RFC 8860




Internet Engineering Task Force (IETF)                     M. Westerlund
Request for Comments: 8860                                      Ericsson
Updates: 3550, 3551                                           C. Perkins
Category: Standards Track                          University of Glasgow
ISSN: 2070-1721                                                J. Lennox
                                                             8x8 / Jitsi
                                                            January 2021


        Sending Multiple Types of Media in a Single RTP Session

Abstract



   This document specifies how an RTP session can contain RTP streams
   with media from multiple media types such as audio, video, and text.
   This has been restricted by the RTP specifications (RFCs 3550 and
   3551), and thus this document updates RFCs 3550 and 3551 to enable
   this behaviour for applications that satisfy the applicability for
   using multiple media types in a single RTP session.

Status of This Memo



   This is an Internet Standards Track document.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Further information on
   Internet Standards is available in Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   https://www.rfc-editor.org/info/rfc8860.

Copyright Notice



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

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

Table of Contents



   1.  Introduction
   2.  Terminology
   3.  Background and Motivation
   4.  Applicability
   5.  Using Multiple Media Types in a Single RTP Session
     5.1.  Allowing Multiple Media Types in an RTP Session
     5.2.  Demultiplexing Media Types within an RTP Session
     5.3.  Per-SSRC Media Type Restrictions
     5.4.  RTCP Considerations
   6.  Extension Considerations
     6.1.  RTP Retransmission Payload Format
     6.2.  RTP Payload Format for Generic FEC
     6.3.  RTP Payload Format for Redundant Audio
   7.  Signalling
   8.  Security Considerations
   9.  IANA Considerations
   10. References
     10.1.  Normative References
     10.2.  Informative References
   Acknowledgements

   Authors' Addresses



1.  Introduction



   The Real-time Transport Protocol [RFC3550] was designed to use
   separate RTP sessions to transport different types of media.  This
   implies that different transport-layer flows are used for different
   RTP streams.  For example, a video conferencing application might
   send audio and video traffic RTP flows on separate UDP ports.  With
   increased use of network address/port translation, firewalls, and
   other middleboxes, it is, however, becoming difficult to establish
   multiple transport-layer flows between endpoints.  Hence, there is
   pressure to reduce the number of concurrent transport flows used by
   RTP applications.

   This memo updates [RFC3550] and [RFC3551] to allow multiple media
   types to be sent in a single RTP session in certain cases, thereby
   reducing the number of transport-layer flows that are needed.  It
   makes no changes to RTP behaviour when using multiple RTP streams
   containing media of the same type (e.g., multiple audio streams or
   multiple video streams) in a single RTP session.  However, [RFC8108]
   provides important clarifications to RTP behaviour in that case.

   This memo is structured as follows.  Section 2 defines terminology.
   Section 3 further describes the background to, and motivation for,
   this memo; Section 4 describes the scenarios where this memo is
   applicable.  Section 5 discusses issues arising from the base RTP and
   RTP Control Protocol (RTCP) specifications [RFC3550] [RFC3551] when
   using multiple types of media in a single RTP session, while
   Section 6 considers the impact of RTP extensions.  We discuss
   signalling in Section 7.  Finally, security considerations are
   discussed in Section 8.

2.  Terminology



   The terms "encoded stream", "endpoint", "media source", "RTP
   session", and "RTP stream" are used as defined in [RFC7656].  We also
   define the following terms:

   Media Type:  The general type of media data used by a real-time
      application.  The media type corresponds to the value used in the
      <media> field of a Session Description Protocol (SDP) "m=" line.
      The media types defined at the time of this writing are "audio",
      "video", "text", "image", "application", and "message" [RFC4566]
      [RFC6466].

   Quality of Service (QoS):  Network mechanisms that are intended to
      ensure that the packets within a flow or with a specific marking
      are transported with certain properties.

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

3.  Background and Motivation



   RTP was designed to support multimedia sessions, containing multiple
   types of media sent simultaneously, by using multiple transport-layer
   flows.  The existence of network address translators, firewalls, and
   other middleboxes complicates this, however, since a mechanism is
   needed to ensure that all the transport-layer flows needed by the
   application can be established.  This has three consequences:

   1.  increased delay to establish a complete session, since each of
       the transport-layer flows needs to be negotiated and established;

   2.  increased state and resource consumption in the middleboxes that
       can lead to unexpected behaviour when middlebox resource limits
       are reached; and

   3.  increased risk that a subset of the transport-layer flows will
       fail to be established, thus preventing the application from
       communicating.

   Using fewer transport-layer flows can hence be seen to reduce the
   risk of communication failure and can lead to improved reliability
   and performance.

   One of the benefits of using multiple transport-layer flows is that
   it makes it easy to use network-layer QoS mechanisms to give
   differentiated performance for different flows.  However, we note
   that many applications that use RTP don't use network QoS features
   and don't expect or desire any separation in network treatment of
   their media packets, independent of whether they are audio, video, or
   text.  When an application has no such desire, it doesn't need to
   provide a transport flow structure that simplifies flow-based QoS.

   Given the above issues, it might seem appropriate for RTP-based
   applications to send all their RTP streams bundled into one RTP
   session, running over a single transport-layer flow.  However, this
   is prohibited by the RTP specifications [RFC3550] [RFC3551], because
   the design of RTP makes certain assumptions that can be incompatible
   with sending multiple media types in a single RTP session.
   Specifically, the RTCP timing rules assume that all RTP media flows
   in a single RTP session have broadly similar RTCP reporting and
   feedback requirements, which can be problematic when different types
   of media are multiplexed together.  Various RTP extensions also make
   assumptions about Synchronisation Source (SSRC) use and RTCP
   reporting that are incompatible with sending different media types in
   a single RTP session.

   This memo updates [RFC3550] and [RFC3551] to allow RTP sessions to
   contain more than one media type in certain circumstances and gives
   guidance on when it is safe to send multiple media types in a single
   RTP session.

4.  Applicability



   This specification has limited applicability, and anyone intending to
   use it needs to ensure that their application and use case meet the
   following criteria:

   Equal treatment of media:  The use of a single RTP session normally
      results in similar network treatment for all types of media used
      within the session.  Applications that require significantly
      different network QoS or RTCP configuration for different RTP
      streams are better suited to sending those RTP streams in separate
      RTP sessions, using separate transport-layer flows for each, since
      that method provides greater flexibility.  Further guidance on how
      to provide differential treatment for some media streams is given
      in [RFC8872] and [RFC7657].

   Compatible RTCP behaviour:  The RTCP timing rules enforce a single
      RTCP reporting interval for all participants in an RTP session.
      Flows with very different media sending rates or RTCP feedback
      requirements cannot be multiplexed together, since this leads to
      either excessive or insufficient RTCP for some flows, depending on
      how the RTCP session bandwidth, and hence the reporting interval,
      are configured.  For example, it is likely infeasible to find a
      single RTCP configuration that simultaneously suits both a low-
      rate audio flow with no feedback and a high-quality video flow
      with sophisticated RTCP-based feedback.  Thus, combining these
      into a single RTP session is difficult and/or inadvisable.

   Signalled support:  The extensions defined in this memo are not
      compatible with unmodified endpoints that are compatible with
      [RFC3550].  Their use requires signalling and mutual agreement by
      all participants within an RTP session.  This requirement can be a
      problem for signalling solutions that can't negotiate with all
      participants.  For declarative signalling solutions, mandating
      that the session use multiple media types in one RTP session can
      be a way of attempting to ensure that all participants in the RTP
      session follow the requirement.  However, for signalling solutions
      that lack methods for enforcing a requirement that a receiver
      support a specific feature, this can still cause issues.

   Consistent support for multiparty RTP sessions:  If it is desired to
      send multiple types of media in a multiparty RTP session, then all
      participants in that session need to support sending multiple
      types of media in a single RTP session.  It is not possible, in
      the general case, to implement a gateway that can interconnect an
      endpoint that uses multiple types of media sent using separate RTP
      sessions with one or more endpoints that send multiple types of
      media in a single RTP session.

      One reason for this is that the same SSRC value can safely be used
      for different streams in multiple RTP sessions, but when collapsed
      to a single RTP session there is an SSRC collision.  This would
      not be an issue, since SSRC collision detection will resolve the
      conflict, except that some RTP payload formats and extensions use
      matching SSRCs to identify related flows and will break when a
      single RTP session is used.

      A middlebox that remaps SSRC values when combining multiple RTP
      sessions into one also needs to be aware of all possible RTCP
      packet types that might be used, so that it can remap the SSRC
      values in those packets.  This is impossible to do without
      restricting the set of RTCP packet types that can be used to those
      that are known by the middlebox.  Such a middlebox might also have
      difficulty due to differences in configured RTCP bandwidth and
      other parameters between the RTP sessions.

      Finally, the use of a middlebox that translates SSRC values can
      negatively impact the possibility of loop detection, as SSRC/CSRC
      (Contributing Source) can't be used to detect the loops; instead,
      some other RTP stream or media source identity namespace that is
      common across all interconnected parts is needed.

   Ability to operate with limited payload type space:  An RTP session
      has only a single 7-bit payload type space for all its payload
      type numbers.  Some applications might find this space to be
      limiting (i.e., overly restrictive) when using different media
      types and RTP payload formats within a single RTP session.

   Avoidance of incompatible extensions:  Some RTP and RTCP extensions
      rely on the existence of multiple RTP sessions and relate RTP
      streams between sessions.  Others report on particular media types
      and cannot be used with other media types.  Applications that send
      multiple types of media into a single RTP session need to avoid
      such extensions.

5.  Using Multiple Media Types in a Single RTP Session



   This section defines what needs to be done or avoided to make an RTP
   session with multiple media types function without issues.

5.1.  Allowing Multiple Media Types in an RTP Session



   Section 5.2 of "RTP: A Transport Protocol for Real-Time Applications"
   [RFC3550] states:

   |  For example, in a teleconference composed of audio and video media
   |  encoded separately, each medium SHOULD be carried in a separate
   |  RTP session with its own destination transport address.
   |
   |  Separate audio and video streams SHOULD NOT be carried in a single
   |  RTP session and demultiplexed based on the payload type or SSRC
   |  fields.

   This specification changes both of these sentences.  The first
   sentence is changed to:

   |  For example, in a teleconference composed of audio and video media
   |  encoded separately, each medium SHOULD be carried in a separate
   |  RTP session with its own destination transport address, unless the
   |  guidelines specified in [RFC8860] are followed and the application
   |  meets the applicability constraints.

   The second sentence is changed to:

   |  Separate audio and video media sources SHOULD NOT be carried in a
   |  single RTP session, unless the guidelines specified in [RFC8860]
   |  are followed.

   The second paragraph of Section 6 of "RTP Profile for Audio and Video
   Conferences with Minimal Control" [RFC3551] says:

   |  The payload types currently defined in this profile are assigned
   |  to exactly one of three categories or media types: audio only,
   |  video only and those combining audio and video.  The media types
   |  are marked in Tables 4 and 5 as "A", "V" and "AV", respectively.
   |  Payload types of different media types SHALL NOT be interleaved or
   |  multiplexed within a single RTP session, but multiple RTP sessions
   |  MAY be used in parallel to send multiple media types.  An RTP
   |  source MAY change payload types within the same media type during
   |  a session.  See the section "Multiplexing RTP Sessions" of RFC
   |  3550 for additional explanation.

   This specification's purpose is to override the above-listed
   "SHALL NOT" under certain conditions.  Thus, this sentence also has
   to be changed to allow for multiple media types' payload types in the
   same session.  The sentence containing "SHALL NOT" in the above
   paragraph is changed to:

   |  Payload types of different media types SHALL NOT be interleaved or
   |  multiplexed within a single RTP session unless [RFC8860] is used
   |  and the application conforms to the applicability constraints.
   |  Multiple RTP sessions MAY be used in parallel to send multiple
   |  media types.

5.2.  Demultiplexing Media Types within an RTP Session



   When receiving packets from a transport-layer flow, an endpoint will
   first separate the RTP and RTCP packets from the non-RTP packets and
   pass them to the RTP/RTCP protocol handler.  The RTP and RTCP packets
   are then demultiplexed into the different RTP streams based on their
   SSRC.  For each RTP stream, incoming RTCP packets are processed, and
   the RTP payload type is used to select the appropriate media decoder.
   This process remains the same irrespective of whether multiple media
   types are sent in a single RTP session or not.

   As explained below, it is important to note that the RTP payload type
   is never used to distinguish RTP streams.  The RTP packets are
   demultiplexed into RTP streams based on their SSRC; the RTP payload
   type is then used to select the correct media-decoding pathway for
   each RTP stream.

5.3.  Per-SSRC Media Type Restrictions



   An SSRC in an RTP session can change between media formats of the
   same type, subject to certain restrictions [RFC7160], but MUST NOT
   change its media type during its lifetime.  For example, an SSRC can
   change between different audio formats, but it cannot start sending
   audio and then change to sending video.  The lifetime of an SSRC ends
   when an RTCP BYE packet for that SSRC is sent or when it ceases
   transmission for long enough that it times out for the other
   participants in the session.

   The main motivation is that a given SSRC has its own RTP timestamp
   and sequence number spaces.  The same way that you can't send two
   encoded streams of audio with the same SSRC, you can't send one
   encoded audio and one encoded video stream with the same SSRC.  Each
   encoded stream, when made into an RTP stream, needs to have sole
   control over the sequence number and timestamp space.  If not, one
   would not be able to detect packet loss for that particular encoded
   stream, nor could one easily determine which clock rate a particular
   SSRC's timestamp will increase with.  For additional arguments
   regarding why multiplexing of multiple media sources that is based on
   RTP payload type doesn't work, see [RFC8872].

   Within an RTP session where multiple media types have been configured
   for use, an SSRC can only send one type of media during its lifetime
   (i.e., it can switch between different audio codecs, since those are
   both the same type of media, but it cannot switch between audio and
   video).  Different SSRCs MUST be used for the different media
   sources, the same way multiple media sources of the same media type
   already have to do.  The payload type will inform a receiver which
   media type the SSRC is being used for.  Thus, the payload type MUST
   be unique across all of the payload configurations, independent of
   the media type that is used in the RTP session.

5.4.  RTCP Considerations



   When sending multiple types of media that have different rates in a
   single RTP session, endpoints MUST follow the guidelines for handling
   RTCP as provided in Section 7 of [RFC8108].

6.  Extension Considerations



   This section outlines known issues and incompatibilities with RTP and
   RTCP extensions when multiple media types are used in a single RTP
   session.  Future extensions to RTP and RTCP need to consider, and
   document, any potential incompatibilities.

6.1.  RTP Retransmission Payload Format



   The RTP retransmission payload format [RFC4588] can operate in either
   SSRC-multiplexed mode or session-multiplexed mode.

   In SSRC-multiplexed mode, retransmitted RTP packets are sent in the
   same RTP session as the original packets but use a different SSRC
   with the same RTCP Source Description (SDES) CNAME.  If each endpoint
   sends only a single original RTP stream and a single retransmission
   RTP stream in the session, this is sufficient.  If an endpoint sends
   multiple original and retransmission RTP streams, as would occur when
   sending multiple media types in a single RTP session, then each
   original RTP stream and the retransmission RTP stream have to be
   associated using heuristics.  By having retransmission requests
   outstanding for only one SSRC not yet mapped, a receiver can
   determine the binding between the original and retransmission RTP
   streams.  Another alternative is the use of different RTP payload
   types, allowing the signalled "apt" (associated payload type)
   parameter [RFC4588] of the RTP retransmission payload format to be
   used to associate retransmitted and original packets.

   Session-multiplexed mode sends the retransmission RTP stream in a
   separate RTP session to the original RTP stream, but using the same
   SSRC for each, with the association being done by matching SSRCs
   between the two sessions.  This is unaffected by the use of multiple
   media types in a single RTP session, since each media type will be
   sent using a different SSRC in the original RTP session, and the same
   SSRCs can be used in the retransmission session, allowing the streams
   to be associated.  This can be signalled using SDP with the BUNDLE
   grouping extension [RFC8843] and the Flow Identification (FID)
   grouping extension [RFC5888].  These SDP extensions require each "m="
   line to only be included in a single FID group, but the RTP
   retransmission payload format uses FID groups to indicate the "m="
   lines that form an original and retransmission pair.  Accordingly,
   when using the BUNDLE extension to allow multiple media types to be
   sent in a single RTP session, each original media source ("m=" line)
   that is retransmitted needs a corresponding "m=" line in the
   retransmission RTP session.  If there are multiple media lines for
   retransmission, these media lines will form an independent BUNDLE
   group from the BUNDLE group with the source streams.

   An example SDP fragment showing the grouping structures is provided
   in Figure 1.  This example is not legal SDP, and only the most
   important attributes have been left in place.  Note that this SDP is
   not an initial BUNDLE offer.  As can be seen in this example, there
   are two bundle groups -- one for the source RTP session and one for
   the retransmissions.  Then, each of the media sources is grouped with
   its retransmission flow using FID, resulting in three more groupings.

          a=group:BUNDLE foo bar fiz
          a=group:BUNDLE zoo kelp glo
          a=group:FID foo zoo
          a=group:FID bar kelp
          a=group:FID fiz glo
          m=audio 10000 RTP/AVP 0
          a=mid:foo
          a=rtpmap:0 PCMU/8000
          m=video 10000 RTP/AVP 31
          a=mid:bar
          a=rtpmap:31 H261/90000
          m=video 10000 RTP/AVP 31
          a=mid:fiz
          a=rtpmap:31 H261/90000
          m=audio 40000 RTP/AVPF 99
          a=rtpmap:99 rtx/90000
          a=fmtp:99 apt=0;rtx-time=3000
          a=mid:zoo
          m=video 40000 RTP/AVPF 100
          a=rtpmap:100 rtx/90000
          a=fmtp:199 apt=31;rtx-time=3000
          a=mid:kelp
          m=video 40000 RTP/AVPF 100
          a=rtpmap:100 rtx/90000
          a=fmtp:199 apt=31;rtx-time=3000
          a=mid:glo

      Figure 1: SDP Example of Session-Multiplexed RTP Retransmission

6.2.  RTP Payload Format for Generic FEC



   The RTP payload format for generic Forward Error Correction (FEC), as
   defined in [RFC5109] (and its predecessor, [RFC2733]), can either
   send the FEC stream as a separate RTP stream or send the FEC combined
   with the original RTP stream as a redundant encoding [RFC2198].

   When sending FEC as a separate stream, the RTP payload format for
   generic FEC requires that FEC stream to be sent in a separate RTP
   session to the original stream, using the same SSRC, with the FEC
   stream being associated by matching the SSRC between sessions.  The
   RTP session used for the original streams can include multiple RTP
   streams, and those RTP streams can use multiple media types.  The
   repair session only needs one RTP payload type to indicate FEC data,
   irrespective of the number of FEC streams sent, since the SSRC is
   used to associate the FEC streams with the original streams.  Hence,
   it is RECOMMENDED that the FEC stream use the "application/ulpfec"
   media type in the case of support for [RFC5109] and the
   "application/parityfec" media type in the case of support for
   [RFC2733].  It is legal, but NOT RECOMMENDED, to send FEC streams
   using media-specific payload format names (e.g., using both the
   "audio/ulpfec" and "video/ulpfec" payload formats for a single RTP
   session containing both audio and video flows), since this
   (1) unnecessarily uses up RTP payload type values and (2) adds no
   value for demultiplexing because there might be multiple streams of
   the same media type).

   The combination of an original RTP session using multiple media types
   with an associated generic FEC session can be signalled using SDP
   with the BUNDLE extension [RFC8843].  In this case, the RTP session
   carrying the FEC streams will be its own BUNDLE group.  The "m=" line
   for each original stream and the "m=" line for the corresponding FEC
   stream are grouped using the SDP Grouping Framework, using either the
   FEC-FR grouping [RFC5956] or, for backwards compatibility, the FEC
   grouping [RFC4756].  This is similar to the situation that arises for
   RTP retransmission with session-based multiplexing as discussed in
   Section 6.1.

   The source-specific media attributes specification [RFC5576] defines
   an SDP extension (the "FEC" semantic of the "ssrc-group" attribute)
   to signal FEC relationships between multiple RTP streams within a
   single RTP session.  This cannot be used with generic FEC, since the
   FEC repair packets need to have the same SSRC value as the source
   packets being protected.  There existed a proposal (now abandoned)
   for an Uneven Level Protection (ULP) extension to enable transmission
   of the FEC RTP streams within the same RTP session as the source
   stream [FEC-Src-Multiplexing].

   When the FEC is sent as a redundant encoding, the considerations in
   Section 6.3 apply.

6.3.  RTP Payload Format for Redundant Audio



   The RTP payload format for redundant audio [RFC2198] can be used to
   protect audio streams.  It can also be used along with the generic
   FEC payload format to send original and repair data in the same RTP
   packets.  Both are compatible with RTP sessions containing multiple
   media types.

   This payload format requires each different redundant encoding to use
   a different RTP payload type number.  When used with generic FEC in
   sessions that contain multiple media types, this requires each media
   type to use a different payload type for the FEC stream.  For
   example, if audio and text are sent in a single RTP session with
   generic ULP FEC sent as a redundant encoding for each, then payload
   types need to be assigned for FEC using the audio/ulpfec and
   text/ulpfec payload formats.  If multiple original payload types are
   used in the session, different redundant payload types need to be
   allocated for each one.  This has potential to rapidly exhaust the
   available RTP payload type numbers.

7.  Signalling



   Establishing a single RTP session using multiple media types requires
   signalling.  This signalling has to:

   1.  ensure that any participant in the RTP session is aware that this
       is an RTP session with multiple media types;

   2.  ensure that the payload types in use in the RTP session are using
       unique values, with no overlap between the media types;

   3.  ensure that RTP session-level parameters -- for example, the RTCP
       RR and RS bandwidth modifiers [RFC3556], the RTP/AVPF trr-int
       parameter [RFC4585], transport protocol, RTCP extensions in use,
       and any security parameters -- are consistent across the session;
       and

   4.  ensure that RTP and RTCP functions that can be bound to a
       particular media type are reused where possible, rather than
       configuring multiple code points for the same thing.

   When using SDP signalling, the BUNDLE extension [RFC8843] is used to
   signal RTP sessions containing multiple media types.

8.  Security Considerations



   RTP provides a range of strong security mechanisms that can be used
   to secure sessions [RFC7201] [RFC7202].  The majority of these are
   independent of the type of media sent in the RTP session; however, it
   is important to check that the security mechanism chosen is
   compatible with all types of media sent within the session.

   Sending multiple media types in a single RTP session will generally
   require that all use the same security mechanism, whereas media sent
   using different RTP sessions can be secured in different ways.  When
   different media types have different security requirements, it might
   be necessary to send them using separate RTP sessions to meet those
   different requirements.  This can have significant costs in terms of
   resource usage, session setup time, etc.

9.  IANA Considerations



   This document has no IANA actions.

10.  References



10.1.  Normative References



   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <https://www.rfc-editor.org/info/rfc2119>.

   [RFC3550]  Schulzrinne, H., Casner, S., Frederick, R., and V.
              Jacobson, "RTP: A Transport Protocol for Real-Time
              Applications", STD 64, RFC 3550, DOI 10.17487/RFC3550,
              July 2003, <https://www.rfc-editor.org/info/rfc3550>.

   [RFC3551]  Schulzrinne, H. and S. Casner, "RTP Profile for Audio and
              Video Conferences with Minimal Control", STD 65, RFC 3551,
              DOI 10.17487/RFC3551, July 2003,
              <https://www.rfc-editor.org/info/rfc3551>.

   [RFC8108]  Lennox, J., Westerlund, M., Wu, Q., and C. Perkins,
              "Sending Multiple RTP Streams in a Single RTP Session",
              RFC 8108, DOI 10.17487/RFC8108, March 2017,
              <https://www.rfc-editor.org/info/rfc8108>.

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

   [RFC8843]  Holmberg, C., Alvestrand, H., and C. Jennings,
              "Negotiating Media Multiplexing Using the Session
              Description Protocol (SDP)", RFC 8843,
              DOI 10.17487/RFC8843, January 2021,
              <https://www.rfc-editor.org/info/rfc8843>.

10.2.  Informative References



   [FEC-Src-Multiplexing]
              Lennox, J., "Supporting Source-Multiplexing of the Real-
              Time Transport Protocol (RTP) Payload for Generic Forward
              Error Correction", Work in Progress, Internet-Draft,
              draft-lennox-payload-ulp-ssrc-mux-00, 18 February 2013,
              <https://tools.ietf.org/html/draft-lennox-payload-ulp-
              ssrc-mux-00>.

   [RFC2198]  Perkins, C., Kouvelas, I., Hodson, O., Hardman, V.,
              Handley, M., Bolot, J.C., Vega-Garcia, A., and S. Fosse-
              Parisis, "RTP Payload for Redundant Audio Data", RFC 2198,
              DOI 10.17487/RFC2198, September 1997,
              <https://www.rfc-editor.org/info/rfc2198>.

   [RFC2733]  Rosenberg, J. and H. Schulzrinne, "An RTP Payload Format
              for Generic Forward Error Correction", RFC 2733,
              DOI 10.17487/RFC2733, December 1999,
              <https://www.rfc-editor.org/info/rfc2733>.

   [RFC3556]  Casner, S., "Session Description Protocol (SDP) Bandwidth
              Modifiers for RTP Control Protocol (RTCP) Bandwidth",
              RFC 3556, DOI 10.17487/RFC3556, July 2003,
              <https://www.rfc-editor.org/info/rfc3556>.

   [RFC4566]  Handley, M., Jacobson, V., and C. Perkins, "SDP: Session
              Description Protocol", RFC 4566, DOI 10.17487/RFC4566,
              July 2006, <https://www.rfc-editor.org/info/rfc4566>.

   [RFC4585]  Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey,
              "Extended RTP Profile for Real-time Transport Control
              Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585,
              DOI 10.17487/RFC4585, July 2006,
              <https://www.rfc-editor.org/info/rfc4585>.

   [RFC4588]  Rey, J., Leon, D., Miyazaki, A., Varsa, V., and R.
              Hakenberg, "RTP Retransmission Payload Format", RFC 4588,
              DOI 10.17487/RFC4588, July 2006,
              <https://www.rfc-editor.org/info/rfc4588>.

   [RFC4756]  Li, A., "Forward Error Correction Grouping Semantics in
              Session Description Protocol", RFC 4756,
              DOI 10.17487/RFC4756, November 2006,
              <https://www.rfc-editor.org/info/rfc4756>.

   [RFC5109]  Li, A., Ed., "RTP Payload Format for Generic Forward Error
              Correction", RFC 5109, DOI 10.17487/RFC5109, December
              2007, <https://www.rfc-editor.org/info/rfc5109>.

   [RFC5576]  Lennox, J., Ott, J., and T. Schierl, "Source-Specific
              Media Attributes in the Session Description Protocol
              (SDP)", RFC 5576, DOI 10.17487/RFC5576, June 2009,
              <https://www.rfc-editor.org/info/rfc5576>.

   [RFC5888]  Camarillo, G. and H. Schulzrinne, "The Session Description
              Protocol (SDP) Grouping Framework", RFC 5888,
              DOI 10.17487/RFC5888, June 2010,
              <https://www.rfc-editor.org/info/rfc5888>.

   [RFC5956]  Begen, A., "Forward Error Correction Grouping Semantics in
              the Session Description Protocol", RFC 5956,
              DOI 10.17487/RFC5956, September 2010,
              <https://www.rfc-editor.org/info/rfc5956>.

   [RFC6466]  Salgueiro, G., "IANA Registration of the 'image' Media
              Type for the Session Description Protocol (SDP)",
              RFC 6466, DOI 10.17487/RFC6466, December 2011,
              <https://www.rfc-editor.org/info/rfc6466>.

   [RFC7160]  Petit-Huguenin, M. and G. Zorn, Ed., "Support for Multiple
              Clock Rates in an RTP Session", RFC 7160,
              DOI 10.17487/RFC7160, April 2014,
              <https://www.rfc-editor.org/info/rfc7160>.

   [RFC7201]  Westerlund, M. and C. Perkins, "Options for Securing RTP
              Sessions", RFC 7201, DOI 10.17487/RFC7201, April 2014,
              <https://www.rfc-editor.org/info/rfc7201>.

   [RFC7202]  Perkins, C. and M. Westerlund, "Securing the RTP
              Framework: Why RTP Does Not Mandate a Single Media
              Security Solution", RFC 7202, DOI 10.17487/RFC7202, April
              2014, <https://www.rfc-editor.org/info/rfc7202>.

   [RFC7656]  Lennox, J., Gross, K., Nandakumar, S., Salgueiro, G., and
              B. Burman, Ed., "A Taxonomy of Semantics and Mechanisms
              for Real-Time Transport Protocol (RTP) Sources", RFC 7656,
              DOI 10.17487/RFC7656, November 2015,
              <https://www.rfc-editor.org/info/rfc7656>.

   [RFC7657]  Black, D., Ed. and P. Jones, "Differentiated Services
              (Diffserv) and Real-Time Communication", RFC 7657,
              DOI 10.17487/RFC7657, November 2015,
              <https://www.rfc-editor.org/info/rfc7657>.

   [RFC8872]  Westerlund, M., Burman, B., Perkins, C., Alvestrand, H.,
              and R. Even, "Guidelines for Using the Multiplexing
              Features of RTP to Support Multiple Media Streams",
              RFC 8872, DOI 10.17487/RFC8872, January 2021,
              <https://www.rfc-editor.org/info/rfc8872>.

Acknowledgements



   The authors would like to thank Christer Holmberg, Gunnar Hellström,
   Charles Eckel, Tolga Asveren, Warren Kumari, and Meral Shirazipour
   for their feedback on this document.

Authors' Addresses



   Magnus Westerlund
   Ericsson
   Torshamnsgatan 23
   SE-164 80 Stockholm
   Sweden

   Email: magnus.westerlund@ericsson.com


   Colin Perkins
   University of Glasgow
   School of Computing Science
   Glasgow
   G12 8QQ
   United Kingdom

   Email: csp@csperkins.org


   Jonathan Lennox
   8x8, Inc. / Jitsi
   Jersey City, NJ 07302
   United States of America

   Email: jonathan.lennox@8x8.com