RFC 9048

Internet Engineering Task Force (IETF)                          J. Arkko
Request for Comments: 9048                                 V. Lehtovirta
Updates: 4187, 5448                                          V. Torvinen
Category: Informational                                         Ericsson
ISSN: 2070-1721                                                P. Eronen
                                                            October 2021

   Improved Extensible Authentication Protocol Method for 3GPP Mobile
          Network Authentication and Key Agreement (EAP-AKA')


   The 3GPP mobile network Authentication and Key Agreement (AKA) is an
   authentication mechanism for devices wishing to access mobile
   networks.  RFC 4187 (EAP-AKA) made the use of this mechanism possible
   within the Extensible Authentication Protocol (EAP) framework.  RFC
   5448 (EAP-AKA') was an improved version of EAP-AKA.

   This document is the most recent specification of EAP-AKA',
   including, for instance, details about and references related to
   operating EAP-AKA' in 5G networks.

   EAP-AKA' differs from EAP-AKA by providing a key derivation function
   that binds the keys derived within the method to the name of the
   access network.  The key derivation function has been defined in the
   3rd Generation Partnership Project (3GPP).  EAP-AKA' allows its use
   in EAP in an interoperable manner.  EAP-AKA' also updates the
   algorithm used in hash functions, as it employs SHA-256 / HMAC-
   SHA-256 instead of SHA-1 / HMAC-SHA-1, which is used in EAP-AKA.

   This version of the EAP-AKA' specification defines the protocol
   behavior for both 4G and 5G deployments, whereas the previous version
   defined protocol behavior for 4G deployments only.  While EAP-AKA' as
   defined in RFC 5448 is not obsolete, this document defines the most
   recent and fully backwards-compatible specification of EAP-AKA'.
   This document updates both RFCs 4187 and 5448.

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Not all documents
   approved by the IESG are candidates for any level of Internet
   Standard; see Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at

Copyright Notice

   Copyright (c) 2021 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (https://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Table of Contents

   1.  Introduction
   2.  Requirements Language
   3.  EAP-AKA'
     3.1.  AT_KDF_INPUT
     3.2.  AT_KDF
     3.3.  Key Derivation
     3.4.  Hash Functions
       3.4.1.  PRF'
       3.4.2.  AT_MAC
       3.4.3.  AT_CHECKCODE
     3.5.  Summary of Attributes for EAP-AKA'
   4.  Bidding Down Prevention for EAP-AKA
     4.1.  Summary of Attributes for EAP-AKA
   5.  Peer Identities
     5.1.  Username Types in EAP-AKA' Identities
     5.2.  Generating Pseudonyms and Fast Re-Authentication Identities
     5.3.  Identifier Usage in 5G
       5.3.1.  Key Derivation
       5.3.2.  EAP Identity Response and EAP-AKA' AT_IDENTITY
   6.  Exported Parameters
   7.  Security Considerations
     7.1.  Privacy
     7.2.  Discovered Vulnerabilities
     7.3.  Pervasive Monitoring
     7.4.  Security Properties of Binding Network Names
   8.  IANA Considerations
     8.1.  Type Value
     8.2.  Attribute Type Values
     8.3.  Key Derivation Function Namespace
   9.  References
     9.1.  Normative References
     9.2.  Informative References
   Appendix A.  Changes from RFC 5448
   Appendix B.  Changes to RFC 4187
   Appendix C.  Importance of Explicit Negotiation
   Appendix D.  Test Vectors

   Authors' Addresses

1.  Introduction

   The 3GPP mobile network Authentication and Key Agreement (AKA) is an
   authentication mechanism for devices wishing to access mobile
   networks.  [RFC4187] (EAP-AKA) made the use of this mechanism
   possible within the Extensible Authentication Protocol (EAP)
   framework [RFC3748].

   EAP-AKA' is an improved version of EAP-AKA.  EAP-AKA' was defined in
   RFC 5448 [RFC5448], and it updated EAP-AKA [RFC4187].

   This document is the most recent specification of EAP-AKA',
   including, for instance, details about and references related to
   operating EAP-AKA' in 5G networks.  This document does not obsolete
   RFC 5448; however, this document is the most recent and fully
   backwards-compatible specification.

   EAP-AKA' is commonly implemented in mobile phones and network
   equipment.  It can be used for authentication to gain network access
   via Wireless LAN networks and, with 5G, also directly to mobile

   EAP-AKA' differs from EAP-AKA by providing a different key derivation
   function.  This function binds the keys derived within the method to
   the name of the access network.  This limits the effects of
   compromised access network nodes and keys.  EAP-AKA' also updates the
   algorithm used for hash functions.

   The EAP-AKA' method employs the derived keys CK' and IK' from the
   3GPP specification [TS-3GPP.33.402] and updates the hash function
   that is used to SHA-256 [FIPS.180-4] and HMAC to HMAC-SHA-256.
   Otherwise, EAP-AKA' is equivalent to EAP-AKA.  Given that a different
   EAP method Type value is used for EAP-AKA and EAP-AKA', a mutually
   supported method may be negotiated using the standard mechanisms in
   EAP [RFC3748].

         Note that any change of the key derivation must be unambiguous
         to both sides in the protocol.  That is, it must not be
         possible to accidentally connect old equipment to new equipment
         and get the key derivation wrong or to attempt to use incorrect
         keys without getting a proper error message.  See Appendix C
         for further information.

         Note also that choices in authentication protocols should be
         secure against bidding down attacks that attempt to force the
         participants to use the least secure function.  See Section 4
         for further information.

   This specification makes the following changes from RFC 5448:

   *  Updates the reference that specifies how the Network Name field is
      constructed in the protocol.  This update ensures that EAP-AKA' is
      compatible with 5G deployments.  RFC 5448 referred to the Release
      8 version of [TS-3GPP.24.302].  This document points to the first
      5G version, Release 16.

   *  Specifies how EAP and EAP-AKA' use identifiers in 5G.  Additional
      identifiers are introduced in 5G, and for interoperability, it is
      necessary that the right identifiers are used as inputs in the key
      derivation.  In addition, for identity privacy it is important
      that when privacy-friendly identifiers in 5G are used, no
      trackable, permanent identifiers are passed in EAP-AKA', either.

   *  Specifies session identifiers and other exported parameters, as
      those were not specified in [RFC5448] despite requirements set
      forward in [RFC5247] to do so.  Also, while [RFC5247] specified
      session identifiers for EAP-AKA, it only did so for the full
      authentication case, not for the case of fast re-authentication.

   *  Updates the requirements on generating pseudonym usernames and
      fast re-authentication identities to ensure identity privacy.

   *  Describes what has been learned about any vulnerabilities in AKA
      or EAP-AKA'.

   *  Describes the privacy and pervasive monitoring considerations
      related to EAP-AKA'.

   *  Adds summaries of the attributes.

   Some of the updates are small.  For instance, the reference update to
   [TS-3GPP.24.302] does not change the 3GPP specification number, only
   the version.  But this reference is crucial for the correct
   calculation of the keys that result from running the EAP-AKA' method,
   so an RFC update pointing to the newest version was warranted.

         Note: Any further updates in 3GPP specifications that affect,
         for instance, key derivation is something that EAP-AKA'
         implementations need to take into account.  Upon such updates,
         there will be a need to update both this specification and the

   It is an explicit non-goal of this specification to include any other
   technical modifications, addition of new features, or other changes.
   The EAP-AKA' base protocol is stable and needs to stay that way.  If
   there are any extensions or variants, those need to be proposed as
   standalone extensions or even as different authentication methods.

   The rest of this specification is structured as follows.  Section 3
   defines the EAP-AKA' method.  Section 4 adds support to EAP-AKA to
   prevent bidding down attacks from EAP-AKA'.  Section 5 specifies
   requirements regarding the use of peer identities, including how 5G
   identifiers are used in the EAP-AKA' context.  Section 6 specifies
   which parameters EAP-AKA' exports out of the method.  Section 7
   explains the security differences between EAP-AKA and EAP-AKA'.
   Section 8 describes the IANA considerations, and Appendix A and
   Appendix B explain the updates to RFC 5448 (EAP-AKA') and RFC 4187
   (EAP-AKA) that have been made in this specification.  Appendix C
   explains some of the design rationale for creating EAP-AKA'.
   Finally, Appendix D provides test vectors.

2.  Requirements Language

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "OPTIONAL" in this document are to be interpreted as described in
   BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
   capitals, as shown here.

3.  EAP-AKA'

   EAP-AKA' is an EAP method that follows the EAP-AKA specification
   [RFC4187] in all respects except the following:

   *  It uses the Type code 0x32, not 0x17 (which is used by EAP-AKA).

   *  It carries the AT_KDF_INPUT attribute, as defined in Section 3.1,
      to ensure that both the peer and server know the name of the
      access network.

   *  It supports key derivation function negotiation via the AT_KDF
      attribute (Section 3.2) to allow for future extensions.

   *  It calculates keys as defined in Section 3.3, not as defined in

   *  It employs SHA-256 / HMAC-SHA-256 [FIPS.180-4], not SHA-1 / HMAC-
      SHA-1 [RFC2104] (see Section 3.4).

   Figure 1 shows an example of the authentication process.  Each
   message AKA'-Challenge and so on represents the corresponding message
   from EAP-AKA, but with the EAP-AKA' Type code.  The definition of
   these messages, along with the definition of attributes AT_RAND,
   AT_AUTN, AT_MAC, and AT_RES can be found in [RFC4187].

    Peer                                                    Server
       |                       EAP-Request/Identity             |
       |                                                        |
       |  EAP-Response/Identity                                 |
       |  (Includes user's Network Access Identifier, NAI)      |
       |         +--------------------------------------------------+
       |         | Server determines the network name and ensures   |
       |         | that the given access network is authorized to   |
       |         | use the claimed name.  The server then runs the  |
       |         | AKA' algorithms generating RAND and AUTN, and    |
       |         | derives session keys from CK' and IK'.  RAND and |
       |         | AUTN are sent as AT_RAND and AT_AUTN attributes, |
       |         | whereas the network name is transported in the   |
       |         | AT_KDF_INPUT attribute.  AT_KDF signals the used |
       |         | key derivation function.  The session keys are   |
       |         | used in creating the AT_MAC attribute.           |
       |         +--------------------------------------------------+
       |                         EAP-Request/AKA'-Challenge     |
       |        (AT_RAND, AT_AUTN, AT_KDF, AT_KDF_INPUT, AT_MAC)|
   +------------------------------------------------------+     |
   | The peer determines what the network name should be, |     |
   | based on, e.g., what access technology it is using.  |     |
   | The peer also retrieves the network name sent by     |     |
   | the network from the AT_KDF_INPUT attribute.  The    |     |
   | two names are compared for discrepancies, and if     |     |
   | necessary, the authentication is aborted.  Otherwise,|     |
   | the network name from AT_KDF_INPUT attribute is      |     |
   | used in running the AKA' algorithms, verifying AUTN  |     |
   | from AT_AUTN and MAC from AT_MAC attributes.  The    |     |
   | peer then generates RES.  The peer also derives      |     |
   | session keys from CK'/IK'.  The AT_RES and AT_MAC    |     |
   | attributes are constructed.                          |     |
   +------------------------------------------------------+     |
       | EAP-Response/AKA'-Challenge                            |
       | (AT_RES, AT_MAC)                                       |
       |         +--------------------------------------------------+
       |         | Server checks the RES and MAC values received    |
       |         | in AT_RES and AT_MAC, respectively.  Success     |
       |         | requires both to be found correct.               |
       |         +--------------------------------------------------+
       |                                           EAP-Success  |

                 Figure 1: EAP-AKA' Authentication Process

   EAP-AKA' can operate on the same credentials as EAP-AKA and employ
   the same identities.  However, EAP-AKA' employs different leading
   characters than EAP-AKA for the conventions given in Section 4.1.1 of
   [RFC4187] for usernames based on International Mobile Subscriber
   Identifier (IMSI).  For 4G networks, EAP-AKA' MUST use the leading
   character "6" (ASCII 36 hexadecimal) instead of "0" for IMSI-based
   permanent usernames.  For 5G networks, the leading character "6" is
   not used for IMSI-based permanent usernames.  Identifier usage in 5G
   is specified in Section 5.3.  All other usage and processing of the
   leading characters, usernames, and identities is as defined by EAP-
   AKA [RFC4187].  For instance, the pseudonym and fast re-
   authentication usernames need to be constructed so that the server
   can recognize them.  As an example, a pseudonym could begin with a
   leading "7" character (ASCII 37 hexadecimal) and a fast re-
   authentication username could begin with "8" (ASCII 38 hexadecimal).
   Note that a server that implements only EAP-AKA may not recognize
   these leading characters.  According to Section 4.1.4 of [RFC4187],
   such a server will re-request the identity via the EAP-Request/AKA-
   Identity message, making obvious to the peer that EAP-AKA and
   associated identity are expected.


   The format of the AT_KDF_INPUT attribute is shown below.

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      | AT_KDF_INPUT  | Length        | Actual Network Name Length    |
      |                                                               |
      .                        Network Name                           .
      .                                                               .
      |                                                               |

   The fields are as follows:

      This is set to 23.

      The length of the attribute, calculated as defined in [RFC4187],
      Section 8.1.

   Actual Network Name Length
      This is a 2-byte actual length field, needed due to the
      requirement that the previous field is expressed in multiples of 4
      bytes per the usual EAP-AKA rules.  The Actual Network Name Length
      field provides the length of the network name in bytes.

   Network Name
      This field contains the network name of the access network for
      which the authentication is being performed.  The name does not
      include any terminating null characters.  Because the length of
      the entire attribute must be a multiple of 4 bytes, the sender
      pads the name with 1, 2, or 3 bytes of all zero bits when

   Only the server sends the AT_KDF_INPUT attribute.  The value is sent
   as specified in [TS-3GPP.24.302] for both non-3GPP access networks
   and for 5G access networks.  Per [TS-3GPP.33.402], the server always
   verifies the authorization of a given access network to use a
   particular name before sending it to the peer over EAP-AKA'.  The
   value of the AT_KDF_INPUT attribute from the server MUST be non-
   empty, with a greater than zero length in the Actual Network Name
   Length field.  If the AT_KDF_INPUT attribute is empty, the peer
   behaves as if AUTN had been incorrect and authentication fails.  See
   Section 3 and Figure 3 of [RFC4187] for an overview of how
   authentication failures are handled.

   In addition, the peer MAY check the received value against its own
   understanding of the network name.  Upon detecting a discrepancy, the
   peer either warns the user and continues, or fails the authentication
   process.  More specifically, the peer SHOULD have a configurable
   policy that it can follow under these circumstances.  If the policy
   indicates that it can continue, the peer SHOULD log a warning message
   or display it to the user.  If the peer chooses to proceed, it MUST
   use the network name as received in the AT_KDF_INPUT attribute.  If
   the policy indicates that the authentication should fail, the peer
   behaves as if AUTN had been incorrect and authentication fails.

   The Network Name field contains a UTF-8 string.  This string MUST be
   constructed as specified in [TS-3GPP.24.302] for "Access Network
   Identity".  The string is structured as fields separated by colons
   (:).  The algorithms and mechanisms to construct the identity string
   depend on the used access technology.

   On the network side, the network name construction is a configuration
   issue in an access network and an authorization check in the
   authentication server.  On the peer, the network name is constructed
   based on the local observations.  For instance, the peer knows which
   access technology it is using on the link, it can see information in
   a link-layer beacon, and so on.  The construction rules specify how
   this information maps to an access network name.  Typically, the
   network name consists of the name of the access technology or the
   name of the access technology followed by some operator identifier
   that was advertised in a link-layer beacon.  In all cases,
   [TS-3GPP.24.302] is the normative specification for the construction
   in both the network and peer side.  If the peer policy allows running
   EAP-AKA' over an access technology for which that specification does
   not provide network name construction rules, the peer SHOULD rely
   only on the information from the AT_KDF_INPUT attribute and not
   perform a comparison.

   If a comparison of the locally determined network name and the one
   received over EAP-AKA' is performed on the peer, it MUST be done as
   follows.  First, each name is broken down to the fields separated by
   colons.  If one of the names has more colons and fields than the
   other one, the additional fields are ignored.  The remaining
   sequences of fields are compared, and they match only if they are
   equal character by character.  This algorithm allows a prefix match
   where the peer would be able to match "", "FOO", and "FOO:BAR"
   against the value "FOO:BAR" received from the server.  This
   capability is important in order to allow possible updates to the
   specifications that dictate how the network names are constructed.
   For instance, if a peer knows that it is running on access technology
   "FOO", it can use the string "FOO" even if the server uses an
   additional, more accurate description, e.g., "FOO:BAR", that contains
   more information.

   The allocation procedures in [TS-3GPP.24.302] ensure that conflicts
   potentially arising from using the same name in different types of
   networks are avoided.  The specification also has detailed rules
   about how a client can determine these based on information available
   to the client, such as the type of protocol used to attach to the
   network, beacons sent out by the network, and so on.  Information
   that the client cannot directly observe (such as the type or version
   of the home network) is not used by this algorithm.

   The AT_KDF_INPUT attribute MUST be sent and processed as explained
   above when AT_KDF attribute has the value 1.  Future definitions of
   new AT_KDF values MUST define how this attribute is sent and

3.2.  AT_KDF

   AT_KDF is an attribute that the server uses to reference a specific
   key derivation function.  It offers a negotiation capability that can
   be useful for future evolution of the key derivation functions.

   The format of the AT_KDF attribute is shown below.

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      | AT_KDF        | Length        |    Key Derivation Function    |

   The fields are as follows:

      This is set to 24.

      The length of the attribute, calculated as defined in [RFC4187],
      Section 8.1.  For AT_KDF, the Length field MUST be set to 1.

   Key Derivation Function
      An enumerated value representing the key derivation function that
      the server (or peer) wishes to use.  Value 1 represents the
      default key derivation function for EAP-AKA', i.e., employing CK'
      and IK' as defined in Section 3.3.

   Servers MUST send one or more AT_KDF attributes in the EAP-Request/
   AKA'-Challenge message.  These attributes represent the desired
   functions ordered by preference, the most preferred function being
   the first attribute.

   Upon receiving a set of these attributes, if the peer supports and is
   willing to use the key derivation function indicated by the first
   attribute, the function is taken into use without any further
   negotiation.  However, if the peer does not support this function or
   is unwilling to use it, it does not process the received EAP-Request/
   AKA'-Challenge in any way except by responding with the EAP-Response/
   AKA'-Challenge message that contains only one attribute, AT_KDF with
   the value set to the selected alternative.  If there is no suitable
   alternative, the peer behaves as if AUTN had been incorrect and
   authentication fails (see Figure 3 of [RFC4187]).  The peer fails the
   authentication also if there are any duplicate values within the list
   of AT_KDF attributes (except where the duplication is due to a
   request to change the key derivation function; see below for further

   Upon receiving an EAP-Response/AKA'-Challenge with AT_KDF from the
   peer, the server checks that the suggested AT_KDF value was one of
   the alternatives in its offer.  The first AT_KDF value in the message
   from the server is not a valid alternative since the peer should have
   accepted it without further negotiation.  If the peer has replied
   with the first AT_KDF value, the server behaves as if AT_MAC of the
   response had been incorrect and fails the authentication.  For an
   overview of the failed authentication process in the server side, see
   Section 3 and Figure 2 of [RFC4187].  Otherwise, the server re-sends
   the EAP-Response/AKA'-Challenge message, but adds the selected
   alternative to the beginning of the list of AT_KDF attributes and
   retains the entire list following it.  Note that this means that the
   selected alternative appears twice in the set of AT_KDF values.
   Responding to the peer's request to change the key derivation
   function is the only legal situation where such duplication may

   When the peer receives the new EAP-Request/AKA'-Challenge message, it
   MUST check that the requested change, and only the requested change,
   occurred in the list of AT_KDF attributes.  If so, it continues with
   processing the received EAP-Request/AKA'-Challenge as specified in
   [RFC4187] and Section 3.1 of this document.  If not, it behaves as if
   AT_MAC had been incorrect and fails the authentication.  If the peer
   receives multiple EAP-Request/AKA'-Challenge messages with differing
   AT_KDF attributes without having requested negotiation, the peer MUST
   behave as if AT_MAC had been incorrect and fail the authentication.

   Note that the peer may also request sequence number resynchronization
   [RFC4187].  This happens after AT_KDF negotiation has already
   completed.  That is, the EAP-Request/AKA'-Challenge and, possibly,
   the EAP-Response/AKA'-Challenge messages are exchanged first to
   determine a mutually acceptable key derivation function, and only
   then is the possible AKA'-Synchronization-Failure message sent.  The
   AKA'-Synchronization-Failure message is sent as a response to the
   newly received EAP-Request/AKA'-Challenge, which is the last message
   of the AT_KDF negotiation.  Note that if the first proposed KDF is
   acceptable, then the first EAP-Request/AKA'-Challenge message is also
   the last message.  The AKA'-Synchronization-Failure message MUST
   contain the AUTS parameter as specified in [RFC4187] and a copy the
   AT_KDF attributes as they appeared in the last message of the AT_KDF
   negotiation.  If the AT_KDF attributes are found to differ from their
   earlier values, the peer and server MUST behave as if AT_MAC had been
   incorrect and fail the authentication.

3.3.  Key Derivation

   Both the peer and server MUST derive the keys as follows.

   AT_KDF parameter has the value 1
      In this case, MK is derived and used as follows:

          MK = PRF'(IK'|CK',"EAP-AKA'"|Identity)
          K_encr = MK[0..127]
          K_aut  = MK[128..383]
          K_re   = MK[384..639]
          MSK    = MK[640..1151]
          EMSK   = MK[1152..1663]

      Here [n..m] denotes the substring from bit n to m, including bits
      n and m.  PRF' is a new pseudorandom function specified in
      Section 3.4.  The first 1664 bits from its output are used for
      K_encr (encryption key, 128 bits), K_aut (authentication key, 256
      bits), K_re (re-authentication key, 256 bits), MSK (Master Session
      Key, 512 bits), and EMSK (Extended Master Session Key, 512 bits).
      These keys are used by the subsequent EAP-AKA' process.  K_encr is
      used by the AT_ENCR_DATA attribute, and K_aut by the AT_MAC
      attribute.  K_re is used later in this section.  MSK and EMSK are
      outputs from a successful EAP method run [RFC3748].

      IK' and CK' are derived as specified in [TS-3GPP.33.402].  The
      functions that derive IK' and CK' take the following parameters:
      CK and IK produced by the AKA algorithm, and value of the Network
      Name field comes from the AT_KDF_INPUT attribute (without length
      or padding).

      The value "EAP-AKA'" is an eight-characters-long ASCII string.  It
      is used as is, without any trailing NUL characters.

      Identity is the peer identity as specified in Section 7 of
      [RFC4187] and in Section 5.3.2 of in this document for the 5G

      When the server creates an AKA challenge and corresponding AUTN,
      CK, CK', IK, and IK' values, it MUST set the Authentication
      Management Field (AMF) separation bit to 1 in the AKA algorithm
      [TS-3GPP.33.102].  Similarly, the peer MUST check that the AMF
      separation bit is set to 1.  If the bit is not set to 1, the peer
      behaves as if the AUTN had been incorrect and fails the

      On fast re-authentication, the following keys are calculated:

          MK = PRF'(K_re,"EAP-AKA' re-auth"|Identity|counter|NONCE_S)
          MSK  = MK[0..511]
          EMSK = MK[512..1023]

      MSK and EMSK are the resulting 512-bit keys, taking the first 1024
      bits from the result of PRF'.  Note that K_encr and K_aut are not
      re-derived on fast re-authentication.  K_re is the re-
      authentication key from the preceding full authentication and
      stays unchanged over any fast re-authentication(s) that may happen
      based on it.  The value "EAP-AKA' re-auth" is a sixteen-
      characters-long ASCII string, again represented without any
      trailing NUL characters.  Identity is the fast re-authentication
      identity, counter is the value from the AT_COUNTER attribute,
      NONCE_S is the nonce value from the AT_NONCE_S attribute, all as
      specified in Section 7 of [RFC4187].  To prevent the use of
      compromised keys in other places, it is forbidden to change the
      network name when going from the full to the fast re-
      authentication process.  The peer SHOULD NOT attempt fast re-
      authentication when it knows that the network name in the current
      access network is different from the one in the initial, full
      authentication.  Upon seeing a re-authentication request with a
      changed network name, the server SHOULD behave as if the re-
      authentication identifier had been unrecognized, and fall back to
      full authentication.  The server observes the change in the name
      by comparing where the fast re-authentication and full
      authentication EAP transactions were received at the
      Authentication, Authorization, and Accounting (AAA) protocol

   AT_KDF has any other value
      Future variations of key derivation functions may be defined, and
      they will be represented by new values of AT_KDF.  If the peer
      does not recognize the value, it cannot calculate the keys and
      behaves as explained in Section 3.2.

   AT_KDF is missing
      The peer behaves as if the AUTN had been incorrect and MUST fail
      the authentication.

   If the peer supports a given key derivation function but is unwilling
   to perform it for policy reasons, it refuses to calculate the keys
   and behaves as explained in Section 3.2.

3.4.  Hash Functions

   EAP-AKA' uses SHA-256 / HMAC-SHA-256, not SHA-1 / HMAC-SHA-1 (see
   [FIPS.180-4] and [RFC2104]) as in EAP-AKA.  This requires a change to
   the pseudorandom function (PRF) as well as the AT_MAC and
   AT_CHECKCODE attributes.

3.4.1.  PRF'

   The PRF' construction is the same one IKEv2 uses (see Section 2.13 of
   [RFC7296]; the definition of this function has not changed since
   [RFC4306], which was referenced by [RFC5448]).  The function takes
   two arguments.  K is a 256-bit value and S is a byte string of
   arbitrary length.  PRF' is defined as follows:

   PRF'(K,S) = T1 | T2 | T3 | T4 | ...

      T1 = HMAC-SHA-256 (K, S | 0x01)
      T2 = HMAC-SHA-256 (K, T1 | S | 0x02)
      T3 = HMAC-SHA-256 (K, T2 | S | 0x03)
      T4 = HMAC-SHA-256 (K, T3 | S | 0x04)

   PRF' produces as many bits of output as is needed.  HMAC-SHA-256 is
   the application of HMAC [RFC2104] to SHA-256.

3.4.2.  AT_MAC

   When used within EAP-AKA', the AT_MAC attribute is changed as
   follows.  The MAC algorithm is HMAC-SHA-256-128, a keyed hash value.
   The HMAC-SHA-256-128 value is obtained from the 32-byte HMAC-SHA-256
   value by truncating the output to the first 16 bytes.  Hence, the
   length of the MAC is 16 bytes.

   Otherwise, the use of AT_MAC in EAP-AKA' follows Section 10.15 of


   When used within EAP-AKA', the AT_CHECKCODE attribute is changed as
   follows.  First, a 32-byte value is needed to accommodate a 256-bit
   hash output:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   | AT_CHECKCODE  | Length        |           Reserved            |
   |                                                               |
   |                     Checkcode (0 or 32 bytes)                 |
   |                                                               |
   |                                                               |
   |                                                               |

   Second, the checkcode is a hash value, calculated with SHA-256
   [FIPS.180-4], over the data specified in Section 10.13 of [RFC4187].

3.5.  Summary of Attributes for EAP-AKA'

   Table 1 identifies which attributes may be found in which kinds of
   messages, and in what quantity.

   Messages are denoted with numbers as follows:

   1  EAP-Request/AKA-Identity

   2  EAP-Response/AKA-Identity

   3  EAP-Request/AKA-Challenge

   4  EAP-Response/AKA-Challenge

   5  EAP-Request/AKA-Notification

   6  EAP-Response/AKA-Notification

   7  EAP-Response/AKA-Client-Error

   8  EAP-Request/AKA-Reauthentication

   9  EAP-Response/AKA-Reauthentication

   10  EAP-Response/AKA-Authentication-Reject

   11  EAP-Response/AKA-Synchronization-Failure

   The column denoted with "E" indicates whether the attribute is a
   nested attribute that MUST be included within AT_ENCR_DATA.

   In addition, the numbered columns indicate the quantity of the
   attribute within the message as follows:

   "0"     Indicates that the attribute MUST NOT be included in the

   "1"     Indicates that the attribute MUST be included in the message.

   "0-1"   Indicates that the attribute is sometimes included in the

   "0+"    Indicates that zero or more copies of the attribute MAY be
           included in the message.

   "1+"    Indicates that there MUST be at least one attribute in the
           message but more than one MAY be included in the message.

   "0*"    Indicates that the attribute is not included in the message
           in cases specified in this document, but MAY be included in
           the future versions of the protocol.

   The attribute table is shown below.  The table is largely the same as
   in the EAP-AKA attribute table ([RFC4187], Section 10.1), but changes
   how many times AT_MAC may appear in an EAP-Response/AKA'-Challenge
   message as it does not appear there when AT_KDF has to be sent from
   the peer to the server.  The table also adds the AT_KDF and
   AT_KDF_INPUT attributes.

   | Attribute            |1  |2  |3  |4  |5  |6  |7|8   | 9   |10|11|E|
   | AT_PERMANENT_ID_REQ  |0-1|0  |0  |0  |0  |0  |0|0   | 0   |0 |0 |N|
   | AT_ANY_ID_REQ        |0-1|0  |0  |0  |0  |0  |0|0   | 0   |0 |0 |N|
   | AT_FULLAUTH_ID_REQ   |0-1|0  |0  |0  |0  |0  |0|0   | 0   |0 |0 |N|
   | AT_IDENTITY          |0  |0-1|0  |0  |0  |0  |0|0   | 0   |0 |0 |N|
   | AT_RAND              |0  |0  |1  |0  |0  |0  |0|0   | 0   |0 |0 |N|
   | AT_AUTN              |0  |0  |1  |0  |0  |0  |0|0   | 0   |0 |0 |N|
   | AT_RES               |0  |0  |0  |1  |0  |0  |0|0   | 0   |0 |0 |N|
   | AT_AUTS              |0  |0  |0  |0  |0  |0  |0|0   | 0   |0 |1 |N|
   | AT_NEXT_PSEUDONYM    |0  |0  |0-1|0  |0  |0  |0|0   | 0   |0 |0 |Y|
   | AT_NEXT_REAUTH_ID    |0  |0  |0-1|0  |0  |0  |0|0-1 | 0   |0 |0 |Y|
   | AT_IV                |0  |0  |0-1|0* |0-1|0-1|0|1   | 1   |0 |0 |N|
   | AT_ENCR_DATA         |0  |0  |0-1|0* |0-1|0-1|0|1   | 1   |0 |0 |N|
   | AT_PADDING           |0  |0  |0-1|0* |0-1|0-1|0|0-1 | 0-1 |0 |0 |Y|
   | AT_CHECKCODE         |0  |0  |0-1|0-1|0  |0  |0|0-1 | 0-1 |0 |0 |N|
   | AT_RESULT_IND        |0  |0  |0-1|0-1|0  |0  |0|0-1 | 0-1 |0 |0 |N|
   | AT_MAC               |0  |0  |1  |0-1|0-1|0-1|0|1   | 1   |0 |0 |N|
   | AT_COUNTER           |0  |0  |0  |0  |0-1|0-1|0|1   | 1   |0 |0 |Y|
   | AT_COUNTER_TOO_SMALL |0  |0  |0  |0  |0  |0  |0|0   | 0-1 |0 |0 |Y|
   | AT_NONCE_S           |0  |0  |0  |0  |0  |0  |0|1   | 0   |0 |0 |Y|
   | AT_NOTIFICATION      |0  |0  |0  |0  |1  |0  |0|0   | 0   |0 |0 |N|
   | AT_CLIENT_ERROR_CODE |0  |0  |0  |0  |0  |0  |1|0   | 0   |0 |0 |N|
   | AT_KDF               |0  |0  |1+ |0+ |0  |0  |0|0   | 0   |0 |1+|N|
   | AT_KDF_INPUT         |0  |0  |1  |0  |0  |0  |0|0   | 0   |0 |0 |N|

                        Table 1: The Attribute Table

4.  Bidding Down Prevention for EAP-AKA

   As discussed in [RFC3748], negotiation of methods within EAP is
   insecure.  That is, a man-in-the-middle attacker may force the
   endpoints to use a method that is not the strongest that they both
   support.  This is a problem, as we expect EAP-AKA and EAP-AKA' to be
   negotiated via EAP.

   In order to prevent such attacks, this RFC specifies a mechanism for
   EAP-AKA that allows the endpoints to securely discover the
   capabilities of each other.  This mechanism comes in the form of the
   AT_BIDDING attribute.  This allows both endpoints to communicate
   their desire and support for EAP-AKA' when exchanging EAP-AKA
   messages.  This attribute is not included in EAP-AKA' messages.  It
   is only included in EAP-AKA messages, which are protected with the
   AT_MAC attribute.  This approach is based on the assumption that EAP-
   AKA' is always preferable (see Section 7).  If during the EAP-AKA
   authentication process it is discovered that both endpoints would
   have been able to use EAP-AKA', the authentication process SHOULD be
   aborted, as a bidding down attack may have happened.

   The format of the AT_BIDDING attribute is shown below.

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      | AT_BIDDING    | Length        |D|          Reserved           |

   The fields are as follows:

      This is set to 136.

      The length of the attribute, calculated as defined in [RFC4187],
      Section 8.1.  For AT_BIDDING, the Length MUST be set to 1.

      This bit is set to 1 if the sender supports EAP-AKA', is willing
      to use it, and prefers it over EAP-AKA.  Otherwise, it should be
      set to zero.

      This field MUST be set to zero when sent and ignored on receipt.

   The server sends this attribute in the EAP-Request/AKA-Challenge
   message.  If the peer supports EAP-AKA', it compares the received
   value to its own capabilities.  If it turns out that both the server
   and peer would have been able to use EAP-AKA' and preferred it over
   EAP-AKA, the peer behaves as if AUTN had been incorrect and fails the
   authentication (see Figure 3 of [RFC4187]).  A peer not supporting
   EAP-AKA' will simply ignore this attribute.  In all cases, the
   attribute is protected by the integrity mechanisms of EAP-AKA, so it
   cannot be removed by a man-in-the-middle attacker.

   Note that we assume (Section 7) that EAP-AKA' is always stronger than
   EAP-AKA.  As a result, this specification does not provide protection
   against bidding "down" attacks in the other direction, i.e.,
   attackers forcing the endpoints to use EAP-AKA'.

4.1.  Summary of Attributes for EAP-AKA

   The appearance of the AT_BIDDING attribute in EAP-AKA exchanges is
   shown below, using the notation from Section 3.5:

     | Attribute  | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | E |
     | AT_BIDDING | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0  | 0  | N |

                 Table 2: AT_BIDDING Attribute Appearance

5.  Peer Identities

   EAP-AKA' peer identities are as specified in [RFC4187], Section 4.1,
   with the addition of some requirements specified in this section.

   EAP-AKA' includes optional identity privacy support that can be used
   to hide the cleartext permanent identity and thereby make the
   subscriber's EAP exchanges untraceable to eavesdroppers.  EAP-AKA'
   can also use the privacy-friendly identifiers specified for 5G

   The permanent identity is usually based on the IMSI.  Exposing the
   IMSI is undesirable because, as a permanent identity, it is easily
   trackable.  In addition, since IMSIs may be used in other contexts as
   well, there would be additional opportunities for such tracking.

   In EAP-AKA', identity privacy is based on temporary usernames or
   pseudonym usernames.  These are similar to, but separate from, the
   Temporary Mobile Subscriber Identities (TMSI) that are used on
   cellular networks.

5.1.  Username Types in EAP-AKA' Identities

   Section of [RFC4187] specifies that there are three types of
   usernames: permanent, pseudonym, and fast re-authentication
   usernames.  This specification extends this definition as follows.
   There are four types of usernames:

   (1)  Regular usernames.  These are external names given to EAP-AKA'
        peers.  The regular usernames are further subdivided into to

        (a)  Permanent usernames, for instance, IMSI-based usernames.

        (b)  Privacy-friendly temporary usernames, for instance, 5G GUTI
             (5G Globally Unique Temporary Identifier) or 5G privacy
             identifiers (see Section 5.3.2) such as SUCI (Subscription
             Concealed Identifier).

   (2)  EAP-AKA' pseudonym usernames.  For example,
        2s7ah6n9q@example.com might be a valid pseudonym identity.  In
        this example, 2s7ah6n9q is the pseudonym username.

   (3)  EAP-AKA' fast re-authentication usernames.  For example,
        43953754@example.com might be a valid fast re-authentication
        identity and 43953754 the fast re-authentication username.

   The permanent, privacy-friendly temporary, and pseudonym usernames
   are only used with full authentication, and fast re-authentication
   usernames only with fast re-authentication.  Unlike permanent
   usernames and pseudonym usernames, privacy-friendly temporary
   usernames and fast re-authentication usernames are one-time
   identifiers, which are not reused across EAP exchanges.

5.2.  Generating Pseudonyms and Fast Re-Authentication Identities

   This section provides some additional guidance to implementations for
   producing secure pseudonyms and fast re-authentication identities.
   It does not impact backwards compatibility because each server
   consumes only the identities that it generates itself.  However,
   adherence to the guidance will provide better security.

   As specified by [RFC4187], Section, pseudonym usernames and
   fast re-authentication identities are generated by the EAP server in
   an implementation-dependent manner.  RFC 4187 provides some general
   requirements on how these identities are transported, how they map to
   the NAI syntax, how they are distinguished from each other, and so

   However, to enhance privacy, some additional requirements need to be

   The pseudonym usernames and fast re-authentication identities MUST be
   generated in a cryptographically secure way so that it is
   computationally infeasible for an attacker to differentiate two
   identities belonging to the same user from two identities belonging
   to different users.  This can be achieved, for instance, by using
   random or pseudorandom identifiers such as random byte strings or
   ciphertexts.  See also [RFC4086] for guidance on random number

   Note that the pseudonym and fast re-authentication usernames also
   MUST NOT include substrings that can be used to relate the username
   to a particular entity or a particular permanent identity.  For
   instance, the usernames cannot include any subscriber-identifying
   part of an IMSI or other permanent identifier.  Similarly, no part of
   the username can be formed by a fixed mapping that stays the same
   across multiple different pseudonyms or fast re-authentication
   identities for the same subscriber.

   When the identifier used to identify a subscriber in an EAP-AKA'
   authentication exchange is a privacy-friendly identifier that is used
   only once, the EAP-AKA' peer MUST NOT use a pseudonym provided in
   that authentication exchange in subsequent exchanges more than once.
   To ensure that this does not happen, the EAP-AKA' server MAY decline
   to provide a pseudonym in such authentication exchanges.  An
   important case where such privacy-friendly identifiers are used is in
   5G networks (see Section 5.3).

5.3.  Identifier Usage in 5G

   In EAP-AKA', the peer identity may be communicated to the server in
   one of three ways:

   *  As a part of link-layer establishment procedures, externally to

   *  With the EAP-Response/Identity message in the beginning of the EAP
      exchange, but before the selection of EAP-AKA'.

   *  Transmitted from the peer to the server using EAP-AKA' messages
      instead of EAP-Response/Identity.  In this case, the server
      includes an identity-requesting attribute (AT_ANY_ID_REQ,
      AKA-Identity message, and the peer includes the AT_IDENTITY
      attribute, which contains the peer's identity, in the EAP-
      Response/AKA-Identity message.

   The identity carried above may be a permanent identity, privacy-
   friendly identity, pseudonym identity, or fast re-authentication
   identity as defined in Section 5.1.

   5G supports the concept of privacy identifiers, and it is important
   for interoperability that the right type of identifier is used.

   5G defines the SUbscription Permanent Identifier (SUPI) and
   SUbscription Concealed Identifier (SUCI) [TS-3GPP.23.501]
   [TS-3GPP.33.501] [TS-3GPP.23.003].  SUPI is globally unique and
   allocated to each subscriber.  However, it is only used internally in
   the 5G network and is privacy sensitive.  The SUCI is a privacy-
   preserving identifier containing the concealed SUPI, using public key
   cryptography to encrypt the SUPI.

   Given the choice between these two types of identifiers, EAP-AKA'
   ensures interoperability as follows:

   *  Where identifiers are used within EAP-AKA' (such as key
      derivation) determine the exact values of the identity to be used,
      to avoid ambiguity (see Section 5.3.1).

   *  Where identifiers are carried within EAP-AKA' packets (such as in
      the AT_IDENTITY attribute) determine which identifiers should be
      filled in (see Section 5.3.2).

   In 5G, the normal mode of operation is that identifiers are only
   transmitted outside EAP.  However, in a system involving terminals
   from many generations and several connectivity options via 5G and
   other mechanisms, implementations and the EAP-AKA' specification need
   to prepare for many different situations, including sometimes having
   to communicate identities within EAP.

   The following sections clarify which identifiers are used and how.

5.3.1.  Key Derivation

   In EAP-AKA', the peer identity is used in the key derivation formula
   found in Section 3.3.

   The identity needs to be represented in exactly the correct format
   for the key derivation formula to produce correct results.

   If the AT_KDF_INPUT parameter contains the prefix "5G:", the AT_KDF
   parameter has the value 1, and this authentication is not a fast re-
   authentication, then the peer identity used in the key derivation
   MUST be as specified in Annex F.3 of [TS-3GPP.33.501] and Clause 2.2
   of [TS-3GPP.23.003].  This is in contrast to [RFC5448], which uses
   the identity as communicated in EAP and represented as a NAI.  Also,
   in contrast to [RFC5448], in 5G EAP-AKA' does not use the "0" nor the
   "6" prefix in front of the identifier.

   For an example of the format of the identity, see Clause 2.2 of

   In all other cases, the following applies:

         The identity used in the key derivation formula MUST be exactly
         the one sent in the EAP-AKA' AT_IDENTITY attribute, if one was
         sent, regardless of the kind of identity that it may have been.
         If no AT_IDENTITY was sent, the identity MUST be exactly the
         one sent in the generic EAP Identity exchange, if one was made.

         If no identity was communicated inside EAP, then the identity
         is the one communicated outside EAP in link-layer messaging.

         In this case, the used identity MUST be the identity most
         recently communicated by the peer to the network, again
         regardless of what type of identity it may have been.

5.3.2.  EAP Identity Response and EAP-AKA' AT_IDENTITY Attribute

   The EAP authentication option is only available in 5G when the new 5G
   core network is also in use.  However, in other networks, an EAP-AKA'
   peer may be connecting to other types of networks and existing

   When the EAP server is in a 5G network, the 5G procedures for EAP-
   AKA' apply.  [TS-3GPP.33.501] specifies when the EAP server is in a
   5G network.

         Note: Currently, the following conditions are specified: when
         the EAP peer uses the 5G Non-Access Stratum (NAS) protocol
         [TS-3GPP.24.501] or when the EAP peer attaches to a network
         that advertises 5G connectivity without NAS [TS-3GPP.23.501].
         Possible future conditions may also be specified by 3GPP.

   When the 5G procedures for EAP-AKA' apply, EAP identity exchanges are
   generally not used as the identity is already made available on
   previous link-layer exchanges.

   In this situation, the EAP Identity Response and EAP-AKA' AT_IDENTITY
   attribute are handled as specified in Annex F.2 of [TS-3GPP.33.501].

   When used in EAP-AKA', the format of the SUCI MUST be as specified in
   [TS-3GPP.23.003], Section 28.7.3, with the semantics defined in
   [TS-3GPP.23.003], Section 2.2B.  Also, in contrast to [RFC5448], in
   5G EAP-AKA' does not use the "0" nor the "6" prefix in front of the

   For an example of an IMSI in NAI format, see [TS-3GPP.23.003],
   Section 28.7.3.

   Otherwise, the peer SHOULD employ an IMSI, SUPI, or NAI [RFC7542] as
   it is configured to use.

6.  Exported Parameters

   When not using fast re-authentication, the EAP-AKA' Session-Id is the
   concatenation of the EAP-AKA' Type value (0x32, one byte) with the
   contents of the RAND field from the AT_RAND attribute followed by the
   contents of the AUTN field in the AT_AUTN attribute:

         Session-Id = 0x32 || RAND || AUTN

   When using fast re-authentication, the EAP-AKA' Session-Id is the
   concatenation of the EAP-AKA' Type value (0x32) with the contents of
   the NONCE_S field from the AT_NONCE_S attribute followed by the
   contents of the MAC field from the AT_MAC attribute from the EAP-

         Session-Id = 0x32 || NONCE_S || MAC

   The Peer-Id is the contents of the Identity field from the
   AT_IDENTITY attribute, using only the Actual Identity Length bytes
   from the beginning.  Note that the contents are used as they are
   transmitted, regardless of whether the transmitted identity was a
   permanent, pseudonym, or fast EAP re-authentication identity.  If no
   AT_IDENTITY attribute was exchanged, the exported Peer-Id is the
   identity provided from the EAP Identity Response packet.  If no EAP
   Identity Response was provided either, the exported Peer-Id is the
   null string (zero length).

   The Server-Id is the null string (zero length).

7.  Security Considerations

   A summary of the security properties of EAP-AKA' follows.  These
   properties are very similar to those in EAP-AKA.  We assume that HMAC
   SHA-256 is at least as secure as HMAC SHA-1 (see also [RFC6194]).
   This is called the SHA-256 assumption in the remainder of this
   section.  Under this assumption, EAP-AKA' is at least as secure as

   If the AT_KDF attribute has value 1, then the security properties of
   EAP-AKA' are as follows:

   Protected ciphersuite negotiation
      EAP-AKA' has no ciphersuite negotiation mechanisms.  It does have
      a negotiation mechanism for selecting the key derivation
      functions.  This mechanism is secure against bidding down attacks
      from EAP-AKA' to EAP-AKA.  The negotiation mechanism allows
      changing the offered key derivation function, but the change is
      visible in the final EAP-Request/AKA'-Challenge message that the
      server sends to the peer.  This message is authenticated via the
      AT_MAC attribute, and carries both the chosen alternative and the
      initially offered list.  The peer refuses to accept a change it
      did not initiate.  As a result, both parties are aware that a
      change is being made and what the original offer was.

      Per assumptions in Section 4, there is no protection against
      bidding down attacks from EAP-AKA to EAP-AKA' should EAP-AKA'
      somehow be considered less secure some day than EAP-AKA.  Such
      protection was not provided in RFC 5448 implementations and
      consequently neither does this specification provide it.  If such
      support is needed, it would have to be added as a separate new

      In general, it is expected that the current negotiation
      capabilities in EAP-AKA' are sufficient for some types of
      extensions, including adding Perfect Forward Secrecy [EMU-AKA-PFS]
      and perhaps others.  However, some larger changes may require a
      new EAP method type, which is how EAP-AKA' itself happened.  One
      example of such change would be the introduction of new

   Mutual authentication
      Under the SHA-256 assumption, the properties of EAP-AKA' are at
      least as good as those of EAP-AKA in this respect.  Refer to
      [RFC4187], Section 12, for further details.

   Integrity protection
      Under the SHA-256 assumption, the properties of EAP-AKA' are at
      least as good (most likely better) as those of EAP-AKA in this
      respect.  Refer to [RFC4187], Section 12, for further details.
      The only difference is that a stronger hash algorithm and keyed
      MAC, SHA-256 / HMAC-SHA-256, is used instead of SHA-1 / HMAC-SHA-

   Replay protection
      Under the SHA-256 assumption, the properties of EAP-AKA' are at
      least as good as those of EAP-AKA in this respect.  Refer to
      [RFC4187], Section 12, for further details.

      The properties of EAP-AKA' are exactly the same as those of EAP-
      AKA in this respect.  Refer to [RFC4187], Section 12, for further

   Key derivation
      EAP-AKA' supports key derivation with an effective key strength
      against brute-force attacks equal to the minimum of the length of
      the derived keys and the length of the AKA base key, i.e., 128
      bits or more.  The key hierarchy is specified in Section 3.3.

      The Transient EAP Keys used to protect EAP-AKA packets (K_encr,
      K_aut, K_re), the MSK, and the EMSK are cryptographically
      separate.  If we make the assumption that SHA-256 behaves as a
      pseudorandom function, an attacker is incapable of deriving any
      non-trivial information about any of these keys based on the other
      keys.  An attacker also cannot calculate the pre-shared secret
      from IK, CK, IK', CK', K_encr, K_aut, K_re, MSK, or EMSK by any
      practically feasible means.

      EAP-AKA' adds an additional layer of key derivation functions
      within itself to protect against the use of compromised keys.
      This is discussed further in Section 7.4.

      EAP-AKA' uses a pseudorandom function modeled after the one used
      in IKEv2 [RFC7296] together with SHA-256.

   Key strength
      See above.

   Dictionary attack resistance
      Under the SHA-256 assumption, the properties of EAP-AKA' are at
      least as good as those of EAP-AKA in this respect.  Refer to
      [RFC4187], Section 12, for further details.

   Fast reconnect
      Under the SHA-256 assumption, the properties of EAP-AKA' are at
      least as good as those of EAP-AKA in this respect.  Refer to
      [RFC4187], Section 12, for further details.  Note that
      implementations MUST prevent performing a fast reconnect across
      method types.

   Cryptographic binding
      Note that this term refers to a very specific form of binding,
      something that is performed between two layers of authentication.
      It is not the same as the binding to a particular network name.
      The properties of EAP-AKA' are exactly the same as those of EAP-
      AKA in this respect, i.e., as it is not a tunnel method, this
      property is not applicable to it.  Refer to [RFC4187], Section 12,
      for further details.

   Session independence
      The properties of EAP-AKA' are exactly the same as those of EAP-
      AKA in this respect.  Refer to [RFC4187], Section 12, for further

      The properties of EAP-AKA' are exactly the same as those of EAP-
      AKA in this respect.  Refer to [RFC4187], Section 12, for further

   Channel binding
      EAP-AKA', like EAP-AKA, does not provide channel bindings as
      they're defined in [RFC3748] and [RFC5247].  New skippable
      attributes can be used to add channel binding support in the
      future, if required.

      However, including the Network Name field in the AKA' algorithms
      (which are also used for other purposes than EAP-AKA') provides a
      form of cryptographic separation between different network names,
      which resembles channel bindings.  However, the network name does
      not typically identify the EAP (pass-through) authenticator.  See
      Section 7.4 for more discussion.

7.1.  Privacy

   [RFC6973] suggests that the privacy considerations of IETF protocols
   be documented.

   The confidentiality properties of EAP-AKA' itself have been discussed
   above under "Confidentiality" (Section 7).

   EAP-AKA' uses several different types of identifiers to identify the
   authenticating peer.  It is strongly RECOMMENDED to use the privacy-
   friendly temporary or hidden identifiers, i.e., the 5G GUTI or SUCI,
   pseudonym usernames, and fast re-authentication usernames.  The use
   of permanent identifiers such as the IMSI or SUPI may lead to an
   ability to track the peer and/or user associated with the peer.  The
   use of permanent identifiers such as the IMSI or SUPI is strongly NOT

   As discussed in Section 5.3, when authenticating to a 5G network,
   only the SUCI identifier is normally used.  The use of EAP-AKA'
   pseudonyms in this situation is at best limited because the SUCI
   already provides a stronger mechanism.  In fact, reusing the same
   pseudonym multiple times will result in a tracking opportunity for
   observers that see the pseudonym pass by.  To avoid this, the peer
   and server need to follow the guidelines given in Section 5.2.

   When authenticating to a 5G network, per Section 5.3.1, both the EAP-
   AKA' peer and server need to employ the permanent identifier SUPI as
   an input to key derivation.  However, this use of the SUPI is only
   internal.  As such, the SUPI need not be communicated in EAP
   messages.  Therefore, SUPI MUST NOT be communicated in EAP-AKA' when
   authenticating to a 5G network.

   While the use of SUCI in 5G networks generally provides identity
   privacy, this is not true if the null-scheme encryption is used to
   construct the SUCI (see [TS-3GPP.33.501], Annex C).  The use of this
   scheme makes the use of SUCI equivalent to the use of SUPI or IMSI.
   The use of the null scheme is NOT RECOMMENDED where identity privacy
   is important.

   The use of fast re-authentication identities when authenticating to a
   5G network does not have the same problems as the use of pseudonyms,
   as long as the 5G authentication server generates the fast re-
   authentication identifiers in a proper manner specified in
   Section 5.2.

   Outside 5G, the peer can freely choose between the use of permanent,
   pseudonym, or fast re-authentication identifiers:

   *  A peer that has not yet performed any EAP-AKA' exchanges does not
      typically have a pseudonym available.  If the peer does not have a
      pseudonym available, then the privacy mechanism cannot be used,
      and the permanent identity will have to be sent in the clear.

      The terminal SHOULD store the pseudonym in nonvolatile memory so
      that it can be maintained across reboots.  An active attacker that
      impersonates the network may use the AT_PERMANENT_ID_REQ attribute
      ([RFC4187], Section 4.1.2) to learn the subscriber's IMSI.
      However, as discussed in [RFC4187], Section 4.1.2, the terminal
      can refuse to send the cleartext permanent identity if it believes
      that the network should be able to recognize the pseudonym.

   *  When pseudonyms and fast re-authentication identities are used,
      the peer relies on the properly created identifiers by the server.

      It is essential that an attacker cannot link a privacy-friendly
      identifier to the user in any way or determine that two
      identifiers belong to the same user as outlined in Section 5.2.
      The pseudonym usernames and fast re-authentication identities MUST
be used for other purposes (e.g., in other protocols).

   If the peer and server cannot guarantee that SUCI can be used or that
   pseudonyms will be available, generated properly, and maintained
   reliably, and identity privacy is required, then additional
   protection from an external security mechanism such as tunneled EAP
   methods like Tunneled Transport Layer Security (TTLS) [RFC5281] or
   Tunnel Extensible Authentication Protocol (TEAP) [RFC7170] may be
   used.  The benefits and the security considerations of using an
   external security mechanism with EAP-AKA are beyond the scope of this

   Finally, as with other EAP methods, even when privacy-friendly
   identifiers or EAP tunneling is used, typically the domain part of an
   identifier (e.g., the home operator) is visible to external parties.

7.2.  Discovered Vulnerabilities

   There have been no published attacks that violate the primary secrecy
   or authentication properties defined for Authentication and Key
   Agreement (AKA) under the originally assumed trust model.  The same
   is true of EAP-AKA'.

   However, there have been attacks when a different trust model is in
   use, with characteristics not originally provided by the design, or
   when participants in the protocol leak information to outsiders on
   purpose, and there have been some privacy-related attacks.

   For instance, the original AKA protocol does not prevent an insider
   supplying keys to a third party, e.g., as described by Mjølsnes and
   Tsay in [MT2012] where a serving network lets an authentication run
   succeed, but then it misuses the session keys to send traffic on the
   authenticated user's behalf.  This particular attack is not different
   from any on-path entity (such as a router) pretending to send
   traffic, but the general issue of insider attacks can be a problem,
   particularly in a large group of collaborating operators.

   Another class of attacks is the use of tunneling of traffic from one
   place to another, e.g., as done by Zhang and Fang in [ZF2005] to
   leverage security policy differences between different operator
   networks, for instance.  To gain something in such an attack, the
   attacker needs to trick the user into believing it is in another
   location.  If policies between locations differ, for instance, if
   payload traffic is not required to be encrypted in some location, the
   attacker may trick the user into opening a vulnerability.  As an
   authentication mechanism, EAP-AKA' is not directly affected by most
   of these attacks.  EAP-AKA' network name binding can also help
   alleviate some of the attacks.  In any case, it is recommended that
   EAP-AKA' configuration not be dependent on the location of request
   origin, unless the location information can be cryptographically
   confirmed, e.g., with the network name binding.

   Zhang and Fang also looked at denial-of-service attacks [ZF2005].  A
   serving network may request large numbers of authentication runs for
   a particular subscriber from a home network.  While the
   resynchronization process can help recover from this, eventually it
   is possible to exhaust the sequence number space and render the
   subscriber's card unusable.  This attack is possible for both
   original AKA and EAP-AKA'.  However, it requires the collaboration of
   a serving network in an attack.  It is recommended that EAP-AKA'
   implementations provide the means to track, detect, and limit
   excessive authentication attempts to combat this problem.

   There have also been attacks related to the use of AKA without the
   generated session keys (e.g., [BT2013]).  Some of those attacks
   relate to the use of HTTP Digest AKAv1 [RFC3310], which was
   originally vulnerable to man-in-the-middle attacks.  This has since
   been corrected in [RFC4169].  The EAP-AKA' protocol uses session keys
   and provides channel binding, and as such, it is resistant to the
   above attacks except where the protocol participants leak information
   to outsiders.

   Basin, et al. [Basin2018] have performed formal analysis and
   concluded that the AKA protocol would have benefited from additional
   security requirements such as key confirmation.

   In the context of pervasive monitoring revelations, there were also
   reports of compromised long-term pre-shared keys used in SIM and AKA
   [Heist2015].  While no protocol can survive the theft of key material
   associated with its credentials, there are some things that alleviate
   the impacts in such situations.  These are discussed further in
   Section 7.3.

   Arapinis, et al. [Arapinis2012] describe an attack that uses the AKA
   resynchronization protocol to attempt to detect whether a particular
   subscriber is in a given area.  This attack depends on the attacker
   setting up a false base station in the given area and on the
   subscriber performing at least one authentication between the time
   the attack is set up and run.

   Borgaonkar, et al. discovered that the AKA resynchronization protocol
   may also be used to predict the authentication frequency of a
   subscriber if a non-time-based sequence number (SQN) generation
   scheme is used [Borgaonkar2018].  The attacker can force the reuse of
   the keystream that is used to protect the SQN in the AKA
   resynchronization protocol.  The attacker then guesses the
   authentication frequency based on the lowest bits of two XORed SQNs.
   The researchers' concern was that the authentication frequency would
   reveal some information about the phone usage behavior, e.g., number
   of phone calls made or number of SMS messages sent.  There are a
   number of possible triggers for authentication, so such an
   information leak is not direct, but it can be a concern.  The impact
   of the attack differs depending on whether the SQN generation scheme
   that is used is time-based or not.

   Similar attacks are possible outside AKA in the cellular paging
   protocols where the attacker can simply send application-layer data,
   send short messages, or make phone calls to the intended victim and
   observe the air interface (e.g., [Kune2012] and [Shaik2016]).
   Hussain, et al. demonstrated a slightly more sophisticated version of
   the attack that exploits the fact that the 4G paging protocol uses
   the IMSI to calculate the paging timeslot [Hussain2019].  As this
   attack is outside AKA, it does not impact EAP-AKA'.

   Finally, bad implementations of EAP-AKA' may not produce pseudonym
   usernames or fast re-authentication identities in a manner that is
   sufficiently secure.  While it is not a problem with the protocol
   itself, following the recommendations in Section 5.2 can mitigate
   this concern.

7.3.  Pervasive Monitoring

   As required by [RFC7258], work on IETF protocols needs to consider
   the effects of pervasive monitoring and mitigate them when possible.

   As described in Section 7.2, after the publication of RFC 5448, new
   information has come to light regarding the use of pervasive
   monitoring techniques against many security technologies, including
   AKA-based authentication.

   For AKA, these attacks relate to theft of the long-term, shared-
   secret key material stored on the cards.  Such attacks are
   conceivable, for instance, during the manufacturing process of cards,
   through coercion of the card manufacturers, or during the transfer of
   cards and associated information to an operator.  Since the
   publication of reports about such attacks, manufacturing and
   provisioning processes have gained much scrutiny and have improved.

   In particular, it is crucial that manufacturers limit access to the
   secret information and the cards only to necessary systems and
   personnel.  It is also crucial that secure mechanisms be used to
   store and communicate the secrets between the manufacturer and the
   operator that adopts those cards for their customers.

   Beyond these operational considerations, there are also technical
   means to improve resistance to these attacks.  One approach is to
   provide Perfect Forward Secrecy (PFS).  This would prevent any
   passive attacks merely based on the long-term secrets and observation
   of traffic.  Such a mechanism can be defined as a backwards-
   compatible extension of EAP-AKA' and is pursued separately from this
   specification [EMU-AKA-PFS].  Alternatively, EAP-AKA' authentication
   can be run inside a PFS-capable, tunneled authentication method.  In
   any case, the use of some PFS-capable mechanism is recommended.

7.4.  Security Properties of Binding Network Names

   The ability of EAP-AKA' to bind the network name into the used keys
   provides some additional protection against key leakage to
   inappropriate parties.  The keys used in the protocol are specific to
   a particular network name.  If key leakage occurs due to an accident,
   access node compromise, or another attack, the leaked keys are only
   useful when providing access with that name.  For instance, a
   malicious access point cannot claim to be network Y if it has stolen
   keys from network X.  Obviously, if an access point is compromised,
   the malicious node can still represent the compromised node.  As a
   result, neither EAP-AKA' nor any other extension can prevent such
   attacks; however, the binding to a particular name limits the
   attacker's choices, allows better tracking of attacks, makes it
   possible to identify compromised networks, and applies good
   cryptographic hygiene.

   The server receives the EAP transaction from a given access network,
   and verifies that the claim from the access network corresponds to
   the name that this access network should be using.  It becomes
   impossible for an access network to claim over AAA that it is another
   access network.  In addition, if the peer checks that the information
   it has received locally over the network-access link-layer matches
   with the information the server has given it via EAP-AKA', it becomes
   impossible for the access network to tell one story to the AAA
   network and another one to the peer.  These checks prevent some
   "lying NAS" (Network Access Server) attacks.  For instance, a roaming
   partner, R, might claim that it is the home network H in an effort to
   lure peers to connect to itself.  Such an attack would be beneficial
   for the roaming partner if it can attract more users, and damaging
   for the users if their access costs in R are higher than those in
   other alternative networks, such as H.

   Any attacker who gets hold of the keys CK and IK, produced by the AKA
   algorithm, can compute the keys CK' and IK' and, hence, the Master
   Key (MK) according to the rules in Section 3.3.  The attacker could
   then act as a lying NAS.  In 3GPP systems in general, the keys CK and
   IK have been distributed to, for instance, nodes in a visited access
   network where they may be vulnerable.  In order to reduce this risk,
   the AKA algorithm MUST be computed with the AMF separation bit set to
   1, and the peer MUST check that this is indeed the case whenever it
   runs EAP-AKA'.  Furthermore, [TS-3GPP.33.402] requires that no CK or
   IK keys computed in this way ever leave the home subscriber system.

   The additional security benefits obtained from the binding depend
   obviously on the way names are assigned to different access networks.
   This is specified in [TS-3GPP.24.302].  See also [TS-3GPP.23.003].
   Ideally, the names allow separating each different access technology,
   each different access network, and each different NAS within a
   domain.  If this is not possible, the full benefits may not be
   achieved.  For instance, if the names identify just an access
   technology, use of compromised keys in a different technology can be
   prevented, but it is not possible to prevent their use by other
   domains or devices using the same technology.

8.  IANA Considerations

   IANA has updated the "Extensible Authentication Protocol (EAP)
   Registry" and the "EAP-AKA and EAP-SIM Parameters" registry so that
   entries that pointed to RFC 5448 now point to this RFC instead.

8.1.  Type Value

   IANA has updated the reference for EAP-AKA' (0x32) in the "Method
   Types" subregistry under the "Extensible Authentication Protocol
   (EAP) Registry" to point to this document.  Per Section 6.2 of
   [RFC3748], this allocation can be made with Specification Required

8.2.  Attribute Type Values

   EAP-AKA' shares its attribute space and subtypes with EAP-SIM
   [RFC4186] and EAP-AKA [RFC4187].  No new registries are needed.

   IANA has updated the reference for AT_KDF_INPUT (23) and AT_KDF (24)
   in the "Attribute Types (Non-Skippable Attributes 0-127)" subregistry
   under the "EAP-AKA and EAP-SIM Parameters" registry to point to this
   document.  AT_KDF_INPUT and AT_KDF are defined in Sections 3.1 and
   3.2, respectively, of this document.

   IANA has also updated the reference for AT_BIDDING (136) in the
   "Attribute Types (Skippable Attributes 128-255)" subregistry of the
   "EAP-AKA and EAP-SIM Parameters" registry to point to this document.
   AT_BIDDING is defined in Section 4.

8.3.  Key Derivation Function Namespace

   IANA has updated the reference for the "EAP-AKA' AT_KDF Key
   Derivation Function Values" subregistry to point to this document.
   This subregistry appears under the "EAP-AKA and EAP-SIM Parameters"
   registry.  The references for following entries have also been
   updated to point to this document.  New values can be created through
   the Specification Required policy [RFC8126].

               | Value | Description           | Reference |
               | 0     | Reserved              | RFC 9048  |
               | 1     | EAP-AKA' with CK'/IK' | RFC 9048  |

                  Table 3: EAP-AKA' AT_KDF Key Derivation
                              Function Values

9.  References

9.1.  Normative References

              National Institute of Standards and Technology, "Secure
              Hash Standard", FIPS PUB 180-4,
              DOI 10.6028/NIST.FIPS.180-4, August 2015,

   [RFC2104]  Krawczyk, H., Bellare, M., and R. Canetti, "HMAC: Keyed-
              Hashing for Message Authentication", RFC 2104,
              DOI 10.17487/RFC2104, February 1997,

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,

   [RFC3748]  Aboba, B., Blunk, L., Vollbrecht, J., Carlson, J., and H.
              Levkowetz, Ed., "Extensible Authentication Protocol
              (EAP)", RFC 3748, DOI 10.17487/RFC3748, June 2004,

   [RFC4187]  Arkko, J. and H. Haverinen, "Extensible Authentication
              Protocol Method for 3rd Generation Authentication and Key
              Agreement (EAP-AKA)", RFC 4187, DOI 10.17487/RFC4187,
              January 2006, <https://www.rfc-editor.org/info/rfc4187>.

   [RFC7542]  DeKok, A., "The Network Access Identifier", RFC 7542,
              DOI 10.17487/RFC7542, May 2015,

   [RFC8126]  Cotton, M., Leiba, B., and T. Narten, "Guidelines for
              Writing an IANA Considerations Section in RFCs", BCP 26,
              RFC 8126, DOI 10.17487/RFC8126, June 2017,

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
              2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
              May 2017, <https://www.rfc-editor.org/info/rfc8174>.

              3GPP, "3rd Generation Partnership Project; Technical
              Specification Group Core Network and Terminals; Numbering,
              addressing and identification (Release 16)", Version
              16.7.0, 3GPP Technical Specification 23.003, June 2021.

              3GPP, "3rd Generation Partnership Project; Technical
              Specification Group Services and System Aspects; System
              architecture for the 5G System (5GS); (Release 16)",
              Version 16.9.0, 3GPP Technical Specification 23.501, June

              3GPP, "3rd Generation Partnership Project; Technical
              Specification Group Core Network and Terminals; Access to
              the 3GPP Evolved Packet Core (EPC) via non-3GPP access
              networks; Stage 3; (Release 16)", Version 16.4.0, 3GPP
              Technical Specification 24.302, July 2020.

              3GPP, "3rd Generation Partnership Project; Technical
              Specification Group Core Network and Terminals; Non-
              Access-Stratum (NAS) protocol for 5G System (5GS); Stage
              3; (Release 16)", Version 16.9.0, 3GPP Draft Technical
              Specification 24.501, June 2021.

              3GPP, "3rd Generation Partnership Project; Technical
              Specification Group Services and System Aspects; 3G
              Security; Security architecture (Release 16)", Version
              16.0.0, 3GPP Technical Specification 33.102, July 2020.

              3GPP, "3GPP System Architecture Evolution (SAE); Security
              aspects of non-3GPP accesses (Release 16)", Version
              16.0.0, 3GPP Technical Specification 33.402, July 2020.

              3GPP, "3rd Generation Partnership Project; Technical
              Specification Group Services and System Aspects; 3G
              Security; Security architecture and procedures for 5G
              System (Release 16)", Version 16.7.1, 3GPP Technical
              Specification 33.501, July 2021.

9.2.  Informative References

              Arapinis, M., Mancini, L., Ritter, E., Ryan, M., Golde,
              N., Redon, R., and R. Borgaonkar, "New Privacy Issues in
              Mobile Telephony: Fix and Verification", in CCS '12:
              Proceedings of the 2012 ACM Conference on Computer and
              Communications Security, Raleigh, North Carolina, USA,
              DOI 10.1145/2382196.2382221, October 2012,

              Basin, D., Dreier, J., Hirschi, L., Radomirović, S.,
              Sasse, R., and V. Stettler, "A Formal Analysis of 5G
              Authentication", arXiv:1806.10360,
              DOI 10.1145/3243734.3243846, August 2018,

              Borgaonkar, R., Hirschi, L., Park, S., and A. Shaik, "New
              Privacy Threat on 3G, 4G, and Upcoming 5G AKA Protocols",
              in IACR Cryptology ePrint Archive, 2018.

   [BT2013]   Beekman, J. G. and C. Thompson, "Breaking Cell Phone
              Authentication: Vulnerabilities in AKA, IMS and Android",
              in 7th USENIX Workshop on Offensive Technologies, WOOT
              '13, August 2013.

              Arkko, J., Norrman, K., and V. Torvinen, "Perfect-Forward
              Secrecy for the Extensible Authentication Protocol Method
              for Authentication and Key Agreement (EAP-AKA' PFS)", Work
              in Progress, Internet-Draft, draft-ietf-emu-aka-pfs-05, 30
              October 2020, <https://datatracker.ietf.org/doc/html/

              National Institute of Standards and Technology, "Secure
              Hash Standard", FIPS PUB 180-1,
              DOI 10.6028/NIST.FIPS.180-1, April 1995,

              National Institute of Standards and Technology, "Secure
              Hash Standard", FIPS PUB 180-2, August 2002,

              Scahill, J. and J. Begley, "How Spies Stole the Keys to
              the Encryption Castle", February 2015,

              Hussain, S., Echeverria, M., Chowdhury, O., Li, N., and E.
              Bertino, "Privacy Attacks to the 4G and 5G Cellular Paging
              Protocols Using Side Channel Information", in the
              proceedings of NDSS '19, held 24-27 February, 2019, San
              Diego, California, 2019.

   [Kune2012] Kune, D., Koelndorfer, J., Hopper, N., and Y. Kim,
              "Location Leaks on the GSM Air Interface", in the
              proceedings of NDSS '12, held 5-8 February, 2012, San
              Diego, California, 2012.

   [MT2012]   Mjølsnes, S. F. and J-K. Tsay, "A Vulnerability in the
              UMTS and LTE Authentication and Key Agreement Protocols",
              in Computer Network Security, Proceedings of the 6th
              International Conference on Mathematical Methods, Models
              and Architectures for Computer Network Security, Lecture
              Notes in Computer Science, Vol. 7531, pp. 65-76,
              DOI 10.1007/978-3-642-33704-8_6, October 2012,

   [RFC3310]  Niemi, A., Arkko, J., and V. Torvinen, "Hypertext Transfer
              Protocol (HTTP) Digest Authentication Using Authentication
              and Key Agreement (AKA)", RFC 3310, DOI 10.17487/RFC3310,
              September 2002, <https://www.rfc-editor.org/info/rfc3310>.

   [RFC4086]  Eastlake 3rd, D., Schiller, J., and S. Crocker,
              "Randomness Requirements for Security", BCP 106, RFC 4086,
              DOI 10.17487/RFC4086, June 2005,

   [RFC4169]  Torvinen, V., Arkko, J., and M. Naslund, "Hypertext
              Transfer Protocol (HTTP) Digest Authentication Using
              Authentication and Key Agreement (AKA) Version-2",
              RFC 4169, DOI 10.17487/RFC4169, November 2005,

   [RFC4186]  Haverinen, H., Ed. and J. Salowey, Ed., "Extensible
              Authentication Protocol Method for Global System for
              Mobile Communications (GSM) Subscriber Identity Modules
              (EAP-SIM)", RFC 4186, DOI 10.17487/RFC4186, January 2006,

   [RFC4284]  Adrangi, F., Lortz, V., Bari, F., and P. Eronen, "Identity
              Selection Hints for the Extensible Authentication Protocol
              (EAP)", RFC 4284, DOI 10.17487/RFC4284, January 2006,

   [RFC4306]  Kaufman, C., Ed., "Internet Key Exchange (IKEv2)
              Protocol", RFC 4306, DOI 10.17487/RFC4306, December 2005,

   [RFC5113]  Arkko, J., Aboba, B., Korhonen, J., Ed., and F. Bari,
              "Network Discovery and Selection Problem", RFC 5113,
              DOI 10.17487/RFC5113, January 2008,

   [RFC5247]  Aboba, B., Simon, D., and P. Eronen, "Extensible
              Authentication Protocol (EAP) Key Management Framework",
              RFC 5247, DOI 10.17487/RFC5247, August 2008,

   [RFC5281]  Funk, P. and S. Blake-Wilson, "Extensible Authentication
              Protocol Tunneled Transport Layer Security Authenticated
              Protocol Version 0 (EAP-TTLSv0)", RFC 5281,
              DOI 10.17487/RFC5281, August 2008,

   [RFC5448]  Arkko, J., Lehtovirta, V., and P. Eronen, "Improved
              Extensible Authentication Protocol Method for 3rd
              Generation Authentication and Key Agreement (EAP-AKA')",
              RFC 5448, DOI 10.17487/RFC5448, May 2009,

   [RFC6194]  Polk, T., Chen, L., Turner, S., and P. Hoffman, "Security
              Considerations for the SHA-0 and SHA-1 Message-Digest
              Algorithms", RFC 6194, DOI 10.17487/RFC6194, March 2011,

   [RFC6973]  Cooper, A., Tschofenig, H., Aboba, B., Peterson, J.,
              Morris, J., Hansen, M., and R. Smith, "Privacy
              Considerations for Internet Protocols", RFC 6973,
              DOI 10.17487/RFC6973, July 2013,

   [RFC7170]  Zhou, H., Cam-Winget, N., Salowey, J., and S. Hanna,
              "Tunnel Extensible Authentication Protocol (TEAP) Version
              1", RFC 7170, DOI 10.17487/RFC7170, May 2014,

   [RFC7258]  Farrell, S. and H. Tschofenig, "Pervasive Monitoring Is an
              Attack", BCP 188, RFC 7258, DOI 10.17487/RFC7258, May
              2014, <https://www.rfc-editor.org/info/rfc7258>.

   [RFC7296]  Kaufman, C., Hoffman, P., Nir, Y., Eronen, P., and T.
              Kivinen, "Internet Key Exchange Protocol Version 2
              (IKEv2)", STD 79, RFC 7296, DOI 10.17487/RFC7296, October
              2014, <https://www.rfc-editor.org/info/rfc7296>.

              Shaik, A., Seifert, J., Borgaonkar, R., Asokan, N., and V.
              Niemi, "Practical attacks against Privacy and Availability
              in 4G/LTE Mobile Communication Systems", in the
              proceedings of NDSS '16 held 21-24 February, 2016, San
              Diego, California, 2012.

              3GPP, "3rd Generation Partnership Project; Technical
              Specification Group Services and System Aspects; 3G
              Security; Specification of the MILENAGE Algorithm Set: An
              example algorithm set for the 3GPP authentication and key
              generation functions f1, f1*, f2, f3, f4, f5 and f5*;
              Document 4: Design Conformance Test Data (Release 14)",
              Version 16.0.0, 3GPP Technical Specification 35.208, July

   [ZF2005]   Zhang, M. and Y. Fang, "Security analysis and enhancements
              of 3GPP authentication and key agreement protocol", IEEE
              Transactions on Wireless Communications, Vol. 4, No. 2,
              DOI 10.1109/TWC.2004.842941, March 2005,

Appendix A.  Changes from RFC 5448

   The change from RFC 5448 was to refer to a newer version of
   [TS-3GPP.24.302].  This RFC includes an updated definition of the
   Network Name field to include 5G.

   Identifier usage for 5G has been specified in Section 5.3.  Also, the
   requirements for generating pseudonym usernames and fast re-
   authentication identities have been updated from the original
   definition in RFC 5448, which referenced RFC 4187.  See Section 5.

   Exported parameters for EAP-AKA' have been defined in Section 6, as
   required by [RFC5247], including the definition of those parameters
   for both full authentication and fast re-authentication.

   The security, privacy, and pervasive monitoring considerations have
   been updated or added.  See Section 7.

   The references to [RFC2119], [RFC4306], [RFC7296], [FIPS.180-1] and
   [FIPS.180-2] have been updated to their most recent versions, and
   language in this document has been changed accordingly.  However,
   these are merely reference updates to newer specifications; the
   actual protocol functions are the same as defined in the earlier

   Similarly, references to all 3GPP technical specifications have been
   updated to their 5G versions (Release 16) or otherwise most recent
   version when there has not been a 5G-related update.

   Finally, a number of clarifications have been made, including a
   summary of where attributes may appear.

Appendix B.  Changes to RFC 4187

   In addition to specifying EAP-AKA', this document also mandates a
   change to another EAP method -- EAP-AKA that was defined in RFC 4187.
   This change was already mandated in RFC 5448 but repeated here to
   ensure that the latest EAP-AKA' specification contains the
   instructions about the necessary bidding down prevention feature in
   EAP-AKA as well.

   The changes to RFC 4187 relate only to the bidding down prevention
   support defined in Section 4.  In particular, this document does not
   change how the Master Key (MK) is calculated or any other aspect of
   EAP-AKA.  The provisions in this specification for EAP-AKA' do not
   apply to EAP-AKA, outside of Section 4.

Appendix C.  Importance of Explicit Negotiation

   Choosing between the traditional and revised AKA key derivation
   functions is easy when their use is unambiguously tied to a
   particular radio access network, e.g., Long Term Evolution (LTE) as
   defined by 3GPP or evolved High Rate Packet Data (eHRPD) as defined
   by 3GPP2.  There is no possibility for interoperability problems if
   this radio access network is always used in conjunction with new
   protocols that cannot be mixed with the old ones; clients will always
   know whether they are connecting to the old or new system.

   However, using the new key derivation functions over EAP introduces
   several degrees of separation, making the choice of the correct key
   derivation functions much harder.  Many different types of networks
   employ EAP.  Most of these networks have no means to carry any
   information about what is expected from the authentication process.
   EAP itself is severely limited in carrying any additional
   information, as noted in [RFC4284] and [RFC5113].  Even if these
   networks or EAP were extended to carry additional information, it
   would not affect millions of deployed access networks and clients
   attaching to them.

   Simply changing the key derivation functions that EAP-AKA [RFC4187]
   uses would cause interoperability problems with all of the existing
   implementations.  Perhaps it would be possible to employ strict
   separation into domain names that should be used by the new clients
   and networks.  Only these new devices would then employ the new key
   derivation function.  While this can be made to work for specific
   cases, it would be an extremely brittle mechanism, ripe to result in
   problems whenever client configuration, routing of authentication
   requests, or server configuration does not match expectations.  It
   also does not help to assume that the EAP client and server are
   running a particular release of 3GPP network specifications.  Network
   vendors often provide features from future releases early or do not
   provide all features of the current release.  And obviously, there
   are many EAP and even some EAP-AKA implementations that are not
   bundled with the 3GPP network offerings.  In general, these
   approaches are expected to lead to hard-to-diagnose problems and
   increased support calls.

Appendix D.  Test Vectors

   Test vectors are provided below for four different cases.  The test
   vectors may be useful for testing implementations.  In the first two
   cases, we employ the MILENAGE algorithm and the algorithm
   configuration parameters (the subscriber key K and operator algorithm
   variant configuration value OP) from test set 19 in [TS-3GPP.35.208].

   The last two cases use artificial values as the output of AKA, which
   are useful only for testing the computation of values within EAP-
   AKA', not AKA itself.

   Case 1

      The parameters for the AKA run are as follows:

         Identity:     "0555444333222111"

         Network name: "WLAN"

         RAND:         81e9 2b6c 0ee0 e12e bceb a8d9 2a99 dfa5

         AUTN:         bb52 e91c 747a c3ab 2a5c 23d1 5ee3 51d5

         IK:           9744 871a d32b f9bb d1dd 5ce5 4e3e 2e5a

         CK:           5349 fbe0 9864 9f94 8f5d 2e97 3a81 c00f

         RES:          28d7 b0f2 a2ec 3de5

      Then the derived keys are generated as follows:

         CK':          0093 962d 0dd8 4aa5 684b 045c 9edf fa04

         IK':          ccfc 230c a74f cc96 c0a5 d611 64f5 a76c

         K_encr:       766f a0a6 c317 174b 812d 52fb cd11 a179

         K_aut:        0842 ea72 2ff6 835b fa20 3249 9fc3 ec23
                       c2f0 e388 b4f0 7543 ffc6 77f1 696d 71ea

         K_re:         cf83 aa8b c7e0 aced 892a cc98 e76a 9b20
                       95b5 58c7 795c 7094 715c b339 3aa7 d17a

         MSK:          67c4 2d9a a56c 1b79 e295 e345 9fc3 d187
                       d42b e0bf 818d 3070 e362 c5e9 67a4 d544
                       e8ec fe19 358a b303 9aff 03b7 c930 588c
                       055b abee 58a0 2650 b067 ec4e 9347 c75a

         EMSK:         f861 703c d775 590e 16c7 679e a387 4ada
                       8663 11de 2907 64d7 60cf 76df 647e a01c
                       313f 6992 4bdd 7650 ca9b ac14 1ea0 75c4
                       ef9e 8029 c0e2 90cd bad5 638b 63bc 23fb

   Case 2

      The parameters for the AKA run are as follows:

         Identity:     "0555444333222111"

         Network name: "HRPD"

         RAND:         81e9 2b6c 0ee0 e12e bceb a8d9 2a99 dfa5

         AUTN:         bb52 e91c 747a c3ab 2a5c 23d1 5ee3 51d5

         IK:           9744 871a d32b f9bb d1dd 5ce5 4e3e 2e5a

         CK:           5349 fbe0 9864 9f94 8f5d 2e97 3a81 c00f

         RES:          28d7 b0f2 a2ec 3de5

      Then the derived keys are generated as follows:

         CK':          3820 f027 7fa5 f777 32b1 fb1d 90c1 a0da

         IK':          db94 a0ab 557e f6c9 ab48 619c a05b 9a9f

         K_encr:       05ad 73ac 915f ce89 ac77 e152 0d82 187b

         K_aut:        5b4a caef 62c6 ebb8 882b 2f3d 534c 4b35
                       2773 37a0 0184 f20f f25d 224c 04be 2afd

         K_re:         3f90 bf5c 6e5e f325 ff04 eb5e f653 9fa8
                       cca8 3981 94fb d00b e425 b3f4 0dba 10ac

         MSK:          87b3 2157 0117 cd6c 95ab 6c43 6fb5 073f
                       f15c f855 05d2 bc5b b735 5fc2 1ea8 a757
                       57e8 f86a 2b13 8002 e057 5291 3bb4 3b82
                       f868 a961 17e9 1a2d 95f5 2667 7d57 2900

         EMSK:         c891 d5f2 0f14 8a10 0755 3e2d ea55 5c9c
                       b672 e967 5f4a 66b4 bafa 0273 79f9 3aee
                       539a 5979 d0a0 042b 9d2a e28b ed3b 17a3
                       1dc8 ab75 072b 80bd 0c1d a612 466e 402c

   Case 3

      The parameters for the AKA run are as follows:

         Identity:     "0555444333222111"

         Network name: "WLAN"

         RAND:         e0e0 e0e0 e0e0 e0e0 e0e0 e0e0 e0e0 e0e0

         AUTN:         a0a0 a0a0 a0a0 a0a0 a0a0 a0a0 a0a0 a0a0

         IK:           b0b0 b0b0 b0b0 b0b0 b0b0 b0b0 b0b0 b0b0

         CK:           c0c0 c0c0 c0c0 c0c0 c0c0 c0c0 c0c0 c0c0

         RES:          d0d0 d0d0 d0d0 d0d0 d0d0 d0d0 d0d0 d0d0

      Then the derived keys are generated as follows:

         CK':          cd4c 8e5c 68f5 7dd1 d7d7 dfd0 c538 e577

         IK':          3ece 6b70 5dbb f7df c459 a112 80c6 5524

         K_encr:       897d 302f a284 7416 488c 28e2 0dcb 7be4

         K_aut:        c407 00e7 7224 83ae 3dc7 139e b0b8 8bb5
                       58cb 3081 eccd 057f 9207 d128 6ee7 dd53

         K_re:         0a59 1a22 dd8b 5b1c f29e 3d50 8c91 dbbd
                       b4ae e230 5189 2c42 b6a2 de66 ea50 4473

         MSK:          9f7d ca9e 37bb 2202 9ed9 86e7 cd09 d4a7
                       0d1a c76d 9553 5c5c ac40 a750 4699 bb89
                       61a2 9ef6 f3e9 0f18 3de5 861a d1be dc81
                       ce99 1639 1b40 1aa0 06c9 8785 a575 6df7

         EMSK:         724d e00b db9e 5681 87be 3fe7 4611 4557
                       d501 8779 537e e37f 4d3c 6c73 8cb9 7b9d
                       c651 bc19 bfad c344 ffe2 b52c a78b d831
                       6b51 dacc 5f2b 1440 cb95 1552 1cc7 ba23

   Case 4

      The parameters for the AKA run are as follows:

         Identity:     "0555444333222111"

         Network name: "HRPD"

         RAND:         e0e0 e0e0 e0e0 e0e0 e0e0 e0e0 e0e0 e0e0

         AUTN:         a0a0 a0a0 a0a0 a0a0 a0a0 a0a0 a0a0 a0a0

         IK:           b0b0 b0b0 b0b0 b0b0 b0b0 b0b0 b0b0 b0b0

         CK:           c0c0 c0c0 c0c0 c0c0 c0c0 c0c0 c0c0 c0c0

         RES:          d0d0 d0d0 d0d0 d0d0 d0d0 d0d0 d0d0 d0d0

      Then the derived keys are generated as follows:

         CK':          8310 a71c e6f7 5488 9613 da8f 64d5 fb46

         IK':          5adf 1436 0ae8 3819 2db2 3f6f cb7f 8c76

         K_encr:       745e 7439 ba23 8f50 fcac 4d15 d47c d1d9

         K_aut:        3e1d 2aa4 e677 025c fd86 2a4b e183 61a1
                       3a64 5765 5714 63df 833a 9759 e809 9879

         K_re:         99da 835e 2ae8 2462 576f e651 6fad 1f80
                       2f0f a119 1655 dd0a 273d a96d 04e0 fcd3

         MSK:          c6d3 a6e0 ceea 951e b20d 74f3 2c30 61d0
                       680a 04b0 b086 ee87 00ac e3e0 b95f a026
                       83c2 87be ee44 4322 94ff 98af 26d2 cc78
                       3bac e75c 4b0a f7fd feb5 511b a8e4 cbd0

         EMSK:         7fb5 6813 838a dafa 99d1 40c2 f198 f6da
                       cebf b6af ee44 4961 1054 02b5 08c7 f363
                       352c b291 9644 b504 63e6 a693 5415 0147
                       ae09 cbc5 4b8a 651d 8787 a689 3ed8 536d


   The authors would like to thank Guenther Horn, Joe Salowey, Mats
   Naslund, Adrian Escott, Brian Rosenberg, Laksminath Dondeti, Ahmad
   Muhanna, Stefan Rommer, Miguel Garcia, Jan Kall, Ankur Agarwal, Jouni
   Malinen, John Mattsson, Jesus De Gregorio, Brian Weis, Russ Housley,
   Alfred Hoenes, Anand Palanigounder, Michael Richardson, Roman
   Danyliw, Dan Romascanu, Kyle Rose, Benjamin Kaduk, Alissa Cooper,
   Erik Kline, Murray Kucherawy, Robert Wilton, Warren Kumari, Andreas
   Kunz, Marcus Wong, Kalle Jarvinen, Daniel Migault, and Mohit Sethi
   for their in-depth reviews and interesting discussions in this
   problem space.


   The test vectors in Appendix D were provided by Yogendra Pal and
   Jouni Malinen, based on two independent implementations of this

   Jouni Malinen provided suggested text for Section 6.  John Mattsson
   provided much of the text for Section 7.1.  Karl Norrman was the
   source of much of the information in Section 7.2.

Authors' Addresses

   Jari Arkko
   FI-02420 Jorvas

   Email: jari.arkko@piuha.net

   Vesa Lehtovirta
   FI-02420 Jorvas

   Email: vesa.lehtovirta@ericsson.com

   Vesa Torvinen
   FI-02420 Jorvas

   Email: vesa.torvinen@ericsson.com

   Pasi Eronen

   Email: pe@iki.fi