MMUSIC Working Group F. Andreasen Internet Draft Cisco Systems Expires: December 2006 June 19, 2006 SDP Capability Negotiation draft-andreasen-mmusic-sdp-capability-negotiation-00.txt Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. 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/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html This Internet-Draft will expire on December 19, 2006. Abstract The Session Description Protocol (SDP) was intended for describing multimedia sessions for the purposes of session announcement, session invitation, and other forms of multimedia session initiation. SDP was not intended to provide capability indication or capability negotiation, however over the years, SDP has seen widespread adoption and as a result it has been gradually extended to provide limited support for these. SDP and its current extensions however do not have the ability to negotiate one or more alternative transport protocols (e.g. RTP profiles) which makes it particularly difficult to deploy new RTP profiles such as secure RTP and RTP with RTCP-based feedback. Andreasen Expires December 19, 2006 [Page 1] Internet-Draft SDP Capability Negotiation June 2006 The purpose of this document is to address that by identifying a set of requirements, evaluate existing work in this area, and provide a recommended solution for extending SDP with capability negotiation. 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 [RFC2119]. Table of Contents 1. Introduction...................................................3 2. Requirements...................................................5 3. Review of Existing Work........................................7 3.1. Grouping of Media Lines...................................7 3.2. Session Description Protocol (SDP) Simple Capability Declaration....................................................9 3.3. Session Description and Capability Negotiation (SDPng)...10 3.4. Multipart/alternative....................................12 3.5. Sharing Ports Between "m=" Lines.........................13 3.6. Opportunistic Encryption Using a Session Attribute.......14 3.7. Opportunistic Encryption using Probing...................14 4. Proposed Solution.............................................15 4.1. Solution Overview........................................15 4.2. Attribute Definitions....................................17 4.2.1. The Transport Protocol Capability Attribute.........17 4.2.2. The Transport Address Capability Attribute..........18 4.2.3. The Potential Configuration Attribute...............19 4.2.4. The Actual Configuration Attribute..................20 4.3. Offer/Answer Model Extensions............................22 4.3.1. Generating the Initial Offer........................22 4.3.2. Generating the Answer...............................23 4.3.3. Offerer Processing of the Answer....................23 4.3.4. Modifying the Session...............................24 5. Examples......................................................24 6. Security Considerations.......................................24 7. IANA Considerations...........................................24 8. To Do.........................................................24 9. Acknowledgments...............................................24 10. References...................................................25 10.1. Normative References....................................25 10.2. Informative References..................................25 Author's Addresses...............................................26 Intellectual Property Statement..................................27 Disclaimer of Validity...........................................27 Andreasen Expires December 19, 2006 [Page 2] Internet-Draft SDP Capability Negotiation June 2006 Copyright Statement..............................................27 Acknowledgment...................................................27 1. Introduction The Session Description Protocol (SDP) was intended for describing multimedia sessions for the purposes of session announcement, session invitation, and other forms of multimedia session initiation. The SDP contains one or more media stream descriptions with information such as IP-address and port, type of media stream (e.g. audio or video), transport protocol (possibly including profile information, e.g. RTP/AVP or RTP/SAVP), media formats (e.g. codecs), and various other session and media stream parameters that define the session. Simply providing media stream descriptions is sufficient for session announcements for a broadcast application, where the media stream parameters are fixed for all participants. When a participant wants to join the session, he obtains the session announcement and uses the media descriptions provided, e.g., joins a multicast group and receives media packets in the encoding format specified. If the media stream description is not supported by the participant, he is unable to receive the media. Such restrictions are not generally acceptable to multimedia session invitations, where two or more entities attempt to establish a media session using a set of media stream parameters acceptable to all participants. First of all, each entity must inform the other of its receive address, and secondly, the entities need to agree on the media stream parameters to use for the session, e.g. transport protocols and codecs. We here make a distinction between the capabilities supported by each participant and the parameters that can actually be used for the session. More generally, we can say that we have the following: o A set of capabilities, or potential configurations of the media stream components, supported by each side. o A set of actual configurations of the media stream components, which specifies which media stream components to use and with what parameters. o A negotiation process that takes the set of potential configurations (capabilities) as input and provides the actual configurations as output. SDP by itself was designed to provide only the second of these, i.e., the actual configurations, however over the years, use of SDP has Andreasen Expires December 19, 2006 [Page 3] Internet-Draft SDP Capability Negotiation June 2006 been extended beyond its original scope. Session negotiation semantics was defined by the offer/answer model in RFC 3264. It defines how two entities, an offerer and an answerer, exchange SDPs to negotiate a session. The offerer can include one or more media formats (codecs) per media stream, and the answerer then selects one or more of those offered and returns them in an answer. Both the offer and the answer contain actual configurations - potential configurations are not supported. The answer however may reduce the set of actual configurations from the offer. Other relevant extensions have been defined. Simple capability declarations, which defines how to provide a simple and limited set of capability descriptions in SDP was defined in RFC 3407. Grouping of media lines, which defines how media lines in SDP can have other semantics than the traditional "simultaneous media streams" semantics, was defined in RFC 3388, etc. Each of these extensions was designed to solve a specific limitation of SDP. Since SDP had already been stretched beyond its original intent, a more comprehensive capability declaration and negotiation process was intentionally not defined. Instead, work on a "next generation" of a protocol to provide session description and capability negotiation was initiated [SDPng]. SDPng however has not gained traction and has remained as work in progress for an extended period of time. Existing real-time multimedia communication protocols such as SIP, RTSP, Megaco, and MGCP continue to use SDP. SDP and its current extensions however do not address an increasingly important problem: the ability to negotiate one or more alternative transport protocols (e.g., RTP profiles). This makes it difficult to deploy new RTP profiles such as secure RTP (SRTP) [RFC3711], RTP with RTCP-Based Feedback [AVPF], etc. This particular problem is exacerbated by the fact that RTP profiles are defined independently. When a new profile is defined and N other profiles already exist, there is a potential need for defining N additional profiles, since profiles cannot be combined automatically. For example, in order to support the plain and secure RTP version of RTP with and without RTCP-based feedback, four separate profiles (and hence profile definitions) are needed: RTP/AVP [RFC3551], RTP/SAVP [RFC3711], RTP/AVPF [AVPF], and RTP/SAVPF [SAVPF]. In addition to the pressing profile negotiation problem, other important real-life constraints have been found as well. The purpose of this document is to define a mechanism that enables SDP to provide limited support for indicating potential configurations (capabilities) and negotiate the use of those potential configurations as actual configurations. It is not the intent to provide a full-fledged capability indication and Andreasen Expires December 19, 2006 [Page 4] Internet-Draft SDP Capability Negotiation June 2006 negotiation mechanism along the lines of SDPng or ITU-T H.245. Instead, the focus is on addressing a set of well-known real-life limitations. As mentioned above, SDP is used by several protocols, and hence the mechanism should be usable by all of these. One particularly important protocol for this problem however is the Session Initiation Protocol (SIP) [RFC3261]. SIP uses the offer/answer model (which is not specific to SIP) to negotiate sessions and hence any mechanism must at least consider how it either interacts with offer/answer, or how it should extend it. The rest of the document is structured as follows. We first provide a set of requirements for the solution in Section 2. In Section 3. we review relevant existing work in this area, how a solution based on that might look, and the pros and cons associated with each. In Section 4. we present our proposed solution in more detail followed examples in Section 5. and security considerations in Section 6. 2. Requirements REQ-10: It MUST be possible to indicate and negotiate alternative media formats on a per media stream basis. For example, many implementations support multiple codecs, but only one at a time. Changes between codecs cannot be done on- the-fly, e.g. when receiving a simple RTP payload type change. REQ-20: It MUST be possible to indicate and negotiate alternative attribute values ("a=") on a per media stream basis. For example, T.38 defines new attributes that may need to be conveyed as part of a capability. REQ-30: It MUST be possible to indicate and negotiate alternative media format parameter values ("a=fmtp") per media format on a per media stream basis. For example, a media format (codec) indicated as an alternative capability may include fmtp parameters. REQ-40: It MUST be possible to indicate and negotiate alternative transport protocols, e.g. different RTP profiles, on a per media stream basis. For example, "RTP/AVP" and "RTP/SAVP" may be alternatives. Andreasen Expires December 19, 2006 [Page 5] Internet-Draft SDP Capability Negotiation June 2006 REQ-50: It MUST be possible to indicate and negotiate alternative transport protocol and media type combinations on a per media stream basis. For example, an entity may support a fax call using either T.38 fax relay ("m=image udptl t38") or PCMU ("m=audio RTP/AVP 0"). REQ-60: It MUST be possible to specify unicast and multicast addresses as alternatives. [Editor's Note: This leads to some interesting issues with respect to the offer/answer model, where the set of parameters of relevance (or even defined) for a unicast stream differs from that of multicast streams. Also, some parameters have different meanings (e.g. direction attributes).] REQ-70: It MUST be possible to specify IPv4 and IPv6 addresses as alternatives. [Editor's note: This duplicates the ANAT grouping semantics - is it needed here ?] REQ-80: The mechanism MUST be backwards compatible for at least SIP and RTSP. Ideally, the mechanism should be completely transparent to entities that do not support it, without the need for any further signaling. REQ-90: The mechanism MUST either be backwards compatible for Megaco and MGCP or it MUST be possible to interwork it with Megaco and MGCP without any additional signaling. For example, if a media gateway controller (MGC) uses SIP to communicate with peers, and the MGC uses Megaco or MGCP to control a media gateway, it must be possible to translate between the mechanism and normal SDP. Avoiding interworking requirements in the MGC is desirable. REQ-100: The mechanism MUST work within the context of the offer/answer model [RFC3264]. Specifically, it MUST be able to negotiate alternatives within a single round-trip. [Editor's note: Should we limit scope to O/A only, or should we include RTSP as well ?] REQ-110: The offer/answer model requires the offerer to be able to receive media for any media streams listed as either "recvonly" or Andreasen Expires December 19, 2006 [Page 6] Internet-Draft SDP Capability Negotiation June 2006 "sendrecv" in an offer, as soon as that offer is generated. The mechanism MUST preserve this capability for all actual configurations included in an offer. Potential configurations do not have such a requirement. REQ-120: The mechanism MUST enable inclusion of potential configurations (alternative capabilities) in the offer - the answer would then indicate which, if any of these potential configurations were accepted. The offerer is not required to process media for a specific potential configuration until the offerer receives an answer showing that potential configuration was accepted. REQ-130: The mechanism MUST work in the presence of SIP forking. REQ-140: The mechanism SHOULD be reasonably efficient in terms of overall message size. This is a relative requirement to evaluate alternative solutions as opposed to an absolute and quantifiable requirement Above, we presented the requirements for the capability negotiation mechanism. Below, we provide a set of features that were considered and then explicitly deemed to be out-of-scope: o Indication of mandatory and optional capabilities. o Constraints on combinations of configurations, e.g. inability to use a video codec together with a particular audio codec, parameter X values that can only be used with parameter Y values, etc. 3. Review of Existing Work In this section, we provide an overview of existing relevant work that has either been completed or is work in progress. For each item, we outline how/if it can be used to address the requirements provided and the pros and cons of doing so. 3.1. Grouping of Media Lines Grouping of Media Lines is defined in [RFC3388]. RFC 3388 defines a framework that enables two or media lines to be grouped together for different purposes. Each media line is assigned an identifier and one or more group attributes then references two or more of those identifiers. Associated with each group attribute is a semantics indication. One semantic indication is the Alternative Network Andreasen Expires December 19, 2006 [Page 7] Internet-Draft SDP Capability Negotiation June 2006 Address Types ("ANAT") [RFC4091] which allows for IPv4 and IPv6 addresses to be specified as alternatives. The requirements presented above go beyond that, however a new semantic to simply indicate alternative media lines and associated negotiation rules could easily be defined. The main advantages of such an approach would be: o Mechanism Reuse: Several semantics have already been defined which increases the likelihood of an implementation supporting the framework. The disadvantages of such an approach would be: o Backwards Compatibility: The mechanism is not transparently backwards compatible. If an entity that does not support the mechanism receives it, the entity may incorrectly interpret the SDP as consisting of multiple media streams. While RFC 3388 defines procedures for recognizing and recover from this when using offer/answer, it can still lead to unintended behavior with endpoints that do not support the mechanism. In practice, it is not clear how much of an issue this is, at least for intelligent SIP endpoints. Most current implementations generally accept only one media stream of a given type (e.g. audio). Use of alternatives with different media stream types (e.g. a fax call using "audio" for voice- band data or "image" for T.38) makes it less clear though. Also, Media Gateway Controllers and Media Gateways that do not support grouping of media lines have been known to encounter problems. o Semantics Combination Issues: Multiple semantics may be provided by use of grouping, however they may interact with each other unintentionally. For example, the "FID" semantics defined in RFC 3388 forbids grouping of media lines with the same transport address, however that would be needed for alternative capabilities. Thus, using "FID" and alternative capabilities together would require special consideration. Andreasen Expires December 19, 2006 [Page 8] Internet-Draft SDP Capability Negotiation June 2006 o Some Combinatoric Explosion: The mechanism is not ideal to indicate alternative capabilities for multiple parameters or media formats within a particular media stream. For example, alternative attribute values and media format parameters for several codecs would lead to combinatoric explosion. [Editor's note: In practice, it is not clear this is a huge issue though.] o Message Size: Each alternative requires full duplication of all the relevant media stream parameters. [Editor's note: In practice, it is not clear this is a huge issue though.] 3.2. Session Description Protocol (SDP) Simple Capability Declaration SDP Simple Capability Declaration (simcap) is defined in [RFC3407]. It defines a set of SDP attributes that enables capabilities to be described at a session level or on a per media stream basis. RFC 3407 defines capability declaration only - actual negotiation procedures taking advantage of such capabilities have not been defined. Also, RFC 3407 does not define a way to indicate alternative transport addresses. Such rules and transport address capabilities however could easily be defined - the negotiation part would extend the offer/answer model to examine alternative configurations (capabilities). In conjunction with that, attributes to indicate the alternative configurations accepted would likely be needed as well. The main advantages of this approach are: o Satisfies all the requirements provided above. In particular, by relying solely on SDP attributes, transparent backwards compatibility is always ensured. The disadvantages of this approach are: o Offered Capabilities Hidden in Attributes: An offer may be accepted by the answerer and a media stream established based on SDP parameters contained in SDP attributes not known to intermediaries. Such intermediaries may be back-to-back user agents, or proxies that need to inspect the SDP, e.g., to authorize Quality of Service, add transcoders, etc. Andreasen Expires December 19, 2006 [Page 9] Internet-Draft SDP Capability Negotiation June 2006 o Maximum of 255 alternative media formats per SDP: RFC 3407 currently allows a maximum of 255 alternative media formats (codecs) per SDP. This may be too restrictive. 3.3. Session Description and Capability Negotiation (SDPng) The Session Description and Capability Negotiation protocol [SDPng] was intended as a replacement for SDP [SDP]. SDPng includes a full capability indication and negotiation framework that would address the shortcomings of SDP and satisfy the requirements provided above. However, SDPng has not gained traction, in large part due to existing widespread adoption of SDP. As a consequence, SDPng has remained as work in progress with limited progress for an extended period of time. SDPng consists of two things: an SDPng description, which is an XML document that describes the actual and/or potential configurations as well as an optional negotiation process (an offer/answer compliant process is included as part of SDPng). The SDPng description consists of up to five parts: o Capabilities: An optional list of capabilities (potential configurations) to be matched with the other parties' capabilities. o Definitions: An optional set of definitions of commonly used parameters for later referencing. o Configurations: A mandatory description of the conference components, each of which can provide a list of alternative configurations. o Constraints: An optional set of constraints of combinations of configurations. Constraints are not defined as part of the base SDPng specification. o Session Information: Optional meta information on the conferences and individual components. SDPng is application-agnostic with the base specification defining a basic XML schema supporting the above. In order to actually use SDPng, application-specific packages are needed. Packages define things such as media types, codecs and their configuration parameters, etc. The base SDPng specification includes a couple of example packages to support audio, video, and RTP. Andreasen Expires December 19, 2006 [Page 10] Internet-Draft SDP Capability Negotiation June 2006 One approach to extending SDP with capability indication and negotiation capabilities could be to adopt the mechanisms defined by SDPng that are necessary to satisfy the requirements provided above. Those areas could then be included within SDP itself, e.g. in the form of one or more SDP attributes ("a=") containing the actual SDPng description. The areas to consider here include: o Capabilities: This would be needed to describe alternative media formats and media format parameters. o Definitions: This would be needed to define alternative transport addresses. o Configurations: This would be needed to define alternative configurations The constraints and session information parts of SDPng would not be used. The main advantages of such an approach would be: o SDPng was designed and intended to solve the general capability negotiation problems faced by SDP. A considerable amount of work has already gone into it and it was originally targeted as the long-term direction (replacement for SDP). The disadvantages of such an approach would be: o Duplicate Encoding and Specification Work: SDPng uses a different coding format than SDP and hence all SDP parameters (incl. codecs and transports) that need to be provided will need to have an equivalent SDPng definition. There is currently no automatic process for translating all SDP parameters or values into corresponding SDPng parameters or values; many existing SDP parameters and values currently have no corresponding SDPng definition. o SDPng is Work in Progress: SDPng is currently work in progress but has seen limited interest and progress for a while. Adoption of a subset of its current definition may end up differing from the final specification. Also, the current SDPng specification needs further clarification and semantic tightening in a number of areas that would be of relevance to this approach. Andreasen Expires December 19, 2006 [Page 11] Internet-Draft SDP Capability Negotiation June 2006 o Negotiation of Transport Parameters: SDPng currently does not support negotiation of transport parameters as individual capabilities. It is however still possible to negotiate different transport parameters by providing alternative configurations. o Verbose Encoding and Large Message Size: SDPng descriptions are XML documents, which are fairly verbose and result in descriptions that are substantially larger than existing SDP. 3.4. Multipart/alternative In [I-D.jenning-sipping-multipart], the use of multipart/alternative MIME is proposed as a way to support multiple alternative offers. Each alternative offer has an id associated with it by use of a new MIME header field called Content-Answering-CID. The answerer chooses one of the offers and performs normal offer/answer operation on that offer, and then sends back a single answer which includes the Content-Answering-CID value of the offer chosen. The main advantages of this approach are: o It allows for use of alternative encodings of the offer, e.g. SDP and SDPng, as well as varying levels of confidentiality and integrity by use of S/MIME [RFC3851]. Use of multipart/alternative to solve the SDP capability negotiation problems however has several shortcomings: o Backwards Compatibility: Neither SIP nor RTSP mandate support for Multipart MIME. In the case of SIP, multipart/alternative is generally incompatible with existing SIP proxies, firewalls, Session Border Controllers, and endpoints. o Heterogeneous Error Response Forking Problem (HERFP): When a SIP proxy forks a request to multiple Contacts, each of which generate a response, the proxy only forwards the "best" of these responses to the request originator. If one or more of the Contacts do not support multipart/alternative, the request originator may never discover this. Instead, only a Contact that supports multipart/alternative will be able to generate an answer that reaches the request originator. o Combinatoric Explosions: Use of multipart/alternative to convey alternatives on a per media stream basis or even per media format parameter basis quickly leads to combinatoric explosions. Andreasen Expires December 19, 2006 [Page 12] Internet-Draft SDP Capability Negotiation June 2006 o Message Size: Each alternative requires full duplication of all the relevant SDP parameters (one complete SDP per alternative). It should be noted, that use of multipart/alternative has been discussed several times before and, in large part due to the problems mentioned above as well as the semantics defined for multipart/alternative [RFC2046], has met with opposition when it comes to addressing the above types of requirements. 3.5. Sharing Ports Between "m=" Lines SDP [SDP] does not state whether two "m=" lines can share the same transport address or not but rather leaves this explicitly undefined. It has been suggested that alternative capabilities for a media stream could be indicated by including multiple media stream descriptions sharing the same transport address (i.e. using the same port number in the "m=" line and sharing the same IP-address). Such practice was not defined in [RFC2327], however it was suggested in an Internet-Draft version of [SDP]. Following discussion of the potential problems it introduced, it was removed. The main advantages of this approach would be: o May not require any additional extensions to SDP - only additional semantics. [Editor's note: It is somewhat unclear how it would work without extensions if we allow for alternative attributes and media format parameters and the offerer needs to always know which ones were accepted] The disadvantages of this approach would be: o Alternative Address Support: It would not be possible to support unicast and multicast transport addresses as alternatives using this method. Similarly, IPv4 and IPv6 addresses as alternatives would not be possible. o Backwards Compatibility Issues: Since sharing of transport addresses between multiple streams was never specified as part of SDP, backwards compatibility is likely to be an issue. Some implementations may support it whereas others may not. The lack of an explicit signaling indication to indicate the desired operation may lead to ungraceful failure scenarios. Offer/answer semantics would be unclear here as well. Andreasen Expires December 19, 2006 [Page 13] Internet-Draft SDP Capability Negotiation June 2006 o Some Combinatoric Explosion: The mechanism is not ideal to indicate alternative capabilities for multiple parameters or media formats within a particular media stream. For example, alternative attribute values and media format parameters for several codecs would lead to combinatoric explosion. o Message Size: Each alternative requires full duplication of all the relevant media stream parameters. [Editor's note: In practice, it is not clear this is a huge issue though.] 3.6. Opportunistic Encryption Using a Session Attribute This approach was suggested to address the specific scenario of negotiating either RTP or SRTP. The endpoints signal their desire to do SRTP by signaling RTP (RTP/AVP), and using an attribute ("a=") in the SDP that indicates whether SRTP is supported or not. The main advantages of this approach are: o Compatible with non-SRTP-aware endpoints. The disadvantages of this approach are: o Violates RFC 3711 by listing an incorrect profile ("RTP/AVP" instead of "RTP/SAVP"). o Addresses only a small subset of the requirements provided above. 3.7. Opportunistic Encryption using Probing This is another approach suggested to address the specific scenario of negotiating either RTP or SRTP. In this case, the endpoints first establish an RTP session using RTP (RTP/AVP). The endpoints send probe messages, over the media path, to determine if the remote endpoint supports their keying technique. The main advantages of this approach are: o Compatible with non-SRTP-aware endpoints. The disadvantages of this approach are: o Addresses only a small subset of the requirements provided above. Andreasen Expires December 19, 2006 [Page 14] Internet-Draft SDP Capability Negotiation June 2006 4. Proposed Solution Based on the requirements provided in Section 2. and the alternatives examined in Section 3. the solution based on the Session Description Protocol (SDP) Simple Capability Declaration (simcap) [RFC3407] with extensions as outlined in Section 3.2. is preferred. In this section we present that solution in detail. 4.1. Solution Overview The solution consists of the following: o The capability declaration mechanism defined by simcap [RFC3407]. o A new attribute ("a=ctrpr") that defines how to list transport protocols as capabilities. o A new attribute ("a=ctrad") that defines how to list transport addresses as capabilities. o A new attribute ("a=pcfg") that lists the potential configurations supported by the entity that generated the SDP. This is done by reference to the simcap capabilities from the SDP in question, and optionally one or more of the transport protocol and transport address capabilities. o A new attribute ("a=acfg") to be used in an answer SDP. The attribute identifies which of the potential configurations from an offer SDP were used as actual configurations to form the answer SDP. This is done by listing the potential configurations that were used from the offer SDP. o Extensions to the offer/answer model that allow for potential configurations to be included in an offer, where they constitute offers that may be accepted by the answerer instead of the actual configuration(s) included in the "m=" line(s). The mechanism is illustrated by the offer/answer exchange below, where Alice sends an offer to Bob: Andreasen Expires December 19, 2006 [Page 15] Internet-Draft SDP Capability Negotiation June 2006 Alice Bob | (1) Offer (SRTP and RTP) | |--------------------------------->| | | | (2) Answer (RTP) | |<---------------------------------| | | Alice's offer includes RTP and SRTP as alternatives, with SRTP being the preferred one: v=0 o=- 25678 753849 IN IP4 128.96.41.1 s= c=IN IP4 128.96.41.1 t=0 0 m=audio 3456 RTP/SAVP 0 18 a=crypto:1 AES_CM_128_HMAC_SHA1_32 inline:NzB4d1BINUAvLEw6UzF3WSJ+PSdFcGdUJShpX1Zj|2^20|1:32 a=sqn: 0 a=cdsc: 1 audio RTP/SAVP 0 18 a=cdsc: 3 audio RTP/AVP 0 18 a=pcfg: c=1,2|3,4 The "m=" line indicates that Alice is offering to use RTP with PCMU or G.729. The "crypto" attribute provides the keying material using SDP security descriptions [SDES]. The simcap capability declaration is provided by the "a=sqn" and "a=cdsc" attributes as defined in [RFC3407]. The capabilities indicate that PCMU and G.729 are supported with either secure RTP (preferred) or RTP. The new "a=pcfg" attribute provides the potential configurations included in the offer by reference to the simcap capability declarations. Two alternatives are provided; the first one using capabilities 1 and 2, i.e. PCMU and G.729 under the RTP/SAVP profile (secure RTP), and the second one using capabilities 3 and 4, i.e. PCMU and G.729 under the RTP/AVP profile. Bob receives the SDP offer from Alice. Bob supports RTP, but not SRTP, and hence he accepts the potential configuration for RTP provided by Alice: v=0 o=- 24351 621814 IN IP4 128.96.41.2 s= c=IN IP4 128.96.41.2 t=0 0 Andreasen Expires December 19, 2006 [Page 16] Internet-Draft SDP Capability Negotiation June 2006 m=audio 4567 RTP/AVP 0 18 a=acfg: c=3,4 Bob includes the new "a=acfg" attribute in the answer to inform Alice that he based his answer on an offer containing the potential configuration with capabilities 3 and 4 from the offer SDP (i.e. PCMU and G.729 under the RTP/AVP profile). This in turn implies that absence of an "a=crypto" attribute in the answer (to convey SRTP keying material) does not constitute an error. 4.2. Attribute Definitions 4.2.1. The Transport Protocol Capability Attribute Transport Protocols can be expressed as capabilities by use of a new Transport Protocol Capability attribute ("a=ctrpr") defined as follows: a=ctrpr: where is an integer between 1 and 255 (both included) used to number the transport address capability for later reference, and is defined as in the SDP "m=" line. The "ctrpr" attribute is a media-level only attribute. Each occurrence of the attribute within a given media description ("m=" line) MUST use a different value of , with the first one being 1, the second one being 2, etc. The values provided are independent of similar values provided for other attributes, i.e., they form a separate name-space for transport protocol capabilities. [Editor's Note: Could easily be a session-level attribute - would potentially be more efficient in case of multiple media descriptions.] Below, we provide examples of the "a=ctrpr" attribute: a=ctrpr: 1 RTP/AVP a=ctrpr: 2 RTP/AVPF The first one provides a capability for the "RTP/AVP" profile defined in [RFC3551] and the second one provides a capability for the RTP with RTCP-Based Feedback profile defined in [AVPF]. Note that the simcap extensions already provide for this capability by inclusion in the "cdsc" attribute, however having this as a Andreasen Expires December 19, 2006 [Page 17] Internet-Draft SDP Capability Negotiation June 2006 separate capability indication can provide significant message size reduction when negotiating alternative profiles (of which there can be many). TO DO: Explain further when it is appropriate to use one mechanism versus the other. For example, when attributes are included as capabilities and those attributes apply to only some profiles, use of this attribute may not be appropriate (e.g. consider "crypto" attribute with "RTP/AVP" and "RTP/SAVP". 4.2.2. The Transport Address Capability Attribute Transport addresses can be expressed as capabilities by use of a new Transport Address Capability attribute ("a=ctrad") defined as follows: a=ctrad:
* where is an integer between 1 and 255 (both included) used to number the transport address capability for later reference, ,
and are defined as in the SDP "c=" line, and the first occurrence of is as defined in the SDP "m=" line. Additional occurrences may be present; the only currently well-defined semantics associated with this is when one additional port is present for RTP-based media streams. In that case, that second port field takes on the meaning of the "a=rtcp" attribute defined in [RFC3605]. The "ctrad" attribute is a media-level only attribute. Each occurrence of the attribute within a given media description ("m=" line) MUST use a different value of , with the first one being 1, the second one being 2, etc. The values provided are independent of similar values provided for other attributes, i.e., they form a separate name-space for transport address capabilities. Below, we provide an example of the "a=ctrad" attribute: a=ctrad: 1 IN IP4 128.96.41.1 3456 a=ctrad: 2 IN IP4 224.2.17.12/127 49170 The first provides a unicast address on port 3456, whereas the second provides a multicast address with a time to live of 127 on port 49170. Andreasen Expires December 19, 2006 [Page 18] Internet-Draft SDP Capability Negotiation June 2006 4.2.3. The Potential Configuration Attribute Potential Configurations can be expressed by use of a new Potential Configuration Attribute ("a=pcfg") defined as follows: a=pcfg: [] The potential configuration attribute includes one or more sets of simcap-capabilities. A list of possible transport address capabilities as well as transport protocol capabilities can optionally be included as well. Together, these values define a set of potential configurations. TO DO: Clean up capability and configuration terminology. is defined by the following ABNF: simcap-capabilities = "c=" cap-list *("|" cap-list) cap-list = cap-num *("," cap-num) cap-num = 1*3DIGIT ; defined in [RFC4234] Each capability list is a comma-separated list of simcap capability numbers where cap-num refers to simcap capability numbers and hence MUST be between 1 and 255 (both included). Alternative potential simcap configurations are separated by a vertical bar ("|"). is defined by the following ABNF: transport-protocol-capabilities = "p=" trpr-cap-list trpr-cap-list = trpr-cap-num *("," trpr-cap-num) trpr-cap-num = 1*3DIGIT ; defined in [RFC4234] The trpr-cap-num refers to transport protocol capability numbers defined above and hence MUST be between 1 and 255 (both included). Multiple transport protocol capabilities are provided in a comma- separated list. When transport protocol capabilities are not included, the transport protocol information from the media description ("m=" line) will be used. is defined by the following ABNF: transport-address-capabilities = "a=" trad-cap-list trad-cap-list = trad-cap-num *("," trad-cap-num) trad-cap-num = 1*3DIGIT ; defined in [RFC4234] Andreasen Expires December 19, 2006 [Page 19] Internet-Draft SDP Capability Negotiation June 2006 The trad-cap-num refers to transport address capability numbers defined above and hence MUST be between 1 and 255 (both included). Multiple transport address capabilities are provided in a comma- separated list. When transport address capabilities are not included, the transport address information from the media description ("m=" line and corresponding "c=" line) will be used. The potential configuration ("a=pcfg") attribute is a media-level only attribute. Each occurrence of the attribute within a given media description ("m=" line) defines a set of potential configurations that can be used for that media description. Below, we provide an example of the "a=pcfg" attribute in a complete media description in order to properly indicate the supporting attributes: v=0 o=- 25678 753849 IN IP4 128.96.41.1 s= c=IN IP4 128.96.41.1 t=0 0 m=audio 3456 RTP/SAVPF 0 18 a=crypto:1 AES_CM_128_HMAC_SHA1_32 inline:NzB4d1BINUAvLEw6UzF3WSJ+PSdFcGdUJShpX1Zj|2^20|1:32 a=sqn: 0 a=cdsc: 1 audio RTP/SAVP 0 4 18 a=ctrad: 1 IN IP4 128.96.41.1 3456 a=ctrad: 2 IN IP4 224.2.17.12/127 49170 a=ctrpr: 1 RTP/AVP a=ctrpr: 2 RTP/AVPF a=ctrpr: 3 RTP/SAVP a=ctrpr: 4 RTP/SAVPF a=pcfg: c=1|3 p=1,2,3,4 a=1 a=pcfg: c=2 p=1 a=1,2 We have two potential configurations listed here. The first one indicates that PCMU (payload type number 0) or G.729 (payload type number 18) can be supported with either of the profiles RTP/AVP, RTP/AVPF, RTP/SAVP, or RTP/SAVPF on the unicast address only. The second potential configuration indicates that G.723 (payload type number 4) can be supported with the RTP/AVP profile only but on either the unicast or multicast address indicated. 4.2.4. The Actual Configuration Attribute The actual configuration attribute identifies which of the potential configurations from an offer SDP were used as actual configurations Andreasen Expires December 19, 2006 [Page 20] Internet-Draft SDP Capability Negotiation June 2006 in an answer SDP. This is done by reference to the simcap capabilities, and the transport protocol and transport address (if included) capabilities from the offer that were actually used by the answerer in his offer/answer procedure. The Actual Configuration Attribute ("a=acfg") is defined as follows: a=acfg: [] is defined by the following ABNF: simcap-capabily-list = "c=" cap-list cap-list = cap-num *("," cap-num) cap-num = 1*3DIGIT ; defined in [RFC4234] Each capability list is a comma-separated list of simcap capability numbers where cap-num refers to simcap capability numbers and hence MUST be between 1 and 255 (both included). is defined by the following ABNF: transport-protocol-capability = "p=" trpr-cap-num trpr-cap-num = 1*3DIGIT ; defined in [RFC4234] The trpr-cap-num refers to transport protocol capability numbers defined above and hence MUST be between 1 and 255 (both included). When a transport protocol capability is not included, the transport protocol information from the media description ("m=" line) in the offer is being used. is defined by the following ABNF: transport-address-capability = "a=" trad-cap-num trad-cap-num = 1*3DIGIT ; defined in [RFC4234] The trad-cap-num refers to transport address capability numbers defined above and hence MUST be between 1 and 255 (both included). When a transport address capability is not included, the transport address information from the media description ("m=" line and corresponding "c=" line) in the corresponding offer is being used. The actual configuration ("a=acfg") attribute is a media-level only attribute. There MUST NOT be more than one occurrence of an actual configuration attribute within a given media description. Andreasen Expires December 19, 2006 [Page 21] Internet-Draft SDP Capability Negotiation June 2006 Below, we provide an example of the "a=acfg" attribute (building on the previous example with the potential configuration attribute): v=0 o=- 24351 621814 IN IP4 128.96.41.2 s= c=IN IP4 128.96.41.2 t=0 0 m=audio 4567 RTP/AVPF 0 a=acfg: c=1 p=2 a=1 It indicates that the answerer used an offer consisting of simcap capability 1 (PCMU), transport protocol capability 2 (RTP/AVPF), and transport address capability 1 (IPv4 address 128.96.41.1 and port 3456). 4.3. Offer/Answer Model Extensions In this section, we define extensions to the offer/answer model defined in [RFC3264] to allow for potential configurations to be included in an offer, where they constitute offers that may be accepted by the answerer instead of the actual configuration(s) included in the "m=" line(s). [Editor's Note: Multicast considerations have been omitted for now.] TO DO: Elaborate and firm up offer/answer procedures. 4.3.1. Generating the Initial Offer An offerer that wants to use capability negotiation extensions defined in this document MUST include the following in the offer: o one or more simcap capability descriptions (as defined in [RFC3407]) for each of the capabilities o optionally, one or more transport protocol capability attributes (as defined in Section 4.2.1. if one or more alternative transport protocols is to be negotiated) o optionally, one or more transport address capability attributes (as defined in Section 4.2.2. if one or more alternative transport addresses is to be negotiated) Andreasen Expires December 19, 2006 [Page 22] Internet-Draft SDP Capability Negotiation June 2006 o one or more potential configuration attributes (as defined in Section 4.2.3. which define the potential configurations supported by the offerer. Each of the potential configurations listed constitutes an alternative offer which may be used to negotiate and establish the session. The current actual configuration is included in the "m=" line (as defined by [RFC3264]). 4.3.2. Generating the Answer When the answerer receives an offer with one or more valid potential configuration information attributes present, it may use any of the potential configurations as an alternative offer. A potential configuration information attribute is valid if all of the capabilities (simcap, transport protocol and transport addresses) it references are present and valid themselves. The actual configuration is contained in the media description's "m=" line. The answerer can send media to the offerer in accordance with the actual configuration, however if it chooses to use one of the alternative potential configurations, media sent to the offerer may be discarded by the offerer until the answer is received. If the answerer chooses to accept one of the alternative potential configurations instead of the actual configuration, the answerer MUST generate an answer as if the offer contained that potential configuration instead of the actual configuration included. The answerer MUST also include an actual configuration attribute in the answer that identifies the potential configuration from the offer used by the answerer. 4.3.3. Offerer Processing of the Answer When the offerer included potential configurations for a media stream, it MUST examine the answer for the presence of an actual configuration attribute for each such media stream. If the attribute is missing, offerer processing of the answer MUST proceed as defined by [RFC3264]. If the attribute is present, processing continues as follows: The actual configuration attribute specifies which of the potential configurations was used by the answerer to generate the answer. The offerer MUST now process the answer as if the offer had contained the potential configuration indicated as the actual configuration in the media description ("m=" line) in the offer. Andreasen Expires December 19, 2006 [Page 23] Internet-Draft SDP Capability Negotiation June 2006 4.3.4. Modifying the Session Potential configurations may be included in subsequent offers as defined in [RFC3264, Section 8]. The procedure for doing so is similar to that described above with the answer including an indication of the actual configuration used by the answerer. 5. Examples TBD. 6. Security Considerations TBD. 7. IANA Considerations TBD. 8. To Do o Simcap allows for the same parameter to be described more than once as a way of indicating alternative parameter values. Whenever this is done, the answer must enable the offerer to determine which of the alternatives was accepted. Similar considerations apply to min and max values supported by simcap. o Lots of other things noted throughout the document. 9. Acknowledgments Thanks to Francois Audet and Dan Wing for comments on this document. Andreasen Expires December 19, 2006 [Page 24] Internet-Draft SDP Capability Negotiation June 2006 10. References 10.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC2234] Crocker, D. and Overell, P.(Editors), "Augmented BNF for Syntax Specifications: ABNF", RFC 2234, Internet Mail Consortium and Demon Internet Ltd., November 1997. [RFC3264] Rosenberg, J., and H. Schulzrinne, "An Offer/Answer Model with Session Description Protocol (SDP)", RFC 3264, June 2002. [RFC3407] F. Andreasen, "Session Description Protocol (SDP) Simple Capability Declaration", RFC 3407, October 2002. [RFC3605] C. Huitema, "Real Time Control Protocol (RTCP) attribute in Session Description Protocol (SDP)", RFC 3605, October 2003. [RFC4234] Crocker, D., and P. Overell, "Augmented BNF for Syntax Specifications: ABNF", RFC 4234, October 2005. [SDP] Handley, M., Jacobson, V., and C. Perkins, "SDP: Session Description Protocol", RFC 4567, June 2006. 10.2. Informative References [RFC2046] Freed, N., and N. Borensteain, "Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types", RFC 2046, November 1996. [RFC2327] Handley, M., Jacobson, V., and C. Perkins, "SDP: Session Description Protocol", RFC 2327, April 1998. [RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, A., Peterson, J., Sparks, R., Handley, M., and E. Schooler, "SIP: Session Initiation Protocol", RFC 3261, June 2002. [RFC3388] Camarillo, G., Eriksson, G., Holler, J., and H. Schulzrinne, "Grouping of Media Lines in the Session Description Protocol (SDP)", RFC 3388, December 2002. Andreasen Expires December 19, 2006 [Page 25] Internet-Draft SDP Capability Negotiation June 2006 [RFC3551] Schulzrinne, H., and S. Casner, "RTP Profile for Audio and Video Conferences with Minimal Control", RFC 3551, July 2003. [RFC3711] Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K. Norrman, "The Secure Real-time Transport Protocol (SRTP)", RFC 3711, March 2004. [RFC3851] B. Ramsdell, "Secure/Multipurpose Internet Mail Extensions (S/MIME) Version 3.1 Message Specification", RFC 3851, July 2004. [RFC4091] Camarillo, G., and J. Rosenberg, The Alternative Network Address Types (ANAT) Semantics for the Session Description Protocol (SDP) Grouping Framework, RFC 4091, June 2005. [AVPF] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, "Extended RTP Profile for RTCP-Based Feedback (RTP/AVPF)", Work in Progress, August 2004. [I-D.jennings-sipping-multipart] Wing, D., and C. Jennings, "Session Initiation Protocol (SIP) Offer/Answer with Multipart Alternative", Work in Progress, March 2006. [SAVPF] Ott, J., and E Carrara, "Extended Secure RTP Profile for RTCP-based Feedback (RTP/SAVPF)", Work in Progress, December 2005. [SDES] Andreasen, F., Baugher, M., and D. Wing, "Session Description Protocol Security Descriptions for Media Streams", Work in Progress, September 2005. [SDPng] Kutscher, D., Ott, J., and C. Bormann, "Session Description and Capability Negotiation", Work in Progress, February 2005. Author's Addresses Flemming Andreasen Cisco Systems Edison, NJ Email: fandreas@cisco.com Andreasen Expires December 19, 2006 [Page 26] Internet-Draft SDP Capability Negotiation June 2006 Intellectual Property Statement The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org. Disclaimer of Validity This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM 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. Copyright Statement Copyright (C) The Internet Society (2006). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. Acknowledgment Funding for the RFC Editor function is currently provided by the Internet Society. Andreasen Expires December 19, 2006 [Page 27] Internet-Draft SDP Capability Negotiation June 2006 Andreasen Expires December 19, 2006 [Page 28]