One document matched: draft-ietf-avtcore-rtp-topologies-update-01.xml


<?xml version="1.0" encoding="US-ASCII"?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<?rfc toc="yes"?>
<?rfc tocompact="yes"?>
<?rfc tocdepth="3"?>
<?rfc tocindent="yes"?>
<?rfc symrefs="yes"?>
<?rfc sortrefs="yes"?>
<?rfc comments="yes"?>
<?rfc inline="yes"?>
<?rfc compact="yes"?>
<?rfc subcompact="no"?>
<rfc category="info" docName="draft-ietf-avtcore-rtp-topologies-update-01"
     ipr="trust200902" obsoletes="5117">
  <front>
    <title abbrev="RTP Topologies">RTP Topologies</title>

    <author fullname="Magnus Westerlund" initials="M." surname="Westerlund">
      <organization>Ericsson</organization>

      <address>
        <postal>
          <street>Farogatan 6</street>

          <city>SE-164 80 Kista</city>

          <country>Sweden</country>
        </postal>

        <phone>+46 10 714 82 87</phone>

        <email>magnus.westerlund@ericsson.com</email>
      </address>
    </author>

    <author fullname="Stephan Wenger" initials="S." surname="Wenger">
      <organization>Vidyo</organization>

      <address>
        <postal>
          <street>433 Hackensack Ave</street>

          <city>Hackensack</city>

          <region>NJ</region>

          <code>07601</code>

          <country>USA</country>
        </postal>

        <email>stewe@stewe.org</email>
      </address>
    </author>

    <date/>

    <abstract>
      <t>This document discusses point to point and multi-endpoint topologies
      used in Real-time Transport Protocol (RTP)-based environments. In
      particular, centralized topologies commonly employed in the video
      conferencing industry are mapped to the RTP terminology.</t>

      <t>This document is updated with additional topologies and is intended
      to replace RFC 5117.</t>
    </abstract>
  </front>

  <middle>
    <section title="Introduction">
      <t><xref target="RFC3550">Real-time Transport Protocol (RTP)</xref>
      topologies describe methods for interconnecting RTP entities and their
      processing behavior of RTP and RTCP. This document tries to address past
      and existing confusion, especially with respect to terms not defined in
      RTP but in common use in the conversational communication industry, such
      as the Multipoint Control Unit or MCU.</t>

      <t>When the <xref target="RFC4585">Audio-Visual Profile with Feedback
      (AVPF)</xref> was developed the main emphasis lay in the efficient
      support of point to point and small multipoint scenarios without
      centralized multipoint control. In practice, however, most multipoint
      conferences operate utilizing centralized units referred to as MCUs.
      MCUs may implement Mixer or Translator functionality (in <xref
      target="RFC3550">RTP</xref> terminology), and signalling support. They
      may also contain additional application layer functionality. This
      document focuses on the media transport aspects of the MCU that can be
      realized using RTP, as discussed below. Further considered are the
      properties of Mixers and Translators, and how some types of deployed
      MCUs deviate from these properties.</t>

      <t>This document also codifies new multipoint architectures that have
      recently been introduced and which were not anticipated in RFC 5117.
      These architectures use scalable video coding and simulcasting, and
      their associated centralized units are referred to as Selective
      Forwarding Units (SFU). This codification provides a common information
      basis for future discussion and specification work.</t>

      <t>The document's attempt to clarify and explain sections of the <xref
      target="RFC3550">Real-time Transport Protocol (RTP) spec</xref> is
      informal. It is not intended to update or change what is normatively
      specified within RFC 3550.</t>
    </section>

    <section title="Definitions">
      <t/>

      <section title="Glossary">
        <t><list style="hanging">
            <t hangText="ASM:">Any Source Multicast</t>

            <t hangText="AVPF:">The Extended RTP Profile for RTCP-based
            Feedback</t>

            <t hangText="CSRC:">Contributing Source</t>

            <t hangText="Link:">The data transport to the next IP hop<!--MW: IS IP hop clear? Is the definition needed?--></t>

            <t hangText="Middlebox:">A device that is on the Path that media
            travel between two Endpoints</t>

            <t hangText="MCU:">Multipoint Control Unit</t>

            <t hangText="Path:">The concatenation of multiple links, resulting
            in an end-to-end data transfer.</t>

            <t hangText="PtM:">Point to Multipoint</t>

            <t hangText="PtP:">Point to Point</t>

            <t hangText="SFU:">Selective Forwarding Unit</t>

            <t hangText="SSM:">Source-Specific Multicast</t>

            <t hangText="SSRC:">Synchronization Source</t>
          </list></t>
      </section>
    </section>

    <section anchor="sec-topologies" title="Topologies">
      <t>This subsection defines several topologies that are relevant for
      codec control but also RTP usage in other contexts. The section starts
      with point to point cases, with or without middleboxes. Then follows a
      number of different methods for establishing point to multipoint
      communication. These are structured around the most fundamental enabler,
      i.e., multicast, a mesh of connections, translators, mixers and finally
      MCUs and SFUs. The section ends by discussing de-composited endpoints,
      asymmetric middlebox behaviors and combining topologies.</t>

      <t>The topologies may be referenced in other documents by a shortcut
      name, indicated by the prefix "Topo-".</t>

      <t>For each of the RTP-defined topologies, we discuss how RTP, RTCP, and
      the carried media are handled. With respect to RTCP, we also discuss the
      handling of RTCP feedback messages as defined in <xref
      target="RFC4585"/> and <xref target="RFC5104"/>.</t>

      <section title="Point to Point">
        <t>Shortcut name: Topo-Point-to-Point</t>

        <t>The <xref target="fig-point-to-point">Point to Point (PtP)
        topology</xref> consists of two endpoints, communicating using
        unicast. Both RTP and RTCP traffic are conveyed endpoint-to-endpoint,
        using unicast traffic only (even if, in exotic cases, this unicast
        traffic happens to be conveyed over an IP-multicast address).</t>

        <figure align="center" anchor="fig-point-to-point"
                title="Point to Point">
          <artwork><![CDATA[
+---+         +---+
| A |<------->| B |
+---+         +---+
]]></artwork>
        </figure>

        <t>The main property of this topology is that A sends to B, and only
        B, while B sends to A, and only A. This avoids all complexities of
        handling multiple endpoints and combining the requirements stemming
        from them. Note that an endpoint can still use multiple RTP
        Synchronization Sources (SSRCs) in an RTP session. The number of RTP
        sessions in use between A and B can also be of any number, subject
        only to system level limitations like the number range of ports.</t>

        <t>RTCP feedback messages for the indicated SSRCs are communicated
        directly between the endpoints. Therefore, this topology poses minimal
        (if any) issues for any feedback messages. For RTP sessions which use
        multiple SSRC per endpoint it can be relevant to implement support for
        cross-reporting suppression as defined in <xref
        target="I-D.ietf-avtcore-rtp-multi-stream">"Sending Multiple Media
        Streams in a Single RTP Session"</xref>.</t>
      </section>

      <section title="Point to Point via Middlebox">
        <t>This section discusses cases where two endpoints communicate but
        have one or more middleboxes involved in the RTP session.</t>

        <section anchor="sec-ptp-translators" title="Translators">
          <t>Shortcut name: Topo-PtP-Translator</t>

          <t>Two main categories of Translators can be distinguished;
          Transport Translators and Media translators. Both Translator types
          share common attributes that separate them from Mixers. For each
          media stream that the Translator receives, it generates an
          individual stream in the other domain. A translator keeps the SSRC
          for a stream across the translation, whereas a Mixer can select a
          single media stream, or send out multiple mixed media streams, but
          always under its own SSRC, possibly using the CSRC field to indicate
          the source(s) of the content. Mixers are more common in point to
          multipoint cases than in PtP. The reason is that in PtP use cases
          the primary focus is interoperability, such as transcoding to a
          codec the receiver supports, which can be done by a media
          translator.</t>

          <t>As specified in Section 7.1 of <xref target="RFC3550"/>, the SSRC
          space is common for all participants in the RTP session, independent
          of on which side of the Translator the session resides. Therefore,
          it is the responsibility of the participants to run SSRC collision
          detection, and the SSRC is thus a field the Translator cannot
          change. Any SDES information associated with a SSRC or CSRC also
          needs to be forwarded between the domains for any SSRC/CSRC used in
          the different domains.</t>

          <t>A Translator commonly does not use an SSRC of its own, and is not
          visible as an active participant in the session. One reason to have
          its own SSRC is when a Translator acts as a quality monitor that
          sends RTCP reports and therefore is required to have an SSRC.
          Another example is the case when a Translator is prepared to use
          RTCP feedback messages. This may, for example, occur in a translator
          configured to detect packet loss of important video packets and
          wants to trigger repair by the media sender, by sending feedback
          messages. While such feedback could use the SSRC of the target for
          the translator, this in turn would require translation of the
          targets RTCP reports to make them consistent. It may be simpler to
          expose an additional SSRC in the session. The only concern is
          endpoints failing to support the full RTP specification, thus having
          issues with multiple SSRCs reporting on the RTP streams sent by that
          endpoint.</t>

          <t>In general, a Translator implementation should consider which
          RTCP feedback messages or codec-control messages it needs to
          understand in relation to the functionality of the Translator
          itself. This is completely in line with the requirement to also
          translate RTCP messages between the domains.</t>

          <section anchor="sec-transport-anchor"
                   title="Transport Relay/Anchoring">
            <t>There exist a number of different types of middleboxes that
            might be inserted between two RTP endpoints on the transport
            level, e.g., to perform changes on the IP/UDP headers, and are,
            therefore, basic transport translators. These middleboxes come in
            many variations including <xref target="RFC3022">NAT</xref>
            traversal by pinning the media path to a public address domain
            relay, network topologies where the media flow is required to pass
            a particular point for audit by employing relaying, or preserving
            privacy by hiding each peer's transport addresses to the other
            party. Other protocols or functionalities that provide this
            behavior are <xref target="RFC5766">TURN</xref> servers, Session
            Border Gateways and Media Processing Nodes with media anchoring
            functionalities.</t>

            <figure align="center" anchor="fig-ptp-translator"
                    title="Point to Point with Translator">
              <artwork><![CDATA[
+---+        +---+         +---+
| A |<------>| T |<------->| B |
+---+        +---+         +---+
]]></artwork>
            </figure>

            <t>A common element in these functions is that they are normally
            transparent at the RTP level, i.e., they perform no changes on any
            RTP or RTCP packet fields and only affect the lower layers. They
            may affect, however, the path the RTP and RTCP packets are routed
            between the endpoints in the RTP session, and thereby only
            indirectly affect the RTP session. For this reason, one could
            believe that transport translator-type middleboxes do not need to
            be included in this document. This topology, however, can raise
            additional requirements in the RTP implementation and its
            interactions with the signalling solution. Both in signalling and
            in certain RTCP fields, network addresses other than those of the
            relay can occur since B has a different network address than the
            relay (T). Implementations that can not support this will also not
            work correctly when endpoints are subject to NAT.</t>

            <t>The transport relay implementation also have some
            considerations, where security considerations are an important
            aspect. Source address filtering of incoming packets are usually
            important in relays, to prevent attackers to inject traffic into a
            session, which one peer will think comes from the other peer. </t>
          </section>

          <section title="Transport Translator">
            <t>Transport Translators (Topo-Trn-Translator) do not modify the
            media stream itself, but are concerned with transport parameters.
            Transport parameters, in the sense of this section, comprise the
            transport addresses (to bridge different domains such unicast to
            multicast) and the media packetization to allow other transport
            protocols to be interconnected to a session (in gateways). Of the
            transport Translators, this memo is primarily interested in those
            that use RTP on both sides, and this is assumed henceforth. </t>

            <t>Translators that bridge between different protocol worlds need
            to be concerned about the mapping of the SSRC/CSRC (Contributing
            Source) concept to the non-RTP protocol. When designing a
            Translator to a non-RTP-based media transport, an important
            consideration is how to handle different sources and their
            identities. This problem space is not discussed henceforth.</t>

            <t>The most basic transport translators that operate below the RTP
            level were already discussed in <xref
            target="sec-transport-anchor"/>.</t>
          </section>

          <section title="Media Translator">
            <t>Media Translators (Topo-Media-Translator) modify the media
            stream itself. This process is commonly known as transcoding. The
            modification of the media stream can be as small as removing parts
            of the stream, and it can go all the way to a full decoding and
            re-encoding (down to the sample level or equivalent) utilizing a
            different media codec. Media Translators are commonly used to
            connect entities without a common interoperability point in the
            media encoding.</t>

            <t>Stand-alone Media Translators are rare. Most commonly, a
            combination of Transport and Media Translator is used to translate
            both the media stream and the transport aspects of a stream
            between two transport domains (or clouds).</t>

            <t>When media translation occurs, the Translator's task regarding
            handling of RTCP traffic becomes substantially more complex. In
            this case, the Translator needs to rewrite B's RTCP Receiver
            Report before forwarding them to A. The rewriting is needed as the
            stream received by B is not the same stream as the other
            participants receive. For example, the number of packets
            transmitted to B may be lower than what A sends, due to the
            different media format and data rate. Therefore, if the Receiver
            Reports were forwarded without changes, the extended highest
            sequence number would indicate that B were substantially behind in
            reception, while most likely it would not be. Therefore, the
            Translator must translate that number to a corresponding sequence
            number for the stream the Translator received. Similar arguments
            can be made for most other fields in the RTCP Receiver
            Reports.</t>

            <t>A media Translator may in some cases act on behalf of the
            "real" source and respond to RTCP feedback messages. This may
            occur, for example, when a receiver requests a bandwidth
            reduction, and the media Translator has not detected any
            congestion or other reasons for bandwidth reduction between the
            media source and itself. In that case, it is sensible that the
            media Translator reacts to the codec control messages itself, for
            example, by transcoding to a lower media rate.</t>

            <t>A variant of translator behaviour worth pointing out is the one
            depicted in <xref target="fig-de-composite-translator"/> of an
            endpoint A sends a media flow to B. On the path there is a device
            T that on A's behalf does something with the media streams, for
            example adds an RTP session with FEC information for A's media
            streams. In this case, T needs to bind the new FEC streams to A's
            media stream, for example by using the same CNAME as A.</t>

            <figure align="center" anchor="fig-de-composite-translator"
                    title="When De-composition is a Translator">
              <artwork><![CDATA[
+------+        +------+         +------+
|      |        |      |         |      |
|  A   |------->|  T   |-------->|  B   |
|      |        |      |---FEC-->|      |
+------+        +------+         +------+]]></artwork>
            </figure>

            <t>This type of functionality where T does something with the
            media stream on behalf of A is covered under the media translator
            definition.</t>
          </section>
        </section>

        <section title="Back to Back RTP sessions">
          <t>There exist middleboxes that interconnect two endpoints through
          themselves, but not by being part of a common RTP session. They
          establish instead two different RTP sessions, one between A and the
          middlebox and another between the middlebox and B.</t>

          <figure align="center" anchor="fig-b2b-session"
                  title="When De-composition is a Translator">
            <artwork><![CDATA[
  |<--Session A-->|  |<--Session B-->|
+------+        +------+         +------+
|  A   |------->|  MB  |-------->|  B   |
+------+        +------+         +------+]]></artwork>
          </figure>

          <t>The middlebox acts as an application-level gateway and bridges
          the two RTP sessions. This bridging can be as basic as forwarding
          the RTP payloads between the sessions, or more complex including
          media transcoding. The difference with the single RTP session
          context is the handling of the SSRCs and the other session-related
          identifiers, such as CNAMEs. With two different RTP sessions these
          can be freely changed and it becomes the middlebox's task to
          maintain the correct relations.</t>

          <t>The signalling or other above-RTP level functionalities
          referencing RTP media streams may be what is most impacted by using
          two RTP sessions and changing identifiers. The structure with two
          RTP sessions also puts a congestion control requirement on the
          middlebox, because it becomes fully responsible for the media stream
          it sources into each of the sessions.</t>

          <t>Adherence to congestion control can be solved locally or by
          bridging also statistics from the receiving endpoint. From an
          implementation point, however, this requires dealing with a number
          of inconsistencies. First, packet loss must be detected for an RTP
          flow sent from A to the middlebox, and that loss must be reported
          through a skipped sequence number in the flow from the middlebox to
          B. This coupling and the resulting inconsistencies is conceptually
          easier to handle when considering the two flows as belonging to a
          single RTP session.</t>
        </section>
      </section>

      <section title="Point to Multipoint Using Multicast">
        <t>Multicast is an IP layer functionality that is available in some
        networks. Two main flavors can be distinguished: <xref
        target="RFC1112">Any Source Multicast (ASM)</xref> where any multicast
        group participant can send to the group address and expect the packet
        to reach all group participants; and <xref target="RFC3569">Source
        Specific Multicast (SSM)</xref>, where only a particular IP host sends
        to the multicast group. Both these models are discussed below in their
        respective sections.</t>

        <section title="Any Source Multicast (ASM)">
          <t>Shortcut name: Topo-ASM (was Topo-Multicast)</t>

          <figure align="center" anchor="fig-ptm-multicast"
                  title="Point to Multipoint Using Multicast ">
            <artwork><![CDATA[
            +-----+          
 +---+     /       \    +---+ 
 | A |----/         \---| B |
 +---+   /   Multi-  \  +---+
        +    Cast     +      
 +---+   \  Network  /  +---+
 | C |----\         /---| D |
 +---+     \       /    +---+
            +-----+          
]]></artwork>
          </figure>

          <t>Point to Multipoint (PtM) is defined here as using a multicast
          topology as a transmission model, in which traffic from any
          participant reaches all the other participants, except for cases
          such as:<list style="symbols">
              <t>packet loss, or</t>

              <t>when a participant does not wish to receive the traffic for a
              specific multicast group and, therefore, has not subscribed to
              the IP multicast group in question. This scenario can occur, for
              example, where a multimedia session is distributed using two or
              more multicast groups and a participant is subscribed only to a
              subset of these sessions.</t>
            </list></t>

          <t>In the above context, "traffic" encompasses both RTP and RTCP
          traffic. The number of participants can vary between one and many,
          as RTP and RTCP scale to very large multicast groups (the
          theoretical limit of the number of participants in a single RTP
          session is in the range of billions). The above can be realized
          using Any Source Multicast (ASM).</t>

          <t>For feedback usage, it is useful to define a "small multicast
          group" as a group where the number of participants is so low (and
          other factors such as the connectivity is so good) that it allows
          the participants to use early or immediate feedback, as defined in
          <xref target="RFC4585">AVPF</xref>. Even when the environment would
          allow for the use of a small multicast group, some applications may
          still want to use the more limited options for RTCP feedback
          available to large multicast groups, for example when there is a
          likelihood that the threshold of the small multicast group (in terms
          of participants) may be exceeded during the lifetime of a
          session.</t>

          <t>RTCP feedback messages in multicast reach, like media data, every
          subscriber (subject to packet losses and multicast group
          subscription). Therefore, the feedback suppression mechanism
          discussed in <xref target="RFC4585"/> is typically required. Each
          individual node needs to process every feedback message it receives,
          not to determine if it is affected or if the feedback message
          applies only to some other participant, but also to derive timing
          restrictions for the sending of its own feedback messages, if
          any.</t>
        </section>

        <section title="Source Specific Multicast (SSM)">
          <t>In Any Source Multicast, any of the participants can send to all
          the other participants, by sending a packet to the multicast group.
          In contrast, <xref target="RFC3569">Source Specific
          Multicast</xref><xref target="RFC4607"/> refers to scenarios where
          only a single source (Distribution Source) can send to the multicast
          group, creating a topology that looks like the one below:</t>

          <figure align="center" anchor="fig-multipoint-ssm"
                  title="Point to Multipoint using Source Specific Multicast">
            <artwork><![CDATA[
+--------+       +-----+
|Media   |       |     |       Source-specific
|Sender 1|<----->| D S |          Multicast
+--------+       | I O |  +--+----------------> R(1)
                 | S U |  |  |                    |
+--------+       | T R |  |  +-----------> R(2)   |
|Media   |<----->| R C |->+  |           :   |    |
|Sender 2|       | I E |  |  +------> R(n-1) |    |
+--------+       | B   |  |  |          |    |    |
    :            | U   |  +--+--> R(n)  |    |    |
    :            | T +-|          |     |    |    |
    :            | I | |<---------+     |    |    |
+--------+       | O |F|<---------------+    |    |
|Media   |       | N |T|<--------------------+    |
|Sender M|<----->|   | |<-------------------------+
+--------+       +-----+       RTCP Unicast

FT = Feedback Target
Transport from the Feedback Target to the Distribution
Source is via unicast or multicast RTCP if they are not
co-located.
]]></artwork>
          </figure>

          <t>In the <xref target="fig-multipoint-ssm">SSM topology</xref> a
          number of RTP sources (1 to M) are allowed to send media to the SSM
          group. These sources send media to a dedicated distribution source,
          which forwards the media streams to the multicast group on behalf of
          the original senders. The media streams reach the Receivers (R(1) to
          R(n)). The Receivers' RTCP messages cannot be sent to the multicast
          group, as the SSM multicast group by definition has only a single
          source. To support RTCP, an <xref target="RFC5760">RTP extension for
          SSM</xref> was defined. It uses unicast transmission to send RTCP
          from each of the receivers to one or more Feedback Targets (FT). The
          feedback targets relay the RTCP unmodified, or provide a summary of
          the participants RTCP reports towards the whole group by forwarding
          the RTCP traffic to the distribution source. <xref
          target="fig-multipoint-ssm"/> only shows a single feedback target
          integrated in the distribution source, but for scalability the FT
          can be many and have responsibility for sub-groups of the receivers.
          For summary reports, however, there must be a single feedback
          aggregating all the summaries to a common message to the whole
          receiver group.</t>

          <t>The RTP extension for SSM specifies how feedback (both reception
          information and specific feedback events) are handled. The more
          general problems associated with the use of multicast, where
          everyone receives what the distribution source sends needs to be
          accounted for.</t>

          <t>Aforementioned situation results in common behavior for RTP
          multicast:<list style="numbers">
              <t>Multicast applications often use a group of RTP sessions, not
              one. Each endpoint needs to be a member of most or all of these
              RTP sessions in order to perform well.</t>

              <t>Within each RTP session, the number of media sinks is likely
              to be much larger than the number of RTP sources.</t>

              <t>Multicast applications need signalling functions to identify
              the relationships between RTP sessions.</t>

              <t>Multicast applications need signalling functions to identify
              the relationships between SSRCs in different RTP sessions.</t>
            </list></t>

          <t>All multicast configurations share a signalling requirement: all
          of the participants need to have the same RTP and payload type
          configuration. Otherwise, A could, for example, be using payload
          type 97 to identify the video codec H.264, while B would identify it
          as MPEG-2.</t>

          <t>Security solutions for this type of group communications are also
          challenging. First, the key-management and the security protocol
          must support group communication. Source authentication becomes more
          difficult and requires special solutions. For more discussion on
          this please review <xref
          target="I-D.ietf-avtcore-rtp-security-options">Options for Securing
          RTP Sessions</xref>.</t>
        </section>

        <section title="SSM with Local Unicast Resources">
          <t>[RFC6285] "Unicast-Based Rapid Acquisition of Multicast RTP
          Sessions" results in additional extensions to SSM Topology.</t>

          <figure anchor="fig-rams">
            <artwork><![CDATA[ -----------                                       --------------
|           |------------------------------------>|              |
|           |.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.->|              |
|           |                                     |              |
| Multicast |          ----------------           |              |
|  Source   |         | Retransmission |          |              |
|           |-------->|  Server  (RS)  |          |              |
|           |.-.-.-.->|                |          |              |
|           |         |  ------------  |          |              |
 -----------          | |  Feedback  | |<.=.=.=.=.|              |
                      | | Target (FT)| |<~~~~~~~~~| RTP Receiver |
PRIMARY MULTICAST     |  ------------  |          |   (RTP_Rx)   |
RTP SESSION with      |                |          |              |
UNICAST FEEDBACK      |                |          |              |
                      |                |          |              |
- - - - - - - - - - - |- - - - - - - - |- - - - - |- - - - - - - |- -
                      |                |          |              |
UNICAST BURST         |  ------------  |          |              |
(or RETRANSMISSION)   | |   Burst/   | |<~~~~~~~~>|              |
RTP SESSION           | |  Retrans.  | |.........>|              |
                      | |Source (BRS)| |<.=.=.=.=>|              |
                      |  ------------  |          |              |
                      |                |          |              |
                       ----------------            --------------

   -------> Multicast RTP Flow
   .-.-.-.> Multicast RTCP Flow
   .=.=.=.> Unicast RTCP Reports
   ~~~~~~~> Unicast RTCP Feedback Messages
   .......> Unicast RTP Flow]]></artwork>
          </figure>

          <t>The Rapid acquisition extension allows an endpoint joining an SSM
          multicast session to request media starting with the last sync-point
          (from where media can be decoded without requiring context
          established by the decoding of prior packets) to be sent at high
          speed until such time where, after decoding of these burst-delivered
          media packets, the correct media timing is established, i.e. media
          packets are received within adequate buffer intervals for this
          application. This is accomplished by first establishing a unicast
          PtP RTP session between the Burst/Retransmission Source (BRS, <xref
          target="fig-rams"/>) and the RTP Receiver. The unicast session is
          used to transmit cached packets from the multicast group at higher
          then normal speed in order to synchronize the receiver to the
          ongoing multicast packet flow. Once the RTP receiver and its decoder
          have caught up with the multicast session's current delivery, the
          receiver switches over to receiving directly from the multicast
          group. The (still existing) PtP RTP session is, in many deployed
          applications, be used as a repair channel, i.e., for RTP
          Retransmission traffic of those packets that were not received from
          the multicast group.</t>
        </section>
      </section>

      <section title="Point to Multipoint Using Mesh">
        <t>Shortcut name: Topo-Mesh</t>

        <figure align="center" anchor="fig-mesh"
                title="Point to Multi-Point using Mesh">
          <artwork><![CDATA[
+---+      +---+
| A |<---->| B |
+---+      +---+
  ^         ^   
   \       /    
    \     /     
     v   v      
     +---+      
     | C |      
     +---+
]]></artwork>
        </figure>

        <t>Based on the RTP session definition, it is clearly possible to have
        a joint RTP session over multiple unicast transport flows like the
        above joint three endpoint session. In this case, A needs to send its'
        media streams and RTCP packets to both B and C over their respective
        transport flows. As long as all participants do the same, everyone
        will have a joint view of the RTP session. </t>

        <t>This does not create any additional requirements beyond the need to
        have multiple transport flows associated with a single RTP session.
        Note that an endpoint may use a single local port to receive all these
        transport flows, or it might have separate local reception ports for
        each of the endpoints.</t>

        <figure anchor="fig-mesh-joint-session"
                title="An Multi-unicast Mesh with a joint RTP session">
          <artwork><![CDATA[
+-A--------------------+                 +-B-----------+
|+---+                 |                 |             |
||CAM|                 |                 |             |
|+---+     +-UDP1------|                 |-UDP1------+ |
|  |       | +-RTP1----|                 |-RTP1----+ | |
|  V       | | +-Video-|                 |-Video-+ | | |
|+----+    | | |       |<----------------|BV1    | | | |
||ENC |----+-+-+--->AV1|---------------->|       | | | |
|+----+    | | +-------|                 |-------+ | | |
|  |       | +---------|                 |---------+ | |
|  |       +-----------|                 |-----------+ |
|  |       ------------|                 |------------ |
|  |                   |                 |-------------+
|  |                   |                                
|  |                   |                 +-C-----------+
|  |                   |                 |             |
|  |       +-UDP2------|                 |-UDP2------+ |
|  |       | +-RTP1----|                 |-RTP1----+ | |
|  |       | | +-Video-|                 |-Video-+ | | |
|  +-------+-+-+--->AV1|---------------->|       | | | |
|          | | |       |<----------------|CV1    | | | |
|          | | +-------|                 |-------+ | | |
|          | +---------|                 |---------+ | |
|          +-----------|                 |-----------+ |
|          ------------|                 |------------ |
+----------------------+                 +-------------+
]]></artwork>
        </figure>

        <t>A joint RTP session from A's perspective for the Mesh depicted in
        <xref target="fig-mesh"/> with a joint RTP session have multiple
        transport flows, here enumerated as UDP1 and UDP2. However, there is
        only one RTP session (RTP1). The media source (CAM) is encoded and
        transmitted over the SSRC (AV1) across both transport layers. However,
        as this is a joint RTP session, the two streams must be the same.
        Thus, an congestion control adaptation needed for the paths A to B and
        A to C needs to use the most restricting path's properties. </t>

        <t>An alternative structure for establishing the above topology is to
        use independent RTP sessions between each pair of peers, i.e., three
        different RTP sessions. In some scenarios, the same RTP media stream
        may be sent from transmitting endpoint, however it also supports local
        adaptation taking place in one or more of the RTP media streams,
        rendering them non-identical. </t>

        <figure anchor="fig-mesh-diff-session"
                title="An Multi-unicast Mesh with independent RTP session">
          <artwork><![CDATA[
+-A----------------------+              +-B-----------+
|+---+                   |              |             |
||MIC|       +-UDP1------|              |-UDP1------+ |
|+---+       | +-RTP1----|              |-RTP1----+ | |
| |  +----+  | | +-Audio-|              |-Audio-+ | | |
| +->|ENC1|--+-+-+--->AA1|------------->|       | | | |
| |  +----+  | | |       |<-------------|BA1    | | | |
| |          | | +-------|              |-------+ | | |
| |          | +---------|              |---------+ | |
| |          +-----------|              |-----------+ |
| |          ------------|              |-------------|
| |                      |              |-------------+
| |                      |
| |                      |              +-C-----------+
| |                      |              |             |
| |          +-UDP2------|              |-UDP2------+ |
| |          | +-RTP2----|              |-RTP2----+ | |
| |  +----+  | | +-Audio-|              |-Audio-+ | | |
| +->|ENC2|--+-+-+--->AA2|------------->|       | | | |
|    +----+  | | |       |<-------------|CA1    | | | |
|            | | +-------|              |-------+ | | |
|            | +---------|              |---------+ | |
|            +-----------|              |-----------+ |
+------------------------+              +-------------+
]]></artwork>
        </figure>

        <t>Lets review the topology when independent RTP sessions are used,
        from A's perspective in <xref target="fig-mesh"/> by considering both
        how the media is a handled and the RTP sessions that are set-up in
        <xref target="fig-mesh-diff-session"/>. A's microphone is captured and
        the digital audio can then be feed into two different encoder
        instances, as each beeing associated with two independent RTP sessions
        (RTP1 and RTP2). The SSRCs (AA1 and AA2) in each RTP session will be
        completely independent and the media bit-rate produced by the encoders
        can also be tuned differently to address any congestion control
        requirements differing for the paths A to B compared to A to C.</t>

        <t>From a topologies viewpoint, an important difference exists in the
        behavior around RTCP. First, when a single RTP session spans all three
        endpoints and their connecting flows, an common RTCP bandwidth is
        calculated and used for this single joint session. In contrast, when
        there are multiple independent RTP sessions, each RTP session has its
        local RTCP bandwidth allocation. </t>

        <t>Further, when multiple sessions are used, endpoints not directly
        involved in a session, do not have any awareness of the conditions in
        those sessions. For example, in the case of the three endpoint
        configuration in <xref target="fig-mesh"/>, endpoint A has no
        awareness of the conditions occurring in the session between endpoints
        B and C (whereas, if a single RTP session were used, it would have
        such awareness).</t>

        <t>Loop detection is also affected. With independent RTP sessions, the
        SSRC/CSRC cannot be used to determine when an endpoint receives its
        own media stream, or a mixed media stream including its own media
        stream (a condition known as a loop). The identification of loops and,
        in most cases, their avoidance, has to be achieved by other means, for
        example through signaling or the use of an RTP external name space
        binding SSRC/CSRC among any communicating RTP sessions in the
        mesh.</t>
      </section>

      <section anchor="sec-ptm-translator"
               title="Point to Multipoint Using the RFC 3550 Translator">
        <t/>

        <t>This section discusses some additional usages related to point to
        multipoint of Translators compared to the point to point only cases in
        <xref target="sec-ptp-translators"/>.</t>

        <section title="Relay - Transport Translator">
          <t>Shortcut name: Topo-PtM-Trn-Translator</t>

          <t>This section discusses Transport Translator only usages to enable
          multipoint sessions.</t>

          <figure align="center" anchor="fig-ptm-multicast-translator"
                  title="Point to Multipoint Using Multicast ">
            <artwork><![CDATA[       
           +-----+                                 
+---+     /       \     +------------+      +---+  
| A |<---/         \    |            |<---->| B |  
+---+   /   Multi-  \   |            |      +---+  
       +    cast     +->| Translator |             
+---+   \  Network  /   |            |      +---+  
| C |<---\         /    |            |<---->| D |  
+---+     \       /     +------------+      +---+  
           +-----+                                 
]]></artwork>
          </figure>

          <t><xref target="fig-ptm-multicast-translator"/> depicts an example
          of a Transport Translator performing at least IP address
          translation. It allows the (non-multicast-capable) participants B
          and D to take part in an any source multicast session by having the
          Translator forward their unicast traffic to the multicast addresses
          in use, and vice versa. It must also forward B's traffic to D, and
          vice versa, to provide each of B and D with a complete view of the
          session.</t>

          <figure align="center" anchor="fig-translator-unicast"
                  title="RTP Translator (Relay) with Only Unicast Paths">
            <artwork><![CDATA[
+---+      +------------+      +---+
| A |<---->|            |<---->| B |
+---+      |            |      +---+
           | Translator |
+---+      |            |      +---+
| C |<---->|            |<---->| D |
+---+      +------------+      +---+
]]></artwork>
          </figure>

          <t>Another Translator scenario is depicted in <xref
          target="fig-translator-unicast"/>. The Translator in this case
          connects multiple users of a conference through unicast. This can be
          implemented using a very simple transport Translator which, in this
          document, is called a relay. The relay forwards all traffic it
          receives, both RTP and RTCP, to all other participants. In doing so,
          a multicast network is emulated without relying on a
          multicast-capable network infrastructure.</t>

          <t>For RTCP feedback this results in a similar set of considerations
          to those described in the ASM RTP topology. It also puts some
          additional signalling requirements onto the session establishment;
          for example, a common configuration of RTP payload types is
          required.</t>

          <t>Transport translators and relays should always consider doing
          source address filtering, to prevent attackers to inject traffic
          using the listening ports on the translator. The translator can
          however go one step further, and especially if explicit SSRC
          signalling is used, prevent other session participants to send SSRCs
          that are used by other participants in the session. This can improve
          the security properties of the session, despite the use of group
          keys that on cryptographic level allows anyone to impersonate
          another in the same RTP session.</t>

          <t>A Translator that doesn't change the RTP/RTCP packets content can
          be operated without the requiring the translator to have access to
          the security contexts used to protect the RTP/RTCP traffic between
          the participants.</t>
        </section>

        <section title="Media Translator">
          <t>In the context of multipoint communications a Media Translator is
          not providing new mechanisms to establish a multipoint session. It
          is more of an enabler, or facilitator, that ensures one or some
          sub-set of session participants can participate in the session.</t>

          <t>If B in <xref target="fig-ptm-multicast-translator"/> were behind
          a limited network path, the Translator may perform media transcoding
          to allow the traffic received from the other participants to reach B
          without overloading the path. This transcoding can help the other
          participants in the Multicast part of the session, by not requiring
          the quality transmitted by A to be lowered to the bitrates that B is
          actually capable of receiving.</t>
        </section>
      </section>

      <section anchor="sec-ptm-mixer"
               title="Point to Multipoint Using the RFC 3550 Mixer Model">
        <t>Shortcut name: Topo-Mixer</t>

        <t>A Mixer is a middlebox that aggregates multiple RTP streams that
        are part of a session by generating a new RTP stream and, in most
        cases, by manipulating the media data. One common application for a
        Mixer is to allow a participant to receive a session with a reduced
        amount of resources.</t>

        <figure align="center" anchor="fig-ptm-mixer"
                title="Point to Multipoint Using the RFC 3550 Mixer Model">
          <artwork><![CDATA[
           +-----+                              
+---+     /       \     +-----------+      +---+
| A |<---/         \    |           |<---->| B |
+---+   /   Multi-  \   |           |      +---+
       +    cast     +->|   Mixer   |           
+---+   \  Network  /   |           |      +---+
| C |<---\         /    |           |<---->| D |
+---+     \       /     +-----------+      +---+
           +-----+                              
]]></artwork>
        </figure>

        <t>A Mixer can be viewed as a device terminating the media streams
        received from other session participants. Using the media data from
        the received media streams, a Mixer generates a media stream that is
        sent to the session participant.</t>

        <t>The content that the Mixer provides is the mixed aggregate of what
        the Mixer receives over the PtP or PtM paths, which are part of the
        same conference session.</t>

        <t>The Mixer is the content source, as it mixes the content (often in
        the uncompressed domain) and then encodes it for transmission to a
        participant. The CSRC Count (CC) and CSRC fields in the RTP header can
        be used to indicate the contributors to the newly generated stream.
        The SSRCs of the to-be-mixed streams on the Mixer input appear as the
        CSRCs at the Mixer output. That output stream uses a unique SSRC that
        identifies the Mixer's stream. The CSRC should be forwarded between
        the different conference participants to allow for loop detection and
        identification of sources that are part of the global session. Note
        that Section 7.1 of RFC 3550 requires the SSRC space to be shared
        between domains for these reasons. This also implies that any SDES
        information normally needs to be forwarded across the mixer.</t>

        <t>The Mixer is responsible for generating RTCP packets in accordance
        with its role. It is a receiver and should therefore send receiver
        reports for the media streams it receives. In its role as a media
        sender, it should also generate sender reports for those media streams
        it sends. As specified in Section 7.3 of RFC 3550, a Mixer must not
        forward RTCP unaltered between the two domains.</t>

        <t>The Mixer depicted in <xref target="fig-ptm-mixer"/> is involved in
        three domains that need to be separated: the any source multicast
        network (including participants A and C), participant B, and
        participant D. Assuming all four participants in the conference are
        interested in receiving content from each other participant, the Mixer
        produces different mixed streams for B and D, as the one to B may
        contain content received from D, and vice versa. However, the Mixer
        may only need one SSRC per media type in each domain where it is the
        receiving entity and transmitter of mixed content.</t>

        <t>In the multicast domain, a Mixer still needs to provide a mixed
        view of the other domains. This makes the Mixer simpler to implement
        and avoids any issues with advanced RTCP handling or loop detection,
        which would be problematic if the Mixer were providing non-symmetric
        behavior. Please see <xref target="sec-asymmetric"/> for more
        discussion on this topic. The mixing operation, however, in each
        domain could potentially be different.</t>

        <t>A Mixer is responsible for receiving RTCP feedback messages and
        handling them appropriately. The definition of "appropriate" depends
        on the message itself and the context. In some cases, the reception of
        a codec-control message by the Mixer may result in the generation and
        transmission of RTCP feedback messages by the Mixer to the
        participants in the other domain(s). In other cases, a message is
        handled by the Mixer itself and therefore not forwarded to any other
        domain.</t>

        <t>When replacing the multicast network in <xref
        target="fig-ptm-mixer"/> (to the left of the Mixer) with individual
        unicast paths as depicted in <xref target="fig-mixer-unicast"/>, the
        Mixer model is very similar to the one discussed in <xref
        target="sec-ptm-mcu"/> below. Please see the discussion in <xref
        target="sec-ptm-mcu"/> about the differences between these two
        models.</t>

        <figure align="center" anchor="fig-mixer-unicast"
                title="RTP Mixer with Only Unicast Paths ">
          <artwork><![CDATA[
+---+      +------------+      +---+
| A |<---->|            |<---->| B |
+---+      |            |      +---+
           |   Mixer    |           
+---+      |            |      +---+
| C |<---->|            |<---->| D |
+---+      +------------+      +---+
]]></artwork>
        </figure>

        <t>We now discuss in more detail the different mixing operations that
        a mixer can perform and how they can affect RTP and RTCP behavior.</t>

        <section title="Media Mixing">
          <t>The media mixing mixer is likely the one that most think of when
          they hear the term "mixer". Its basic mode of operation is that it
          receives media streams from several participants and selects the
          stream(s) to be included in a media-domain mix. The selection can be
          through static configuration or by dynamic, content dependent means
          such as voice activation. The mixer then creates a single outgoing
          stream from this mix.</t>

          <t>The most commonly deployed media mixer is probably the audio
          mixer, used in voice conferencing, where the output consists of a
          mixture of all the input streams; this needs minimal signalling to
          be successfully set up. Audio mixing is relatively straightforward
          and commonly possible for a reasonable number of participants.
          Assume, for example, that one wants to mix N streams from different
          participants. The mixer needs to decode those N streams, typically
          into the sample domain, and then produce N or N+1 mixes. Different
          mixes are needed so that each contributing source gets a mix of all
          other sources except its own, as this would result in an echo. When
          N is lower than the number of all participants one may produce a Mix
          of all N streams for the group that are currently not included in
          the mix, thus N+1 mixes. These audio streams are then encoded again,
          RTP packetized and sent out. In many cases, audio level
          normalization is also required before the actual mixing process.</t>

          <t>In video, the term "mixing" has a different interpretation than
          audio. It is commonly used to refer to the process of spatially
          combining contributed video streams is known as "tiling". The
          reconstructed, appropriately scaled down videos can be spatially
          arranged in a set of tiles, each tile containing the video from a
          participant. Tiles can be of different sizes, so that, for example,
          a particularly important participant, or the loudest speaker, is
          being shown on in larger tile than other participants. A self-view
          picture can be included in the tiling, which can either be locally
          produced or be a feedback from a received and reconstructed video
          image. Such remote loopback allows for confidence monitoring, i.e.,
          it enables the participant to see himself/herself just as other
          participants see him/her. The tiling normally operates on
          reconstructed video in the sample domain. The tiled image is
          encoded, packetized, and sent by the mixer. It is possible that a
          middlebox with media mixing duties contains only a single mixer of
          the aforementioned type, in which case all participants necessarily
          see the same tiled video, even if it is being sent over different
          RTP streams. More common, however, are mixing arrangement where an
          individual mixer is available for each outgoing port of the
          middlebox, allowing individual compositions for each participant (a
          feature referred to as personalized layout).</t>

          <t>One problem with media mixing is that it consumes both large
          amount of media processing (for the actual mixing process in the
          uncompressed domain) and encoding resources (for the encoding of the
          mixed signal). Another problem is the quality degradation created by
          decoding and re-encoding the media that is encapsulated in the RTP
          media stream, which is the result of the lossy nature of most
          commonly used media codecs. A third problem is the latency
          introduced by the media mixing, which can be substantial and
          annoyingly noticeable in case of video, or in case of audio if that
          mixed audio is lip-sychronized with high latency video. The
          advantage of media mixing is that it is straightforward for the
          clients to handle the single media stream (which includes the mixed
          aggregate of many sources), as they don't need to handle multiple
          decodings, local mixing and composition. In fact, mixers were
          introduced in pre-RTP times so that legacy, single stream receiving
          endpoints could successfully participate in what a user would
          recognize as a multiparty video conference.</t>

          <figure align="center" anchor="fig-media-mixer"
                  title="Session and SSRC details for Media Mixer">
            <artwork><![CDATA[+-A---------+          +-MIXER----------------------+
| +-RTP1----|          |-RTP1------+        +-----+ |
| | +-Audio-|          |-Audio---+ | +---+  |     | |
| | |    AA1|--------->|---------+-+-|DEC|->|     | |
| | |       |<---------|MA1 <----+ | +---+  |     | |
| | |       |          |(BA1+CA1)|\| +---+  |     | |
| | +-------|          |---------+ +-|ENC|<-| B+C | |
| +---------|          |-----------+ +---+  |     | |
+-----------+          |                    |     | |
                       |                    |  M  | |
+-B---------+          |                    |  E  | |
| +-RTP2----|          |-RTP2------+        |  D  | |
| | +-Audio-|          |-Audio---+ | +---+  |  I  | |
| | |    BA1|--------->|---------+-+-|DEC|->|  A  | |
| | |       |<---------|MA2 <----+ | +---+  |     | |
| | +-------|          |(BA1+CA1)|\| +---+  |     | |
| +---------|          |---------+ +-|ENC|<-| A+C | |
+-----------+          |-----------+ +---+  |     | |
                       |                    |  M  | |
+-C---------+          |                    |  I  | |
| +-RTP3----|          |-RTP3------+        |  X  | |
| | +-Audio-|          |-Audio---+ | +---+  |  E  | |
| | |    CA1|--------->|---------+-+-|DEC|->|  R  | |
| | |       |<---------|MA3 <----+ | +---+  |     | |
| | +-------|          |(BA1+CA1)|\| +---+  |     | |
| +---------|          |---------+ +-|ENC|<-| A+B | |
+-----------+          |-----------+ +---+  +-----+ |
                       +----------------------------+
]]></artwork>
          </figure>

          <t>From an RTP perspective media mixing can be a very simple
          process, as can be seen in <xref target="fig-media-mixer"/>. The
          mixer presents one SSRC towards the receiving client, e.g., MA1 to
          Peer A, where the associated stream is the media mix of the other
          participants. As each peer, in this example, receives a different
          version of a mix from the mixer, there is no actual relation between
          the different RTP sessions in terms of actual media or transport
          level information. There are, however, common relationships between
          RTP1-RTP3, namely SSRC space and identity information. When A
          receives the MA1 stream which is a combination of BA1 and CA1
          streams, the mixer may include CSRC information in the MA1 stream to
          identify the contributing source BA1 and CA1, allowing the receiver
          to identify the contributing sources even if this were not possible
          through the media itself or through other signaling means.</t>

          <t>The CSRC has, in turn, utility in RTP extensions, like the <xref
          target="RFC6465">Mixer to Client audio levels RTP header
          extension</xref>. If the SSRCs from the endpoint to mixer paths are
          used as CSRCs in another RTP session, then RTP1, RTP2 and RTP3
          become one joint session as they have a common SSRC space. At this
          stage, the mixer also needs to consider which RTCP information it
          needs to expose in the different paths. In the above scenario, a
          mixer would normally expose nothing more than the Source Description
          (SDES) information and RTCP BYE for a CSRC leaving the session. The
          main goal would be to enable the correct binding against the
          application logic and other information sources. This also enables
          loop detection in the RTP session.</t>
        </section>

        <section anchor="sec-media-switching" title="Media Switching">
          <t>Media switching mixers are used from limited functionality
          scenarios where no, or only very limited, concurrent presentation of
          multiple sources is required by the application to more complex
          multi-stream usages with receiver mixing or tiling, including
          combined with simulcast and/or scalability between source and mixer.
          An RTP Mixer based on media switching avoids the media decoding and
          encoding operations in the mixer, as it conceptually forwards the
          encoded media stream as it was being sent to the mixer. It does not
          avoid, however, the decryption and re-encryption cycle as it
          rewrites RTP headers. Forwarding media (in contrast to
          reconstructing-mixing-encoding media) reduces the amount of
          computational resources needed in the mixer and increases the media
          quality (both in terms of fidelity and reduced latency).</t>

          <t>A media switching mixer maintains a pool of SSRCs representing
          conceptual or functional streams that the mixer can produce. These
          streams are created by selecting media from one of the RTP media
          streams received by the mixer and forwarded to the peer using the
          mixer's own SSRCs. The mixer can switch between available sources if
          that is required by the concept for the source, like the currently
          active speaker. Note that the mixer, in most cases, still needs to
          perform a certain amount of media processing, as many media formats
          do not allow to "tune into" the stream at arbitrary points of their
          bitstream.</t>

          <t>To achieve a coherent RTP media stream from the mixer's SSRC, the
          mixer needs to rewrite the incoming RTP packet's header. First the
          SSRC field must be set to the value of the Mixer's SSRC. Second, the
          sequence number must be the next in the sequence of outgoing packets
          it sent. Third, the RTP timestamp value needs to be adjusted using
          an offset that changes each time one switches media source. Finally,
          depending on the negotiation of the RTP payload type, the value
          representing this particular RTP payload configuration may have to
          be changed if the different endpoint mixer paths have not arrived on
          the same numbering for a given configuration. This also requires
          that the different endpoints support a common set of codecs,
          otherwise media transcoding for codec compatibility would still be
          required.</t>

          <t>We now consider the operation of a media switching mixer that
          supports a video conference with six participants (A-F) where the
          two most recent speakers in the conference are shown to each
          participant. The mixer has thus two SSRCs sending video to each
          peer, and each peer is capable of locally handling two video streams
          simultaneously.</t>

          <figure align="center" anchor="fig-media-switching"
                  title="Media Switching RTP Mixer">
            <artwork><![CDATA[+-A---------+             +-MIXER----------------------+ 
| +-RTP1----|             |-RTP1------+        +-----+ | 
| | +-Video-|             |-Video---+ |        |     | | 
| | |    AV1|------------>|---------+-+------->|  S  | | 
| | |       |<------------|MV1 <----+-+-BV1----|  W  | | 
| | |       |<------------|MV2 <----+-+-EV1----|  I  | | 
| | +-------|             |---------+ |        |  T  | | 
| +---------|             |-----------+        |  C  | | 
+-----------+             |                    |  H  | | 
                          |                    |     | | 
+-B---------+             |                    |  M  | | 
| +-RTP2----|             |-RTP2------+        |  A  | | 
| | +-Video-|             |-Video---+ |        |  T  | | 
| | |    BV1|------------>|---------+-+------->|  R  | | 
| | |       |<------------|MV3 <----+-+-AV1----|  I  | | 
| | |       |<------------|MV4 <----+-+-EV1----|  X  | | 
| | +-------|             |---------+ |        |     | | 
| +---------|             |-----------+        |     | | 
+-----------+             |                    |     | | 
                          :                    :     : : 
                          :                    :     : : 
+-F---------+             |                    |     | | 
| +-RTP6----|             |-RTP6------+        |     | | 
| | +-Video-|             |-Video---+ |        |     | | 
| | |    CV1|------------>|---------+-+------->|     | | 
| | |       |<------------|MV11 <---+-+-AV1----|     | | 
| | |       |<------------|MV12 <---+-+-EV1----|     | | 
| | +-------|             |---------+ |        |     | | 
| +---------|             |-----------+        +-----+ | 
+-----------+             +----------------------------+ 

]]></artwork>
          </figure>

          <t>The Media Switching RTP mixer can, similarly to the Media Mixing
          Mixer, reduce the bit-rate required for media transmission towards
          the different peers by selecting and forwarding only a sub-set of
          RTP media streams it receives from the conference participants. In
          cases the mixer receives simulcast transmissions or a scalable
          encoding of the media source, the mixer has more degrees of freedom
          to select streams or sub-sets of stream to forward to a receiver,
          both based on transport or client restrictions as well as
          application logic. </t>

          <t>To ensure that a media receiver can correctly decode the RTP
          media stream after a switch, a codec that uses temporal prediction
          needs to start its decoding from independent refresh points, or
          similar points in the bitstream. For some codecs, for example frame
          based speech and audio codecs, this is easily achieved by starting
          the decoding at RTP packet boundaries, as each packet boundary
          provides a refresh point (assuming proper packetization on the
          encoder side). For other codecs, particularly in video, refresh
          points are less common in the bitstream or may not be present at all
          without an explicit request to the respective encoder. The <xref
          target="RFC5104">Full Intra Request</xref> RTCP codec control
          message has been defined for this purpose.</t>

          <t>In this type of mixer one could consider to fully terminate the
          RTP sessions between the different endpoint and mixer paths. The
          same arguments and considerations as discussed in <xref
          target="sec-ptm-mcu"/> need to be taken into consideration and apply
          here.</t>
        </section>
      </section>

      <section title="Selective Forwarding Middlebox">
        <t>Another method for handling media in the RTP mixer is to "project",
        or make available, all potential RTP sources (SSRCs) into a
        per-endpoint, independent RTP session. The middlebox can select which
        of the potential sources that are currently actively transmitting
        media will be sent to each of the endpoints. This is similar to the
        media switching Mixer but has some important differences in RTP
        details.</t>

        <figure align="center" anchor="fig-projecting"
                title="Selective Forwarding Middlebox">
          <artwork><![CDATA[+-A---------+             +-Middlebox-----------------+
| +-RTP1----|             |-RTP1------+       +-----+ |
| | +-Video-|             |-Video---+ |       |     | |
| | |    AV1|------------>|---------+-+------>|     | |
| | |       |<------------|BV1 <----+-+-------|  S  | |
| | |       |<------------|CV1 <----+-+-------|  W  | |
| | |       |<------------|DV1 <----+-+-------|  I  | |
| | |       |<------------|EV1 <----+-+-------|  T  | |
| | |       |<------------|FV1 <----+-+-------|  C  | |
| | +-------|             |---------+ |       |  H  | |
| +---------|             |-----------+       |     | |
+-----------+             |                   |  M  | |
                          |                   |  A  | |
+-B---------+             |                   |  T  | |
| +-RTP2----|             |-RTP2------+       |  R  | |
| | +-Video-|             |-Video---+ |       |  I  | |
| | |    BV1|------------>|---------+-+------>|  X  | |
| | |       |<------------|AV1 <----+-+-------|     | |
| | |       |<------------|CV1 <----+-+-------|     | |
| | |       | :    :    : |: :  : : : : :  : :|     | |
| | |       |<------------|FV1 <----+-+-------|     | |
| | +-------|             |---------+ |       |     | |
| +---------|             |-----------+       |     | |
+-----------+             |                   |     | |
                          :                   :     : :
                          :                   :     : :
+-F---------+             |                   |     | |
| +-RTP6----|             |-RTP6------+       |     | |
| | +-Video-|             |-Video---+ |       |     | |
| | |    FV1|------------>|---------+-+------>|     | |
| | |       |<------------|AV1 <----+-+-------|     | |
| | |       | :    :    : |: :  : : : : :  : :|     | |
| | |       |<------------|EV1 <----+-+-------|     | |
| | +-------|             |---------+ |       |     | |
| +---------|             |-----------+       +-----+ |
+-----------+             +---------------------------+
]]></artwork>
        </figure>

        <t>In the six participant conference depicted above <xref
        target="fig-projecting">in</xref> one can see that end-point A is
        aware of five incoming SSRCs, BV1-FV1. If this middlebox intends to
        have a similar behavior as in <xref target="sec-media-switching"/>
        where the mixer provides the end-points with the two latest speaking
        end-points, then only two out of these five SSRCs need concurrently
        transmit media to A. As the middlebox selects the source in the
        different RTP sessions that transmit media to the end-points, each RTP
        media stream requires some rewriting of RTP header fields when being
        projected from one session into another. In particular, the sequence
        number needs to be consecutively incremented based on the packet
        actually being transmitted in each RTP session. Therefore, the RTP
        sequence number offset will change each time a source is turned on in
        a RTP session. The timestamp (possibly offset) stays the same.</t>

        <t>As the RTP sessions are independent, the SSRC numbers used can also
        be handled independently, thereby bypassing the requirement for SSRC
        collision detection and avoidance. On the other hand, tools such as
        remapping tables between the RTP sessions are required. For example,
        the stream that is being sent by endpoint B to the middlebox (BV1) may
        use an SSRC value of 12345678. When that media stream is sent to
        endpoint F by the middlebox, it can use any SSRC value, e.g. 87654321.
        As a result, each endpoint may have a different view of the
        application usage of a particular SSRC. Any RTP level identity
        information, such as SDES items also needs to update the SSRC
        referenced, if the included SDES items are intended to be global. Thus
        the application must not use SSRC as references to RTP media streams
        when communicating with other peers directly. This also affects loop
        detection which will fail to work, as there is no common namespace and
        identities across the different legs in the communication session on
        RTP level. Instead this responsibility falls onto higher layers.</t>

        <t>The middlebox is also responsible to receive any RTCP codec control
        requests coming from an end-point, and decide if it can act on the
        request locally or needs to translate the request into the RTP session
        that contains the media source. Both end-points and the middlebox need
        to implement conference related codec control functionalities to
        provide a good experience. Commonly used are Full Intra Request to
        request from the media source to provide switching points between the
        sources, and Temporary Maximum Media Bit-rate Request (TMMBR) to
        enable the middlebox to aggregate congestion control responses towards
        the media source so to enable it to adjust its bit-rate (obviously
        only in case the limitation is not in the source to middlebox
        link).</t>

        <t>The selective forwarding middlebox has been introduced in recently
        developed videoconferencing systems in conjunction with, and to
        capitalize on, scalable video coding as well as simulcasting. An
        example of scalable video coding is Annex G of H.264, but other
        codecs, including H.264 AVC and VP8 also exhibit scalability, albeit
        only in the temporal dimension. In both scalable coding and simulcast
        cases the video signal is represented by a set of two or more
        bitstreams, providing a corresponding number of distinct fidelity
        points. The middlebox selects which parts of a scalable bitstream (or
        which bitstream, in the case of simulcasting) to forward to each of
        the receiving endpoints. The decision may be driven by a number of
        factors, such as available bit rate, desired layout, etc. Contrary to
        transcoding MCUs, these "Selective Forwarding Units" (SFUs) have
        extremely low delay, and provide features that are typically
        associated with high-end systems (personalized layout, error
        localization) without any signal processing at the middlebox. They are
        also capable of scaling to a large number of concurrent users,
        and--due to their very low delay--can also be cascaded. </t>

        <t>This version of the middlebox also puts different requirements on
        the endpoint when it comes to decoder instances and handling of the
        RTP media streams providing media. As each projected SSRC can, at any
        time, provide media, the endpoint either needs to be able to handle as
        many decoder instances as the middlebox received, or have efficient
        switching of decoder contexts in a more limited set of actual decoder
        instances to cope with the switches. The application also gets more
        responsibility to update how the media provided is to be presented to
        the user.</t>

        <t>Note that this topology could potentially be seen as a media
        translator which include an on/off logic as part of its media
        translation. The main difference would be a common global SSRC space
        in the case of the Media Translator and the mapped one used in the
        above. It also has mixer aspects, as the streams it provides are not
        basically translated version, but instead they have conceptual
        property assigned to them. Thus this topology appears to be some
        hybrid between the translator and mixer model.</t>

        <t>The differences between selective forwarding middlebox and a <xref
        target="sec-media-switching">switching mixer</xref> are minor, and
        they share most properties. The above requirement on having a large
        number of decoding instances or requiring efficient switching of
        decoder contexts, are one point of difference. The other is how the
        identification is performed, where the Mixer uses CSRC to provide info
        what is included in a particular RTP packet stream that represent a
        particular concept. Selective forwarding gets the source information
        through the SSRC, and instead have to use other mechanism to make
        clear the streams current purpose.</t>
      </section>

      <section anchor="sec-ptm-switch-mcu"
               title="Point to Multipoint Using Video Switching MCUs ">
        <t>Shortcut name: Topo-Video-switch-MCU</t>

        <figure align="center" anchor="fig-ptm-switching-mcu"
                title="Point to Multipoint Using a Video Switching MCU">
          <artwork><![CDATA[
+---+      +------------+      +---+
| A |------| Multipoint |------| B |
+---+      |  Control   |      +---+
           |   Unit     |           
+---+      |   (MCU)    |      +---+
| C |------|            |------| D |
+---+      +------------+      +---+
]]></artwork>
        </figure>

        <t>This PtM topology was popular in early implementations of
        multipoint videoconferencing systems due to its simplicity, and the
        corresponding middlebox design has been known as a "video switching
        MCU". The more complex RTCP-terminating MCUs, discussed in the next
        section, became the norm, however, when technology allowed
        implementations at acceptable costs.</t>

        <t>A video switching MCU forwards to a participant a single media
        stream, selected from the available streams. The criteria for
        selection are often based on voice activity in the audio-visual
        conference, but other conference management mechanisms (like
        presentation mode or explicit floor control) are known to exist as
        well.</t>

        <t>The video switching MCU may also perform media translation to
        modify the content in bit-rate, encoding, or resolution. However, it
        still may indicate the original sender of the content through the
        SSRC. In this case, the values of the CC and CSRC fields are
        retained.</t>

        <t>If not terminating RTP, the RTCP Sender Reports are forwarded for
        the currently selected sender. All RTCP Receiver Reports are freely
        forwarded between the participants. In addition, the MCU may also
        originate RTCP control traffic in order to control the session and/or
        report on status from its viewpoint.</t>

        <t>The video switching MCU has most of the attributes of a Translator.
        However, its stream selection is a mixing behavior. This behavior has
        some RTP and RTCP issues associated with it. The suppression of all
        but one media stream results in most participants seeing only a subset
        of the sent media streams at any given time, often a single stream per
        conference. Therefore, RTCP Receiver Reports only report on these
        streams. Consequently, the media senders that are not currently
        forwarded receive a view of the session that indicates their media
        streams disappear somewhere en route. This makes the use of RTCP for
        congestion control, or any type of quality reporting, very
        problematic.</t>

        <t>To avoid the aforementioned issues, the MCU needs to implement two
        features. First, it needs to act as a Mixer (see <xref
        target="sec-ptm-mixer"/>) and forward the selected media stream under
        its own SSRC and with the appropriate CSRC values. Second, the MCU
        needs to modify the RTCP RRs it forwards between the domains. As a
        result, it is recommended that one implement a centralized video
        switching conference using a Mixer according to RFC 3550, instead of
        the shortcut implementation described here.</t>
      </section>

      <section anchor="sec-ptm-mcu"
               title="Point to Multipoint Using RTCP-Terminating MCU">
        <t>Shortcut name: Topo-RTCP-terminating-MCU</t>

        <figure align="center" anchor="fig-ptm-terminating-mcu"
                title="Point to Multipoint Using Content Modifying MCUs ">
          <artwork><![CDATA[
+---+      +------------+      +---+
| A |<---->| Multipoint |<---->| B |
+---+      |  Control   |      +---+
           |   Unit     |           
+---+      |   (MCU)    |      +---+
| C |<---->|            |<---->| D |
+---+      +------------+      +---+
]]></artwork>
        </figure>

        <t>In this PtM scenario, each participant runs an RTP point-to-point
        session between itself and the MCU. This is a very commonly deployed
        topology in multipoint video conferencing. The content that the MCU
        provides to each participant is either:<list style="letters">
            <t>a selection of the content received from the other
            participants, or</t>

            <t>the mixed aggregate of what the MCU receives from the other PtP
            paths, which are part of the same conference session.</t>
          </list></t>

        <t>In case (a), the MCU may modify the content in terms of bit-rate,
        encoding format, or resolution. No explicit RTP mechanism is used to
        establish the relationship between the original media sender and the
        version the MCU sends. In other words, the outgoing sessions typically
        use a different SSRC, and may well use a different payload type (PT),
        even if this different PT happens to be mapped to the same media type.
        This is a result of the individually negotiated session for each
        participant.</t>

        <t>In case (b), the MCU is the content source as it mixes the content
        and then encodes it for transmission to a participant. According to
        <xref target="RFC3550">RTP</xref>, the SSRC of the contributors are to
        be signalled using the CSRC/CC mechanism. In practice, today, most
        deployed MCUs do not implement this feature. Instead, the
        identification of the participants whose content is included in the
        Mixer's output is not indicated through any explicit RTP mechanism.
        That is, most deployed MCUs set the CSRC Count (CC) field in the RTP
        header to zero, thereby indicating no available CSRC information, even
        if they could identify the content sources as suggested in RTP.</t>

        <t>The main feature that sets this topology apart from what RFC 3550
        describes is the breaking of the common RTP session across the
        centralized device, such as the MCU. This results in the loss of
        explicit RTP-level indication of all participants. If one were using
        the mechanisms available in RTP and RTCP to signal this explicitly,
        the topology would follow the approach of an RTP Mixer. The lack of
        explicit indication has at least the following potential
        problems:<list style="numbers">
            <t>Loop detection cannot be performed on the RTP level. When
            carelessly connecting two misconfigured MCUs, a loop could be
            generated.</t>

            <t>There is no information about active media senders available in
            the RTP packet. As this information is missing, receivers cannot
            use it. It also deprives the client of information related to
            currently active senders in a machine-usable way, thus preventing
            clients from indicating currently active speakers in user
            interfaces, etc.</t>
          </list></t>

        <t>Note that deployed MCUs (and endpoints) rely on signalling layer
        mechanisms for the identification of the contributing sources, for
        example, a <xref target="RFC4575">SIP conferencing package</xref>.
        This alleviates, to some extent, the aforementioned issues resulting
        from ignoring RTP's CSRC mechanism.</t>
      </section>

      <section title="Split Component Endpoint">
        <t>Shortcut name: Topo-Split-Endpoint</t>

        <t>The implementation of an application may desire to send a subset of
        the application's data to each of multiple devices, each with its own
        network address. A very basic use case for this would be to separate
        audio and video processing for a particular endpoint into different
        components. For example, in a video conference room system the
        endpoint could be considered as being composed of one device handling
        the audio and another handling the video, interconnected by some
        control functions allowing them to behave as a single endpoint in all
        aspects except for transport as depicted in <xref
        target="fig-de-composite"/>.</t>

        <t>Which decomposition scheme is possible is highly dependent on the
        RTP session usage. It is not really feasible to decompose one logical
        end-point into two different transport nodes in one RTP session. A
        third party monitor would report such an attempt as two entities being
        two different end-points with a CNAME collision. As a result, a fully
        RTP conformant de-composited endpoint is one where the different
        decomposed parts use separate RTP sessions to send and/or receive
        media streams intended for them.</t>

        <figure align="center" anchor="fig-de-composite"
                title="Split Component Endpoint">
          <artwork><![CDATA[
+---------------------+
| Endpoint A          |
| Local Area Network  |
|      +------------+ |
|   +->| Audio      |<+-RTP---\
|   |  +------------+ |        \    +------+
|   |  +------------+ |         +-->|      |
|   +->| Video      |<+-RTP-------->|  B   |
|   |  +------------+ |         +-->|      |
|   |  +------------+ |        /    +------+
|   +->| Control    |<+-SIP---/
|      +------------+ |
+---------------------+
]]></artwork>
        </figure>

        <t>In the above usage, let us assume that the different RTP sessions
        are used for audio and video. The audio and video parts, however, use
        a common CNAME and also have a common clock to ensure that
        synchronization and clock drift handling works, despite the fact that
        the components are separated. Also, RTCP handling works correctly as
        long as only one part of the split endpoint is part of each RTP
        session. That way any differences in the path between A's audio entity
        and B and A's video and B are related to different SSRCs in different
        RTP sessions.</t>

        <t>The requirement that can be derived from the above usage is that
        the transport flows for each RTP session might be under common
        control, but still are addressed to what looks like different
        endpoints (based on addresses and ports). This connection diagram
        cannot be accomplished using one RTP session and thus multiple RTP
        sessions are needed.</t>
      </section>

      <section anchor="sec-asymmetric" title="Non-Symmetric Mixer/Translators">
        <t>Shortcut name: Topo-Asymmetric</t>

        <t>It is theoretically possible to construct an MCU that is a Mixer in
        one direction and a Translator in another. The main reason to consider
        this would be to allow topologies similar to <xref
        target="fig-ptm-mixer"/>, where the Mixer does not need to mix in the
        direction from B or D towards the multicast domains with A and C.
        Instead, the media streams from B and D are forwarded without changes.
        Avoiding this mixing would save media processing resources that
        perform the mixing in cases where it isn't needed. However, there
        would still be a need to mix B's stream towards D. Only in the
        direction B -> multicast domain or D -> multicast domain would
        it be possible to work as a Translator. In all other directions, it
        would function as a Mixer.</t>

        <t>The Mixer/Translator would still need to process and change the
        RTCP before forwarding it in the directions of B or D to the multicast
        domain. One issue is that A and C do not know about the mixed-media
        stream the Mixer sends to either B or D. Therefore, any reports
        related to these streams must be removed. Also, receiver reports
        related to A and C's media stream would be missing. To avoid A and C
        thinking that B and D aren't receiving A and C at all, the Mixer needs
        to insert locally generated reports reflecting the situation for the
        streams from A and C into B and D's Sender Reports. In the opposite
        direction, the Receiver Reports from A and C about B's and D's stream
        also need to be aggregated into the Mixer's Receiver Reports sent to B
        and D. Since B and D only have the Mixer as source for the stream, all
        RTCP from A and C must be suppressed by the Mixer.</t>

        <t>This topology is so problematic and it is so easy to get the RTCP
        processing wrong, that it is not recommended for implementation.</t>
      </section>

      <section anchor="sec-combining-topologies" title="Combining Topologies">
        <t>Topologies can be combined and linked to each other using Mixers or
        Translators. However, care must be taken in handling the SSRC/CSRC
        space. A Mixer does not forward RTCP from sources in other domains,
        but instead generates its own RTCP packets for each domain it mixes
        into, including the necessary Source Description (SDES) information
        for both the CSRCs and the SSRCs. Thus, in a mixed domain, the only
        SSRCs seen will be the ones present in the domain, while there can be
        CSRCs from all the domains connected together with a combination of
        Mixers and Translators. The combined SSRC and CSRC space is common
        over any Translator or Mixer. It is important to facilitate loop
        detection, something that is likely to be even more important in
        combined topologies due to the mixed behavior between the domains. Any
        hybrid, like the Topo-Video-switch-MCU or Topo-Asymmetric, requires
        considerable thought on how RTCP is dealt with.</t>
      </section>
    </section>

    <section title="Comparing Topologies">
      <t>The topologies discussed in <xref target="sec-topologies"/> have
      different properties. This section first describes these properties and
      then analyzes how these properties are supported by the different
      topologies. Note that, even if a certain property is supported within a
      particular topology concept, the necessary functionality may be optional
      to implement.</t>

      <t>Note: This section has not yet been updated with the new additions of
      topologies.</t>

      <section title="Topology Properties">
        <t/>

        <section title="All to All Media Transmission">
          <t>Multicast, at least Any Source Multicast (ASM), provides the
          functionality that everyone may send to, or receive from, everyone
          else within the session. Mesh, MCUs, Mixers, and Translators may all
          provide that functionality at least on some basic level. However,
          there are some differences in which type of reachability they
          provide.</t>

          <t>The transport Translator function called "relay", in <xref
          target="sec-ptm-translator"/>, as well as the Mesh is the ones that
          provides the emulation of ASM that is closest to true
          IP-multicast-based, all to all transmission. Media Translators,
          Mixers, and the MCU variants do not provide a fully meshed
          forwarding on the transport level; instead, they only allow limited
          forwarding of content from the other session participants.</t>

          <t>The "all to all media transmission" requires that any media
          transmitting entity considers the path to the least capable
          receiver. Otherwise, the media transmissions may overload that path.
          Therefore, a media sender needs to monitor the path from itself to
          any of the participants, to detect the currently least capable
          receiver, and adapt its sending rate accordingly. As multiple
          participants may send simultaneously, the available resources may
          vary. RTCP's Receiver Reports help performing this monitoring, at
          least on a medium time scale.</t>

          <t>The resource consumption for performing all to all transmission
          varies, where the benefit of ASM is that only one copy of each
          packet traverse a particular link. Using a relay, causes one copy
          per client to relay path and packet transmitted, however, in most
          cases the links with the multiple copies will be the ones close to
          the relay, rather than the clients unless they share LAN segment.
          The Mesh causes N-1 copies of of each transmitted packet to traverse
          the first hop link from the client, in a N client mesh. How long the
          different paths are common, is highly situation dependent.</t>

          <t>The transmission of RTCP automatically adapts to any changes in
          the number of participants due to the transmission algorithm,
          defined in the <xref target="RFC3550">RTP specification</xref>, and
          the extensions in <xref target="RFC4585">AVPF</xref> (when
          applicable). That way, the resources utilized for RTCP stay within
          the bounds configured for the session.</t>
        </section>

        <section title="Transport or Media Interoperability">
          <t>Translators, Mixers, and RTCP-terminating MCU, and Mesh with
          individual RTP sessions, all allow changing the media encoding or
          the transport to other properties of the other domain, thereby
          providing extended interoperability in cases where the participants
          lack a common set of media codecs and/or transport protocols.</t>
        </section>

        <section title="Per Domain Bit-Rate Adaptation">
          <t>Participants are most likely to be connected to each other with a
          heterogeneous set of paths. This makes congestion control in a Point
          to Multipoint set problematic. For the ASM, Mesh with common RTP
          session, and Relay scenario, each individual sender has to adapt to
          the receiver with the least capable path. This is no longer
          necessary when Media Translators, Mixers, or MCUs are involved, as
          each participant only needs to adapt to the slowest path within its
          own domain. The Translator, Mixer, or MCU topologies all require
          their respective outgoing streams to adjust the bit-rate,
          packet-rate, etc., to adapt to the least capable path in each of the
          other domains. That way one can avoid lowering the quality to the
          least-capable participant in all the domains at the cost
          (complexity, delay, equipment) of the Mixer or Translator.</t>
        </section>

        <section title="Aggregation of Media">
          <t>In the all to all media property mentioned above and provided by
          ASM, all simultaneous media transmissions share the available
          bit-rate. For participants with limited reception capabilities, this
          may result in a situation where even a minimal acceptable media
          quality cannot be accomplished. This is the result of multiple media
          streams needing to share the available resources. The solution to
          this problem is to provide for a Mixer or MCU to aggregate the
          multiple streams into a single one. This aggregation can be
          performed according to different methods. Mixing or selection are
          two common methods.</t>
        </section>

        <section title="View of All Session Participants">
          <t>The RTP protocol includes functionality to identify the session
          participants through the use of the SSRC and CSRC fields. In
          addition, it is capable of carrying some further identity
          information about these participants using the RTCP Source
          Descriptors (SDES). To maintain this functionality, it is necessary
          that RTCP is handled correctly in domain bridging function. This is
          specified for Translators and Mixers. The MCU described in <xref
          target="sec-ptm-switch-mcu"/> does not entirely fulfill this. The
          one described in <xref target="sec-ptm-mcu"/> does not support this
          at all.</t>
        </section>

        <section title="Loop Detection">
          <t>In complex topologies with multiple interconnected domains, it is
          possible to form media loops. RTP and RTCP support detecting such
          loops, as long as the SSRC and CSRC identities are correctly set in
          forwarded packets. It is likely that loop detection works for the
          MCU, described in <xref target="sec-ptm-switch-mcu"/>, at least as
          long as it forwards the RTCP between the participants. However, the
          MCU in <xref target="sec-ptm-mcu"/> will definitely break the loop
          detection mechanism.</t>

          <!--MW: Considering adding security with several aspects, source authentication, 
confidentiality, need to trust middlebox. Or is security consideration 
complete in this regards, but should be included below?-->
        </section>
      </section>

      <section title="Comparison of Topologies">
        <t>The table below attempts to summarize the properties of the
        different topologies. The legend to the topology abbreviations are:
        Topo-Point-to-Point (PtP), Topo-Multicast (Multic),
        Topo-Trns-Translator (TTrn), Topo-Media-Translator (including
        Transport Translator) (MTrn), Topo-Mixer (Mixer), Topo-Asymmetric
        (ASY), Topo-Video-switch-MCU (MCUs), and Topo-RTCP-terminating-MCU
        (MCUt). In the table below, Y indicates Yes or full support, N
        indicates No support, (Y) indicates partial support, and N/A indicates
        not applicable.</t>

        <figure>
          <artwork><![CDATA[
Property               PtP  Multic TTrn MTrn Mixer ASY MCUs MCUt  
------------------------------------------------------------------
All to All media        N    Y      Y    Y   (Y)   (Y) (Y)  (Y)   
Interoperability        N/A  N      Y    Y    Y     Y   N    Y    
Per Domain Adaptation   N/A  N      N    Y    Y     Y   N    Y    
Aggregation of media    N    N      N    N    Y    (Y)  Y    Y    
Full Session View       Y    Y      Y    Y    Y     Y  (Y)   N    
Loop Detection          Y    Y      Y    Y    Y     Y  (Y)   N    
]]></artwork>
        </figure>

        <t>Please note that the Media Translator also includes the transport
        Translator functionality.</t>

        <!--MW: Needs update with additional scenarios. -->
      </section>
    </section>

    <section title="Security Considerations">
      <t>The use of Mixers and Translators has impact on security and the
      security functions used. The primary issue is that both Mixers and
      Translators modify packets, thus preventing the use of integrity and
      source authentication, unless they are trusted devices that take part in
      the security context, e.g., the device can send <xref
      target="RFC3711">Secure Realtime Transport Protocol (SRTP) and Secure
      Realtime Transport Control Protocol (SRTCP)</xref> packets to session
      endpoints. If encryption is employed, the media Translator and Mixer
      need to be able to decrypt the media to perform its function. A
      transport Translator may be used without access to the encrypted payload
      in cases where it translates parts that are not included in the
      encryption and integrity protection, for example, IP address and UDP
      port numbers in a media stream using <xref target="RFC3711">SRTP</xref>.
      However, in general, the Translator or Mixer needs to be part of the
      signalling context and get the necessary security associations (e.g.,
      SRTP crypto contexts) established with its RTP session participants.</t>

      <t>Including the Mixer and Translator in the security context allows the
      entity, if subverted or misbehaving, to perform a number of very serious
      attacks as it has full access. It can perform all the attacks possible
      (see RFC 3550 and any applicable profiles) as if the media session were
      not protected at all, while giving the impression to the session
      participants that they are protected.</t>

      <t>Transport Translators have no interactions with cryptography that
      works above the transport layer, such as SRTP, since that sort of
      Translator leaves the RTP header and payload unaltered. Media
      Translators, on the other hand, have strong interactions with
      cryptography, since they alter the RTP payload. A media Translator in a
      session that uses cryptographic protection needs to perform
      cryptographic processing to both inbound and outbound packets.</t>

      <t>A media Translator may need to use different cryptographic keys for
      the inbound and outbound processing. For SRTP, different keys are
      required, because an RFC 3550 media Translator leaves the SSRC unchanged
      during its packet processing, and SRTP key sharing is only allowed when
      distinct SSRCs can be used to protect distinct packet streams.</t>

      <t>When the media Translator uses different keys to process inbound and
      outbound packets, each session participant needs to be provided with the
      appropriate key, depending on whether they are listening to the
      Translator or the original source. (Note that there is an architectural
      difference between RTP media translation, in which participants can rely
      on the RTP Payload Type field of a packet to determine appropriate
      processing, and cryptographically protected media translation, in which
      participants must use information that is not carried in the
      packet.)</t>

      <t>When using security mechanisms with Translators and Mixers, it is
      possible that the Translator or Mixer could create different security
      associations for the different domains they are working in. Doing so has
      some implications:</t>

      <t>First, it might weaken security if the Mixer/Translator accepts a
      weaker algorithm or key in one domain than in another. Therefore, care
      should be taken that appropriately strong security parameters are
      negotiated in all domains. In many cases, "appropriate" translates to
      "similar" strength. If a key management system does allow the
      negotiation of security parameters resulting in a different strength of
      the security, then this system should notify the participants in the
      other domains about this.</t>

      <t>Second, the number of crypto contexts (keys and security related
      state) needed (for example, in <xref target="RFC3711">SRTP</xref>) may
      vary between Mixers and Translators. A Mixer normally needs to represent
      only a single SSRC per domain and therefore needs to create only one
      security association (SRTP crypto context) per domain. In contrast, a
      Translator needs one security association per participant it translates
      towards, in the opposite domain. Considering <xref
      target="fig-ptm-multicast-translator"/>, the Translator needs two
      security associations towards the multicast domain, one for B and one
      for D. It may be forced to maintain a set of totally independent
      security associations between itself and B and D respectively, so as to
      avoid two-time pad occurrences. These contexts must also be capable of
      handling all the sources present in the other domains. Hence, using
      completely independent security associations (for certain keying
      mechanisms) may force a Translator to handle N*DM keys and related
      state; where N is the total number of SSRCs used over all domains and DM
      is the total number of domains.</t>

      <t>There exist a number of different mechanisms to provide keys to the
      different participants. One example is the choice between group keys and
      unique keys per SSRC. The appropriate keying model is impacted by the
      topologies one intends to use. The final security properties are
      dependent on both the topologies in use and the keying mechanisms'
      properties, and need to be considered by the application. Exactly which
      mechanisms are used is outside of the scope of this document. Please
      review <xref target="I-D.ietf-avtcore-rtp-security-options">RTP Security
      Options</xref> to get a better understanding of most of the available
      options.</t>
    </section>

    <section anchor="IANA" title="IANA Considerations">
      <t>This document makes no request of IANA.</t>

      <t>Note to RFC Editor: this section may be removed on publication as an
      RFC.</t>
    </section>

    <section title="Acknowledgements">
      <t>The authors would like to thank Bo Burman, Umesh Chandra, Roni Even,
      Keith Lantz, Ladan Gharai, Geoff Hunt, Mark Baugher, and Alex
      Eleftheriadis for their help in reviewing this document.</t>
    </section>
  </middle>

  <back>
    <references title="Normative References">
      <?rfc include='reference.RFC.3550'?>

      <?rfc include='reference.RFC.3711'?>

      <?rfc include='reference.RFC.4575'?>

      <?rfc include='reference.RFC.4585'?>
    </references>

    <references title="Informative References">
      <?rfc include='reference.RFC.1112'?>

      <?rfc include='reference.RFC.3022'?>

      <?rfc include='reference.RFC.3569'?>

      <?rfc include='reference.RFC.4607'?>

      <?rfc include="reference.RFC.5104"?>

      <?rfc include="reference.RFC.5760"?>

      <?rfc include='reference.RFC.5766'?>

      <?rfc include='reference.RFC.6285'?>

      <?rfc include='reference.RFC.6465'?>

      <?rfc include='reference.I-D.ietf-avtcore-rtp-security-options'?>

      <?rfc include='reference.I-D.ietf-avtcore-rtp-multi-stream'?>
    </references>
  </back>
</rfc>

PAFTECH AB 2003-20262026-04-23 20:42:43