PPPEXT Working Group Paul Funk Internet-Draft Funk Software, Inc. Category: Standards Track Simon Blake-Wilson Basic Commerce & Industries, Inc. July 2004 EAP Tunneled TLS Authentication Protocol (EAP-TTLS) Status of this Memo This document is an Internet-Draft and is subject to all provisions of Section 10 of RFC2026. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/1id-abstracts.html The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html Copyright Notice Copyright (C) The Internet Society (2001-2004). All Rights Reserved. Abstract EAP-TTLS is an EAP protocol that extends EAP-TLS. In EAP-TLS, a TLS handshake is used to mutually authenticate a client and server. EAP- TTLS extends this authentication negotiation by using the secure connection established by the TLS handshake to exchange additional information between client and server. In EAP-TTLS, the TLS handshake may be mutual; or it may be one-way, in which only the server is authenticated to the client. The secure connection established by the handshake may then be used to allow the server to authenticate the client using existing, widely-deployed authentication infrastructures such as RADIUS. The authentication of Internet-Draft April 2004 the client may itself be EAP, or it may be another authentication protocol such as PAP, CHAP, MS-CHAP or MS-CHAP-V2. Thus, EAP-TTLS allows legacy password-based authentication protocols to be used against existing authentication databases, while protecting the security of these legacy protocols against eavesdropping, man-in-the-middle and other cryptographic attacks. EAP-TTLS also allows client and server to establish keying material for use in the data connection between the client and access point. The keying material is established implicitly between client and server based on the TLS handshake. This document describes two versions of EAP-TTLS - version 0 and version 1. Most of the document concerns EAP-TTLS v0, a form of the protocol that has been implemented by multiple vendors. Section 11 defines EAP-TTLS v1, an enhanced version of the protocol that utilizes the TLS extensions mechanism to allow authentications to occur within, rather than after, the TLS handshake. The TLS extension that is defined is believed to useful in its own right, and may be used in other contexts in addition to EAP-TTLS v1. Table of Contents 1. Introduction......................................................3 2. Motivation........................................................4 3. Terminology.......................................................6 4. Architectural Model...............................................8 4.1 Carrier Protocols.............................................9 4.2 Security Relationships.......................................10 4.3 Messaging....................................................10 4.4 Resulting Security...........................................11 5. Protocol Layering Model..........................................11 6. EAP-TTLS version 0 Overview......................................12 6.1 Phase 1: Handshake...........................................13 6.2 Phase 2: Tunnel..............................................14 6.3 Piggybacking.................................................14 6.4 Session Resumption...........................................15 6.4.1 TTLS Server Guidelines for Session Resumption............16 7. Generating Keying Material.......................................16 8. EAP-TTLS Encoding................................................17 8.1 EAP-TTLS Start Packet........................................17 8.2 EAP-TTLS Packets with No Data................................18 9. Encapsulation of AVPs within the TLS Record Layer................18 9.1 AVP Format...................................................19 9.2 AVP Sequences................................................20 9.3 Guidelines for Maximum Compatibility with AAA Servers........20 10. Tunneled Authentication..........................................20 10.1 Implicit challenge...........................................21 10.2 Tunneled Authentication Protocols............................21 10.2.1 EAP ......................................................22 Paul Funk expires January 2005 [Page 2] Internet-Draft April 2004 10.2.2 CHAP .....................................................23 10.2.3 MS-CHAP..................................................23 10.2.4 MS-CHAP-V2...............................................24 10.2.5 PAP ......................................................25 10.3 Performing Multiple Authentications..........................26 11. EAP-TTLS Version 1...............................................27 11.1 EAP-TTLS v1 Introduction.....................................27 11.2 Intentions Beyond EAP-TTLS...................................28 11.3 The InnerApplication Extension to TLS........................28 11.3.1 TLS/IA Overview..........................................29 11.3.2 Message Exchange.........................................31 11.3.3 Master Key Permutation...................................31 11.3.4 Session Resumption.......................................33 11.3.5 Error Termination........................................33 11.3.6 Application Session Key Material.........................33 11.3.7 Computing Verification Data..............................34 11.3.8 Attribute-Value Pairs (AVPs).............................36 11.3.9 TLS/IA Messages..........................................36 11.3.10 The InnerApplication Extension...........................37 11.3.11 The PhaseFinished Handshake Message......................37 11.3.12 The ApplicationPayload Handshake Message.................37 11.3.13 The InnerApplicationFailure Alert........................38 11.4 Binding of TLS/IA to EAP-TTLS v1.............................38 11.4.1 Flags Octet..............................................38 11.4.2 Version Negotiation......................................39 11.4.3 Acknowledgement Packets..................................39 11.4.4 Generating Keying Material...............................40 12. Discussion of Certificates and PKI...............................40 13. Message Sequences................................................42 13.1 Successful authentication via tunneled CHAP..................42 13.2 Successful authentication via tunneled EAP/MD5-Challenge.....44 13.3 Successful session resumption................................46 14. Security Considerations..........................................47 15. Changes since previous drafts....................................49 16. References.......................................................50 17. Authors' Addresses...............................................51 18. Full Copyright Statement.........................................51 1. Introduction Extensible Authentication Protocol (EAP) [2] defines a standard message exchange that allows a server to authenticate a client based on an authentication protocol agreed upon by both parties. EAP may be extended with additional authentication protocols by registering such protocols with IANA or by defining vendor specific protocols. Transport Layer Security (TLS) [3] is an authentication protocol that provides for client authentication of a server or mutual authentication of client and server, as well as secure ciphersuite negotiation and key exchange between the parties. TLS has been Paul Funk expires January 2005 [Page 3] Internet-Draft April 2004 defined as an authentication protocol for use within EAP (EAP-TLS) [1]. Other authentication protocols are also widely deployed. These are typically password-based protocols, and there is a large installed base of support for these protocols in the form of credential databases that may be accessed by RADIUS, Diameter or other AAA servers. These include non-EAP protocols such as PAP, CHAP, MS-CHAP and MS-CHAP-V2, as well as EAP protocols such as MD5-Challenge. EAP-TTLS is an EAP protocol that extends EAP-TLS. In EAP-TLS, a TLS handshake is used to mutually authenticate a client and server. EAP- TTLS extends this authentication negotiation by using the secure connection established by the TLS handshake to exchange additional information between client and server. In EAP-TTLS, the TLS handshake may be mutual; or it may be one-way, in which only the server is authenticated to the client. The secure connection established by the handshake may then be used to allow the server to authenticate the client using existing, widely-deployed authentication infrastructures such as RADIUS. The authentication of the client may itself be EAP, or it may be another authentication protocol such as PAP, CHAP, MS-CHAP or MS-CHAP-V2. Thus, EAP-TTLS allows legacy password-based authentication protocols to be used against existing authentication databases, while protecting the security of these legacy protocols against eavesdropping, man-in-the-middle and other cryptographic attacks. EAP-TTLS also allows client and server to establish keying material for use in the data connection between the client and access point. The keying material is established implicitly between client and server based on the TLS handshake. In EAP-TTLS, client and server communicate using attribute-value pairs encrypted within TLS. This generality allows arbitrary functions beyond authentication and key exchange to be added to the EAP negotiation, in a manner compatible with the AAA infrastructure. 2. Motivation Most password-based protocols in use today rely on a hash of the password with a random challenge. Thus, the server issues a challenge, the client hashes that challenge with the password and forwards a response to the server, and the server validates that response against the user's password retrieved from its database. This general approach describes CHAP, MS-CHAP, MS-CHAP-V2, EAP/MD5- Challenge and EAP/One-Time Password. An issue with such an approach is that an eavesdropper that observes both challenge and response may be able to mount a dictionary attack, in which random passwords are tested against the known Paul Funk expires January 2005 [Page 4] Internet-Draft April 2004 challenge to attempt to find one which results in the known response. Because passwords typically have low entropy, such attacks can in practice easily discover many passwords. While this vulnerability has long been understood, it has not been of great concern in environments where eavesdropping attacks are unlikely in practice. For example, users with wired or dial-up connections to their service providers have not been concerned that such connections may be monitored. Users have also been willing to entrust their passwords to their service providers, or at least to allow their service providers to view challenges and hashed responses which are then forwarded to their home authentication servers using, for example, proxy RADIUS, without fear that the service provider will mount dictionary attacks on the observed credentials. Because a user typically has a relationship with a single service provider, such trust is entirely manageable. With the advent of wireless connectivity, however, the situation changes dramatically: - Wireless connections are considerably more susceptible to eavesdropping and man-in-the-middle attacks. These attacks may enable dictionary attacks against low-entropy passwords. In addition, they may enable channel hijacking, in which an attacker gains fraudulent access by seizing control of the communications channel after authentication is complete. - Existing authentication protocols often begin by exchanging the clientĘs username in the clear. In the context of eavesdropping on the wireless channel, this can compromise the clientĘs anonymity and locational privacy. - Often in wireless networks, the access point does not reside in the administrative domain of the service provider with which the user has a relationship. For example, the access point may reside in an airport, coffee shop, or hotel in order to provide public access via 802.11. Even if password authentications are protected in the wireless leg, they may still be susceptible to eavesdropping within the untrusted wired network of the access point. - In the traditional wired world, the user typically intentionally connects with a particular service provider by dialing an associated phone number; that service provider may be required to route an authentication to the user's home domain. In a wireless network, however, the user does not get to choose an access domain, and must connect with whichever access point is nearby; providing for the routing of the authentication from an arbitrary access point to the user's home domain may pose a challenge. Paul Funk expires January 2005 [Page 5] Internet-Draft April 2004 Thus, the authentication requirements for a wireless environment that EAP-TTLS attempts to address can be summarized as follows: - Legacy password protocols must be supported, to allow easy deployment against existing authentication databases. - Password-based information must not be observable in the communications channel between the client node and a trusted service provider, to protect the user against dictionary attacks. - The user's identity must not be observable in the communications channel between the client node and a trusted service provider, to protect the user's locational privacy against surveillance, undesired acquisition of marketing information, and the like. - The authentication process must result in the distribution of shared keying information to the client and access point to permit encryption and validation of the wireless data connection subsequent to authentication, to secure it against eavesdroppers and prevent channel hijacking. - The authentication mechanism must support roaming among small access domains with which the user has no relationship and which will have limited capabilities for routing authentication requests. 3. Terminology AAA Authentication, Authorization and Accounting - functions that are generally required to control access to a network and support billing and auditing. AAA protocol A network protocol used to communicate with AAA servers; examples include RADIUS and Diameter. AAA server A server which performs one or more AAA functions: authenticating a user prior to granting network service, providing authorization (policy) information governing the type of network service the user is to be granted, and accumulating accounting information about actual usage. AAA/H A AAA server in the user's home domain, where authentication and authorization for that user are administered. Paul Funk expires January 2005 [Page 6] Internet-Draft April 2004 access point A network device providing users with a point of entry into the network, and which may enforce access control and policy based on information returned by a AAA server. For the purposes of this document, "access point" and "NAS" are architecturally equivalent. "Access point" is used throughout because it is suggestive of devices used for wireless access; "NAS" is used when more traditional forms of access, such as dial-up, are discussed. access domain The domain, including access points and other devices, that provides users with an initial point of entry into the network; for example, a wireless hot spot. client A host or device that connects to a network through an access point. domain A network and associated devices that are under the administrative control of an entity such as a service provider or the user's home organization. link layer protocol A protocol used to carry data between hosts that are connected within a single network segment; examples include PPP and Ethernet. NAI A Network Access Identifier [7], normally consisting of the name of the user and, optionally, the user's home realm. NAS A network device providing users with a point of entry into the network, and which may enforce access control and policy based on information returned by a AAA server. For the purposes of this document, "access point" and "NAS" are architecturally equivalent. "Access point" is used throughout because it is suggestive of devices used for wireless access; "NAS" is used when more traditional forms of access, such as dial-up, are discussed. proxy Paul Funk expires January 2005 [Page 7] Internet-Draft April 2004 A server that is able to route AAA transactions to the appropriate AAA server, possibly in another domain, typically based on the realm portion of an NAI. realm The optional part of an NAI indicating the domain to which a AAA transaction is to be routed, normally the user's home domain. service provider An organization with which a user has a business relationship, that provides network or other services. The service provider may provide the access equipment with which the user connects, may perform authentication or other AAA functions, may proxy AAA transactions to the user's home domain, etc. TTLS server A AAA server which implements EAP-TTLS. This server may also be capable of performing user authentication, or it may proxy the user authentication to a AAA/H. user The person operating the client device. Though the line is often blurred, "user" is intended to refer to the human being who is possessed of an identity (username), password or other authenticating information, and "client" is intended to refer to the device which makes use of this information to negotiate network access. There may also be clients with no human operators; in this case the term "user" is a convenient abstraction. 4. Architectural Model The network architectural model for EAP-TTLS usage and the type of security it provides is shown below. +----------+ +----------+ +----------+ +----------+ | | | | | | | | | client |<---->| access |<---->| TTLS AAA |<---->| AAA/H | | | | point | | server | | server | | | | | | | | | +----------+ +----------+ +----------+ +----------+ <---- secure password authentication tunnel ---> <---- secure data tunnel ----> Paul Funk expires January 2005 [Page 8] Internet-Draft April 2004 The entities depicted above are logical entities and may or may not correspond to separate network components. For example, the TTLS server and AAA/H server might be a single entity; the access point and TTLS server might be a single entity; or, indeed, the functions of the access point, TTLS server and AAA/H server might be combined into a single physical device. The above diagram illustrates the division of labor among entities in a general manner and shows how a distributed system might be constructed; however, actual systems might be realized more simply. Note also that one or more AAA proxy servers might be deployed between access point and TTLS server, or between TTLS server and AAA/H server. Such proxies typically perform aggregation or are required for realm-based message routing. However, such servers play no direct role in EAP-TTLS and are therefore not shown. 4.1 Carrier Protocols The entities shown above communicate with each other using carrier protocols capable of encapsulating EAP. The client and access point communicate using a link layer carrier protocol such as PPP or EAPOL. The access point, TTLS server and AAA/H server communicate using a AAA carrier protocol such as RADIUS or Diameter. EAP, and therefore EAP-TTLS, must be initiated via the link layer protocol. In PPP or EAPOL, for example, EAP is initiated when the access point sends an EAP-Request/Identity packet to the client. The keying material used to encrypt and authenticate the data connection between the client and access point is developed implicitly between the client and TTLS server as a result of EAP- TTLS negotiation. This keying material must be communicated to the access point by the TTLS server using the AAA carrier protocol. The client and access point must also agree on an encryption/validation algorithm to be used based on the keying material. In some systems, both these devices may be preconfigured with this information, and distribution of the keying material alone is sufficient. Or, the link layer protocol may provide a mechanism for client and access point to negotiate an algorithm. In the most general case, however, it may be necessary for both client and access point to communicate their algorithm preferences to the TTLS server, and for the TTLS server to select one and communicate its choice to both parties. This information would be transported between access point and TTLS server via the AAA protocol, and between client and TTLS server via EAP-TTLS in encrypted form. Paul Funk expires January 2005 [Page 9] Internet-Draft April 2004 4.2 Security Relationships The client and access point have no pre-existing security relationship. The access point, TTLS server and AAA/H server are each assumed to have a pre-existing security association with the adjacent entity with which it communicates. With RADIUS, for example, this is achieved using shared secrets. It is essential for such security relationships to permit secure key distribution. The client and AAA/H server have a security relationship based on the user's credentials such as a password. The client and TTLS server may have a one-way security relationship based on the TTLS server's possession of a private key guaranteed by a CA certificate which the user trusts, or may have a mutual security relationship based on certificates for both parties. 4.3 Messaging The client and access point initiate an EAP conversation to negotiate the client's access to the network. Typically, the access point issues an EAP-Request/Identity to the client, which responds with an EAP-Response/Identity. Note that the client does not include the user's actual identity in this EAP-Response/Identity packet; the user's identity will not be transmitted until an encrypted channel has been established. The access point now acts as a passthrough device, allowing the TTLS server to negotiate EAP-TTLS with the client directly. During the first phase of the negotiation, the TLS handshake protocol is used to authenticate the TTLS server to the client and, optionally, to authenticate the client to the TTLS server, based on public/private key certificates. As a result of the handshake, client and TTLS server now have shared keying material and an agreed upon TLS record layer cipher suite with which to secure subsequent EAP-TTLS communication. During the second phase of negotiation, client and TTLS server use the secure TLS record layer channel established by the TLS handshake as a tunnel to exchange information encapsulated in attribute-value pairs, to perform additional functions such as client authentication and key distribution for the subsequent data connection. If a tunneled client authentication is performed, the TTLS server de-tunnels and forwards the authentication information to the AAA/H. If the AAA/H performs a challenge, the TTLS server tunnels the challenge information to the client. The AAA/H server may be a legacy device and needs to know nothing about EAP-TTLS; it only Paul Funk expires January 2005 [Page 10] Internet-Draft April 2004 needs to be able to authenticate the client based on commonly used authentication protocols. Keying material for the subsequent data connection between client and access point may be generated based on secret information developed during the TLS handshake between client and TTLS server. At the conclusion of a successful authentication, the TTLS server may transmit this keying material to the access point, encrypted based on the existing security associations between those devices (e.g., RADIUS). The client and access point now share keying material which they can use to encrypt data traffic between them. 4.4 Resulting Security As the diagram above indicates, EAP-TTLS allows user identity and password information to be securely transmitted between client and TTLS server, and performs key distribution to allow network data subsequent to authentication to be securely transmitted between client and access point. 5. Protocol Layering Model EAP-TTLS packets are encapsulated within EAP, and EAP in turn requires a carrier protocol to transport it. EAP-TTLS packets themselves encapsulate TLS, which is then used to encapsulate user authentication information. Thus, EAP-TTLS messaging can be described using a layered model, where each layer is encapsulated by the layer beneath it. The following diagram clarifies the relationship between protocols: +--------------------------------------------------------+ | User Authentication Protocol (PAP, CHAP, MS-CHAP, etc.)| +--------------------------------------------------------+ | TLS | +--------------------------------------------------------+ | EAP-TTLS | +--------------------------------------------------------+ | EAP | +--------------------------------------------------------+ | Carrier Protocol (PPP, EAPOL, RADIUS, Diameter, etc.) | +--------------------------------------------------------+ When the user authentication protocol is itself EAP, the layering is as follows: Paul Funk expires January 2005 [Page 11] Internet-Draft April 2004 +--------------------------------------------------------+ | User EAP Authentication Protocol (MD-Challenge, etc.) | +--------------------------------------------------------+ | EAP | +--------------------------------------------------------+ | TLS | +--------------------------------------------------------+ | EAP-TTLS | +--------------------------------------------------------+ | EAP | +--------------------------------------------------------+ | Carrier Protocol (PPP, EAPOL, RADIUS, Diameter, etc.) | +--------------------------------------------------------+ Methods for encapsulating EAP within carrier protocols are already defined. For example, PPP [5] or EAPOL [4] may be used to transport EAP between client and access point; RADIUS [6] or Diameter [8] are used to transport EAP between access point and TTLS server. 6. EAP-TTLS version 0 Overview [Authors' note: This section as well as sections 7, 8, 9 and 10, describe version 0 of the EAP-TTLS protocol. Section 11 describes version 1 of EAP-TTLS. Much of the material describing version 0 also applies to version 1; the version 1 documentation will refer to the version 0 material as required. The intention is to provide a separate draft for each of the two versions in the near future.] A EAP-TTLS negotiation comprises two phases: the TLS handshake phase and the TLS tunnel phase. During phase 1, TLS is used to authenticate the TTLS server to the client and, optionally, the client to the TTLS server. Phase 1 results in the activation of a cipher suite, allowing phase 2 to proceed securely using the TLS record layer. (Note that the type and degree of security in phase 2 depends on the cipher suite negotiated during phase 1; if the null cipher suite is negotiated, there will be no security!) During phase 2, the TLS record layer is used to tunnel information between client and TTLS server to perform any of a number of functions. These might include user authentication, negotiation of data communication security capabilities, key distribution, communication of accounting information, etc.. Information between client and TTLS server is exchanged via attribute-value pairs (AVPs) compatible with RADIUS and Diameter; thus, any type of function that can be implemented via such AVPs may easily be performed. EAP-TTLS specifies how user authentication may be performed during phase 2. The user authentication may itself be EAP, or it may be a legacy protocol such as PAP, CHAP, MS-CHAP or MS-CHAP-V2. Phase 2 Paul Funk expires January 2005 [Page 12] Internet-Draft April 2004 user authentication may not always be necessary, since the user may already have been authenticated via the mutual authentication option of the TLS handshake protocol. EAP-TTLS is also intended for use in key distribution, and specifies how keying material for the data connection between client and access point is generated. The keying material is developed implicitly between client and TTLS server based on the results of the TLS handshake; the TTLS server will communicate the keying material to the access point over the carrier protocol However, EAP-TTLS does not specify particular key distribution AVPs and their use, since the needs of various systems will be different. Instead, a general model for key distribution is suggested. Organizations may define their own AVPs for this use, possibly using vendor-specific AVPs, either in conformance with the suggested model or otherwise. 6.1 Phase 1: Handshake In phase 1, the TLS handshake protocol is used to authenticate the TTLS server to the client and, optionally, to authenticate the client to the TTLS server. Phase 1 is initiated when the client sends an EAP-Response/Identity packet to the TTLS server. This packet specifically should not include the name of the user; however, it may include the name of the realm of a trusted provider to which EAP-TTLS packets should be forwarded; for example, "@myisp.com". The TTLS server responds to the EAP-Response/Identity packet with a EAP-TTLS/Start packet, which is an EAP-Request with Type = EAP-TTLS, the S (Start) bit set, and no data. This indicates to the client that it should begin TLS handshake by sending a ClientHello message. EAP packets continue to be exchanged between client and TTLS server to complete the TLS handshake, as described in [1]. Phase 1 is completed when the client and TTLS server exchange ChangeCipherSpec and Finished messages. At this point, additional information may be securely tunneled. As part of the TLS handshake protocol, the TTLS server will send its certificate along with a chain of certificates leading to the certificate of a trusted CA. The client will need to be configured with the certificate of the trusted CA in order to perform the authentication. If certificate-based authentication of the client is desired, the client must have been issued a certificate and must have the private key associated with that certificate Paul Funk expires January 2005 [Page 13] Internet-Draft April 2004 6.2 Phase 2: Tunnel In phase 2, the TLS Record Layer is used to securely tunnel information between client and TTLS server. This information is encapsulated in sequences of attribute-value pairs (AVPS), whose use and format are described in later sections. Any type of information may be exchanged during phase 2, according to the requirements of the system. (It is expected that applications utilizing EAP-TTLS will specify what information must be exchanged and therefore which AVPs must be supported.) The client begins the phase 2 exchange by encoding information in a sequence of AVPs, passing this sequence to the TLS record layer for encryption, and sending the resulting data to the TTLS server. The TTLS server recovers the AVPs in clear text from the TLS record layer. If the AVP sequence includes authentication information, it forwards this information to the AAA/H server using the AAA carrier protocol. Note that the EAP-TTLS and AAA/H servers may be one and the same, in which case it simply processes the information locally. The TTLS server may respond with its own sequence of AVPs. The TTLS server passes the AVP sequence to the TLS record layer for encryption and sends the resulting data to the client. For example, the TTLS server may send key distribution information, or it may forward an authentication challenge received from the AAA/H. This process continues until the TTLS server has enough information to issue either an EAP-Success or EAP-Failure. Thus, if the AAA/H rejects the client based on forwarded authentication information, the TTLS server would issue an EAP-Failure. If the AAA/H accepts the client, the TTLS server would issue an EAP-Success. The TTLS server distributes data connection keying information and other authorization information to the access point in the same AAA carrier protocol message that carries the EAP-Success. 6.3 Piggybacking While it is convenient to describe EAP-TTLS messaging in terms of two phases, it is sometimes required that a single EAP-TTLS packet to contain both phase 1 and phase 2 TLS messages. Such "piggybacking" occurs when the party that completes the handshake also has AVPs to send. For example, when negotiating a resumed TLS session, the TTLS server sends its ChangeCipherSpec and Finished messages first, then the client sends its own ChangeCipherSpec and Finished messages to conclude the handshake. If the client has authentication or other AVPs to send to the TTLS server, it must tunnel those AVPs within the same EAP-TTLS packet Paul Funk expires January 2005 [Page 14] Internet-Draft April 2004 immediately following its Finished message. If the client fails to do this, the TTLS server will incorrectly assume that the client has no AVPs to send, and the outcome of the negotiation could be affected. 6.4 Session Resumption When a client and TTLS server that have previously negotiated a EAP- TTLS session begin a new EAP-TTLS negotiation, the client and TTLS server may agree to resume the previous session. This significantly reduces the time required to establish the new session. This could occur when the client connects to a new access point, or when an access point requires reauthentication of a connected client. Session resumption is accomplished using the standard TLS mechanism. The client signals its desire to resume a session by including the session ID of the session it wishes to resume in the ClientHello message; the TTLS server signals its willingness to resume that session by echoing that session ID in its ServerHello message. If the TTLS server elects not to resume the session, it simply does not echo the session ID and a new session will be negotiated. This could occur if the TTLS server is configured not to resume sessions, if it has not retained the requested session's state, or if the session is considered stale. A TTLS server may consider the session stale based on its own configuration, or based on session-limiting information received from the AAA/H (e.g., the RADIUS Session- Timeout attribute). Tunneled authentication is specifically not performed for resumed sessions; the presumption is that the knowledge of the master secret as evidenced by the ability to resume the session is authentication enough. This allows session resumption to occur without any messaging between the TTLS server and the AAA/H. If periodic reauthentication to the AAA/H is desired, the AAA/H must indicate this to the TTLS server when the original session is established, for example, using the RADIUS Session-Timeout attribute. The client must, however, send other required AVPs, in particular key distribution AVPs, that are not associated with tunneled authentication in its first EAP-TTLS packet to the server that is capable of containing phase 2 TLS messages. The TTLS server does not retain client AVPs or key distribution preferences as part of session state, and the client is expected to resend those AVPs in each negotiation. Thus phase 2 of a resumed session proceeds just as would a new session, minus tunneled authentication AVPs. For example, the client would send its key distribution preferences, and the TTLS server would respond with its key distribution selection. Paul Funk expires January 2005 [Page 15] Internet-Draft April 2004 While the TTLS server does not retain client AVPs from session to session, it must retain authorization information returned by the AAA/H for use in resumed sessions. A resumed session must operate under the same authorizations as the original session, and the TTLS server must be prepared to send the appropriate information back to the access point. Authorization information might include the maximum time for the session, the maximum allowed bandwidth, packet filter information and the like. The TTLS server is responsible for modifying time values, such as Session-Timeout, appropriately for each resumed session. A TTLS server must not permit a session to be resumed if that session did not result in a successful authentication of the user during phase 2. The consequence of incorrectly implementing this aspect of session resumption would be catastrophic; any attacker could easily gain network access by first initiating a session that succeeds in the TLS handshake but fails during phase 2 authentication, and then resuming that session. [Implementation note: Toolkits that implement TLS often cache resumable TLS sessions automatically. Implementers must take care to override such automatic behavior, and prevent sessions from being cached for possible resumption until the user has been positively authenticated during phase 2.] 6.4.1 TTLS Server Guidelines for Session Resumption When a domain comprises multiple TTLS servers, a client's attempt to resume a session may fail because each EAP-TTLS negotiation may be routed to a different TTLS server. One strategy to ensure that subsequent EAP-TTLS negotiations are routed to the original TTLS server is for each TTLS server to encode its own identifying information, for example, IP address, in the session IDs that it generates. This would allow any TTLS server receiving a session resumption request to forward the request to the TTLS server that established the original session. 7. Generating Keying Material When record layer security is instantiated at the end of a TLS handshake, a pseudo-random function (PRF) is used to expand the negotiated master secret, server random value and client random value into a sequence of octets that is used as keying material for the record layer. The length of this sequence depends on the negotiated cipher suite, and contains the following components: Paul Funk expires January 2005 [Page 16] Internet-Draft April 2004 client_write_MAC_secret server_write_MAC_secret client_write_key server_write_key client_write_IV (optional) server_write_IV (optional) The ASCII-encoded constant string "key expansion" is used as input to the pseudo-random function to generate this sequence. EAP-TTLS leverages this technique to create keying material for use in the data connection between client and access point. Exactly the same PRF is used to generate as much keying material as required, with the constant string set to "ttls keying material", as follows: EAP-TTLS_keying_material = PRF(SecurityParameters.master_secret, "ttls keying material", SecurityParameters.client_random + SecurityParameters.server_random); The master secret, client random and server random used to generate the data connection keying material must be those established during the TLS handshake. Both client and TTLS server generate this keying material, and they are guaranteed to be the same if the handshake succeeded. The TTLS server distributes this keying material to the access point via the AAA carrier protocol. [Note that the order of client_random and server_random for EAP-TTLS is reversed from that of the TLS protocol [3]. This ordering follows the key derivation method of EAP-TLS [1]. Altering the order of randoms avoids namespace collisions between constant strings defined for EAP-TTLS and those defined for the TLS protocol.] 8. EAP-TTLS Encoding EAP-TTLS is a protocol within EAP. Its assigned EAP number is 21. Except as described in the subsections below, EAP-TTLS's encoding of TLS messages within EAP is identical to EAP-TLS's encoding of TLS messages within EAP. See [1] for details. 8.1 EAP-TTLS Start Packet The EAP-TTLS Start packet (with S-bit set) may, in a future specification, be allowed to contain data (the EAP-TLS Start packet does not). Thus, the data contents of an EAP-TTLS Start packet are reserved for future standardization; in the meantime, servers must not include any data in an EAP-TTLS Start packet, and clients must ignore such data but must not reject a Start packet that contains data. Paul Funk expires January 2005 [Page 17] Internet-Draft April 2004 8.2 EAP-TTLS Packets with No Data One point of clarification has to do with an EAP-TTLS packet (other than a Start packet) that contains no data. EAP-TLS defines the use of such a packet as a fragment ACK. When either party must fragment an EAP-TLS packet, the other party responds with a fragment ACK to allow the original party to send the next fragment. EAP-TTLS uses the fragment ACK in the same way. There are also other instances where a EAP-TTLS packet with no data might be sent: - When the final EAP packet of the EAP-TTLS negotiation is sent by the TTLS server, the client must respond with a EAP-TTLS packet with no data, to allow the TTLS server to issue its final EAP- Success or EAP-Failure packet. - It is possible for a EAP-TTLS packet with no data to be sent in the middle of a negotiation. Such a packet is simply interpreted as packet with no AVPs. For example, during session resumption, the client sends its Finished message first, then the TTLS server replies with its Finished message. The TTLS server cannot piggyback key distribution AVPs within the Record Layer in the same EAP-TTLS packet containing its Finished message, because it must wait for the client to indicate its key distribution preferences. But it is possible that the client has no preferences, and thus has no AVPs to send. The client simply sends a EAP-TTLS packet with no data, to allow the server to continue the negotiation by sending its key distribution selection. 9. Encapsulation of AVPs within the TLS Record Layer Subsequent to the TLS handshake, information is tunneled between client and TTLS server through the use of attribute-value pairs (AVPs) encrypted within the TLS record layer. The AVP format chosen for EAP-TTLS is compatible with the Diameter AVP format. This does not at all represent a requirement that Diameter be supported by any of the devices or servers participating in a EAP-TTLS negotiation. Use of this format is merely a convenience. Diameter is a superset of RADIUS and includes the RADIUS attribute namespace by definition, though it does not limit the size of an AVP as does RADIUS; RADIUS, in turn, is a widely deployed AAA protocol and attribute definitions exist for all commonly used password authentication protocols, including EAP. Thus, Diameter is not considered normative except as specified in this document. Specifically, the AVP Codes used in EAP-TTLS are semantically equivalent to those defined for Diameter, and, by Paul Funk expires January 2005 [Page 18] Internet-Draft April 2004 extension, RADIUS. Also, the representation of the Data field of an AVP in EAP-TTLS is identical to that of Diameter. Use of the RADIUS/Diameter namespace allows a TTLS server to easily translate between AVPs it uses to communicate to clients and the protocol requirements of AAA servers that are widely deployed. Plus, it provides a well-understood mechanism to allow vendors to extend that namespace for their particular requirements. 9.1 AVP Format The format of an AVP is shown below. All items are in network, or big-endian, order; that is, they have most significant octet first. 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | AVP Code | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |V M r r r r r r| AVP Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Vendor-ID (opt) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Data ... +-+-+-+-+-+-+-+-+ AVP Code The AVP Code is four octets and, combined with the Vendor-ID field if present, identifies the attribute uniquely. The first 256 AVP numbers represent attributes defined in RADIUS. AVP numbers 256 and above are defined in Diameter. AVP Flags The AVP Flags field is one octet, and provides the receiver with information necessary to interpret the AVP. The 'V' (Vendor-Specific) bit indicates whether the optional Vendor-ID field is present. When set to 1, the Vendor-ID field is present and the AVP Code is interpreted according to the namespace defined by the vendor indicated in the Vendor-ID field. The 'M' (Mandatory) bit indicates whether support of the AVP is required. If this bit is set to 0, this indicates that the AVP may be safely ignored if the receiving party does not understand or support it. If set to 1, this indicates that the receiving party must fail the negotiation if it does not understand the AVP; for a TTLS server, this would imply returning EAP-Failure, for a client, this would imply abandoning the negotiation. Paul Funk expires January 2005 [Page 19] Internet-Draft April 2004 The 'r' (reserved) bits are unused and must be set to 0. AVP Length The AVP Length field is three octets, and indicates the length of this AVP including the AVP Code, AVP Length, AVP Flags, Vendor-ID (if present) and Data. Vendor-ID The Vendor-ID field is present if the 'V' bit is set in the AVP Flags field. It is four octets, and contains the vendor's IANA- assigned "SMI Network Management Private Enterprise Codes" [9] value. Vendors defining their own AVPs must maintain a consistent namespace for use of those AVPs within RADIUS, Diameter and EAP- TTLS. A Vendor-ID value of zero is equivalent to absence of the Vendor- ID field altogether. 9.2 AVP Sequences Data encapsulated within the TLS Record Layer must consist entirely of a sequence of zero or more AVPs. Each AVP must begin on a 4-octet boundary relative to the first AVP in the sequence. If an AVP is not a multiple of 4 octets, it must be padded with 0s to the next 4- octet boundary. Note that the AVP Length does not include the padding. 9.3 Guidelines for Maximum Compatibility with AAA Servers For maximum compatibility, the following guidelines for AVP usage are suggested: - Non-vendor-specific AVPs should be selected from the set of attributes defined for RADIUS; that is, attributes with codes less than 256. This provides compatibility with both RADIUS and Diameter. - Vendor-specific AVPs should be defined in terms of RADIUS. Vendor-specific RADIUS attributes translate to Diameter (and, hence, to EAP-TTLS) automatically; the reverse is not true. RADIUS vendor-specific attributes use RADIUS attribute 26 and include vendor ID, vendor-specific attribute code and length; see [6] for details. 10. Tunneled Authentication EAP-TTLS permits user authentication information to be tunneled within the TLS record layer between client and TTLS server, Paul Funk expires January 2005 [Page 20] Internet-Draft April 2004 guaranteeing the security of the authentication information against active and passive attack between the client and TTLS server. The TTLS server decrypts and forwards this information to the AAA/H over the AAA carrier protocol. Any type of password or other authentication may be tunneled. Also, multiple tunneled authentications may be performed. Normally, tunneled authentication is used when the client has not been issued a certificate and the TLS handshake provides only one-way authentication of the TTLS server to the client; however, in certain cases it may be desired to perform certificate authentication of the client during the TLS handshake as well as tunneled user authentication afterwards. 10.1 Implicit challenge Certain authentication protocols that use a challenge/response mechanism rely on challenge material that is not generated by the authentication server, and therefore require special handling. In CHAP, MS-CHAP and MS-CHAP-V2, for example, the NAS issues a challenge to the client, the client then hashes the challenge with the password and forwards the response to the NAS. The NAS then forwards both challenge and response to a AAA server. But because the AAA server did not itself generate the challenge, such protocols are susceptible to replay attack. If the client were able to create both challenge and response, anyone able to observe a CHAP or MS-CHAP exchange could pose as that user, even using EAP-TTLS. To make these protocols secure under EAP-TTLS, it is necessary to provide a mechanism to produce a challenge that the client cannot control or predict. This is accomplished using the same technique described above for generating data connection keying material. When a challenge-based authentication mechanism is used, both client and TTLS server use the pseudo-random function to generate as many octets as are required for the challenge, using the constant string "ttls challenge", based on the master secret and random values established during the handshake: EAP-TTLS_challenge = PRF(SecurityParameters.master_secret, "ttls challenge", SecurityParameters.client_random + SecurityParameters.server_random); 10.2 Tunneled Authentication Protocols This section describes the methods for tunneling specific authentication protocols within EAP-TTLS. Paul Funk expires January 2005 [Page 21] Internet-Draft April 2004 For the purpose of explication, it is assumed that the TTLS server and AAA/H use RADIUS as a AAA carrier protocol between them. However, this is not a requirement, and any AAA protocol capable of carrying the required information may be used. 10.2.1 EAP When EAP is the tunneled authentication protocol, each tunneled EAP packet between the client and TTLS server is encapsulated in an EAP- Message AVP, prior to tunneling via the TLS record layer. The client's first tunneled EAP packet within phase 2 will contain the EAP-Response/Identity. The client places the actual username in this packet; the privacy of the user's identity is now guaranteed by the TLS encryption. This username must be a Network Access Identifier (NAI) [7]; that is, it must be in the following format: username@realm The @realm portion is optional, and is used to allow the TTLS server to forward the EAP packet to the appropriate AAA/H. Note that the client has two opportunities to specify realms. The first, in the initial EAP-Response/Identity packet, indicates the realm of the TTLS server. The second, in the tunneled authentication, indicates the realm of the client's home network. Thus, the access point need only know how to route to the realm of the TTLS server; the TTLS server is assumed to know how to route to the client's home realm. This serial routing architecture is anticipated to be useful in roaming environments, allowing access points or AAA proxies behind access points to be configured only with a small number of realms. Upon receipt of the tunneled EAP-Response/Identity, the TTLS server forwards it to the AAA/H in a RADIUS Access-Request. The AAA/H may immediately respond with an Access-Reject, in which case the TTLS server completes the negotiation by sending an EAP- Failure to the access point. This could occur if the AAA/H does not recognize the user's identity, or if it does not support EAP. If the AAA/H does recognize the user's identity and does support EAP, it responds with an Access-Challenge containing an EAP-Request, with the Type and Type-Data fields set according to the EAP protocol with which the AAA/H wishes to authenticate the client; for example MD-Challenge, OTP or Generic Token Card. The EAP authentication between client and AAA/H proceeds normally, as described in [2], with the TTLS server acting as a passthrough device. Each EAP-Request sent by the AAA/H in an Access-Challenge is tunneled by the TTLS server to the client, and each EAP-Response Paul Funk expires January 2005 [Page 22] Internet-Draft April 2004 tunneled by the client is decrypted and forwarded by the TTLS server to the AAA/H in an Access-Request. This process continues until the AAA/H issues an Access-Accept or Access-Reject, at which point the TTLS server completes the negotiation by sending an EAP-Success or EAP-Failure to the access point using the AAA carrier protocol. 10.2.2 CHAP The CHAP algorithm is described in [5]; RADIUS attribute formats are described in [6]. Both client and TTLS server generate 17 octets of challenge material, using the constant string "ttls challenge" as described above. These octets are used as follows: CHAP-Challenge [16 octets] CHAP Identifier [1 octet] The client tunnels User-Name, CHAP-Challenge and CHAP-Password AVPs to the TTLS server. The CHAP-Challenge value is taken from the challenge material. The CHAP-Password consists of CHAP Identifier, taken from the challenge material; and CHAP response, computed according to the CHAP algorithm. Upon receipt of these AVPs from the client, the TTLS server must verify that the value of the CHAP-Challenge AVP and the value of the CHAP Identifier in the CHAP-Password AVP are equal to the values generated as challenge material. If either item does not match exactly, the TTLS server must reject the client. Otherwise, it forwards the AVPs to the AAA/H in an Access-Request. The AAA/H will respond with an Access-Accept or Access-Reject. The TTLS server will then issue an EAP-Success or EAP-Failure to the access point. 10.2.3 MS-CHAP The MS-CHAP algorithm is described in [10]; RADIUS attribute formats are described in [12]. Both client and TTLS server generate 9 octets of challenge material, using the constant string "ttls challenge" as described above. These octets are used as follows: MS-CHAP-Challenge [8 octets] Ident [1 octet] The client tunnels User-Name, MS-CHAP-Challenge and MS-CHAP-Response AVPs to the TTLS server. The MS-CHAP-Challenge value is taken from Paul Funk expires January 2005 [Page 23] Internet-Draft April 2004 the challenge material. The MS-CHAP-Response consists of Ident, taken from the challenge material; Flags, set according the client preferences; and LM-Response and NT-Response, computed according to the MS-CHAP algorithm. Upon receipt of these AVPs from the client, the TTLS server must verify that the value of the MS-CHAP-Challenge AVP and the value of the Ident in the client's MS-CHAP-Response AVP are equal to the values generated as challenge material. If either item does not match exactly, the TTLS server must reject the client. Otherwise, it forwards the AVPs to the AAA/H in an Access-Request. The AAA/H will respond with an Access-Accept or Access-Reject. The TTLS server will then issue an EAP-Success or EAP-Failure to the access point. 10.2.4 MS-CHAP-V2 The MS-CHAP-V2 algorithm is described in [11]; RADIUS attribute formats are described in [12]. Both client and TTLS server generate 17 octets of challenge material, using the constant string "ttls challenge" as described above. These octets are used as follows: MS-CHAP-Challenge [16 octets] Ident [1 octet] The client tunnels User-Name, MS-CHAP-Challenge and MS-CHAP2- Response AVPs to the TTLS server. The MS-CHAP-Challenge value is taken from the challenge material. The MS-CHAP2-Response consists of Ident, taken from the challenge material; Flags, set to 0; Peer- Challenge, set to a random value; and Response, computed according to the MS-CHAP-V2 algorithm. Upon receipt of these AVPs from the client, the TTLS server must verify that the value of the MS-CHAP-Challenge AVP and the value of the Ident in the client's MS-CHAP2-Response AVP are equal to the values generated as challenge material. If either item does not match exactly, the TTLS server must reject the client. Otherwise, it forwards the AVPs to the AAA/H in an Access-Request. If the authentication is successful, the AAA/H will respond with an Access-Accept containing the MS-CHAP2-Success attribute. This attribute contains a 42-octet string that authenticates the AAA/H to the client based on the Peer-Challenge. The TTLS server tunnels this AVP to the client. Note that the authentication is not yet complete; the client must still accept the authentication response of the AAA/H. Paul Funk expires January 2005 [Page 24] Internet-Draft April 2004 Upon receipt of the MS-CHAP2-Success AVP, the client is able to authenticate the AAA/H. If the authentication succeeds, the client sends an EAP-TTLS packet to the TTLS server containing no data. Upon receipt of the empty EAP-TTLS packet from the client, the TTLS server now issues an EAP-Success. If the authentication fails, the AAA/H will respond with an Access- Challenge containing the MS-CHAP2-Error attribute. This attribute contains a new Ident and a string with addition information such as error reason and whether a retry is allowed. If the error reason is an expired password and a retry is allowed, the client may proceed to change the user's password. If the error reason is not an expired password or if the client does not wish to change the user's password, it simply abandons the EAP-TTLS negotiation. If the client does wish to change the password, it tunnels MS-CHAP- NT-Enc-PW, MS-CHAP2-CPW, and MS-CHAP-Challenge AVPs to the TTLS server. The MS-CHAP2-CPW AVP is derived from from the new Ident and Challenge received in the MS-CHAP2-Error AVP. The MS-CHAP-Challenge AVP simply echoes the new Challenge. Upon receipt of these AVPs from the client, the TTLS server must verify that the value of the MS-CHAP-Challenge AVP and the value of the Ident in the client's MS-CHAP2-CPW AVP match the values it sent in the MS-CHAP2-Error AVP. If either item does not match exactly, the TTLS server must reject the client. Otherwise, it forwards the AVPs to the AAA/H in an Access-Request. If the authentication is successful, the AAA/H will respond with an Access-Accept containing the MS-CHAP2-Success attribute. At this point, the negotiation proceeds as described above; the TTLS server tunnels the MS-CHAP2-Success to the client, the client authenticates the AAA/H based on this AVP, it either abandons the negotation on failure or sends an EAP-TTLS packet to the TTLS server containing no data, the TTLS server issues an EAP-Success. Note that additional AVPs associated with MS-CHAP-V2 may be sent by the AAA/H; for example, MS-CHAP-Domain. The TTLS server must tunnel such authentication-related attributes along with the MS-CHAP2- Success. 10.2.5 PAP The client tunnels User-Name and User-Password AVPs to the TTLS server. Normally, in RADIUS, User-Password is padded with nulls to a multiple of 16 octets, then encrypted using a shared secret and other packet information. Paul Funk expires January 2005 [Page 25] Internet-Draft April 2004 An EAP-TTLS client, however, does not RADIUS-encrypt the password since no such RADIUS variables are available; this is not a security weakness since the password will be encrypted via TLS anyway. The client should, however, null-pad the password to a multiple of 16 octets, to obfuscate its length. Upon receipt of these AVPs from the client, the TTLS server forwards them to the AAA/H in a RADIUS Access-Request. (Note that in the Access-Request, the TTLS server must encrypt the User-Password attribute using the shared secret between the TTLS server and AAA/H.) The AAA/H may immediately respond with an Access-Accept or Access- Reject. The TTLS server then completes the negotiation by sending an EAP-Success or EAP-Failure to the access point using the AAA carrier protocol. The AAA/H may also respond with an Access-Challenge. The TTLS server then tunnels the AVPs from the AAA/H's challenge to the client. Upon receipt of these AVPs, the client tunnels User-Name and User- Password again, with User-Password containing new information in response to the challenge. This process continues until the AAA/H issues an Access-Accept or Access-Reject. At least one of the AVPs tunneled to the client upon challenge must be Reply-Message. Normally this is sent by the AAA/H as part of the challenge. However, if the AAA/H has not sent a Reply-Message, the TTLS server must issue one, with null value. This allows the client to determine that a challenge response is required. Note that if the AAA/H includes a Reply-Message as part of an Access-Accept or Access-Reject, the TTLS server does not tunnel this AVP to the client. Rather, this AVP and all other AVPs sent by the AAA/H as part of Access-Accept or Access-Reject are sent to the access point via the AAA carrier protocol. 10.3 Performing Multiple Authentications In some cases, it is desirable to perform multiple user authentications. For example, a AAA/H may want first to authenticate the user by password, then by token card. The AAA/H may perform any number of additional user authentications using EAP, simply by issuing a EAP-Request with a new protocol type once the previous authentication succeeded but prior to issuing an EAP-Success or accepting the user via the AAA carrier protocol. For example, an AAA/H wishing to perform MD5-Challenge followed by Generic Token Card would first issue an EAP-Request/MD5-Challenge and receive a response. If the response is satisfactory, it would Paul Funk expires January 2005 [Page 26] Internet-Draft April 2004 then issue EAP-Request/Generic Token Card and receive a response. If that response were also satisfactory, it would issue EAP-Success. 11. EAP-TTLS Version 1 Version 1 of EAP-TTLS improves upon the original version 0 protocol in several ways. - Session keys developed from inner authentications are mixed with the master secret developed during the initial TLS handshake. This eliminates the Man-in-the-Middle (MitM) attack against tunneled protocols for inner authentications that generate session keys. See [15] and [16] for information about this attack. - A secure final exchange of the result of inner authentication is exchanged between client and server to conclude the EAP-TTLS exchange. This precludes any possibility of truncation attack that could occur when the client relies solely on an unprotected EAP-Success message to determine that the server has completed its authentication. - Inner authentication occurs within the TLS handshake, rather than after it. Thus, the TLS handshake itself includes both a standard TLS authentication as well as tunneled inner authentication(s) using EAP or legacy protocols, as well as any other tunneled communications required between client and server. 11.1 EAP-TTLS v1 Introduction Version 1 of EAP-TTLS utilizes the TLS extensions mechanism to extend the TLS handshake to include exchange of inner AVPs prior to completion of the TLS handshake by exchange of Finished messages. The TLS protocol provides a handshake phase and a data phase. EAP- TTLS v0, as well as other proposed tunneled EAP types such as EAP- PEAP and EAP-FAST, share a common strategy of utilizing the handshake phase to establish a tunnel and the data phase to perform protected authentication. In EAP-TTLS v1, the AVP exchange is folded into the TLS handshake itself; in other words, the inner authentication precedes the conclusion of the TLS handshake, rather following it. An advantage of this arrangement is a certain amount of cryptographic integration of inner authentication with standard TLS mechanisms. For example, mixing of inner session keys to thwart MitM attacks is easily performed in such a way that both the authentication result and the final session key is conditioned upon these inner session keys. Paul Funk expires January 2005 [Page 27] Internet-Draft April 2004 The definition of EAP-TTLS v1 proceeds by first defining the InnerApplication extension to TLS, and then by defining the binding of the extended TLS to EAP via EAP-TTLS v1, which in effect serves as a carrier protocol. 11.2 Intentions Beyond EAP-TTLS The use of TLS for EAP is a relative newcomer. TLS has long used for many other purposes, most notably for protecting HTTP traffic. However, TLS used in these contexts has no mechanism for authentication beyond the certificate mechanisms that have been defined. Any additional authentication, say in HTTP, must use relatively primitive mechanisms defined in the HTTP protocol. It would be very useful for the TLS protocol to provide more general authentication mechanisms for subsequent authentication, for example EAP. The InnerApplication extension allows TLS to provide inner authentication during the handshake, rather than after it. The EAP- TTLS version 1 protocol is in fact just a binding of this extended TLS to EAP; that it, EAP-TTLS is a carrier protocol for the extended TLS. TLS with the InnerApplication extension can just as easily be bound to TCP, to enable its use in HTTP. The applicability of TLS with the InnerApplication extension includes setting up HTTP connections (including SSL VPN connections), establishing IPsec connections as an alternative to IKE, obtaining credentials for single sign-on, providing for client integrity verification, etc. The inner AVP mechanism offers both legacy and EAP authentication capabilities, natural compatibility with RADIUS and Diameter servers, and the flexibility to allow arbitrary client-server exchanges for various purposes. The authors' intention is to separately propose the TLS InnerApplication extension as an enhancement to TLS, and then define EAP-TTLS version 1 as a carrier protocol, or binding, of that extended TLS to EAP. For reasons of timing, the TLS InnerApplication extension is defined in this draft for now. 11.3 The InnerApplication Extension to TLS The InnerApplication extension to TLS follows the guidelines of RFC 3546. The client proposes use of this extension by including an InnerApplication message in its ClientHello handshake message, and the server confirms its use by including an InnerApplication message in its ServerHello handshake message. In this document, the term "TLS/IA" shall refer to TLS with the InnerApplication extension. Two new handshake messages are defined for use in TLS/IA: Paul Funk expires January 2005 [Page 28] Internet-Draft April 2004 - The PhaseFinished message. This message is similar to the standard TLS Finished message; it allows the TLS/IA handshake to operate in phases, with message and key confirmation occurring at the end of each phase. - The ApplicationPayload message. This message is used to carry AVP (Attribute-Value Pair) sequences within the TLS/IA handshake, in support of client-server applications such as authentication. A new alert code is also defined for use in TLS/IA: - The InnerApplicationFailure alert. This error alert allows either party to terminate the handshake due to a failure in an application implemented via AVP sequences carried in ApplicationPayload messages. 11.3.1 TLS/IA Overview In TLS/IA, the handshake is divided into phases. The first phase is called the "initial phase", and consists of a standard TLS handshake with PhaseFinished substituted for Finished as the concluding message. There are one or more subsequent phases, called "application phases". The last application phase is called the "final phase"; any application phase prior to the final phase is called an "intermediate phase". Each application phase consists of ApplicationPayload messages exchanged by client and server to implement applications such as authentication, plus concluding messages for cryptographic confirmation. Thus, the entire handshake consists of a initial phase, zero or more intermediate phases, and a final phase. Intermediate phases are only necessary if interim confirmation of key material generated during an application phase is desired. In each application phase, the client sends the first ApplicationPayload message. ApplicationPayload messages are then traded one at a time between client and server, until the server concludes the phase by sending a ChangeCipherSpec and PhaseFinished sequence to conclude an intermediate phase, or a ChangeCipherSpec and Finished sequence to conclude the final phase. The client then responds with its own ChangeCipherSpec and PhaseFinished sequence, or ChangeCipherSpec and Finished sequence. The server determines which type of concluding message is used, either PhaseFinished or Finished, and the client MUST echo the same type of concluding message. Each PhaseFinished or Finished message Paul Funk expires January 2005 [Page 29] Internet-Draft April 2004 provides cryptographic confirmation of the integrity of all handshake messages and keys generated from the start of the handshake through the current phase. Each ApplicationPayload message contains opaque data interpreted as an AVP (Attribute-Value Pair) sequence. Each AVP in the sequence contains a typed data element. The exchanged AVPs allow client and server to implement "applications" within a secure tunnel. An application may be any procedure that someone may usefully define. A typical application might be authentication; for example, the server may authenticate the client based on password credentials using EAP. Other possible applications include distribution of keys, validating client integrity, setting up IPsec parameters, setting up SSL VPNs, and so on. In TLS/IA, the TLS master secret undergoes multiple permutations until a final master secret is computed at the end of the entire handshake. Each phase of the handshake results in a new master secret; the master secret for each phase is confirmed by the PhaseFinished or Finished message exchange that concludes that phase. The initial master secret is computed during the initial phase of the handshake, using the usual TLS algorithm, namely, that a premaster secret is established and the TLS PRF function is used to compute the initial master secret. This initial master secret is confirmed via the first exchange of ChangeCipherSpec and PhaseFinished messages. Each subsequent master secret for an application phase is computed using a PRF based on the current master secret, then mixing into the result any session key material generated during authentications during that phase. Each party computes a new master secret prior to the conclusion of each application phase, and uses that new master secret is to compute fresh keying material (that is, a TLS "key_block", consisting of client and server MAC secrets, write keys and IVs). The new master secret and keying material become part of the pending read and write connection states. Following standard TLS procedures, these connection states become current states upon sending or receiving ChangeCipherSpec, and are confirmed via the PhaseFinished or Finished message. The final master secret, computed during the final handshake phase and confirmed by an exchange of ChangeCipherSpec and Finished messages, becomes the actual TLS master secret that defines the session. This final master secret is the surviving master secret, and each prior master secrets SHOULD be discarded when a new connection state is instantiated. The final master secret is used for session resumption, as well as for any session key derivation that protocols defined over TLS may require. Paul Funk expires January 2005 [Page 30] Internet-Draft April 2004 11.3.2 Message Exchange Each intermediate handshake phase consists of ApplicationPayload messages sent alternately by client and server, and a concluding exchange of {ChangeCipherSpec, PhaseFinished} messages. The first ApplicationPayload message in the each intermediate phase is sent by the client; the first {ChangeCipherSpec, PhaseFinished} message sequence is sent by the server. Thus the client begins the exchange with an ApplicationPayload message and the server determines when to conclude it by sending {ChangeCipherSpec, PhaseFinished}. When it receives the server's {ChangeCipherSpec, PhaseFinished} messages, the client sends its own {ChangeCipherSpec, PhaseFinished} messages. The client then sends an ApplicationPayload message to begin the next handshake phase. The final handshake proceeds in the same manner as the intermediate handshake, except that the Finished message is used rather than the PhaseFinished message, and the client does not send an ApplicationPayload message for the next phase because there is no next phase. At the start of each application handshake phase, the server MUST wait for the client's opening ApplicationPayload message before it sends its own ApplicationPayload message to the client. The client MAY NOT initiate conclusion of an application handshake phase by sending the first {ChangeCipherSpec, PhaseFinished} or {ChangeCipherSpec, Finished message} sequence; it MUST allow the server to initiate the conclusion of the phase. 11.3.3 Master Key Permutation Each permutation of the master secret from one phase to the next begins with the calculation of a preliminary 48 octet vector based on the current master secret: preliminary_vector = PRF(master_secret, "InnerApplication preliminary vector", server_random + client_random) [0..48]; Session key material generated by applications during the current application phase are mixed into the preliminary vector by arithmetically adding each session key to the preliminary vector to compute the new master secret. The preliminary vector is treated as a 48-octet integer in big-endian order; that is, the first octet is of the highest significance. Each session key is also treated as a big-endian integer of whatever size it happens to be. Arithmetic carry past the most significant octet is discarded; that is, the addition is performed modulo 2 ^ 384. Paul Funk expires January 2005 [Page 31] Internet-Draft April 2004 Thus, the logical procedure for computing the next master secret (which may also be a convenient implementation procedure) is as follows: 1 At the start of each application handshake phase, use the current master secret to compute the preliminary vector for the next master secret. 2 Each time session key material is generated from an authentication or other exchange, arithmetically add that session key material to the preliminary vector. 3 At the conclusion of the application handshake phase, copy the current contents of the preliminary vector (which now includes addition of all session key material) into the new master secret, prior to computing verify_data. The purpose of using a PRF to compute a preliminary vector is to ensure that, even in the absence of session keys, the master secret is cryptographically distinct in each phase of the handshake. The purpose of adding session keys into the preliminary vector is to ensure that the same client entity that negotiated the original master secret also negotiated the inner authentication(s). In the absence of such mixing of keys generated from the standard TLS handshake with keys generated from inner authentication, it is possible for a hostile agent to mount a man-in-the-middle attack, acting as server to an unsuspecting client to induce it to perform an authentication with it, which it can then pass through the TLS tunnel to allow it to pose as that client. An application phase may include no authentications that produce a session key, may include one such authentication, or may include several. Arithmetic addition was chosen as the mixing method because it is commutative, that is, it does not depend on the order of operations. This allows multiple authentications to proceed concurrently if desired, without having to synchronize the order of master secret updates between client and server. Addition was chosen rather than XOR in order to avoid what is probably a highly unlikely problem; namely, that two separate authentications produce the same session key, which, if XORed, would mutually cancel. This might occur, for example, if two instances of an authentication method were to be applied against different forms of a user identity that turn out in a some cases to devolve to the same identity. Finally, it was decided that a more complex mixing mechanism for session key material, such as hashing, besides not being commutative, would not provide any additional security, due to the Paul Funk expires January 2005 [Page 32] Internet-Draft April 2004 effectively random character of the preliminary vector and the powerful PRF function which is applied to create derivative keys. 11.3.4 Session Resumption A TLS/IA handshake may be resumed using standard mechanisms defined in RFC 2246. In TLS/IA, session resumption is simply an alternative form of the initial handshake phase, after which subsequent application phases proceed. When the initial handshake phase is resumed, client and server may not deem it necessary to perform the same type of AVP exchange that they might after a full handshake. In fact, the resumption itself might provide all the security needed and no AVPs need be exchanged at all. If the client determines that it has no need for AVP negotiation, it sends an ApplicationPayload message with no data as its first application phase message. If the server concurs, it may conclude the handshake with ChangeCipherSpec and Finished immediately upon receiving the empty ApplicationPayload message. Alternatively, either party may initiate AVP exchange if inner applications must execute upon session resumption. For example, authentication exchanges might be omitted but key distribution for some purpose might still occur. [Author's note: A future draft may provide a mechanism to avoid the extra round trip incurred when neither party has a requirement to send AVPs after session resumption.] 11.3.5 Error Termination The TLS/IA handshake may be terminated by either party sending a fatal alert, following standard TLS procedures. 11.3.6 Application Session Key Material Many authentication mechanisms generate session keying material as a by-product of authentication. Such keying material is normally intended for use in a subsequent data connection for encryption and validation. For example, EAP-TLS, MS-CHAP-V2 and its alter ego EAP- MS-CHAP-V2 each generate keying material. When encapsulated within TLS/IA, such keying material MUST NOT be used to set up data connections; the TLS/IA master secret is a better basis for this use. However, such keying material generated during an application phase MUST be used to permute the TLS/IA master secret between on phase and the next. The purpose of this is to preclude man-in-the-middle Paul Funk expires January 2005 [Page 33] Internet-Draft April 2004 attacks, in which an unsuspecting client is induced to perform an authentication outside a tunnel with an attacker posing as a server; the attacker can then introduce the authentication protocol into a tunnel such as provided by TLS/IA, fooling an authentic server into believing that the attacker is the authentic user. By mixing keying material generating during application phase authentication into the master secret, such attacks are thwarted, since only a single client identity could both authenticate successfully and have derived the session keying material. Note that the keying material generated during authentication must be cryptographically related to the authentication and not derivable from data exchanged during authentication in order for the keying material to be useful in thwarting such attacks. The RECOMMENDED amount of keying material to mix into the master secret is 32 octets. Up to 48 octets MAY be used. Each authentication protocol may define how the keying material it generates is mapped to an octet sequence of some length for the purpose of TLS/IA mixing. However, for protocols which do not specify this (including the multitude of protocols that pre-date TLS/IA) the following rules are defined. The first rule that applies SHALL be the method for determining keying material: - If the authentication protocol maps its keying material to the RADIUS attributes MS-MPPE-Receive-Key and MS-MPPE-Send-Key, then the keying material for those attributes are concatenated (with MS-MPPE-Receive-Key first), the concatenated sequence is truncated to 32 octets if longer, and the result is used as keying material. (Note that this rule applies to MS-CHAP-V2 and EAP-MS-CHAP-V2.) - If the authentication protocol uses a pseudo-random function to generate keying material, that function is used to generate 32 octets for use as keying material. 11.3.7 Computing Verification Data In standard TLS, the "verify_data" vector of the Finished message is computed as follows: PRF(master_secret, finished_label, MD5(handshake_messages) + SHA-1(handshake_messages)) [0..11]; This allows both parties to confirm the master secret as well as the integrity of all handshake messages that have been exchanged. Paul Funk expires January 2005 [Page 34] Internet-Draft April 2004 In TLS/IA, verify_data for the initial handshake phase is computed in exactly the same manner, though verify_data is encapsulated in a PhaseFinished, rather than Finished, message. In the subsequent application phases, a slight variation to this formula is used. For each hash, the handshake messages of the current phase are appended to the hash of the handshake messages of the previous phase. Thus, for each application phase, the MD5 hash input to the PRF is a hash of the MD5 hash computed for the previous phase concatenated with the handshake messages of the current phase; the SHA-1 hash is computed in the same way, but using the SHA-1 hash computed for the previous phase. Also, the master secret used in the PRF computation in each application phase is the new master secret generated at the conclusion of that phase. For clarity, this is best expressed in formal notation. Let phases be numbered from 0, where phase 0 is the initial phase. Let: Secret[n] be the master secret determined at the conclusion of phase n. Messages[n] be the handshake messages in phase n. MD5[n] be the MD5 hash of handshake message material in phase n. SHA-1[n] be the SHA-1 hash of handshake message material in phase n. PRF[n] be the verify_data generated via PRF in phase n. Hash computations for phase 0 are as follows: MD5[0] = MD5(Messages[0]) SHA-1[0] = SHA-1(Messages[0]) PRF[0] = PRF(master_secret, finished_label, MD5[0] + SHA-1[0]) [0..11] Hash computations for phase i, where i > 0 (i.e. application phases) are as follows: MD5[i] = MD5(MD5[i-1] + Messages[i]) SHA-1[i] = SHA-1(SHA-1[i-1] + Messages[i]) Paul Funk expires January 2005 [Page 35] Internet-Draft April 2004 The PRF computation to generate verify_data for any phase i (including i = 0) is as follows: PRF[i] = PRF(Secret[i], finished_label, MD5[i] + SHA-1[i]) [0..11] Note that for phase 0, the PRF computation is identical to the standard TLS computation. Variations to the algorithm occur only in application phases, in the use of new master secrets and the inclusion of hashes of previous handshake messages as input to the hashing algorithms. Note that the only handshake messages that appear in an application phase are InnerApplication messages and Finished or Phase Finished messages. During an application phase, the handshake messages input to the hashing algorithm by the server will include all InnerApplication messages exchanged during that phase; the handshake messages input to the hashing algorithm by the client will include all InnerApplication messages exchanged during that phase plus the server's PhaseFinished or Finished message. 11.3.8 Attribute-Value Pairs (AVPs) AVPs used in InnerApplication messages are exactly as defined in Section 9 of this document; that is, they are Diameter-style AVPs and use the RADIUS-Diameter namespace. Rules for performing authentications using these AVPs are exactly as defined in Section 10 of this document. This includes rules for creating implicit challenges, and rules for use of inner EAP authentications as well as legacy protocols such as PAP, CHAP and MS-CHAP-V1/V2. Note that all implicit challenges are based on the then-current master secret. 11.3.9 TLS/IA Messages All specifications of TLS/IA messages follow the usage defined in RFC 2246. TLS/IA defines a new TLS extension - "InnerApplication"; two new handshake messages - "PhaseFinished" and "ApplicationPayload"; and a new alert code - "InnerApplicationFailure". The InnerApplication extension type is 9347 (hex). In order to avoid potential type-assignment problems, the new handshake message types and alert code are dynamically defined within the InnerApplication extension message. Client and server independently specify the values they will send. Thus, the client assigns its own message type and alert code values for use in its own transmissions, and includes these values in its InnerApplication Paul Funk expires January 2005 [Page 36] Internet-Draft April 2004 message within ClientHello. Similarly, the server assigns its own message type and alert code values for use in its own transmissions, and includes these values in its InnerApplication message within ServerHello. Each party must note the message type and alert code values assigned by the other party and interpret messages from the other party accordingly. Both client and server assign message types and alert code so as not to conflict with values that that it might otherwise send. There is no requirement that client and server assign identical values for these items. 11.3.10 The InnerApplication Extension Use of the InnerApplication extension follows RFC 3546. The client proposes use of this extension by including the InnerApplication extension in the client_hello_extension_list vector of the extended ClientHello. If the extension is included in the ClientHello, the server MAY accept the proposal by including the InnerApplication extension in the server_hello_extension_list of the extended ServerHello. If use of this extension is either not proposed by the client or not confirmed by the server, the variations to the TLS handshake described here MUST NOT be used. The "extension_data" field of the Extension structure for the InnerApplication extension SHALL contain "InnerApplication" where: struct { uint8 PhaseFinishedType; uint8 ApplicationPayloadType; uint8 InnerApplicationFailureAlertCode; } InnerApplication; 11.3.11 The PhaseFinished Handshake Message The PhaseFinished message concludes the initial handshake phase and each intermediate handshake phase. It MUST be immediately preceded by a ChangeCipherSpec message. It is defined as follows: struct { opaque verify_data[12]; } PhaseFinished; 11.3.12 The ApplicationPayload Handshake Message The ApplicationPayload message carries an AVP sequence during an application handshake phase. It is defined as follows: struct { opaque avps[Handshake.length]; } ApplicationPayload; Paul Funk expires January 2005 [Page 37] Internet-Draft April 2004 where Handshake.length is the 24-bit length field in the encapsulating Handshake message. Note that the "avps" element has its length defined in square bracket rather than angle bracket notation, implying a fixed rather than variable length vector. This avoids the having the length of the AVP sequence specified redundantly both in the encapsulating Handshake message and as a length prefix in the avps element itself. 11.3.13 The InnerApplicationFailure Alert An InnerApplicationFailure error alert may be sent by either party during an application phase. This indicates that the sending party considers the negotiation to have failed due to an application carried in the AVP sequences, for example, a failed authentication. The AlertLevel for an InnerApplicationFailure alert MUST be set to "fatal". Note that other alerts are possible during an application phase; for example, decrypt_error. The InnerApplicationFailure alert relates specifically to the failure of an application implemented via AVP sequences; for example, failure of an EAP or other authentication method, or information passed within the AVP sequence that is found unsatisfactory. 11.4 Binding of TLS/IA to EAP-TTLS v1 EAP-TTLS v1 encapsulates a TLS handshake with the InnerApplication extension (TLS/IA). EAP-TTLS v1 acts as a carrier protocol for TLS/IA, and uses cryptographic information developed during the TLS/IA exchange to create session keys for encrypting subsequent data transmission between client and access point. The format for encapsulated TLS/IA messages in EAP-TTLS v1 is identical to the formats described for EAP-TTLS v0 in Section 8, unless otherwise specified 11.4.1 Flags Octet Use of version 1 of EAP-TTLS is negotiated through a new 3-bit "Version" field in the Flags octet of the EAP-TTLS request/response header. The Flags octet is the first octet of each EAP-TTLS message, following immediately after the EAP type. The Version field uses bits of the Flags octet that were formerly reserved and required to be 0. The new bit field definitions for the Flags octet are as follows: Paul Funk expires January 2005 [Page 38] Internet-Draft April 2004 0 1 2 3 4 5 6 7 +---+---+---+---+---+---+---+---+ | L M S R R | Version | +---+---+---+---+---+---+---+---+ where: L = Length included M = More fragments S = Start R = Reserved Version = EAP-TTLS version number For EAP-TTLS v1, Version is set to 1; that is, the bit sequence 001. Interpretation of L, M and S are as in EAP-TTLS v0. 11.4.2 Version Negotiation The version of EAP-TTLS is negotiated in the first exchange between server and client. The server sets the highest version number of EAP-TTLS that it supports in the Version field of its Start message (in the case of EAP-TTLS v1, this is 1). In its first EAP message in response, the client sets the Version field to the highest version number that it supports that is no higher than the version number offered by the server. If the client version is not acceptable to the server, it sends an EAP-Failure to terminate the EAP session. Otherwise, the version sent by the client is the version of EAP-TTLS that MUST be used, and both server and client set the version field to that version number in all subsequent EAP messages. 11.4.3 Acknowledgement Packets An Acknowledgement packet is an EAP-TTLS v1 packet with no additional data beyond the Flags octet, and with the L, M and S bits of the Flags octet set to 0. (Note, however, that the Version field MUST still be set to the appropriate version number.) An Acknowledgement packet is sent for the following purposes: - Fragment acknowledgement - Error alert acknowledgement Note that in EAP-TTLS v0 there are other cases in which a packet with no data must be sent by the client for the simple reason that the client has no AVPs to send. This situation does not arise in Paul Funk expires January 2005 [Page 39] Internet-Draft April 2004 EAP-TTLS v1. If no AVPs are to be sent, there will nevertheless be an ApplicationPayload message containing no data, which the client must send. - Fragment Acknowledgement Each EAP-TTLS v1 message contains a sequence of TLS/IA messages that represent a single leg of a half-duplex conversation. The EAP carrier protocol (e.g., PPP, EAPOL, RADIUS) may impose constraints on the length of of an EAP message. Therefore it may be necessary to fragment an EAP-TTLS v1 message across multiple EAP messages. Each fragment except for the last MUST have the M bit set, to indicate that more data is to follow; the final fragment MUST NOT have the M bit set. The party that receives a message with the M bit set MUST respond with an Acknowledgement packet. - Error Alert Acknowledgement Either party may at any time send a TLS error alert to fail the TLS/IA handshake. If the client sends an error alert to the server, no further EAP- TTLS messages are exchanged, and the server sends an EAP-Failure to terminate the conversation. If the server sends an error alert to the client, the client MUST respond with an Acknowledgement packet to allow the conversation to continue. Upon receipt of the Acknowledgement packet, the server sends an EAP-Failure to terminate the conversation. 11.4.4 Generating Keying Material EAP-TTLS v1 uses the same mechanism as EAP-TTLS v0 to generate keying material (session keys) for use in the data connection between client and access point. Note that it is the final master secret of the TLS/IA exchange that is used to generate keying material for use in the subsequent data connection. 12. Discussion of Certificates and PKI Public-key cryptography, certificates, and the associated PKI are used in EAP-TTLS to authenticate the EAP-TTLS server to the client, and optionally the client to the EAP-TTLS server. Previous experience with the deployment of PKI in applications has shown that its implementation requires care. This section provides a brief discussion of the issues implementers will face when deploying PKI for EAP-TTLS. Paul Funk expires January 2005 [Page 40] Internet-Draft April 2004 The traditional use of TLS for securing e-commerce transactions over the Internet is perhaps the best-known deployment of PKI, and it serves to illustrate several of the issues relevant here. In the case of e-commerce: - The environment is many-to-many - many consumers do business with many merchants. Typically there is no relationship in advance between a consumer and a merchant. - Users are "notoriously bad" about following security guidelines. When presented with a dialogue saying "the name in the certificate is different from the name you requested", most users will simply continue with the transaction. - Support for revocation is limited. It is important to understand that the environments in which EAP-TTLS are likely to be deployed will typically be very different from e-commerce. In particular, many deployments will be comparable to deploying wireless LAN within an enterprise. In this case, the communications topology is essentially many-to-one or many-to-few - many employees talking to a few EAP-TTLS servers - and all clients are essentially governed by their employer rather than autonomous. This means: - It may be unnecessary to rely on a public CA. Instead the enterprise could choose to run its own CA (either insourced or outsourced). - The enterprise could choose to enforce stringent policies on certificate validation and processing - for example simply insisting connections are dropped if the correct name does not appear in the server certificate. Such policies could be enforced via extensions in the root certificate of the enterprise CA. However it also means: - EAP-TTLS servers may receive considerably less attention than the web servers of large e-commerce sites. As a result, compromise of EAP-TTLS servers may be more common, and therefore deployment and use of revocation solutions may be more relevant. One open question in the area of PKI on which the authors would like to promote discussion is the following: - Should EAP-TTLS enforce rules on name matching regarding the EAP- TTLS server? For example, EAP-TTLS could mandate that radius.xyz.realm or diameter.xyz.realm be used as the name in the EAP-TTLS server's certificate, and that the client must match this name with the realm it sent in the initial EAP- Paul Funk expires January 2005 [Page 41] Internet-Draft April 2004 Response/Identity. 13. Message Sequences [Author's note: The message sequences in these sections apply to version 0 of the EAP-TTLS protocol. Messages sequences for version 1 have not yet been completed.] This section presents EAP-TTLS message sequences for various negotiation scenarios. These examples do not attempt to exhaustively depict all possible scenarios. It is assumed that RADIUS is the AAA carrier protocol both between access point and TTLS server, and between TTLS server and AAA/H. EAP packets that are passed unmodified between client and TTLS server by the access point are indicated as "passthrough". AVPs that are securely tunneled within the TLS record layer are enclosed in curly braces ({}). Items that are optional are suffixed with question mark (?). Items that may appear multiple times are suffixed with plus sign (+). 13.1 Successful authentication via tunneled CHAP In this example, the client performs one-way TLS authentication of the TTLS server nad CHAP is used as a tunneled user authentication mechanism. client access point TTLS server AAA/H ------ ------------ ----------- ----- EAP-Request/Identity <-------------------- EAP-Response/Identity --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge: EAP-Request/TTLS-Start <-------------------- EAP-Request passthrough <-------------------- EAP-Response/TTLS: ClientHello --------------------> Paul Funk expires January 2005 [Page 42] Internet-Draft April 2004 RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge: EAP-Request/TTLS: ServerHello Certificate ServerKeyExchange ServerHelloDone <-------------------- EAP-Request passthrough <-------------------- EAP-Response/TTLS: ClientKeyExchange ChangeCipherSpec Finished --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge: EAP-Request/TTLS: ChangeCipherSpec Finished <-------------------- EAP-Request passthrough <-------------------- EAP-Response/TTLS: {User-Name} {CHAP-Challenge} {CHAP-Password} --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Request: User-Name CHAP-Challenge CHAP-Password --------------------> Paul Funk expires January 2005 [Page 43] Internet-Draft April 2004 RADIUS Access-Accept <-------------------- RADIUS Access-Accept: EAP-Success <-------------------- EAP-Success passthrough <-------------------- 13.2 Successful authentication via tunneled EAP/MD5-Challenge In this example, the client performs one-way TLS authentication of the TTLS server and EAP/MD5-Challenge is used as a tunneled user authentication mechanism. client access point TTLS server AAA/H ------ ------------ ----------- ----- EAP-Request/Identity <-------------------- EAP-Response/Identity --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge: EAP-Request/TTLS-Start <-------------------- EAP-Request passthrough <-------------------- EAP-Response/TTLS: ClientHello --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge: EAP-Request/TTLS: ServerHello Certificate ServerKeyExchange ServerHelloDone <-------------------- Paul Funk expires January 2005 [Page 44] Internet-Draft April 2004 EAP-Request passthrough <-------------------- EAP-Response/TTLS: ClientKeyExchange ChangeCipherSpec Finished --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge: EAP-Request/TTLS: ChangeCipherSpec Finished <-------------------- EAP-Request passthrough <-------------------- EAP-Response/TTLS: {EAP-Response/Identity} --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Request: EAP-Response/Identity --------------------> RADIUS Access-Challenge EAP-Request/ MD5-Challenge --------------------> RADIUS Access-Challenge: EAP-Request/TTLS: {EAP-Request/MD5-Challenge} <-------------------- EAP-Request passthrough <-------------------- EAP-Response/TTLS: {EAP-Response/MD5-Challenge} --------------------> Paul Funk expires January 2005 [Page 45] Internet-Draft April 2004 RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge EAP-Response/ MD5-Challenge --------------------> RADIUS Access-Accept <-------------------- RADIUS Access-Accept: EAP-Success <-------------------- EAP-Success passthrough <-------------------- 13.3 Successful session resumption In this example, the client and server resume a previous TLS session. The ID of the session to be resumed is sent as part of the ClientHello, and the server agrees to resume this session by sending the same session ID as part of ServerHello. client access point TTLS server AAA/H ------ ------------ ----------- ----- EAP-Request/Identity <-------------------- EAP-Response/Identity --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge: EAP-Request/TTLS-Start <-------------------- EAP-Request passthrough <-------------------- EAP-Response/TTLS: ClientHello --------------------> Paul Funk expires January 2005 [Page 46] Internet-Draft April 2004 RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Challenge: EAP-Request/TTLS: ServerHello ChangeCipherSpec Finished <-------------------- EAP-Request passthrough <-------------------- EAP-Response/TTLS: ChangeCipherSpec Finished --------------------> RADIUS Access-Request: EAP-Response passthrough --------------------> RADIUS Access-Accept: EAP-Success <-------------------- EAP-Success passthrough <-------------------- 14. Security Considerations This draft is entirely about security and the security considerations associated with the mechanisms employed in this document should be considered by implementers. The following additional issues are relevant: - Anonymity and privacy. Unlike other EAP methods, EAP-TTLS does not communicate a username in the clear in the initial EAP- Response/Identity. This feature is designed to support anonymity and location privacy from attackers eavesdropping the network path between the client and the TTLS server. However implementers should be aware that other factors - both within EAP-TTLS and elsewhere - may compromise a user's identity. For example, if a user authenticates with a certificate during phase 1 of EAP-TTLS, the subject name in the certificate may reveal the user's identity. Outside of EAP-TTLS, the client's fixed MAC address, or in the case of wireless connections, the client's radio signature, may also reveal information. Additionally, implementers should be aware that a user's identity is not hidden Paul Funk expires January 2005 [Page 47] Internet-Draft April 2004 from the EAP-TTLS server and may be included in the clear in AAA messages between the access point, the EAP-TTLS server, and the AAA/H server. - Trust in the EAP-TTLS server. EAP-TTLS is designed to allow the use of legacy authentication methods to be extended to mediums like wireless in which eavesdropping the link between the client and the access point is easy. However implementers should be aware of the possibility of attacks by rogue EAP-TTLS servers - for example in the event that the phase 2 authentication method within EAP-TTLS is susceptible to dictionary attacks. These threats can be mitigated through the use of authentication methods like one-time passwords which are not susceptible to dictionary attacks, or by ensuring that clients connect only to trusted EAP-TTLS servers. - EAP-TTLS server certificate compromise. The use of EAP-TTLS server certificates within EAP-TTLS makes EAP-TTLS susceptible to attack in the event that an EAP-TTLS server's certificate is compromised. EAP-TTLS servers should therefore take care to protect their private key. In addition, certificate revocation methods may be used to mitigate against the possibility of key compromise. [13] describes a way to integrate one such method - OCSP [14] - into the TLS handshake - use of this approach may be appropriate within EAP-TTLS. - Negotiation of link encryption. EAP-TTLS includes a method to negotiate data cipher suites. It also allows data cipher suites to be negotiated by other means - for example by having client and access point exchange their preferences using the link layer protocol. However the use of the EAP-TTLS negotiation is strongly recommended because it provides a secured negotiation. In contrast, simple unsecured preference exchange over the link layer is susceptible to a man-in-the-middle attack that forces the parties to use the weakest, rather than the strongest, mutually acceptable data cipher suite. The potential of this problem is well-illustrated by wireless LAN where for interoperability purposes many entities will have to continue to support WEP encryption for some time. In the event that the data link protocol already includes a negotiation exchange, it is recommended that the EAP-TTLS exchange still be used, with the link layer exchange simply confirming the data cipher suite selected using EAP-TTLS. - Listing of data cipher preferences. EAP-TTLS negotiates data cipher suites by having the EAP-TTLS server select the first cipher suite appearing on the client list that also appears on the access point list. In order to maximize security, it is therefore recommended that the client order its list according to security - most secure acceptable cipher suite first, least secure acceptable cipher suite last. Paul Funk expires January 2005 [Page 48] Internet-Draft April 2004 - Forward secrecy. With forward secrecy, revelation of a secret does not compromise session keys previously negotiated based on that secret. Thus, when the TLS key exchange algorithm provides forward secrecy, if a TTLS server certificate's private key is eventually stolen or cracked, tunneled user password information will remain secure as long as that certificate is no longer in use. Diffie-Hellman key exchange is an example of an algorithm that provides forward secrecy. A forward secrecy algorithm should be considered if attacks against recorded authentication or data sessions are considered to pose a significant threat. 15. Changes since previous drafts Other than minor editorial changes, the following changes have been made to this draft: Since version 04: - An enhanced version of EAP-TTLS, called version 1, has been defined in section 11. Since version 03: - Removed section on keying information. Since version 02: - Added password change for MS-CHAP-V2. Since version 01: - In section 11, the TTLS server's response with data cipher suites has been made conditional on receiving data cipher suite preferences from both client and access point. Also, implicit acceptance of the client's preferred data cipher suite has been eliminated in favor of explicitly returning the data cipher suite selection. Since version 00: - A Table of Contents has been added. - In section 3, a definition of "access domain" has been added. - In section 6.4, the requirement has been added that TLS session resumption must not be allowed for any negotiation that succeeds in phase 1 TLS handshake but does not successfully complete phase 2 authentication. - In sections 7 and 10.1, reversed the order of randoms used in PRF, to follow EAP-TLS practice and avoid namespace collisions Paul Funk expires January 2005 [Page 49] Internet-Draft April 2004 with TLS. - In section 8, specified the assigned EAP-TTLS number. - Added section 8.1, reserving for future standardization the ability to add data to an EAP-TTLS Start packet. 16. References [1] Aboba, B., and D. Simon, "PPP EAP TLS Authentication Protocol", RFC 2716, October 1999. [2] Aboba, B., Blunk, L., Vollbrecht, J., Carlson, J., and H. Levkowetz, "PPP Extensible Authentication Protocol (EAP)", RFC 3784, June 2004. [3] Dierks, T., and C. Allen, "The TLS Protocol Version 1.0", RFC 2246, November 1998. [4] Institute for Electrical and Electronics Engineers, "IEEE 802.1X, Standard for Port Based Network Access Control", 2001. [5] Simpson, W., Editor, "The Point-to-Point Protocol (PPP)", STD 51, RFC 1661, July 1994. [6] Rigney, C., Rubens, A., Simpson, W., and S. Willens, "Remote Authentication Dial In User Service (RADIUS)", RFC 2865, June 2000. [7] Aboba, B., and M. Beadles, "The Network Access Identifier", RFC 2486, January 1999. [8] Calhoun, P., Loughney, J., Guttman, E., Zorn, G., and J. Arkko, "Diameter Base Protocol", RFC 3588, July 2001. [9] Reynolds, J., and J. Postel, "Assigned Numbers", RFC 1700, October 1994. [10] Zorn, G., and S. Cobb, "Microsoft PPP CHAP Extensions", RFC 2433, October 1998. [11] Zorn, G., "Microsoft PPP CHAP Extensions, Version 2", RFC 2759, January 2000. [12] Zorn, G., "Microsoft Vendor-specific RADIUS Attributes", RFC 2548, March 1999. [13] Blake-Wilson, S., Nystrom, M., Hopwood, D., Mikkelsen, J., and T. Wright, "Transport Layer Security (TLS) Extensions", RFC 3546, June 2003. Paul Funk expires January 2005 [Page 50] Internet-Draft April 2004 [14] Myers, M., Ankney, R., Malpani, A., Galperin, S., and C. Adams, "Internet X.509 Public Key Infrastructure: Online Certificate Status Protocol - OCSP", RFC 2560, June 1999. [15] Asokan, N., Niemi, V., and Nyberg, K., "Man-in-the-Middle in Tunneled Authentication", http://www.saunalahti.fi/~asokan/research/mitm.html, Nokia Research Center, Finland, October 24 2002. [16] Puthenkulam, J., "The Compound Authentication Binding Problem", draft-puthenkulam-eap-binding-04.txt, October 2003. 17. Authors' Addresses Questions about this memo can be directed to: Paul Funk Funk Software, Inc. 222 Third Street Cambridge, MA 02142 USA Phone: +1 617 497-6339 E-mail: paul@funk.com Simon Blake-Wilson Basic Commerce & Industries, Inc. 304 Harper Drive, Suite 203 Moorestown, NJ 08057 Phone: +1 856 778-1660 E-mail: sblakewilson@bcisse.com 18. Full Copyright Statement Copyright (C) The Internet Society (2001-2004). All Rights Reserved. This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations, except as needed for the purpose of developing Internet standards in which case the procedures for copyrights defined in the Internet Standards process must be followed, or as required to translate it into languages other than English. Paul Funk expires January 2005 [Page 51] Internet-Draft April 2004 The limited permissions granted above are perpetual and will not be revoked by the Internet Society or its successors or assigns. This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Paul Funk expires January 2005 [Page 52]