   An attack has been shown on DES-EDE2 (using only two independent
   keys) [Tuchman79] that is somewhat (sixteen times) faster than
   exhaustive search [OW91].  Again, DES-EDE3 with three independant
   keys is actually needed to provide the expected level of security.

   Although it is widely believed that DES-EDE3 is substantially
   stronger than single DES alone, as it is less amenable to brute force
   attack, it should be noted that real cryptanalysis of DES-EDE3 might
   not use brute force methods at all.  Instead, it might be performed
   using variants on differential [BS93] or linear [Matsui94] cryptanal-
   ysis.  It should also be noted that no encryption algorithm is perma-
   nently safe from brute force attack, because of the increasing speed
   of modern computers.

   As with all cryptosystems, those responsible for applications with
   substantial risk when security is breeched should pay close attention
   to developments in cryptology, and especially cryptanalysis, and

Doraswamy, Metzger & S... expires in six months                 [Page 7]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

   switch to other transforms should DES-EDE3 prove weak.

Changes from RFC-1851:

   This specification results in the same default bits-on-the-wire as
   the 32-bit IV calculation method of RFC-1851.  The 32-bit field is
   semantically identical to a Sequence Number when implemented as a
   counter (the recommended method).

   The 64-bit explicit IV option is deprecated, as no hardware manufac-
   turers were found that required it.  It does not meet 64-bit field
   alignment expectations of IPv6, it is a cryptographically weaker con-
   struct than a calculated IV [Bellovin96], and it conflicts with the
   use of a Sequence Number immediately following the SPI.

   Clarified to specify "outer" CBC, as originally intended.

   Updated performance estimates.  Replaced erroneous text about paral-
   lel computation.

   Padding is a known series of integers, that may be checked upon

   Many implementation details by Karn were found to be common to all
   ESP Ciphers, and are awaiting consolidation in the ESP specification.

   Added an operational section.

   Updated acknowledgements, references, and contacts.

   Reorganized according to the new "road map" document.

Doraswamy, Metzger & S... expires in six months                 [Page 8]

DRAFT                       ESP DES-EDE3-CBC                   July 1997


   The basic field naming and layout is based on "swIPe" [IBK93, IB93].

   Some of the text of this specification was derived from work by Ran-
   dall Atkinson for the SIP, SIPP, and IPv6 Working Groups.

   Perry Metzger provided the original Security Considerations text,
   some of which is distributed throughout the document.

   William Allen Simpson was responsible for the name and semantics of
   the SPI, the IV calculation technique(s), editing and formatting.

   The use of known padding values was suggested in various forms by
   Robert Baldwin, Phil Karn, and David Wagner.  This specification uses
   Self-Describing-Padding [RFC-1570].

   Steve Bellovin, Angelos Keromytis, Holger Kummert, and Rodney Thayer
   provided useful critiques of earlier versions of this document.


            Bellovin, S., "Problem Areas for the IP Security Protocols",
            Proceedings of the Sixth Usenix Security Symposium, July

   [BS93]   Biham, E., and Shamir, A., "Differential Cryptanalysis of
            the Data Encryption Standard", Berlin: Springer-Verlag,

   [CN94]   Carroll, J.M., and Nudiati, S., "On Weak Keys and Weak Data:
            Foiling the Two Nemeses", Cryptologia, Vol. 18 No. 23 pp.
            253-280, July 1994.

   [CW92]   Campbell, K.W., and Wiener, M.J., "Proof that DES Is Not a
            Group", Advances in Cryptology -- Crypto '92 Proceedings,
            Berlin: Springer-Verlag, 1993, pp 518-526.

            US National Bureau of Standards, "Data Encryption Standard",
            Federal Information Processing Standard (FIPS) Publication
            46, January 1977.

            US National Bureau of Standards, "Data Encryption Standard",
            Federal Information Processing Standard (FIPS) Publication

Doraswamy, Metzger & S... expires in six months                 [Page 9]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

            46-1, January 1988.

            US National Bureau of Standards, "Guidelines for Implement-
            ing and Using the Data Encryption Standard", Federal Infor-
            mation Processing Standard (FIPS) Publication 74, April

            US National Bureau of Standards, "DES Modes of Operation"
            Federal Information Processing Standard (FIPS) Publication
            81, December 1980.

   [IB93]   Ioannidis, J., and Blaze, M., "The Architecture and Imple-
            mentation of Network-Layer Security Under Unix", Proceedings
            of the Fourth Usenix Security Symposium, Santa Clara Cali-
            fornia, October 1993.

   [IBK93]  Ioannidis, J., Blaze, M., and Karn, P., "swIPe: Network-
            Layer Security for IP", Presentation at the 26th Internet
            Engineering Task Force, Columbus Ohio, March 1993.

            Matsui, M., "Linear Cryptanalysis method for DES Cipher,"
            Advances in Cryptology -- Eurocrypt '93 Proceedings, Berlin:
            Springer-Verlag, 1994.

   [MH81]   Merkle, R.C., and Hellman, M., "On the Security of Multiple
            Encryption", Communications of the ACM, v. 24 n. 7, 1981,
            pp. 465-467.

   [OW91]   van Oorschot, P.C., and Weiner, M.J.  "A Known-Plaintext
            Attack on Two-Key Triple Encryption", Advances in Cryptology
            -- Eurocrypt '90 Proceedings, Berlin: Springer-Verlag, 1991,
            pp. 318-325.

            Simpson, W., "PPP LCP Extensions", DayDreamer, January 1994.

            Atkinson, R., "Security Architecture for the Internet Proto-
            col", Naval Research Laboratory, July 1995.

            Simpson, W., "IP Encapsulating Security Protocol (ESP) for

Doraswamy, Metzger & S... expires in six months                [Page 10]

DRAFT                       ESP DES-EDE3-CBC                   July 1997

            Karn, P., Metzger, P., Simpson, W.A., "The ESP DES-CBC
            Transform", work in progress.

            Bradner, S., "Key words for use in RFCs to Indicate Require-
            ment Levels", BCP 14, Harvard University, March 1997.

            Simpson, W.A, "ESP with Cipher Block Chaining (CBC)", work
            in progress.

            Schneier, B., "Applied Cryptography Second Edition", John
            Wiley & Sons, New York, NY, 1995.  ISBN 0-471-12845-7.

            Tuchman, W, "Hellman Presents No Shortcut Solutions to DES",
            IEEE Spectrum, v. 16 n. 7, July 1979, pp. 40-41.

Doraswamy, Metzger & S... expires in six months                [Page 11]

DRAFT                       ESP DES-EDE3-CBC                   July 1997


   Comments about this document should be discussed on the
   mailing list.

   Questions about this document can also be directed to:

      Naganand Doraswamy
      Bay Networks
      3 Federal Street  #BL3-04
      Billerica, Massachusetts  01821


      Perry Metzger
      Piermont Information Systems Inc.
      160 Cabrini Blvd., Suite #2
      New York, NY  10033

      William Allen Simpson
      Computer Systems Consulting Services
      1384 Fontaine
      Madison Heights, Michigan  48071


Doraswamy, Metzger & S... expires in six months                [Page 12]