Authorization for NSIS Signaling Layer Protocols
draft-ietf-nsis-nslp-auth-05.txt

Abstract

Signaling layer protocols specified within the NSIS framework may rely on the GIST (General Internet Signaling Transport) protocol to handle authorization. Still, the signaling layer protocol above GIST itself may require separate authorization to be performed when a node receives a request for a certain kind of service or resources. This draft presents a generic model and object formats for session authorization within the NSIS Signaling Layer Protocols. The goal of session authorization is to allow the exchange of information between network elements in order to authorize the use of resources for a service and to coordinate actions between the signaling and transport planes.

Status of this Memo

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.

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.”

This Internet-Draft will expire on January 29, 2011.

Copyright Notice

Copyright (c) 2010 IETF Trust and the persons identified as the document authors. All rights reserved.

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

Table of Contents

1.  Conventions used in this document
2.  Introduction
3.  Session Authorization Object
    3.1.  Session Authorization Object format
    3.2.  Session Authorization Attributes
        3.2.1.  Authorizing Entity Identifier
        3.2.2.  Source Address
        3.2.3.  Destination Address
        3.2.4.  Start time
        3.2.5.  End time
        3.2.6.  NSLP Object List
        3.2.7.  Authentication data
4.  Integrity of the AUTH_SESSION policy element
    4.1.  Shared symmetric keys
        4.1.1.  Operational Setting using shared symmetric keys
    4.2.  Kerberos
    4.3.  Public Key
        4.3.1.  Operational Setting for public key based authentication
    4.4.  HMAC Signed
5.  Framework
    5.1.  The Coupled Model
    5.2.  The associated model with one policy server
    5.3.  The associated model with two policy servers
    5.4.  The non-associated model
6.  Message Processing Rules
    6.1.  Generation of the AUTH_SESSION by the authorizing entity
    6.2.  Processing within the QoS NSLP
        6.2.1.  Message Generation
        6.2.2.  Message Reception
        6.2.3.  Authorization (QNE/PDP)
        6.2.4.  Error Signaling
    6.3.  Processing with the NAT/FW NSLP
        6.3.1.  Message Generation
        6.3.2.  Message Reception
        6.3.3.  Authorization (Router/PDP)
        6.3.4.  Error Signaling
    6.4.  Integrity Protection of NSLP messages
7.  Security Considerations
8.  IANA Considerations
9.  Acknowledgments
10.  References
    10.1.  Normative References
    10.2.  Informative References
Appendix A.  Changes
§  Authors' Addresses

1.  Conventions used in this document

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14, RFC 2119 [RFC2119] (Bradner, S., “Key words for use in RFCs to Indicate Requirement Levels,” March 1997.).

The term "NSLP node" (NN) is used to refer to an NSIS node running an NSLP protocol that can make use of the authorization object discussed in this document. Currently, this node would run either the QoS NSLP [I‑D.ietf‑nsis‑qos‑nslp] (Manner, J., Karagiannis, G., and A. McDonald, “NSLP for Quality-of-Service Signaling,” January 2010.) or the NAT/FW NSLP [I‑D.ietf‑nsis‑nslp‑natfw] (Stiemerling, M., Tschofenig, H., Aoun, C., and E. Davies, “NAT/Firewall NSIS Signaling Layer Protocol (NSLP),” April 2010.) service.

2.  Introduction

The Next Steps in Signaling (NSIS) framework [RFC4080] (Hancock, R., Karagiannis, G., Loughney, J., and S. Van den Bosch, “Next Steps in Signaling (NSIS): Framework,” June 2005.) defines a suite of protocols for the next generation in Internet signaling. The design is based on a generalized transport protocol for signaling applications, the General Internet Signaling Transport (GIST) [I‑D.ietf‑nsis‑ntlp] (Schulzrinne, H. and M. Stiemerling, “GIST: General Internet Signalling Transport,” June 2009.), and various kinds of signaling applications. Two signaling applications and their NSIS Signaling Layer Protocol (NSLP) have been designed, a Quality of Service application (QoS NSLP) [I‑D.ietf‑nsis‑qos‑nslp] (Manner, J., Karagiannis, G., and A. McDonald, “NSLP for Quality-of-Service Signaling,” January 2010.) and a NAT/firewall application (NAT/FW) [I‑D.ietf‑nsis‑nslp‑natfw] (Stiemerling, M., Tschofenig, H., Aoun, C., and E. Davies, “NAT/Firewall NSIS Signaling Layer Protocol (NSLP),” April 2010.).

The basic security architecture for NSIS is based on a chain-of-trust model, where each GIST hop may chose the appropriate security protocol, taking into account the signaling application requirements. For instance, communication between two directly adjacent GIST peers may be secured via TCP/TLS. On the one hand this model is appropriate for a number of different use cases, and allows the signaling applications to leave the handling of security to GIST. On the other hand, several sessions of different signaling applications are then multiplexed onto the same GIST TLS connection.

Yet, in order to allow for finer-grain per-session or per-user admission control, it is necessary to provide a mechanism for ensuring that the use of resources by a host has been properly authorized before allowing the signaling application to commit the resource request, e.g., a QoS reservation or mappings for NAT traversal. In order to meet this requirement, there must be information in the NSLP message which may be used to verify the validity of the request. This can be done by providing the host with a session authorization policy element which is inserted into the message and verified by the respective network elements.

This document describes a generic NSLP layer session authorization policy object (AUTH_SESSION) used to convey authorization information for the request. Generic in this context means that it is usable by all NSLPs. The scheme is based on third-party tokens. A trusted third party provides authentication tokens to clients and allows verification of the information by the network elements. The requesting host inserts its authorization information acquired from the trusted third party into the NSLP message to allow verification of the network resource request. Network elements verify the request and then process it based on admission policy (e.g., they perform a resource reservation or change bindings or firewall filter). This work is based on RFC 3520 [RFC3520] (Hamer, L-N., Gage, B., Kosinski, B., and H. Shieh, “Session Authorization Policy Element,” April 2003.) and RFC 3521 [RFC3521] (Hamer, L-N., Gage, B., and H. Shieh, “Framework for Session Set-up with Media Authorization,” April 2003.).

The default operation of the authorization is to add one authorization policy object. Yet, in order to support end-to-end signaling and request authorization from different networks, a host initiating an NSLP signaling session may add more than one AUTH_SESSION object in the message. The identifier of the authorizing entity can be used by the network elements to use the third party they trust to verify the request.

3.  Session Authorization Object

This section presents a new NSLP layer object called session authorization (AUTH_SESSION). The AUTH_SESSION object can be used in the currently specified and future NSLP protocols.

The authorization attributes follow the format and specification given in RFC3520 [RFC3520] (Hamer, L-N., Gage, B., Kosinski, B., and H. Shieh, “Session Authorization Policy Element,” April 2003.).

3.1.  Session Authorization Object format

The AUTH_SESSION object contains a list of fields which describe the session, along with other attributes. The object header follows the generic NSLP object header, therefore it can be used together with any NSLP.

 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|A|B|r|r|         Type          |r|r|r|r|        Length         |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+                                                               +
//         Session Authorization Attribute List                //
+                                                               +
+---------------------------------------------------------------+

The value for the Type field comes from shared NSLP object type space. The Length field is given in units of 32 bit words and measures the length of the Value component of the TLV object (i.e. it does not include the standard header).

The bits marked 'A' and 'B' are extensibility flags, and used to signal the desired treatment for objects whose treatment has not been defined in the protocol specification (i.e. whose Type field is unknown at the receiver). The following four categories of object have been identified, and are described here for informational purposes only, i.e., for normative behavior refer to the particular NSLP documents (e.g., [I‑D.ietf‑nsis‑qos‑nslp] (Manner, J., Karagiannis, G., and A. McDonald, “NSLP for Quality-of-Service Signaling,” January 2010.) [I‑D.ietf‑nsis‑nslp‑natfw] (Stiemerling, M., Tschofenig, H., Aoun, C., and E. Davies, “NAT/Firewall NSIS Signaling Layer Protocol (NSLP),” April 2010.)).

AB=00 ("Mandatory"): If the object is not understood, the entire message containing it MUST be rejected, and an error message sent back (usually of class/code "Protocol Error/Unknown object present").

AB=01 ("Ignore"): If the object is not understood, it MUST be deleted and the rest of the message processed as usual.

AB=10 ("Forward"): If the object is not understood, it MUST be retained unchanged in any message forwarded as a result of message processing, but not stored locally.

AB=11 ("Refresh"): If the object is not understood, it should be incorporated into the locally stored signaling application state for this flow/session, forwarded in any resulting message, and also used in any refresh or repair message which is generated locally. This flag combination is not used by all NSLPs, e.g., it is not used in NATFW NSLP.

The remaining bits marked 'r' are reserved. The extensibility flags follow the definition in the GIST specification. The AUTH_SESSION object defines in this specification MUST have the AB-bits set to "10". An NSLP Node (NN) may use the authorization information if it is configured to do so, but may also just skip the object.

Type: SESSION_AUTH_OBJ (IANA-TBD)

Length: Variable, contains length of Session authorization object list in units of 32 bit words.

Session Authorization Attribute List: variable length

The session authorization attribute list is a collection of objects that describes the session and provides other information necessary to verify resource request (e.g., a resource reservation, binding, or firewall filter change request). An initial set of valid objects is described in Section 3.2 (Session Authorization Attributes).

3.2.  Session Authorization Attributes

A session authorization attribute may contain a variety of information and has both an attribute type and subtype. The attribute itself MUST be a multiple of 4 octets in length, and any attributes that are not a multiple of 4 octets long MUST be padded to a 4-octet boundary. All padding bytes MUST have a value of zero.

 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |    X-Type     |   SubType     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
//                           Value ...                         //
+---------------------------------------------------------------+

Length: 16 bits

The Length field is two octets and indicates the actual length of the attribute (including Length, X-Type and SubType fields) in number of octets. The length does NOT include any bytes padding to the value field to make the attribute a multiple of 4 octets long.

X-Type: 8 bits

Session authorization attribute type (X-Type) field is one octet. IANA acts as a registry for X-Types as described in Section 8 (IANA Considerations), IANA Considerations. This specification uses the following X-Types:

1. AUTH_ENT_ID The unique identifier of the entity that authorized the session.
2. SESSION_ID Unique identifier for this session. See also RFC 3520 [RFC3520] (Hamer, L-N., Gage, B., Kosinski, B., and H. Shieh, “Session Authorization Policy Element,” April 2003.), not to be confused with the SESSIONID of GIST/NSIS.
3. SOURCE_ADDR Address specification for the signaling session initiator, i.e., the source address of the signaling message originator.
4. DEST_ADDR Address specification for the signaling session end-point.
5. START_TIME The starting time for the session.
6. END_TIME The end time for the session.
7. AUTHENTICATION_DATA Authentication data of the session authorization policy element.

SubType: 8 bits

Session authorization attribute sub-type is one octet in length. The value of the SubType depends on the X-Type.

Value: variable length

The attribute specific information.

3.2.1.  Authorizing Entity Identifier

AUTH_ENT_ID is used to identify the entity that authorized the initial service request and generated the session authorization policy element. The AUTH_ENT_ID may be represented in various formats, and the SubType is used to define the format for the ID. The format for AUTH_ENT_ID is as follows:

 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |    X-Type     |   SubType     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
//                        OctetString ...                      //
+---------------------------------------------------------------+

Length: Length of the attribute, which MUST be > 4.

X-Type: AUTH_ENT_ID

SubType:

The following sub-types for AUTH_ENT_ID are defined. IANA acts as a registry for AUTH_ENT_ID sub-types as described in Section 8 (IANA Considerations), IANA Considerations. Initially, the registry contains the following sub-types of AUTH_ENT_ID:

1. IPV4_ADDRESS IPv4 address represented in 32 bits
2. IPV6_ADDRESS IPv6 address represented in 128 bits
3. FQDN Fully Qualified Domain Name as defined in [RFC1034] (Mockapetris, P., “Domain names - concepts and facilities,” November 1987.) as an ASCII string.
4. ASCII_DN X.500 Distinguished name as defined in [RFC4514] (Zeilenga, K., “Lightweight Directory Access Protocol (LDAP): String Representation of Distinguished Names,” June 2006.) as an ASCII string.
5. UNICODE_DN X.500 Distinguished name as defined in [RFC4514] (Zeilenga, K., “Lightweight Directory Access Protocol (LDAP): String Representation of Distinguished Names,” June 2006.) as a UTF-8 string.
6. URI Universal Resource Identifier, as defined in [RFC3986] (Berners-Lee, T., Fielding, R., and L. Masinter, “Uniform Resource Identifier (URI): Generic Syntax,” January 2005.).
7. KRB_PRINCIPAL Fully Qualified Kerberos Principal name represented by the ASCII string of a principal followed by the @ realm name as defined in [RFC4120] (Neuman, C., Yu, T., Hartman, S., and K. Raeburn, “The Kerberos Network Authentication Service (V5),” July 2005.) (e.g., johndoe@nowhere).
8. X509_V3_CERT The Distinguished Name of the subject of the certificate as defined in [RFC4514] (Zeilenga, K., “Lightweight Directory Access Protocol (LDAP): String Representation of Distinguished Names,” June 2006.) as a UTF-8 string.
9. PGP_CERT The OpenPGP certificate of the authorizing entity as defined as Public-Key Packet in [RFC4880] (Callas, J., Donnerhacke, L., Finney, H., Shaw, D., and R. Thayer, “OpenPGP Message Format,” November 2007.).
10. HMAC_SIGNED Indicates that the AUTHENTICATION_DATA attribute (IANA-TBD) contains a self-signed HMAC signature [RFC2104] (Krawczyk, H., Bellare, M., and R. Canetti, “HMAC: Keyed-Hashing for Message Authentication,” February 1997.) that ensures the integrity of the NSLP message. The HMAC is calculated over all NSLP objects given in the NSLP_OBJECT_LIST attribute that MUST also be present. The AUTH_ENT_ID contains the Hash Algorithm that is used for calculation of the HMAC as Transform ID from Transform Type 3 of the IKEv2 registry [RFC4306] (Kaufman, C., “Internet Key Exchange (IKEv2) Protocol,” December 2005.).

OctetString: Contains the authorizing entity identifier.

3.2.2.  Source Address

SOURCE_ADDR is used to identify the source address specification of the authorized session. This X-Type may be useful in some scenarios to make sure the resource request has been authorized for that particular source address and/or port. Usually, it corresponds to the signaling source, e.g., the IP source address of the GIST packet, or flow source or flow destination address respectively, which are contained in the GIST MRI (Message Routing Information) object.

 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |    X-Type     |   SubType     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
//                        OctetString ...                      //
+---------------------------------------------------------------+

Length: Length of the attribute, which MUST be > 4.

X-Type: SOURCE_ADDR

SubType:

The following sub types for SOURCE_ADDR are defined. IANA acts as a registry for SOURCE_ADDR sub-types as described in Section 8 (IANA Considerations), IANA Considerations. Initially, the registry contains the following sub types for SOURCE_ADDR:

1. IPV4_ADDRESS IPv4 address represented in 32 bits
2. IPV6_ADDRESS IPv6 address represented in 128 bits
3. UDP_PORT_LIST list of UDP port specifications, represented as 16 bits per list entry.
4. TCP_PORT_LIST list of TCP port specifications, represented as 16 bits per list entry.
5. SPI Security Parameter Index represented in 32 bits

OctetString: The OctetString contains the source address information.

In scenarios where a source address is required (see Section 5 (Framework)), at least one of the subtypes 1 or 2 MUST be included in every Session Authorization Data Policy Element. Multiple SOURCE_ADDR attributes MAY be included if multiple addresses have been authorized. The source address of the request (e.g., a QoS NSLP RESERVE) MUST match one of the SOURCE_ADDR attributes contained in this Session Authorization Data Policy Element.

At most, one instance of subtype 3 MAY be included in every Session Authorization Data Policy Element. At most, one instance of subtype 4 MAY be included in every Session Authorization Data Policy Element. Inclusion of a subtype 3 attribute does not prevent inclusion of a subtype 4 attribute (i.e., both UDP and TCP ports may be authorized).

If no PORT attributes are specified, then all ports are considered valid; otherwise, only the specified ports are authorized for use. Every source address and port list must be included in a separate SOURCE_ADDR attribute.

3.2.3.  Destination Address

DEST_ADDR is used to identify the destination address of the authorized session. This X-Type may be useful in some scenarios to make sure the resource request has been authorized for that particular destination address and/or port.

 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |    X-Type     |   SubType     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
//                        OctetString ...                      //
+---------------------------------------------------------------+

Length: Length of the attribute in number of octets, which MUST be > 4.

X-Type: DEST_ADDR

SubType:

The following sub types for DEST_ADDR are defined. IANA acts as a registry for DEST_ADDR sub-types as described in Section 8 (IANA Considerations), IANA Considerations. Initially, the registry contains the following sub types for DEST_ADDR:

1. IPV4_ADDRESS IPv4 address represented in 32 bits
2. IPV6_ADDRESS IPv6 address represented in 128 bits
3. UDP_PORT_LIST list of UDP port specifications, represented as 16 bits per list entry.
4. TCP_PORT_LIST list of TCP port specifications, represented as 16 bits per list entry.
5. SPI Security Parameter Index represented in 32 bits

OctetString: The OctetString contains the destination address specification.

In scenarios where a destination address is required (see Section 5 (Framework)), at least one of the subtypes 1 or 2 MUST be included in every Session Authorization Data Policy Element. Multiple DEST_ADDR attributes MAY be included if multiple addresses have been authorized. The destination address field of the resource reservation datagram (e.g., QoS NSLP Reserve) MUST match one of the DEST_ADDR attributes contained in this Session Authorization Data Policy Element.

At most, one instance of subtype 3 MAY be included in every Session Authorization Data Policy Element. At most, one instance of subtype 4 MAY be included in every Session Authorization Data Policy Element. Inclusion of a subtype 3 attribute does not prevent inclusion of a subtype 4 attribute (i.e., both UDP and TCP ports may be authorized).

If no PORT attributes are specified, then all ports are considered valid; otherwise, only the specified ports are authorized for use.

Every destination address and port list must be included in a separate DEST_ADDR attribute.

3.2.4.  Start time

START_TIME is used to identify the start time of the authorized session and can be used to prevent replay attacks. If the AUTH_SESSION policy element is presented in a resource request, the network SHOULD reject the request if it is not received within a few seconds of the start time specified.

 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |    X-Type     |   SubType     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
//                        OctetString ...                      //
+---------------------------------------------------------------+

Length: Length of the attribute, which MUST be > 4.

X-Type: START_TIME

SubType:

The following sub types for START_TIME are defined. IANA acts as a registry for START_TIME sub-types as described in Section 8 (IANA Considerations), IANA Considerations. Initially, the registry contains the following sub types for START_TIME:

1. 1 NTP_TIMESTAMP NTP Timestamp Format as defined in RFC 5905 [RFC5905] (Mills, D., Martin, J., Burbank, J., and W. Kasch, “Network Time Protocol Version 4: Protocol and Algorithms Specification,” June 2010.).

OctetString: The OctetString contains the start time.

3.2.5.  End time

END_TIME is used to identify the end time of the authorized session and can be used to limit the amount of time that resources are authorized for use (e.g., in prepaid session scenarios).

 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |    X-Type     |   SubType     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
//                        OctetString ...                      //
+---------------------------------------------------------------+

Length: Length of the attribute, which MUST be > 4.

X-Type: END_TIME

SubType:

The following sub types for END_TIME are defined. IANA acts as a registry for END_TIME sub-types as described in Section 8 (IANA Considerations), IANA Considerations. Initially, the registry contains the following sub types for END_TIME:

1. NTP_TIMESTAMP NTP Timestamp Format as defined in RFC 5905 [RFC5905] (Mills, D., Martin, J., Burbank, J., and W. Kasch, “Network Time Protocol Version 4: Protocol and Algorithms Specification,” June 2010.).

OctetString: The OctetString contains the end time.

3.2.6.  NSLP Object List

The NSLP_OBJECT_LIST attribute contains a list of NSLP objects types that are used in the keyed-hash computation whose result is given in the AUTHENTICATION_DATA attribute. This allows for an integrity protection of NSLP PDUs. If an NSLP_OBJECT_LIST attribute has been included in the AUTH_SESSION policy element, an AUTHENTICATION_DATA attribute MUST also be present.

The creator of this attribute lists every NSLP object type whose NSLP PDU object was included in the computation of the hash. The receiver can verify the integrity of the NSLP PDU by computing a hash over all NSLP objects that are listed in this attribute including all the attributes of the authorization object. Since all NSLP object types are unique over all different NSLPs, this will work for any NSLP.

Basic NTLP/NSLP objects like the session ID, the NSLPID and the MRI MUST be always included in the HMAC. Since they are not carried within the NSLP itself, but only within GIST, they must be delivered via the GIST API and normalized to their network representation from [I‑D.ietf‑nsis‑ntlp] (Schulzrinne, H. and M. Stiemerling, “GIST: General Internet Signalling Transport,” June 2009.) again before calculating the hash. These values are hashed first, before any other NSLP object values that are included in the hash computation.

A summary of the NSLP_OBJECT_LIST attribute format is described below.

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+---------------+---------------+---------------+---------------+
| Length                        | NSLP_OBJ_LIST |     zero      |
+---------------+---------------+-------+-------+---------------+
|No. of signed NSLP objects = n |  rsv  |  NSLP object type (1) |
+-------+-------+---------------+-------+-------+---------------+
|  rsv  | NSLP object type (2)  |             .....            //
+-------+-------+---------------+---------------+---------------+
|  rsv  | NSLP object type (n)  |     (padding if required)     |
+--------------+----------------+---------------+---------------+

Length: Length of the attribute, which MUST be > 4.

X-Type: NSLP_OBJECT_LIST

SubType: No sub types for NSLP_OBJECT_LIST are currently defined. This field MUST be set to 0 and ignored upon reception.

OctetString: The OctetString contains the authentication data of the AUTH_SESSION.

No. of signed NSLP objects: The number n of NSLP object types that follow. n=0 is allowed, i.e., only a padding field is contained then.

rsv: reserved bits and must be set to 0 (zero) and ignored upon reception.

NSLP object type: the NSLP 12-bit object type identifier of the object that was included in the hash calculation. The NSLP object type values comprise only 12 bit, so four bits per type value are currently not used within the list. Depending on the number of signed objects, a corresponding padding word of 16 bit must be supplied.

padding: padding MUST be added if the number of NSLP objects is even and MUST NOT be added if the number of NSLP objects is odd. If padding has to be applied the padding field MUST be 16 bit set to 0 and its contents MUST be ignored upon reception.

3.2.7.  Authentication data

The AUTHENTICATION_DATA attribute contains the authentication data of the AUTH_SESSION policy element and signs all the data in the policy element up to the AUTHENTICATION_DATA. If the AUTHENTICATION_DATA attribute has been included in the AUTH_SESSION policy element, it MUST be the last attribute in the list. The algorithm used to compute the authentication data depends on the AUTH_ENT_ID SubType field. See Section 4 (Integrity of the AUTH_SESSION policy element) entitled Integrity of the AUTH_SESSION policy element.

A summary of AUTHENTICATION_DATA attribute format is described below.

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |    X-Type     |   SubType     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
//                        OctetString ...                      //
+---------------------------------------------------------------+

Length: Length of the attribute, which MUST be > 4.

X-Type: AUTHENTICATION_DATA

SubType: No sub types for AUTHENTICATION_DATA are currently defined. This field MUST be set to 0 and ignored upon reception.

OctetString: The OctetString contains the authentication data of the AUTH_SESSION.

4.  Integrity of the AUTH_SESSION policy element

This section describes how to ensure the integrity of the policy element is preserved.

4.1.  Shared symmetric keys

In shared symmetric key environments, the AUTH_ENT_ID MUST be of subtypes: IPV4_ADDRESS, IPV6_ADDRESS, FQDN, ASCII_DN, UNICODE_DN or URI. An example AUTH_SESSION object is shown below.

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|1|0|0|0| Type = AUTH_SESSION   |0|0|0|0|    Object   Length    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |   AUTH_ENT_ID | IPV4_ADDRESS  |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|   OctetString ...   (The authorizing entity's Identifier)     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |   AUTH_DATA   |     zero      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                            KEY_ID                             |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|   OctetString ...   (Authentication data)                     |
+---------------------------------------------------------------+

4.1.1.  Operational Setting using shared symmetric keys

This assumes both the Authorizing Entity and the Network router/PDP (Policy Decision Point) are provisioned with shared symmetric keys and with policies detailing which algorithm to be used for computing the authentication data along with the expected length of the authentication data for that particular algorithm.

Key maintenance is outside the scope of this document, but AUTH_SESSION implementations MUST at least provide the ability to manually configure keys and their parameters. The key used to produce the authentication data is identified by the AUTH_ENT_ID field. Since multiple keys may be configured for a particular AUTH_ENT_ID value, the first 32 bits of the AUTH_DATA field MUST be a key ID to be used to identify the appropriate key. Each key must also be configured with lifetime parameters for the time period within which it is valid as well as an associated cryptographic algorithm parameter specifying the algorithm to be used with the key. At a minimum, all AUTH_SESSION implementations MUST support the HMAC-MD5-128 [RFC1321] (Rivest, R., “The MD5 Message-Digest Algorithm,” April 1992.) [RFC2104] (Krawczyk, H., Bellare, M., and R. Canetti, “HMAC: Keyed-Hashing for Message Authentication,” February 1997.) cryptographic algorithm for computing the authentication data.

It is good practice to regularly change keys. Keys MUST be configurable such that their lifetimes overlap allowing smooth transitions between keys. At the midpoint of the lifetime overlap between two keys, senders should transition from using the current key to the next/longer-lived key. Meanwhile, receivers simply accept any identified key received within its configured lifetime and reject those that are not.

4.2.  Kerberos

Since Kerberos [RFC4120] (Neuman, C., Yu, T., Hartman, S., and K. Raeburn, “The Kerberos Network Authentication Service (V5),” July 2005.) is widely used for end-user authorization, e.g., in Windows domains, it is well suited for being used in the context of user-based authorization for NSIS sessions. For instance, a user may request a ticket for authorization of installing rules in an NATFW-capable router.

In a Kerberos environment, it is assumed that the user of the requesting NSLP host requests a ticket from the (the Kerberos Key Distribution Center - KDC) for using the NSLP Node (router) as resource (target service). The ticket can be presented to the NSLP node via Kerberos by sending a KRB_CRED message to the NSLP node independently but prior to the NSLP exchange. Thus, the principal name of the service must be known in advance, though the exact IP address may not be known in advance. How the name is assigned and made available to the client is implementation specific. The extracted common session key can subsequently be used for using the HMAC_SIGNED variant of the AUTH_SESSION object. The

Another option is to encapsulate the credentials in the AUTH_DATA portion of the AUTH_SESSION object. In this case the AUTH_ENT_ID MUST be of the subtype KRB_PRINCIPAL. The KRB_PRINCIPAL field is defined as the Fully Qualified Kerberos Principal name of the authorizing entity. The AUTH_DATA portion of the AUTH_SESSION object contains the KRB_CRED message that the receiving NSLP node has to extract and verify. A second AUTH_SESSION object of type HMAC_SIGNED SHOULD protect the integrity of the NSLP message, including the prior AUTH_SESSION object. The session key included in the first AUTH_SESSION object has to be used for HMAC calculation.

An example of the Kerberos AUTH_DATA policy element is shown below.

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|1|0|0|0| Type = AUTH_SESSION   |0|0|0|0|    Object   Length    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |   AUTH_ENT_ID |  KERB_P.      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|   OctetString ...   (The principal@realm name)                |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |   AUTH_DATA   |     zero      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|   OctetString ...   (KRB_CRED Data)                           |
+---------------------------------------------------------------+

4.3.  Public Key

In a public key environment, the AUTH_ENT_ID MUST be of the subtypes: X509_V3_CERT or PGP_CERT. The authentication data is used for authenticating the authorizing entity. An example of the public key AUTH_SESSION policy element is shown below.

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|1|0|0|0| Type = AUTH_SESSION   |0|0|0|0|    Object   Length    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |   AUTH_ENT_ID |   PGP_CERT    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|   OctetString ...   (Authorizing entity Digital Certificate)  |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |   AUTH_DATA   |     zero      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|   OctetString ...   (Authentication data)                     |
+---------------------------------------------------------------+

4.3.1.  Operational Setting for public key based authentication

Public key based authentication assumes the following:

• Authorizing entities have a pair of keys (private key and public key).
• Private key is secured with the authorizing entity.
• Public keys are stored in digital certificates and a trusted party, certificate authority (CA) issues these digital certificates.
• The verifier (PDP or router) has the ability to verify the digital certificate.

Authorizing entity uses its private key to generate AUTHENTICATION_DATA. Authenticators (router, PDP) use the authorizing entity's public key (stored in the digital certificate) to verify and authenticate the policy element.

4.3.1.1.  X.509 V3 digital certificates

When the AUTH_ENT_ID is of type X509_V3_CERT, AUTHENTICATION_DATA MUST be generated following these steps:

• A Signed-data is constructed as defined in RFC5652 [RFC5652] (Housley, R., “Cryptographic Message Syntax (CMS),” September 2009.) . A digest is computed on the content (as specified in Section 6.1 (Generation of the AUTH_SESSION by the authorizing entity)) with a signer-specific message-digest algorithm. The certificates field contains the chain of authorizing entity's X.509 V3 digital certificates. The certificate revocation list is defined in the crls field. The digest output is digitally signed following Section 8 of RFC 3447 [RFC3447] (Jonsson, J. and B. Kaliski, “Public-Key Cryptography Standards (PKCS) #1: RSA Cryptography Specifications Version 2.1,” February 2003.), using the signer's private key.

When the AUTH_ENT_ID is of type X509_V3_CERT, verification MUST be done following these steps:

• Parse the X.509 V3 certificate to extract the distinguished name of the issuer of the certificate.
• Certification Path Validation is performed as defined in Section 6 of RFC 5280 [RFC5280] (Cooper, D., Santesson, S., Farrell, S., Boeyen, S., Housley, R., and W. Polk, “Internet X.509 Public Key Infrastructure Certificate and Certificate Revocation List (CRL) Profile,” May 2008.).
• Parse through the Certificate Revocation list to verify that the received certificate is not listed.
• Once the X.509 V3 certificate is validated, the public key of the authorizing entity can be extracted from the certificate.
• Extract the digest algorithm and the length of the digested data by parsing the CMS signed-data.
• The recipient independently computes the message digest. This message digest and the signer's public key are used to verify the signature value.

This verification ensures integrity, non-repudiation and data origin.

4.3.1.2.  PGP digital certificates

When the AUTH_ENT_ID is of type PGP_CERT, AUTHENTICATION_DATA MUST be generated following these steps:

• AUTHENTICATION_DATA contains a Signature Packet as defined in Section 5.2.3 of RFC 4880 [RFC4880] (Callas, J., Donnerhacke, L., Finney, H., Shaw, D., and R. Thayer, “OpenPGP Message Format,” November 2007.). In summary:
• Compute the hash of all data in the AUTH_SESSION policy element up to the AUTHENTICATION_DATA.
• The hash output is digitally signed following Section 8 of RFC 3447, using the signer's private key.

When the AUTH_ENT_ID is of type PGP_CERT, verification MUST be done following these steps:

• Validate the certificate.
• Once the PGP certificate is validated, the public key of the authorizing entity can be extracted from the certificate.
• Extract the hash algorithm and the length of the hashed data by parsing the PGP signature packet.
• The recipient independently computes the message digest. This message digest and the signer's public key are used to verify the signature value.

This verification ensures integrity, non-repudiation and data origin.

4.4.  HMAC Signed

An AUTH_SESSION object that carries an AUTH_ENT_ID of HMAC_SIGNED is used as integrity protection for NSLP messages. The AUTH_SESSION object MUST contain the following attributes:

• SOURCE_ADDR the source address of the entity that created the HMAC
• START_TIME the timestamp when the HMAC signature was calculated. This MUST be different for any two messages in sequence in order to prevent replay attacks. Since the NTP timestamp provides currently a resolution of 200 pico seconds this should be sufficient.
• NSLP_OBJECT_LIST this attribute lists all NSLP objects that are included into HMAC calculation.
• AUTHENTICATION_DATA this attribute contains the Key-ID that is used for HMAC calculation as well as the HMAC data itself [RFC2104] (Krawczyk, H., Bellare, M., and R. Canetti, “HMAC: Keyed-Hashing for Message Authentication,” February 1997.).

The key used for HMAC calculation must be exchanged securely by some other means, e.g., a Kerberos Ticket or pre-shared manual installation etc. The Key-ID in the AUTHENTICATION_DATA allows to refer to the appropriate key and also to periodically change signing keys within a session. The key length MUST be 64-bit at least, but it is ideally longer in order to defend against brute force attacks. It is recommended to use a per-user key for signing NSLP messages. This allows for verification of the authenticity of the message and thus provides a basis for a session-based per-user authorization.

Figure 1 shows an example of an object that is used for integrity protection of NSLP messages.

 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|1|0|0|0| Type = AUTH_SESSION   |0|0|0|0|    Object   Length    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |   AUTH_ENT_ID | HMAC_SIGNED   |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                   reserved                    | Transform ID  |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             | SOURCE_ADDR   |  IPV4_ADDRESS |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                IPv4 Source Address of NSLP sender             |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             |  START_TIME   | NTP_TIME_STAMP|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                        NTP Time Stamp (1)                     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                        NTP Time Stamp (2)                     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Length             | NTLP_OBJ_LIST |     zero      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|No. of signed NSLP objects = n |  rsv  |  NSLP object type (1) |
+-------+-------+---------------+-------+-------+---------------+
|  rsv  | NSLP object type (2)  |             .....            //
+-------+-------+---------------+---------------+---------------+
|  rsv  | NSLP object type (n)  |     (padding if required)     |
+--------------+----------------+---------------+---------------+
|            Length             |   AUTH_DATA   |     zero      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                            KEY_ID                             |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|          Message Authentication Code HMAC Data                |
+---------------------------------------------------------------+

Example of an AUTH_SESSION_OBJECT that provides integrity protection for NSLP messages

5.  Framework

RFC3521 [RFC3521] (Hamer, L-N., Gage, B., and H. Shieh, “Framework for Session Set-up with Media Authorization,” April 2003.) describes a framework in which the AUTH_SESSION policy element may be utilized to transport information required for authorizing resource reservation for media flows. RFC3521 introduces 4 different models:

1. The coupled model
2. The associated model with one policy server
3. The associated model with two policy servers
4. The non-associated model.

The fields that are required in an AUTH_SESSION policy element depend on which of the models is used.

5.1.  The Coupled Model

In the coupled model, the only information that MUST be included in the policy element is the SESSION_ID; it is used by the Authorizing Entity to correlate the resource reservation request with the media authorized during session set up. Since the End Host is assumed to be untrusted, the Policy Server SHOULD take measures to ensure that the integrity of the SESSION_ID is preserved in transit; the exact mechanisms to be used and the format of the SESSION_ID are implementation dependent.

5.2.  The associated model with one policy server

In this model, the contents of the AUTH_SESSION policy element MUST include:

• A session identifier - SESSION_ID. This is information that the authorizing entity can use to correlate the resource request with the media authorized during session set up.
• The identity of the authorizing entity - AUTH_ENT_ID. This information is used by an NN to determine which authorizing entity (Policy Server) should be used to solicit resource policy decisions.

In some environments, an NN may have no means for determining if the identity refers to a legitimate Policy Server within its domain. In order to protect against redirection of authorization requests to a bogus authorizing entity, the AUTH_SESSION MUST also include:

AUTHENTICATION_DATA. This authentication data is calculated over all other fields of the AUTH_SESSION policy element.

5.3.  The associated model with two policy servers

The content of the AUTH_SESSION Policy Element is identical to the associated model with one policy server.

5.4.  The non-associated model

In this model, the AUTH_SESSION MUST contain sufficient information to allow the Policy Server to make resource policy decisions autonomously from the authorizing entity. The policy element is created using information about the session by the authorizing entity. The information in the AUTH_SESSION policy element MUST include:

• Initiating party IP address or Identity (e.g., FQDN) - SOURCE_ADDR X-TYPE
• Responding party IP address or Identity (e.g., FQDN) - DEST_ADDR X-TYPE
• The authorization lifetime - START_TIME X-TYPE
• The identity of the authorizing entity to allow for validation of the token in shared symmetric key and Kerberos schemes - AUTH_ENT_ID X-TYPE
• The credentials of the authorizing entity in a public-key scheme - AUTH_ENT_ID X-TYPE
• Authentication data used to prevent tampering with the AUTH_SESSION policy element - AUTHENTICATION_DATA

Furthermore, the AUTH_SESSION policy element MAY contain:

• The lifetime of (each of) the media stream(s) - END_TIME X-TYPE
• Initiating party port number - SOURCE_ADDR X-TYPE
• Responding party port number - DEST_ADDR X-TYPE

All AUTH_SESSION fields MUST match with the resource request. If a field does not match, the request SHOULD be denied.

6.  Message Processing Rules

This section discusses the message processing related to the AUTH_SESSION object. Details of the processing the AUTH_SESSION object within QoS NSLP and NAT/FW NSLP are described. New NSLP protocols should use the same logic in making use of the AUTH_SESSION object.

6.1.  Generation of the AUTH_SESSION by the authorizing entity

1. Generate the AUTH_SESSION policy element with the appropriate contents as specified in Section 3 (Session Authorization Object).
2. If authentication is needed, the entire AUTH_SESSION policy element is constructed, excluding the length, type and subtype fields of the AUTH_SESSION field. Note that the message MUST include a START_TIME to prevent replay attacks. The output of the authentication algorithm, plus appropriate header information, is appended as AUTHENTICATION_DATA attribute to the AUTH_SESSION policy element.

6.2.  Processing within the QoS NSLP

The AUTH_SESSION object may be used with QoS NSLP QUERY and RESERVE messages to authorize the query operation for network resources, and a resource reservation request, respectively.

Moreover, the AUTH_SESSION object may also be used with RESPONSE messages in order to indicate that the authorizing entity changed the original request. For example, the session start or end times may have been modified, or the client may have requested authorization for all ports, but the authorizing entity only allowed the use of certain ports.

If the QoS NSIS Initiator (QNI) receives a RESPONSE message with an AUTH_SESSION object, the QNI MUST inspect the AUTH_SESSION object to see what authentication attribute was changed by an authorizing entity. The QNI SHOULD also silently accept AUTH_SESSION objects in RESPONSE message, which do not indicate any change to the original authorization request.

6.2.1.  Message Generation

A QoS NSLP message is created as specified in [I‑D.ietf‑nsis‑qos‑nslp] (Manner, J., Karagiannis, G., and A. McDonald, “NSLP for Quality-of-Service Signaling,” January 2010.).

1. The policy element received from the authorizing entity MUST be copied without modification into the AUTH_SESSION object.
2. The AUTH_SESSION object (containing the policy element) is inserted in the NSLP message in the appropriate place.

6.2.2.  Message Reception

The QoS NSLP message is processed as specified in [I‑D.ietf‑nsis‑qos‑nslp] (Manner, J., Karagiannis, G., and A. McDonald, “NSLP for Quality-of-Service Signaling,” January 2010.) with following modifications.

1. If the QNE is policy aware then it SHOULD use the Diameter QoS application or the RADIUS QoS protocol to communicate with the PDP. To construct the AAA message it is necessary to extract the AUTH_SESSION object and the QoS related objects from the QoS NSLP message and to craft the respective RADIUS or Diameter message. The message processing and object format is described in the respective RADIUS or Diameter QoS protocol, respectively. If the QNE is policy unaware then it ignores the policy data objects and continues processing the NSLP message.
2. If the response from the PDP is negative the request must be rejected. A negative response in RADIUS is an Access-Reject and in Diameter is based on the 'DIAMETER_SUCCESS' value in the Result-Code AVP of the QoS-Authz-Answer (QAA) message. The QNE must construct and send a RESPONSE message with the status of authorization failure as specified in [I‑D.ietf‑nsis‑qos‑nslp] (Manner, J., Karagiannis, G., and A. McDonald, “NSLP for Quality-of-Service Signaling,” January 2010.).
3. Continue processing the NSIS message.

6.2.3.  Authorization (QNE/PDP)

1. Retrieve the policy element from the AUTH_SESSION object. Check the AUTH_ENT_ID type and SubType fields and return an error if the identity type is not supported.
2. Verify the message integrity.
   • Shared symmetric key authentication: The QNE/PDP uses the AUTH_ENT_ID field to consult a table keyed by that field. The table should identify the cryptographic authentication algorithm to be used along with the expected length of the authentication data and the shared symmetric key for the authorizing entity. Verify that the indicated length of the authentication data is consistent with the configured table entry and validate the authentication data.
   • Public Key: Validate the certificate chain against the trusted Certificate Authority (CA) and validate the message signature using the public key.
   • Kerberos based usage is not yet provided by this document.
3. Once the identity of the authorizing entity and the validity of the service request has been established, the authorizing router/PDP MUST then consult its authorization policy in order to determine whether or not the specific request is authorized (e.g., based on available credits, information in the subscriber's database). To the extent to which these access control decisions require supplementary information, routers/PDPs MUST ensure that supplementary information is obtained securely.
4. Verify the requested resources do not exceed the authorized QoS.

6.2.4.  Error Signaling

When the PDP (e.g., a RADIUS or Diameter server) fails to verify the policy element then the appropriate actions described the respective AAA document need to be taken.

The QNE node MUST return a RESPONSE message with the INFO_SPEC error code Authorization Failure as defined in the QoS NSLP specification. The QNE MAY include an INFO_SPEC Object Value Info to indicate which AUTH_SESSION attribute created the error.

6.3.  Processing with the NAT/FW NSLP

This section presents processing rules for the NAT/FW NSLP [I‑D.ietf‑nsis‑nslp‑natfw] (Stiemerling, M., Tschofenig, H., Aoun, C., and E. Davies, “NAT/Firewall NSIS Signaling Layer Protocol (NSLP),” April 2010.).

6.3.1.  Message Generation

A NAT/FW NSLP message is created as specified in [I‑D.ietf‑nsis‑nslp‑natfw] (Stiemerling, M., Tschofenig, H., Aoun, C., and E. Davies, “NAT/Firewall NSIS Signaling Layer Protocol (NSLP),” April 2010.).

1. The policy element received from the authorizing entity MUST be copied without modification into the AUTH_SESSION object.
2. The AUTH_SESSION object (containing the policy element) is inserted in the NATFW NSLP message in the appropriate place.

6.3.2.  Message Reception

The NAT/FW NSLP message is processed as specified in [I‑D.ietf‑nsis‑nslp‑natfw] (Stiemerling, M., Tschofenig, H., Aoun, C., and E. Davies, “NAT/Firewall NSIS Signaling Layer Protocol (NSLP),” April 2010.) with following modifications.

1. If the router is policy aware then it SHOULD use the Diameter application or the RADIUS protocol to communicate with the PDP. To construct the AAA message it is necessary to extract the AUTH_SESSION element and the NATFW policy rule related objects from the NSLP message and to craft the respective RADIUS or Diameter message. The message processing and object format is described in the respective RADIUS or Diameter protocols, respectively. If the router is policy unaware then it ignores the policy data objects and continues processing the NSLP message.
2. Reject the message if the response from the PDP is negative. A negative response in RADIUS is an Access-Reject and in Diameter is based on the 'DIAMETER_SUCCESS' value in the Result-Code AVP.
3. Continue processing the NSIS message.

6.3.3.  Authorization (Router/PDP)

1. Retrieve the AUTH_SESSION object and the policy element. Check the PE type field and return an error if the identity type is not supported.
2. Verify the message integrity.
   • Shared symmetric key authentication: The Network router/PDP uses the AUTH_ENT_ID field to consult a table keyed by that field. The table should identify the cryptographic authentication algorithm to be used along with the expected length of the authentication data and the shared symmetric key for the authorizing entity. Verify that the indicated length of the authentication data is consistent with the configured table entry and validate the authentication data.
   • Public Key: Validate the certificate chain against the trusted Certificate Authority (CA) and validate the message signature using the public key.
   • Kerberos based usage is not provided by this document.
3. Once the identity of the authorizing entity and the validity of the service request has been established, the authorizing router/PDP MUST then consult its authorization policy in order to deter mine whether or not the specific request is authorized. To the extent to which these access control decisions require supplementary information, routers/PDPs MUST ensure that supplementary information is obtained securely.

6.3.4.  Error Signaling

When the PDP (e.g., a RADIUS or Diameter server) fails to verify the AUTH_SESSION element then the appropriate actions described the respective AAA document need to be taken. The NATFW NSLP node MUST return an error message of class 'Permanent failure' (0x5) with error code 'Authorization failed' (0x02).

6.4.  Integrity Protection of NSLP messages

The AUTH_SESSION object can also be used to provide an integrity protection for every NSLP signaling message, thereby also authorizing requests or responses. Assume that a user has deposited a shared key at some NN. This NN can then verify the integrity of every NSLP message sent by the user to the NN, thereby authorizing actions like resource reservations or opening firewall pinholes according to policy decisions earlier made.

The sender of an NSLP message creates an AUTH_SESSION object that contains AUTH_ENT_ID attribute set to HMAC_SIGNED (cf. Section 4.4 (HMAC Signed)) and hashes with the shared key over all NSLP objects that need to be protected and lists them in the NSLP_OBJECT_LIST. The AUTH_SESSION object itself is also protected by the HMAC. By inclusion of the AUTH_SESSION object into the NSLP message, the receiver of this NSLP message can verify its integrity if it has the suitable shared key for the HMAC. Any response to the sender should also be protected by inclusion of an AUTH_SESSION object in order to prevent attackers sending unauthorized responses on behalf of the real NN.

If an AUTH_SESSION object is present that has an AUTH_ENT_ID attribute set to HMAC_SIGNED, the integrity of all NSLP elements listed in the NSLP_OBJECT_LIST has to be checked, including the AUTH_SESSION object contents itself. Furthermore, session ID, MRI, and NSLP ID have to be included into the HMAC calculation, too, as specified in Section 3.2.6 (NSLP Object List). The key that is used to calculate the HMAC is referred to by the Key ID included in the AUTH_DATA attribute. If the provided timestamp in START_TIME is not recent enough or the calculated HMAC differs from the one provided in AUTH_DATA the message must be discarded silently and an error should be logged locally.

7.  Security Considerations

This document describes a mechanism for session authorization to prevent theft of service. There are three types of security issues to consider: protection against replay attacks, integrity of the AUTH_SESSION object, and the choice of the authentication algorithms and keys.

The first issue, replay attacks, MUST be prevented. In the non-associated model, the AUTH_SESSION object MUST include a START_TIME field and the NNs as well as Policy Servers MUST support NTP to ensure proper clock synchronization. Failure to ensure proper clock synchronization will allow replay attacks since the clocks of the different network entities may not be in sync. The start time is used to verify that the request is not being replayed at a later time. In all other models, the SESSION_ID is used by the Policy Server to ensure that the resource request successfully correlates with records of an authorized session. If a AUTH_SESSION object is replayed, it MUST be detected by the policy server (using internal algorithms) and the request MUST be rejected.

The second issue, the integrity of the policy element, is preserved in untrusted environments by including the AUTHENTICATION_DATA attribute. Therefore, this attribute MUST always be included.

In environments where shared symmetric keys are possible, they should be used in order to keep the AUTH_SESSION policy element size to a strict minimum, e.g., when wireless links are used. A secondary option would be PKI authentication, which provides a high level of security and good scalability. However, it requires the presence of credentials in the AUTH_SESSION policy element which impacts its size.

Further security issues are outlined in RFC 4081 [RFC4081] (Tschofenig, H. and D. Kroeselberg, “Security Threats for Next Steps in Signaling (NSIS),” June 2005.).

8.  IANA Considerations

This specification makes the following request to IANA:

1. Assign a new object value (AUTH_SESSION_OBJECT) for the AUTH_SESSION object from the shared NSLP object value space.
2. All AUTH_SESSION object internal values and numbers should be taken from the allocations already done for RFC 3520 [RFC3520] (Hamer, L-N., Gage, B., Kosinski, B., and H. Shieh, “Session Authorization Policy Element,” April 2003.). A new value must be assigned for the newly defined subtype HMAC_SIGNED of X-Type AUTH_ENT_ID.

9.  Acknowledgments

This document is based on the RFC 3520 [RFC3520] (Hamer, L-N., Gage, B., Kosinski, B., and H. Shieh, “Session Authorization Policy Element,” April 2003.) and credit therefore goes to the authors of RFC 3520, namely Louis-Nicolas Hamer, Brett Kosinski, Bill Gage and Hugh Shieh. Part of this work was funded by Deutsche Telekom Laboratories within the context of the ScaleNet project.

10.  References

10.1. Normative References

10.2. Informative References

Appendix A.  Changes

Note to the RFC Editor: this appendix to be removed before publication as an RFC.

This section describes changes between draft versions.

-00: based on draft-manner-nsis-nslp-auth-04

• removed extensibility flag handling directives as the NSLPs are responsible
• added IANA-TBD flag and SESSION_AUTH_OBJ
• changed Kerberos section
• removed calling/called party
• updated text in IANA section: removed "This specification uses two X-types introduced by RFC3520: Session_ID and Resources." as it may worry IANA (no action required)
• other small additions and fixes
• Updated Jukka's contact info

-01: addressed Xiaoming's comments of 2010-02-17 http://www.ietf.org/mail-archive/web/nsis/current/msg08726.html

• removed resource reservation specific text and used them as examples
• removed referral to checksum and used MAC instead
• specified action if AUTH_ENT_ID or sub type are not known
• added missing _ in AUTH_SESSION

-02:

• changed intended category to experimental, because other NSIS protocols are now in this category.
• added text in Section 4.2 (Kerberos) for Kerberos usage
• added more references to quoted RFCs
• moved Changes to Appendix

-03:

• Incorporated Lars Eggert's comments from AD review.
• added SESSION_ID to 3.2 with some clarifying text
• removed RESOURCES from section 5.4 since it is not directly applicable in the NSIS context

-04:

• Updated references to new RFC 5905 (NTP), RFC 4880 (OpenPGP Message Format), RFC 5280 (PKIX Certificate and CRL Profile)
• changed IPR to trust200902

-05:

• Replaced one remnant of RFC 2440 by RFC 4880

Authors' Addresses