Cross Session Stream Identification in the
Session Description ProtocolGoogleKungsbron 2Stockholm11122Swedenharald@alvestrand.noThis document specifies a grouping mechanism for RTP media streams
that can be used to specify relations between media streams within
different RTP sessions as well as within a single RTP session.This mechanism is used to signal the association between the RTP
concept of SSRC and the WebRTC concept of "media stream" / "media stream
track" using SDP signaling.This document is an input document for discussion. It should be
discussed in the MMUSIC WG list, mmusic@ietf.org.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 RFC 2119.This document extends the SSRC grouping framework by adding a new grouping relation that can
cross RTP session boundaries if needed. gives the background on why a new
mechanism is needed. gives the definition of the
new mechanism. gives the application of
the new mechanism for providing necessary semantic information for the
association of MediaStreamTracks to MediaStreams in the WebRTC API
.When media is carried by RTP , each
RTP media stream is distinguished inside an RTP session by its SSRC;
each RTP session is distinguished from all other RTP sessions by being
on a different transport association (strictly speaking, 2 transport
associations, one used for RTP and one used for RTCP, unless RTCP
multiplexing is used).There exist cases where an application using RTP and SDP needs to
signal some relationship between RTP media streams that may be carried
in either the same RTP session or different RTP sessions. For
instance, there may be a need to signal a relationship between a video
track in one RTP session and an audio track in another RTP session. In
traditional SDP, it is not possible to signal that these two tracks
should be carried in one session, so they are carried in different RTP
sessions.The SSRC grouping mechanism ("a=ssrc-group") can be used to associate RTP media streams
when those RTP media streams are part of the same RTP session. The
semantics of this mechanism prevent the association of RTP media
streams that are spread across different RTP sessions.The SDP grouping framework can be
used to group RTP sessions. When an RTP session carries one and only
one RTP media stream, it is possible to associate RTP media streams
across different RTP sessions. However, if an RTP session has multiple
RTP media streams, using multiple SSRCs, the SDP grouping framework
cannot be used for this purpose.There are use cases (some of which are discussed in ) where
neither of these approaches is appropriate; In those cases, a new
mechanism is needed.In addition, there is sometimes the need for an application to
specify some application-level information about the association
between the SSRC and the group. This is not possible using either of
the frameworks above.The W3C WebRTC API specification specifies that communication
between WebRTC entities is done via MediaStreams, which contain
MediaStreamTracks. A MediaStreamTrack is generally carried using a
single SSRC in an RTP session (forming an RTP media stream. The
collision of terminology is unfortunate.) There might possibly be
additional SSRCs, possibly within additional RTP sessions, in order to
support functionality like forward error correction or simulcast. This
complication is ignored below.In the RTP specification, media streams are identified using the
SSRC field. Streams are grouped into RTP Sessions, and also carry a
CNAME. Neither CNAME nor RTP session correspond to a MediaStream.
Therefore, the association of an RTP media stream to MediaStreams need
to be explicitly signaled.The marking needs to be on a per-SSRC basis, since one RTP session
can carry media from multiple MediaStreams, and one MediaStream can
have media in multiple RTP sessions. This means that the "label" attribute, which is used to label RTP
sessions, is not usable for this purpose.The marking needs to also carry the unique identifier of the RTP
media stream as a MediaStreamTrack within the media stream; this is
done using a single letter to identify whether it belongs in the video
or audio track list, and the MediaStreamTrack's position within that
array.This usage is described in .This document extends the Source-Specific Media Attributes framework
by adding a new "msid" attribute that can
be used with the "a=ssrc" SDP attribute. This new attribute allows
endpoints to associate RTP media streams that are carried in the same or
different RTP sessions, as well as allowing application-specific
information to the association.The value of the "msid" attribute consists of an identifier and
optional application-specific data, according to the following ABNF
grammar:An example MSID value for the SSRC 1234 might look like this:The identifier is a string of ASCII characters chosen from 0-9, a-z,
A-Z and - (hyphen), consisting of between 1 and 64 characters. It MUST
be unique among the identifier values used in the same SDP session. It
is RECOMMENDED that is generated using a random-number generator.Application data is carried on the same line as the identifier,
separated from the identifier by a space.The identifier uniquely identifies a group within the scope of an SDP
description.There may be multiple msid attributes on a single SSRC. There may
also be multiple SSRCs that have the same value for identifier and
application data.Endpoints can update the associations between SSRCs as expressed by
msid attributes at any time; the semantics and restrictions of such
grouping and ungrouping are application dependent.In order to fully reproduce the semantics of the SDP and SSRC
grouping frameworks, a session-level attribute is defined for signaling
the semantics associated with an msid grouping.This OPTIONAL attribute gives the group identifier and its group
semantic; it carries the same meaning as the ssrc-group-attr of RFC 5576
section 4.2, but uses the identifier of the group rather than a list of
SSRC values.An empty list of identifiers is an indication that the sender
understands the indicated semantic, but has no msid groupings of the
given type in the present SDP.The ABNF of msid-semantic is:The semantic field may hold values from the IANA registries
"Semantics for the "ssrc-group" SDP Attribute" and "Semantics for the
"group" SDP Attribute".An example msid-semantic might look like this:This means that the SDP description has two lip sync groups, with the
group identifiers xyzzy and forolow, respectively.This section creates a new semantic for use with the framework
defined in , to be used for
associating SSRCs representing MediaStreamTracks within MediaStreams as
defined in .The semantic token for this semantic is "WMS" (short for WebRTC Media
Stream).The value of the msid corresponds to the "id" attribute of a
MediaStream.In a WebRTC-compatible SDP description, all SSRCs intending to be
sent from one peer will be identified in the SDP generated by that
entity.The appdata for a WebRTC MediaStreamTrack consists of the "id"
attribute of a MediaStreamTrack.If two different SSRCs have the same value for identifier and
appdata, it means that these two SSRCs are both intended for the same
MediaStreamTrack. This may occur if the sender wishes to use simulcast
or forward error correction, or if the sender intends to switch between
multiple codecs on the same MediaStreamTrack.When an SDP description is updated, a specific msid continues to
refer to the same MediaStream. Once negotiation has completed on a
session, there is no memory; an msid value that appears in a later
negotiation will be taken to refer to a new MediaStream.The following are the rules for handling updates of the list of SSRCs
and their msid values.When a new msid value occurs in the description, the recipient
can signal to its application that a new MediaStream has been
added.When a description is updated to have more SSRCs with the same
msid value, but different appdata values, the recipient can signal
to its application that new media stream tracks have been added to
the media stream.When a description is updated to no longer list the msid value on
a specific ssrc, the recipient can signal to its application that
the corresponding media stream track has been closed.When a description is updated to no longer list the msid value on
any ssrc, the recipient can signal to its application that the media
stream has been closed.In addition to signaling that the track is closed when it
disappears from the SDP, the track will also be signaled as being closed
when the SSRC disappears by the rules of
section 6.3.4 (BYE packet received) and 6.3.5 (timeout).Pre-WebRTC entities will not send msid. This means that there will
be some incoming RTP packets with SSRCs where the recipient does not
know about a corresponding MediaStream id.Handling will depend on whether or not any SSRCs are signaled in
the relevant RTP session. There are two cases:No SSRC is signaled with an msid attribute. The SDP session is
assumed to be a backwards-compatible session. All incoming SSRCs,
on all RTP sessions that are part of the SDP session, are assumed
to belong to a single media stream. The identifier of this media
stream is "default".Some SSRCs are signaled with an msid attribute. In this case,
the session is WebRTC compatible, and the newly arrived SSRCs are
either caused by a bug or by timing skew between the arrival of
the media packets and the SDP description. These packets MAY be
discarded, or they MAY be buffered for a while in order to allow
immediate startup of the media stream when the SDP description is
updated. The arrival of media packets MUST NOT cause a new
MediaStreamTrack to be signaled.If a WebRTC entity sends a description, it MUST include the
msid-semantic: WMS attribute, even if no media streams are sent. This
allows us to distinguish between the case of no media streams at the
moment and the case of legacy SDP generation.It follows from the above that media stream tracks in the "default"
media stream cannot be closed by signaling; the application must
instead signal these as closed when the SSRC disappears according to
the rules of RFC 3550 section 6.3.4 and 6.3.5.This document requests IANA to register the "msid" attribute in the
"att-field (source level)" registry within the SDP parameters registry,
according to the procedures of The required information is:Contact name, email: IETF, contacted via rtcweb@ietf.org, or a
successor address designated by IESGAttribute name: msidLong-form attribute name: Media stream group IdentifierThe attribute value contains only ASCII characters, and is
therefore not subject to the charset attribute.The attribute gives an association over a set of SSRCs,
potentially in different RTP sessions. It can be used to signal the
relationship between a WebRTC MediaStream and a set of SSRCs.The details of appropriate values are given in RFC XXXX.This document requests IANA to create a new registry called
"Semantics for the msid-semantic SDP attribute", which should have
exactly the same rules as for the "Semantics for the ssrc-group SDP
attribute" registry, and to register the "WMS" semantic within this new
registry.The required information is:Description: WebRTC Media Stream, as given in RFC XXXX.Token: WMSStandards track reference: RFC XXXXIANA is requested to replace "RFC XXXX" with the RFC number of
this document upon publication.An adversary with the ability to modify SDP descriptions has the
ability to switch around tracks between media streams. This is a special
case of the general security consideration that modification of SDP
descriptions needs to be confined to entities trusted by the
application.No attacks that are relevant to the browser's security have been
identified that depend on this mechanism.This note is based on sketches from, among others, Justin Uberti and
Cullen Jennings.Special thanks to Miguel Garcia and Paul Kyzivat for their work in
reviewing this draft, with many specific language suggestions.This appendix should be deleted before publication as an RFC.One suggested mechanism has been to use CNAME instead of a new
attribute. This was abandoned because CNAME identifies a synchronization
context; one can imagine both wanting to have tracks from the same
synchronization context in multiple media streams and wanting to have
tracks from multiple synchronization contexts within one media
stream.Another suggestion has been to put the msid value within an attribute
of RTCP SR (sender report) packets. This doesn't offer the ability to
know that you have seen all the tracks currently configured for a media
stream.There has been a suggestion that this mechanism could be used to mute
tracks too. This is not done at the moment.An alternative to the "default" media stream is to let each new media
stream track without a msid attribute create its own media stream. Input
on this question is sought.Discarding of incoming data when the SDP description isn't updated
yet (section 3) may cause clipping. However, the same issue exists when
crypto keys aren't available. Input sought.There's been a suggestion that acceptable SSRCs should be signaled in
a response, giving a recipient the ability to say "no" to certain SSRCs.
This is not supported in the current version of this document.This appendix should be deleted before publication as an RFC.Added track identifier.Added inclusion-by-reference of
draft-lennox-mmusic-source-selection for track muting.Some rewording.Split document into sections describing a generic grouping
mechanism and sections describing the application of this grouping
mechanism to the WebRTC MediaStream concept.Removed the mechanism for muting tracks, since this is not central
to the MSID mechanism.Changed the draft name according to the wishes of the MMUSIC group
chairs.Added text indicting cases where it's appropriate to have the same
appdata for multiple SSRCs.Minor textual updates.Increased the amount of explanatory text, much based on a review by
Miguel Garcia.Removed references to BUNDLE, since that spec is under active
discussion.Removed distinguished values of the MSID identifier.Changed the order of the "msid-semantic: " attribute's value fields
and allowed multiple identifiers. This makes the attribute useful as a
marker for "I understand this semantic".Changed the syntax for "identifier" and "appdata" to be
"token".Changed the registry for the "msid-semantic" attribute values to be
a new registry, based on advice given in Atlanta.