Web Real-Time Communication (WebRTC): Media
Transport and Use of RTPUniversity of GlasgowSchool of Computing ScienceGlasgowG12 8QQUnited Kingdomcsp@csperkins.orgEricssonFarogatan 6SE-164 80 KistaSweden+46 10 714 82 87magnus.westerlund@ericsson.comAalto UniversitySchool of Electrical EngineeringEspoo02150Finlandjorg.ott@aalto.fiThe Web Real-Time Communication (WebRTC) framework provides support
for direct interactive rich communication using audio, video, text,
collaboration, games, etc. between two peers' web-browsers. This memo
describes the media transport aspects of the WebRTC framework. It
specifies how the Real-time Transport Protocol (RTP) is used in the
WebRTC context, and gives requirements for which RTP features, profiles,
and extensions need to be supported.The Real-time Transport Protocol (RTP)
provides a framework for delivery of audio and video teleconferencing
data and other real-time media applications. Previous work has defined
the RTP protocol, along with numerous profiles, payload formats, and
other extensions. When combined with appropriate signalling, these form
the basis for many teleconferencing systems.The Web Real-Time communication (WebRTC) framework provides the
protocol building blocks to support direct, interactive, real-time
communication using audio, video, collaboration, games, etc., between
two peers' web-browsers. This memo describes how the RTP framework is to
be used in the WebRTC context. It proposes a baseline set of RTP
features that are to be implemented by all WebRTC-aware end-points,
along with suggested extensions for enhanced functionality.The WebRTC overview
outlines the complete WebRTC framework, of which this memo is a
part.The structure of this memo is as follows. outlines our rationale in preparing this memo
and choosing these RTP features.
defines requirement terminology. Requirements for core RTP protocols are
described in and recommended RTP
extensions are described in . outlines mechanisms that can increase
robustness to network problems, while
describes the required congestion control and rate adaptation
mechanisms. The discussion of mandated RTP mechanisms concludes in with a review of performance monitoring and network
management tools that can be used in the WebRTC context. gives some guidelines for future incorporation of
other RTP and RTP Control Protocol (RTCP) extensions into this
framework. describes requirements placed
on the signalling channel. discusses the
relationship between features of the RTP framework and the WebRTC
application programming interface (API), and discusses RTP implementation considerations.
This memo concludes with an appendix discussing several different RTP
Topologies, and how they affect the RTP session(s) and various
implementation details of possible realization of central nodes.The RTP framework comprises the RTP data transfer protocol, the RTP
control protocol, and numerous RTP payload formats, profiles, and
extensions. This range of add-ons has allowed RTP to meet various needs
that were not envisaged by the original protocol designers, and to
support many new media encodings, but raises the question of what
extensions are to be supported by new implementations. The development
of the WebRTC framework provides an opportunity for us to review the
available RTP features and extensions, and to define a common baseline
feature set for all WebRTC implementations of RTP. This builds on the
past 15 years development of RTP to mandate the use of extensions that
have shown widespread utility, while still remaining compatible with the
wide installed base of RTP implementations where possible.RTP and RTCP extensions not discussed in this document can still be
implemented by a WebRTC end-point, but they are considered optional, are
not required for interoperability, and do not provide features needed to
address the WebRTC use cases and requirements .While the baseline set of RTP features and extensions defined in this
memo is targeted at the requirements of the WebRTC framework, it is
expected to be broadly useful for other conferencing-related uses of
RTP. In particular, it is likely that this set of RTP features and
extensions will be appropriate for other desktop or mobile video
conferencing systems, or for room-based high-quality telepresence
applications.This memo specifies various requirements levels for implementation or
use of RTP features and extensions. When we describe the importance of
RTP extensions, or the need for implementation support, we use the
following requirement levels to specify the importance of the feature in
the WebRTC framework: This word, or the terms "REQUIRED" or "SHALL",
mean that the definition is an absolute requirement of the
specification.This word, or the adjective "RECOMMENDED",
mean that there may exist valid reasons in particular circumstances
to ignore a particular item, but the full implications must be
understood and carefully weighed before choosing a different
course.This word, or the adjective "OPTIONAL", mean that
an item is truly optional. One vendor may choose to include the item
because a particular marketplace requires it or because the vendor
feels that it enhances the product while another vendor may omit the
same item. An implementation which does not include a particular
option MUST be prepared to interoperate with another implementation
which does include the option, though perhaps with reduced
functionality. In the same vein an implementation which does include
a particular option MUST be prepared to interoperate with another
implementation which does not include the option (except, of course,
for the feature the option provides.) These key words are used in a manner consistent with their
definition in . The above interpretation of
these key words applies only when written in ALL CAPS. Lower- or
mixed-case uses of these key words are not to be interpreted as carrying
special significance in this memo.We define the following terms:A sequence of RTP packets, and
associated RTCP packets, using a single synchronisation source
(SSRC) that together carries part or all of the content of a
specific Media Type from a specific sender source within a given RTP
session.As defined by ,
the endpoints belonging to the same RTP Session are those that share
a single SSRC space. That is, those endpoints can see an SSRC
identifier transmitted by any one of the other endpoints. An
endpoint can see an SSRC either directly in RTP and RTCP packets, or
as a contributing source (CSRC) in RTP packets from a mixer. The RTP
Session scope is hence decided by the endpoints' network
interconnection topology, in combination with RTP and RTCP
forwarding strategies deployed by endpoints and any interconnecting
middle nodes.The MediaStream concept defined by
the W3C in the API.Other terms are used according to their definitions from the RTP Specification and WebRTC overview documents.The following sections describe the core features of RTP and RTCP
that need to be implemented, along with the mandated RTP profiles and
payload formats. Also described are the core extensions providing
essential features that all WebRTC implementations need to implement to
function effectively on today's networks.The Real-time Transport Protocol (RTP)
is REQUIRED to be implemented as the media transport protocol
for WebRTC. RTP itself comprises two parts: the RTP data transfer
protocol, and the RTP control protocol (RTCP). RTCP is a fundamental
and integral part of RTP, and MUST be implemented in all WebRTC
applications.The following RTP and RTCP features are sometimes omitted in
limited functionality implementations of RTP, but are REQUIRED in all
WebRTC implementations: Support for use of multiple simultaneous SSRC values in a
single RTP session, including support for RTP end-points that send
many SSRC values simultaneously.Random choice of SSRC on joining a session; collision detection
and resolution for SSRC values (but see also ).Support for reception of RTP data packets containing CSRC
lists, as generated by RTP mixers, and RTCP packets relating to
CSRCs.Support for sending correct synchronization information in the
RTCP Sender Reports, to allow a receiver to implement lip-sync,
with RECOMMENDED support for the rapid RTP synchronisation
extensions (see ).Support for sending and receiving RTCP SR, RR, SDES, and BYE
packet types, with OPTIONAL support for other RTCP packet types;
implementations MUST ignore unknown RTCP packet types.Support for multiple end-points in a single RTP session, and
for scaling the RTCP transmission interval according to the number
of participants in the session; support for randomised RTCP
transmission intervals to avoid synchronisation of RTCP reports;
support for RTCP timer reconsideration.Support for configuring the RTCP bandwidth as a fraction of the
media bandwidth, and for configuring the fraction of the RTCP
bandwidth allocated to senders, e.g., using the SDP "b=" line.It is known that a significant number of legacy RTP
implementations, especially those targeted at VoIP-only systems, do
not support all of the above features, and in some cases do not
support RTCP at all. Implementers are advised to consider the
requirements for graceful degradation when interoperating with legacy
implementations.Other implementation considerations are discussed in .The complete specification of RTP for a particular application
domain requires the choice of an RTP Profile. For WebRTC use, the
"Extended Secure RTP Profile for Real-time
Transport Control Protocol (RTCP)-Based Feedback (RTP/SAVPF)"
is REQUIRED to be implemented. This builds on the basic RTP/AVP profile, the RTP profile for RTCP-based feedback
(RTP/AVPF), and the secure RTP profile
(RTP/SAVP).The RTCP-based feedback extensions are needed for the improved RTCP
timer model, that allows more flexible transmission of RTCP packets in
response to events, rather than strictly according to bandwidth. This
is vital for being able to report congestion events. These extensions
also save RTCP bandwidth, and will commonly only use the full RTCP
bandwidth allocation if there are many events that require feedback.
They are also needed to make use of the RTP conferencing extensions
discussed in .Note: The enhanced RTCP timer model defined in the RTP/AVPF
profile is backwards compatible with legacy systems that implement
only the base RTP/AVP profile, given some constraints on parameter
configuration such as the RTCP bandwidth value and "trr-int" (the
most important factor for interworking with RTP/AVP end-points via
a gateway is to set the trr-int parameter to a value representing
4 seconds).The secure RTP profile is needed to provide SRTP media encryption,
integrity protection, replay protection and a limited form of source
authentication.WebRTC implementations MUST NOT send packets using the basic
RTP/AVP profile or the RTP/AVPF profile; they MUST employ the full
RTP/SAVPF profile to protect all RTP and RTCP packets that are
generated. The default and mandatory-to-implement transforms listed in
Section 5 of SHALL apply.Implementations MUST support
DTLS-SRTP for key-management. Other key management schemes MAY
be supported.The requirement from Section 6 of that
"Audio applications operating under this profile SHOULD, at a minimum,
be able to send and/or receive payload types 0 (PCMU) and 5 (DVI4)"
applies, since of this memo mandates the
use of the RTP/SAVPF profile, which inherits this restriction from the
RTP/AVP profile.(tbd: there is ongoing discussion on whether support for other
audio and video codecs is to be mandated)Endpoints MAY signal support for multiple media formats, or
multiple configurations of a single format, provided each uses a
different RTP payload type number. An endpoint that has signalled its
support for multiple formats is REQUIRED to accept data in any of
those formats at any time, unless it has previously signalled
limitations on its decoding capability.This requirement is constrained if several media types are sent in
the same RTP session. In such a case, a source (SSRC) is restricted to
switching only between the RTP payload formats signalled for the media
type that is being sent by that source; see . To support rapid rate adaptation, RTP does
not require signalling in advance for changes between payload formats
that were signalled during session setup.An RTP sender that changes between two RTP payload types that use
different RTP clock rates MUST follow the recommendations in Section
4.1 of . RTP
receivers MUST follow the recommendations in Section 4.3 of , in order to support
sources that switch between clock rates in an RTP session (these
recommendations for receivers are backwards compatible with the case
where senders use only a single clock rate).An association amongst a set of participants communicating with RTP
is known as an RTP session. A participant can be involved in multiple
RTP sessions at the same time. In a multimedia session, each medium
has typically been carried in a separate RTP session with its own RTCP
packets (i.e., one RTP session for the audio, with a separate RTP
session using a different transport address for the video; if SDP is
used, this corresponds to one RTP session for each "m=" line in the
SDP). WebRTC implementations of RTP are REQUIRED to implement support
for multimedia sessions in this way, for compatibility with legacy
systems.In today's networks, however, with the widespread use of Network
Address/Port Translators (NAT/NAPT) and Firewalls (FW), it is
desirable to reduce the number of transport addresses used by
real-time media applications using RTP by combining multimedia traffic
in a single RTP session. (Details of how this is to be done are tbd,
but see , and .) Using a
single RTP session also effects the possibility for differentiated
treatment of media flows. This is further discussed in .WebRTC implementations of RTP are REQUIRED to support multiplexing
of a multimedia session onto a single RTP session according to (tbd).
If such RTP session multiplexing is to be used, this MUST be
negotiated during the signalling phase. Support for multiple RTP
sessions over a single UDP flow as defined by is
RECOMMENDED/OPTIONAL.(tbd: No consensus on the level of including support of Multiple
RTP sessions over a single UDP flow.)Historically, RTP and RTCP have been run on separate transport
layer addresses (e.g., two UDP ports for each RTP session, one port
for RTP and one port for RTCP). With the increased use of Network
Address/Port Translation (NAPT) this has become problematic, since
maintaining multiple NAT bindings can be costly. It also complicates
firewall administration, since multiple ports need to be opened to
allow RTP traffic. To reduce these costs and session setup times,
support for multiplexing RTP data packets and RTCP control packets on
a single port for each RTP session is REQUIRED, as specified in . For backwards compatibility, implementations are
also REQUIRED to support sending of RTP and RTCP to separate
destination ports.Note that the use of RTP and RTCP multiplexed onto a single
transport port ensures that there is occasional traffic sent on that
port, even if there is no active media traffic. This can be useful to
keep NAT bindings alive, and is the recommend method for application
level keep-alives of RTP sessions.RTCP packets are usually sent as compound RTCP packets, and requires that those compound packets start with an
Sender Report (SR) or Receiver Report (RR) packet. When using frequent
RTCP feedback messages, these general statistics are not needed in
every packet and unnecessarily increase the mean RTCP packet size.
This can limit the frequency at which RTCP packets can be sent within
the RTCP bandwidth share.To avoid this problem, specifies how to
reduce the mean RTCP message size and allow for more frequent
feedback. Frequent feedback, in turn, is essential to make real-time
applications quickly aware of changing network conditions, and to
allow them to adapt their transmission and encoding behaviour. Support
for sending RTCP feedback packets as
non-compound packets is REQUIRED when signalled. For backwards
compatibility, implementations are also REQUIRED to support the use of
compound RTCP feedback packets.To ease traversal of NAT and firewall devices, implementations are
REQUIRED to implement and use Symmetric
RTP. This requires that the IP address and port used for
sending and receiving RTP and RTCP packets are identical. The reasons
for using symmetric RTP is primarily to avoid issues with NAT and
Firewalls by ensuring that the flow is actually bi-directional and
thus kept alive and registered as flow the intended recipient actually
wants. In addition, it saves resources, specifically ports at the
end-points, but also in the network as NAT mappings or firewall state
is not unnecessary bloated. Also the amount of QoS state is
reduced.Implementations are REQUIRED to support signalled RTP SSRC values,
using the "a=ssrc:" SDP attribute defined in Sections 4.1 and 5 of
, and MUST also support the "previous-ssrc"
source attribute defined in Section 6.2 of .
Other attributes defined in MAY be
supported.Use of the "a=ssrc:" attribute is OPTIONAL. Implementations MUST
support random SSRC assignment, and MUST support SSRC collision
detection and resolution, both according to .The RTCP Canonical Name (CNAME) provides a persistent
transport-level identifier for an RTP endpoint. While the
Synchronisation Source (SSRC) identifier for an RTP endpoint can
change if a collision is detected, or when the RTP application is
restarted, its RTCP CNAME is meant to stay unchanged, so that RTP
endpoints can be uniquely identified and associated with their RTP
media streams within a set of related RTP sessions. For proper
functionality, each RTP endpoint needs to have a unique RTCP CNAME
value.The RTP specification includes
guidelines for choosing a unique RTP CNAME, but these are not
sufficient in the presence of NAT devices. In addition, long-term
persistent identifiers can be problematic from a privacy viewpoint.
Accordingly, support for generating a short-term persistent RTCP
CNAMEs following method (b) specified in Section 4.2 of "Guidelines for Choosing RTP Control Protocol (RTCP)
Canonical Names (CNAMEs)" is RECOMMENDED. Note, however, that
this does not resolve the privacy concern as there is not sufficient
randomness to avoid tracking of an end-point.An WebRTC end-point MUST support reception of any CNAME that
matches the syntax limitations specified by the RTP specification and cannot assume that any
CNAME will be according to the recommended form above.(tbd: there seems to be a growing consensus that the working group
wants randomly-chosen CNAME values; need to reference a draft that
describes how this is to be done)There are a number of RTP extensions that are either needed to obtain
full functionality, or extremely useful to improve on the baseline
performance, in the WebRTC application context. One set of these
extensions is related to conferencing, while others are more generic in
nature. The following subsections describe the various RTP extensions
mandated or suggested for use within the WebRTC context.RTP is inherently a group communication protocol. Groups can be
implemented using a centralised server, multi-unicast, or using IP
multicast. While IP multicast was popular in early deployments, in
today's practice, overlay-based conferencing dominates, typically
using one or more central servers to connect endpoints in a star or
flat tree topology. These central servers can be implemented in a
number of ways as discussed in , and in
the memo on RTP Topologies.As discussed in Section 3.5 of , the use of
a video switching MCU makes the use of RTCP for congestion control, or
any type of quality reports, very problematic. Also, as discussed in
section 3.6 of , the use of a content
modifying MCU with RTCP termination breaks RTP loop detection and
removes the ability for receivers to identify active senders. RTP
Transport Translators (Topo-Translator) are not of immediate interest
to WebRTC, although the main difference compared to point to point is
the possibility of seeing multiple different transport paths in any
RTCP feedback. Accordingly, only Point to Point (Topo-Point-to-Point),
Multiple concurrent Point to Point (Mesh) and RTP Mixers (Topo-Mixer)
topologies are needed to achieve the use-cases to be supported in
WebRTC initially. These RECOMMENDED topologies are expected to be
supported by all WebRTC end-points (these topologies require no
special RTP-layer support in the end-point if the RTP features
mandated in this memo are implemented).The RTP extensions described below to be used with centralised
conferencing -- where one RTP Mixer (e.g., a conference bridge)
receives a participant's RTP media streams and distributes them to the
other participants -- are not necessary for interoperability; an RTP
endpoint that does not implement these extensions will work correctly,
but may offer poor performance. Support for the listed extensions will
greatly improve the quality of experience and, to provide a reasonable
baseline quality, some these extensions are mandatory to be supported
by WebRTC end-points.The RTCP packets assisting in such operation are defined in the
Extended RTP Profile for Real-time Transport
Control Protocol (RTCP)-Based Feedback (RTP/AVPF) and the "Codec Control Messages in the RTP Audio-Visual
Profile with Feedback (AVPF)" (CCM) and are fully usable by the
Secure variant of this profile
(RTP/SAVPF).The Full Intra Request is defined in Sections 3.5.1 and 4.3.1 of
the Codec Control Messages. This
message is used to make the mixer request a new Intra picture from a
participant in the session. This is used when switching between
sources to ensure that the receivers can decode the video or other
predictive media encoding with long prediction chains. It is
REQUIRED that this feedback message is supported by RTP senders in
WebRTC, since it greatly improves the user experience when using
centralised mixers-based conferencing.The Picture Loss Indication is defined in Section 6.3.1 of the
RTP/AVPF profile. It is used by a
receiver to tell the sending encoder that it lost the decoder
context and would like to have it repaired somehow. This is
semantically different from the Full Intra Request above as there
there may be multiple methods to fulfill the request. It is REQUIRED
that senders understand and react to this feedback message as a loss
tolerance mechanism; receivers MAY send PLI messages.The Slice Loss Indicator is defined in Section 6.3.2 of the RTP/AVPF profile. It is used by a receiver
to tell the encoder that it has detected the loss or corruption of
one or more consecutive macroblocks, and would like to have these
repaired somehow. The use of this feedback message is OPTIONAL as a
loss tolerance mechanism.Reference Picture Selection Indication (RPSI) is defined in
Section 6.3.3 of the RTP/AVPF profile
. Some video coding standards allow the use of older
reference pictures than the most recent one for predictive coding.
If such a codec is in used, and if the encoder has learned about a
loss of encoder-decoder synchronisation, a known-as-correct
reference picture can be used for future coding. The RPSI message
allows this to be signalled.Support for RPSI messages is OPTIONAL.The temporal-spatial trade-off request and notification are
defined in Sections 3.5.2 and 4.3.2 of .
This request can be used to ask the video encoder to change the
trade-off it makes between temporal and spatial resolution, for
example to prefer high spatial image quality but low frame rate.Support for TSTR requests and notifications is OPTIONAL.This feedback message is defined in Sections 3.5.4 and 4.2.1 of
the Codec Control Messages. This
message and its notification message are used by a media receiver to
inform the sending party that there is a current limitation on the
amount of bandwidth available to this receiver. This may have
various reasons; for example, an RTP mixer may use this message to
limit the media rate of the sender being forwarded by the mixer
(without doing media transcoding) to fit the bottlenecks existing
towards the other session participants. It is REQUIRED that this
feedback message is supported. A RTP media stream sender receiving a
TMMBR for its SSRC MUST follow the limitations set by the message;
the sending of TMMBR requests is OPTIONAL.The RTP specification provides the
capability to include RTP header extensions containing in-band data,
but the format and semantics of the extensions are poorly specified.
The use of header extensions is OPTIONAL in the WebRTC context, but if
they are used, they MUST be formatted and signalled following the
general mechanism for RTP header extensions defined in , since this gives well-defined semantics to RTP
header extensions.As noted in , the requirement from the RTP
specification that header extensions are "designed so that the header
extension may be ignored" stands. To be
specific, header extensions MUST only be used for data that can safely
be ignored by the recipient without affecting interoperability, and
MUST NOT be used when the presence of the extension has changed the
form or nature of the rest of the packet in a way that is not
compatible with the way the stream is signalled (e.g., as defined by
the payload type). Valid examples might include metadata that is
additional to the usual RTP information.Many RTP sessions require synchronisation between audio, video,
and other content. This synchronisation is performed by receivers,
using information contained in RTCP SR packets, as described in the
RTP specification. This basic
mechanism can be slow, however, so it is RECOMMENDED that the rapid
RTP synchronisation extensions described in
be implemented. The rapid synchronisation extensions use the general
RTP header extension mechanism , which
requires signalling, but are otherwise backwards compatible.The Client to Mixer Audio Level
extension is an RTP header extension used by a client to
inform a mixer about the level of audio activity in the packet to
which the header is attached. This enables a central node to make
mixing or selection decisions without decoding or detailed
inspection of the payload, reducing the complexity in some types of
central RTP nodes. It can also save decoding resources in receivers,
which can choose to decode only the most relevant RTP media streams
based on audio activity levels.The Client-to-Mixer Audio Level
extension is RECOMMENDED to be implemented. If it is implemented, it
is REQUIRED that the header extensions are encrypted according to
since
the information contained in these header extensions can be
considered sensitive.The Mixer to Client Audio Level header
extension provides the client with the audio level of the
different sources mixed into a common mix by a RTP mixer. This
enables a user interface to indicate the relative activity level of
each session participant, rather than just being included or not
based on the CSRC field. This is a pure optimisations of non
critical functions, and is hence OPTIONAL to implement. If it is
implemented, it is REQUIRED that the header extensions are encrypted
according to since the
information contained in these header extensions can be considered
sensitive.There are some tools that can make RTP flows robust against Packet
loss and reduce the impact on media quality. However, they all add extra
bits compared to a non-robust stream. These extra bits need to be
considered, and the aggregate bit-rate must be rate-controlled. Thus,
improving robustness might require a lower base encoding quality, but
has the potential to deliver that quality with fewer errors. The
mechanisms described in the following sub-sections can be used to
improve tolerance to packet loss.As a consequence of supporting the RTP/SAVPF profile,
implementations will support negative acknowlegdements (NACKs) for RTP
data packets . This feedback can be used to
inform a sender of the loss of particular RTP packets, subject to the
capacity limitations of the RTCP feedback channel. A sender can use
this information to optimise the user experience by adapting the media
encoding to compensate for known lost packets, for example.Senders are REQUIRED to understand the Generic NACK message defined
in Section 6.2.1 of , but MAY choose to ignore
this feedback (following Section 4.2 of ).
Receivers MAY send NACKs for missing RTP packets; provides some guidelines on when to send NACKs. It
is not expected that a receiver will send a NACK for every lost RTP
packet, rather it should consider the cost of sending NACK feedback,
and the importance of the lost packet, to make an informed decision on
whether it is worth telling the sender about a packet loss event.The RTP Retransmission Payload Format
offers the ability to retransmit lost packets based on NACK feedback.
Retransmission needs to be used with care in interactive real-time
applications to ensure that the retransmitted packet arrives in time
to be useful, but can be effective in environments with relatively low
network RTT (an RTP sender can estimate the RTT to the receivers using
the information in RTCP SR and RR packets). The use of retransmissions
can also increase the forward RTP bandwidth, and can potentially
worsen the problem if the packet loss was caused by network
congestion. We note, however, that retransmission of an important lost
packet to repair decoder state may be lower cost than sending a full
intra frame. It is not appropriate to blindly retransmit RTP packets
in response to a NACK. The importance of lost packets and the
likelihood of them arriving in time to be useful needs to be
considered before RTP retransmission is used.Receivers are REQUIRED to implement support for RTP retransmission
packets . Senders MAY send RTP retransmission
packets in response to NACKs if the RTP retransmission payload format
has been negotiated for the session, and if the sender believes it is
useful to send a retransmission of the packet(s) referenced in the
NACK. An RTP sender is not expected to retransmit every NACKed
packet.The use of Forward Error Correction (FEC) can provide an effective
protection against some degree of packet loss, at the cost of steady
bandwidth overhead. There are several FEC schemes that are defined for
use with RTP. Some of these schemes are specific to a particular RTP
payload format, others operate across RTP packets and can be used with
any payload format. It should be noted that using redundancy encoding
or FEC will lead to increased playout delay, which should be
considered when choosing the redundancy or FEC formats and their
respective parameters.If an RTP payload format negotiated for use in a WebRTC session
supports redundant transmission or FEC as a standard feature of that
payload format, then that support MAY be used in the WebRTC session,
subject to any appropriate signalling.There are several block-based FEC schemes that are designed for use
with RTP independent of the chosen RTP payload format. At the time of
this writing there is no consensus on which, if any, of these FEC
schemes is appropriate for use in the WebRTC context. Accordingly,
this memo makes no recommendation on the choice of block-based FEC for
WebRTC use.WebRTC will be used in very varied network environment with a
heterogeneous set of link technologies, including wired and wireless,
interconnecting peers at different topological locations resulting in
network paths with widely varying one way delays, bit-rate capacity,
load levels and traffic mixes. In addition, individual end-points will
open one or more WebRTC sessions between one or more peers. Each of
these session may contain different mixes of media and data flows.
Asymmetric usage of media bit-rates and number of RTP media streams is
also to be expected. A single end-point may receive zero to many
simultaneous RTP media streams while itself transmitting one or more
streams.The WebRTC application is very dependent from a quality perspective
on the media adaptation working well so that an end-point doesn't
transmit significantly more than the path is capable of handling. If it
would, the result would be high levels of packet loss or delay spikes
causing media quality degradation.WebRTC applications using more than a single RTP media stream of any
media type or data flows have an additional concern. In this case, the
different flows should try to avoid affecting each other negatively. In
addition, in case there is a resource limitation, the available
resources need to be shared. How to share them is something the
application should prioritize so that the limitations in quality or
capabilities are those that have the least impact on the
application.Overall, the diversity of operating environments lead to the need for
functionality that adapts to the available capacity and that competes
fairly with other network flows. If it would not compete fairly enough
WebRTC could be used as an attack method for starving out other traffic
on specific links as long as the attacker is able to create traffic
across the links in question. A possible attack scenario is to use a
web-service capable of attracting large numbers of end-points, combined
with BGP routing state to have the server pick client pairs to drive
traffic to specific paths.The above clearly motivates the need for a well working media
adaptation mechanism. This mechanism also have a number of requirements
on what services it should provide and what performance it needs to
provide.The biggest issue is that there are no standardised and ready to use
mechanism that can simply be included in WebRTC. Thus, there will be a
need for the IETF to produce such a specification. Therefore, the
suggested way forward is to specify requirements on any solution for the
media adaptation. For now, we propose that these requirements be
documented in this specification. In addition, a proposed detailed
solution will be developed, but is expected to take longer time to
finalize than this document.Requirements for congestion control of WebRTC sessions are
discussed in .Implementations are REQUIRED to implement the RTP circuit breakers
described in .(tbd: Should add the RTP/RTCP Mechanisms that an WebRTC
implementation is required to support. Potential candidates include
Transmission Timestamps (RFC 5450).)The session establishment signalling will establish certain
boundary that the media bit-rate adaptation can act within. First of
all the set of media codecs provide practical limitations in the
supported bit-rate span where it can provide useful quality, which
packetization choices that exist. Next the signalling can establish
maximum media bit-rate boundaries using SDP b=AS or b=CT.(tbd: This section needs expanding on how to use these limits)Experience with the congestion control algorithms of TCP , TFRC , and DCCP , , , has shown that feedback on packet arrivals needs
to be sent roughly once per round trip time. We note that the
real-time media traffic may not have to adapt to changing path
conditions as rapidly as needed for the elastic applications TCP was
designed for, but frequent feedback is still required to allow the
congestion control algorithm to track the path dynamics.The total RTCP bandwidth is limited in its transmission rate to a
fraction of the RTP traffic (by default 5%). RTCP packets are larger
than, e.g., TCP ACKs (even when non-compound RTCP packets are used).
The RTP media stream bit rate thus limits the maximum feedback rate as
a function of the mean RTCP packet size.Interactive communication may not be able to afford waiting for
packet losses to occur to indicate congestion, because an increase in
playout delay due to queuing (most prominent in wireless networks) may
easily lead to packets being dropped due to late arrival at the
receiver. Therefore, more sophisticated cues may need to be reported
-- to be defined in a suitable congestion control framework as noted
above -- which, in turn, increase the report size again. For example,
different RTCP XR report blocks (jointly) provide the necessary
details to implement a variety of congestion control algorithms, but
the (compound) report size grows quickly.In group communication, the share of RTCP bandwidth needs to be
shared by all group members, reducing the capacity and thus the
reporting frequency per node.Example: assuming 512 kbit/s video yields 3200 bytes/s RTCP
bandwidth, split across two entities in a point-to-point session. An
endpoint could thus send a report of 100 bytes about every 70ms or for
every other frame in a 30 fps video.There are legacy implementations that do not implement RTCP, and
hence do not provide any congestion feedback. Congestion control
cannot be performed with these end-points. WebRTC implementations that
must interwork with such end-points MUST limit their transmission to a
low rate, equivalent to a VoIP call using a low bandwidth codec, that
is unlikely to cause any significant congestion.When interworking with legacy implementations that support RTCP
using the RTP/AVP profile, congestion
feedback is provided in RTCP RR packets every few seconds.
Implementations that are required to interwork with such end-points
MUST ensure that they keep within the RTP circuit
breaker constraints to limit the congestion they can cause.If a legacy end-point supports RTP/AVPF, this enables negotiation
of important parameters for frequent reporting, such as the "trr-int"
parameter, and the possibility that the end-point supports some useful
feedback format for congestion control purpose such as TMMBR. Implementations that are required to
interwork with such end-points MUST ensure that they stay within the
RTP circuit
breaker constraints to limit the congestion they can cause, but
may find that they can achieve better congestion response depending on
the amount of feedback that is available.RTCP does contains a basic set of RTP flow monitoring metrics like
packet loss and jitter. There are a number of extensions that could be
included in the set to be supported. However, in most cases which RTP
monitoring that is needed depends on the application, which makes it
difficult to select which to include when the set of applications is
very large.Exposing some metrics in the WebRTC API should be considered allowing
the application to gather the measurements of interest. However,
security implications for the different data sets exposed will need to
be considered in this.(tbd: If any RTCP XR metrics should be added is still an open
question, but possible to extend at a later stage)It is possible that the core set of RTP protocols and RTP extensions
specified in this memo will prove insufficient for the future needs of
WebRTC applications. In this case, future updates to this memo MUST be
made following the Guidelines for Writers of RTP
Payload Format Specifications and
Guidelines for Extending the RTP Control Protocol, and SHOULD
take into account any future guidelines for extending RTP and related
protocols that have been developed.Authors of future extensions are urged to consider the wide range of
environments in which RTP is used when recommending extensions, since
extensions that are applicable in some scenarios can be problematic in
others. Where possible, the WebRTC framework should adopt RTP extensions
that are of general utility, to enable easy gatewaying to other
applications using RTP, rather than adopt mechanisms that are narrowly
targeted at specific WebRTC use cases.RTP is built with the assumption of an external signalling channel
that can be used to configure the RTP sessions and their features. The
basic configuration of an RTP session consists of the following
parameters:The name of the RTP profile to be used in
session. The RTP/AVP and RTP/AVPF profiles can interoperate on basic
level, as can their secure variants RTP/SAVP and RTP/SAVPF. The secure variants of the
profiles do not directly interoperate with the non-secure variants,
due to the presence of additional header fields in addition to any
cryptographic transformation of the packet content. As WebRTC
requires the usage of the RTP/SAVPF profile this can be inferred as
there is only a single profile, but in SDP this is still required
information to be signalled. Interworking functions may transform
this into RTP/SAVP for a legacy use case by indicating to the WebRTC
end-point a RTP/SAVPF end-point and limiting the usage of the a=rtcp
attribute to indicate a trr-int value of 4 seconds.Source and destination IP
address(s) and ports for RTP and RTCP MUST be signalled for each RTP
session. In WebRTC these transport addresses will be provided by ICE
that signals candidates and arrives at nominated candidate address
pairs. If RTP and RTCP multiplexing is
to be used, such that a single port is used for RTP and RTCP flows,
this MUST be signalled (see ). If
several RTP sessions are to be multiplexed onto a single transport
layer flow, this MUST also be signalled (see ).The
mapping between media type names (and hence the RTP payload formats
to be used) and the RTP payload type numbers MUST be signalled. Each
media type MAY also have a number of media type parameters that MUST
also be signalled to configure the codec and RTP payload format (the
"a=fmtp:" line from SDP).The RTP extensions to be used SHOULD
be agreed upon, including any parameters for each respective
extension. At the very least, this will help avoiding using
bandwidth for features that the other end-point will ignore. But for
certain mechanisms there is requirement for this to happen as
interoperability failure otherwise happens.Support for exchanging RTCP Bandwidth
values to the end-points will be necessary. This SHALL be done as
described in "Session Description Protocol
(SDP) Bandwidth Modifiers for RTP Control Protocol (RTCP)
Bandwidth", or something semantically equivalent. This also
ensures that the end-points have a common view of the RTCP
bandwidth, this is important as too different view of the bandwidths
may lead to failure to interoperate.These parameters are often expressed in SDP messages conveyed within
an offer/answer exchange. RTP does not depend on SDP or on the
offer/answer model, but does require all the necessary parameters to be
agreed upon, and provided to the RTP implementation. We note that in the
WebRTC context it will depend on the signalling model and API how these
parameters need to be configured but they will be need to either set in
the API or explicitly signalled between the peers.The following sections describe how the WebRTC API features map onto
the RTP mechanisms described in this memo.The WebRTC API and its media function have the concept of a WebRTC
MediaStream that consists of zero or more tracks. A track is an
individual stream of media from any type of media source like a
microphone or a camera, but also conceptual sources, like a audio mix
or a video composition, are possible. The tracks within a WebRTC
MediaStream are expected to be synchronized.A track correspond to the media received with one particular SSRC.
There might be additional SSRCs associated with that SSRC, like for
RTP retransmission or Forward Error Correction. However, one SSRC will
identify an RTP media stream and its timing.As a result, a WebRTC MediaStream is a collection of SSRCs carrying
the different media included in the synchronised aggregate. Therefore,
also the synchronization state associated with the included SSRCs are
part of concept. It is important to consider that there can be
multiple different WebRTC MediaStreams containing a given Track
(SSRC). To avoid unnecessary duplication of media at the transport
level in such cases, a need arises for a binding defining which WebRTC
MediaStreams a given SSRC is associated with at the signalling
level.A proposal for how the binding between WebRTC MediaStreams and SSRC
can be done is specified in "Cross Session Stream
Identification in the Session Description Protocol".(tbd: This text must be improved and achieved consensus on. Interim
meeting in June 2012 shows large differences in opinions.)The following provide some guidance on the implementation of the RTP
features described in this memo.This section discusses RTP functionality that is part of the RTP
standard, required by decisions made, or to enable use cases raised and
their motivations. This discussion is from an WebRTC end-point
perspective. It will occasionally talk about central nodes, but as this
specification is for an end-point, this is where the focus lies. For
more discussion on the central nodes and details about RTP topologies
please see .The section will touch on the relation with certain RTP/RTCP
extensions, but will focus on the RTP core functionality. The definition
of what functionalities and the level of requirement on implementing it
is defined in .An RTP session is an association among RTP nodes, which have one
common SSRC space. An RTP session can include any number of end-points
and nodes sourcing, sinking, manipulating or reporting on the RTP
media streams being sent within the RTP session. A PeerConnection
being a point-to-point association between an end-point and another
node. That peer node may be both an end-point or centralized
processing node of some type; thus, the RTP session may terminate
immediately on the far end of the PeerConnection, but it may also
continue as further discussed below in Multiparty and Multiple RTP End-points.A PeerConnection can contain one or more RTP session depending on
how it is setup and how many UDP flows it uses. A common usage has
been to have one RTP session per media type, e.g. one for audio and
one for video, each sent over different UDP flows. However, the
default usage in WebRTC will be to use one RTP session for all media
types. This usage then uses only one UDP flow, as also RTP and RTCP multiplexing is mandated.
However, for legacy interworking and network prioritization based on flows, a
WebRTC end-point needs to support a mode of operation where one RTP
session per media type is used. Currently, each RTP session must use
its own UDP flow. Discussions are ongoing if a solution enabling
multiple RTP sessions over a single UDP flow, see .The multi-unicast- or mesh-based multi-party topology is a good example for
this section as it concerns the relation between RTP sessions and
PeerConnections. In this topology, each participant sends individual
unicast RTP/UDP/IP flows to each of the other participants using
independent PeerConnections in a full mesh. This topology has the
benefit of not requiring central nodes. The downside is that it
increases the used bandwidth at each sender by requiring one copy of
the RTP media streams for each participant that are part of the same
session beyond the sender itself. Hence, this topology is limited to
scenarios with few participants unless the media is very low
bandwidth.The multi-unicast topology could be implemented as a single RTP
session, spanning multiple peer-to-peer transport layer connections,
or as several pairwise RTP sessions, one between each pair of peers.
To maintain a coherent mapping between the relation between RTP
sessions and PeerConnections we recommend that one implements this as
individual RTP sessions. The only downside is that end-point A will
not learn of the quality of any transmission happening between B and C
based on RTCP. This has not been seen as a significant downside as no
one has yet seen a clear need for why A would need to know about the
B's and C's communication. An advantage of using separate RTP sessions
is that it enables using different media bit-rates to the different
peers, thus not forcing B to endure the same quality reductions if
there are limitations in the transport from A to C as C will.A WebRTC end-point may have multiple cameras, microphones or audio
inputs and thus a single end-point can source multiple RTP media
streams of the same media type concurrently. Even if an end-point does
not have multiple media sources of the same media type it will be
required to support transmission using multiple SSRCs concurrently in
the same RTP session. This is due to the requirement on an WebRTC
end-point to support multiple media types in one RTP session. For
example, one audio and one video source can result in the end-point
sending with two different SSRCs in the same RTP session. As
multi-party conferences are supported, as discussed below in , a WebRTC end-point will need to be capable
of receiving, decoding and playout multiple RTP media streams of the
same type concurrently.tbd: Are any mechanism needed to signal limitations in the number
of SSRC that an end-point can handle?There are numerous situations and clear use cases for WebRTC
supporting RTP sessions supporting multi-party. This can be realized
in a number of ways using a number of different implementation
strategies. In the following, the focus is on the different set of
WebRTC end-point requirements that arise from different sets of
multi-party topologies.The multi-unicast mesh-based
multi-party topology discussed above provides a non-centralized
solution but may incur a heavy tax on the end-points' outgoing paths.
It may also consume large amount of encoding resources if each
outgoing stream is specifically encoded. If an encoding is transmitted
to multiple parties, as in some implementations of the mesh case, a
requirement on the end-point becomes to be able to create RTP media
streams suitable for multiple destinations requirements. These
requirements may both be dependent on transport path and the different
end-points preferences related to playout of the media.A Mixer is an RTP end-point
that optimizes the transmission of RTP media streams from certain
perspectives, either by only sending some of the received RTP media
stream to any given receiver or by providing a combined RTP media
stream out of a set of contributing streams. There are various methods
of implementation as discussed in . A common
aspect is that these central nodes may use a number of tools to
control the media encoding provided by a WebRTC end-point. This
includes functions like requesting breaking the encoding chain and
have the encoder produce a so called Intra frame. Another is limiting
the bit-rate of a given stream to better suit the mixer view of the
multiple down-streams. Others are controlling the most suitable
frame-rate, picture resolution, the trade-off between frame-rate and
spatial quality.A mixer gets a significant responsibility to correctly perform
congestion control, source identification, manage synchronization
while providing the application with suitable media optimizations.Mixers also need to be trusted nodes when it comes to security as
it manipulates either RTP or the media itself before sending it on
towards the end-point(s), thus they must be able to decrypt and then
encrypt it before sending it out.The RTP standard requires any RTP
implementation to have support for detecting and handling SSRC
collisions, i.e., resolve the conflict when two different end-points
use the same SSRC value. This requirement also applies to WebRTC
end-points. There are several scenarios where SSRC collisions may
occur.In a point-to-point session where each SSRC is associated with
either of the two end-points and where the main media carrying SSRC
identifier will be announced in the signalling channel, a collision is
less likely to occur due to the information about used SSRCs provided
by Source-Specific SDP Attributes. Still
if both end-points start uses an new SSRC identifier prior to having
signalled it to the peer and received acknowledgement on the
signalling message, there can be collisions. The Source-Specific SDP Attributes contains no
mechanism to resolve SSRC collisions or reject a end-points usage of
an SSRC.There could also appear unsignalled SSRCs. This is more likely than
it appears as certain RTP functions need extra SSRCs to provide
functionality related to another (the "main") SSRC, for example, SSRC multiplexed RTP retransmission. In those
cases, an end-point can create a new SSRC that strictly doesn't need
to be announced over the signalling channel to function correctly on
both RTP and PeerConnection level.The more likely case for SSRC collision is that multiple end-points
in a multiparty conference create new sources and signals those
towards the central server. In cases where the SSRC/CSRC are
propagated between the different end-points from the central node
collisions can occur.Another scenario is when the central node manages to connect an
end-point's PeerConnection to another PeerConnection the end-point
already has, thus forming a loop where the end-point will receive its
own traffic. While is is clearly considered a bug, it is important
that the end-point is able to recognise and handle the case when it
occurs.Contributing Sources (CSRC) is a functionality in the RTP header
that allows an RTP node to combine media packets from multiple sources
into one and to identify which sources yielded the result. For WebRTC
end-points, supporting contributing sources is trivial. The set of
CSRCs is provided in a given RTP packet. This information can then be
exposed to the applications using some form of API, possibly a mapping
back into WebRTC MediaStream identities to avoid having to expose two
namespaces and the handling of SSRC collision handling to the
JavaScript.(tbd: should the API provide the ability to add a CSRC list to an
outgoing packet? this is only useful if the sender is mixing
content)There are also at least one extension that depends on the CRSRC
list being used: the Mixer-to-client audio
level, which enhances the information provided by the CSRC to
actual energy levels for audio for each contributing source.When an end-point sends media from more than one media source, it
needs to consider if (and which of) these media sources are to be
synchronized. In RTP/RTCP, synchronisation is provided by having a set
of RTP media streams be indicated as coming from the same
synchronisation context and logical end-point by using the same CNAME
identifier.The next provision is that the internal clocks of all media
sources, i.e., what drives the RTP timestamp, can be correlated to a
system clock that is provided in RTCP Sender Reports encoded in an NTP
format. By correlating all RTP timestamps to a common system clock for
all sources, the timing relation of the different RTP media streams,
also across multiple RTP sessions can be derived at the receiver and,
if desired, the streams can be synchronized. The requirement is for
the media sender to provide the correlation information; it is up to
the receiver to use it or not.Some usages of RTP beyond the recommend topologies result in that
an WebRTC end-point sending media in an RTP session out over a single
PeerConnection will receive receiver reports from multiple RTP
receivers. Note that receiving multiple receiver reports is expected
because any RTP node that has multiple SSRCs is required to report to
the media sender. The difference here is that they are multiple nodes,
and thus will likely have different path characteristics.RTP Mixers may create a situation where an end-point experiences a
situation in-between a session with only two end-points and multiple
end-points. Mixers are expected to not forward RTCP reports regarding
RTP media streams across themselves. This is due to the difference in
the RTP media streams provided to the different end-points. The
original media source lacks information about a mixer's manipulations
prior to sending it the different receivers. This setup also results
in that an end-point's feedback or requests goes to the mixer. When
the mixer can't act on this by itself, it is forced to go to the
original media source to fulfill the receivers request. This will not
necessarily be explicitly visible any RTP and RTCP traffic, but the
interactions and the time to complete them will indicate such
dependencies.The topologies in which an end-point receives receiver reports from
multiple other end-points are the centralized relay, multicast and an
end-point forwarding an RTP media stream. Having multiple RTP nodes
receive an RTP flow and send reports and feedback about it has several
impacts. As previously discussed
any codec control and rate control needs to be capable of merging the
requirements and preferences to provide a single best encoding
according to the situation RTP media stream. Specifically, when it
comes to congestion control it needs to be capable of identifying the
different end-points to form independent congestion state information
for each different path.Providing source authentication in multi-party scenarios is a
challenge. In the mixer-based topologies, end-points source
authentication is based on, firstly, verifying that media comes from
the mixer by cryptographic verification and, secondly, trust in the
mixer to correctly identify any source towards the end-point. In RTP
sessions where multiple end-points are directly visible to an
end-point, all end-points will have knowledge about each others'
master keys, and can thus inject packets claimed to come from another
end-point in the session. Any node performing relay can perform
non-cryptographic mitigation by preventing forwarding of packets that
have SSRC fields that came from other end-points before. For
cryptographic verification of the source SRTP would require additional
security mechanisms, like TESLA for
SRTP.This section discusses simulcast in the meaning of providing a
node, for example a Mixer, with multiple different encoded versions of
the same media source. In the WebRTC context, this could be
accomplished in two ways. One is to establish multiple PeerConnection
all being feed the same set of WebRTC MediaStreams. Another method is
to use multiple WebRTC MediaStreams that are differently configured
when it comes to the media parameters. This would result in that
multiple different RTP Media Streams (SSRCs) being in used with
different encoding based on the same media source (camera,
microphone).When intending to use simulcast it is important that this is made
explicit so that the end-points don't automatically try to optimize
away the different encodings and provide a single common version.
Thus, some explicit indications that the intent really is to have
different media encodings is likely required. It should be noted that
it might be a central node, rather than an WebRTC end-point that would
benefit from receiving simulcasted media sources.tbd: How to perform simulcast needs to be determined and the
appropriate API or signalling for its usage needs to be defined.There are use cases for differentiated treatment of RTP media
streams. Such differentiation can happen at several places in the
system. First of all is the prioritization within the end-point
sending the media, which controls, both which RTP media streams that
will be sent, and their allocation of bit-rate out of the current
available aggregate as determined by the congestion control.Secondly, the network can prioritize packet flows, including RTP
media streams. Typically, differential treatment includes two steps,
the first being identifying whether an IP packet belongs to a class
which should be treated differently, the second the actual mechanism
to prioritize packets. This is done according to three methods; The end-point marks a packet with a
diffserv code point to indicate to the network that the packet
belongs to a particular class.Packets that shall be given a particular
treatment are identified using a combination of IP and port
address.A network classifier (DPI)
inspects the packet and tries to determine if the packet
represents a particular application and type that is to be
prioritized.With the exception of diffserv both flow based and DPI have issues
with running multiple media types and flows on a single UDP flow,
especially when combined with data transport (SCTP/DTLS). DPI has
issues because multiple types of flows are aggregated and thus it
becomes more difficult to analyse them. The flow-based differentiation
will provide the same treatment to all packets within the flow, i.e.,
relative prioritization is not possible. Moreover, if the resources
are limited it may not be possible to provide differential treatment
compared to best-effort for all the flows in a WebRTC application.When flow-based differentiation is available the WebRTC application
needs to know about it so that it can provide the separation of the
RTP media streams onto different UDP flows to enable a more granular
usage of flow based differentiation.Diffserv assumes that either the end-point or a classifier can mark
the packets with an appropriate DSCP so that the packets are treated
according to that marking. If the end-point is to mark the traffic two
requirements arise in the WebRTC context: 1) The WebRTC application or
browser has to know which DSCP to use and that it can use them on some
set of RTP media streams. 2) The information needs to be propagated to
the operating system when transmitting the packet.tbd: The model for providing differentiated treatment needs to be
evolved. This includes:How the application can prioritize MediaStreamTracks
differently in the APIHow the browser or application determine availability of
transport differentiationHow to learn about any configuration information for transport
differentiation, such as DSCPs.This memo makes no request of IANA.Note to RFC Editor: this section may be removed on publication as an
RFC.RTP and its various extensions each have their own security
considerations. These should be taken into account when considering the
security properties of the complete suite. We currently don't think this
suite creates any additional security issues or properties. The use of
SRTP will provide protection or mitigation
against most of the fundamental issues by offering confidentiality,
integrity and partial source authentication. A mandatory to implement
media security solution will be required to be picked. We currently
don't discuss the key-management aspect of SRTP in this memo, that needs
to be done taking the WebRTC communication model into account.Privacy concerns are under discussion and the generation of
non-trackable CNAMEs are under discussion.The guidelines in apply when using variable
bit rate (VBR) audio codecs, for example Opus or the Mixer audio level
header extensions.Security considerations for the WebRTC work are discussed in .The authors would like to thank Harald Alvestrand, Cary Bran, Charles
Eckel and Cullen Jennings for valuable feedback.RTP
supports both unicast and group communication, with participants being
connected using wide range of transport-layer topologies. Some of these
topologies involve only the end-points, while others use RTP translators
and mixers to provide in-network processing. Properties of some RTP
topologies are discussed in , and we further
describe those expected to be useful for WebRTC in the following. We
also goes into important RTP session aspects that the topology or
implementation variant can place on a WebRTC end-point.This section includes RTP topologies beyond the recommended ones.
This in an attempt to highlight the differencies and the in many case
small differences in implementation to support a larger set of possible
topologies.The point-to-point RTP topology is
the simplest scenario for WebRTC applications. This is going to be
very common for user to user calls.This being the basic one lets use the topology to high-light a
couple of details that are common for all RTP usage in the WebRTC
context. First is the intention to multiplex RTP and RTCP over the
same UDP-flow. Secondly is the question of using only a single RTP
session or one per media type for legacy interoperability. Thirdly is
the question of using multiple sender sources (SSRCs) per
end-point.Historically, RTP and RTCP have been run on separate UDP ports.
With the increased use of Network Address/Port Translation (NAPT) this
has become problematic, since maintaining multiple NAT bindings can be
costly. It also complicates firewall administration, since multiple
ports must be opened to allow RTP traffic. To reduce these costs and
session setup times, support for multiplexing
RTP data packets and RTCP control packets on a single port will
be supported.In cases where there is only one type of media (e.g., a voice-only
call) this topology will be implemented as a single RTP session, with
bidirectional flows of RTP and RTCP packets, all then multiplexed onto
a single 5-tuple. If multiple types of media are to be used (e.g.,
audio and video), then each type media can be sent as a separate RTP
session using a different 5-tuple, allowing for separate transport
level treatment of each type of media. Alternatively, all types of
media can be multiplexed onto a single 5-tuple as a single RTP
session, or as several RTP sessions if using a demultiplexing shim.
Multiplexing different types of media onto a single 5-tuple places
some limitations on how RTP is used, as described in "RTP
Multiplexing Architecture". It is not expected that these
limitations will significantly affect the scenarios targeted by
WebRTC, but they may impact interoperability with legacy systems.An RTP session have good support for simultanously transport
multiple media sources. Each media source uses an unique SSRC
identifier and each SSRC has independent RTP sequence number and
timestamp spaces. This is being utilized in WebRTC for several cases.
One is to enable multiple media sources of the same type, an end-point
that has two video cameras can potentially transmitt video from both
to its peer(s). Another usage is when a single RTP session is being
used for both multiple media types, thus an end-point can transmit
both audio and video to the peer(s). Thirdly to support multi-party
cases as will be discussed below support for multiple SSRC of the same
media type are required.Thus we can introduce a couple of different notiations in the below
two alternate figures of a single peer connection in a a point to
point setup. The first depicting a setup where the peer connection
established has two different RTP sessions, one for audio and one for
video. The second one using a single RTP session. In both cases A has
two video streams to send and one audio stream. B has only one audio
and video stream. These are used to illustrate the relation between a
peerConnection, the UDP flow(s), the RTP session(s) and the SSRCs that
will be used in the later cases also. In the below figures RTCP flows
are not included. They will flow bi-directionally between any RTP
session instances in the different nodes.As can be seen above in the Point to
Point: Multiple RTP sessions the single Peer Connection
contains two RTP sessions over different UDP flows UDP 1 and UDP 2,
i.e. their 5-tuples will be different, normally on source and
destination ports. The first RTP session (RTP1) carries audio, one
stream in each direction AA1 and BA1. The second RTP session contains
two video streams from A (AV1 and AV2) and one from B to A (BV1).In there is only a single
UDP flow and RTP session (RTP1). This RTP session carries a total of
five (5) RTP media streams (SSRCs). From A to B there is Audio (AA1)
and two video (AV1 and AV2). From B to A there is Audio (BA1) and
Video (BV1).For small multiparty calls, it is practical to set up a
multi-unicast topology (); unfortunately
not discussed in the RTP Topologies RFC.
In this topology, each participant sends individual unicast RTP/UDP/IP
flows to each of the other participants using independent
PeerConnections in a full mesh.This topology has the benefit of not requiring central nodes. The
downside is that it increases the used bandwidth at each sender by
requiring one copy of the RTP media streams for each participant that
are part of the same session beyond the sender itself. Hence, this
topology is limited to scenarios with few participants unless the
media is very low bandwidth. The multi-unicast topology could be
implemented as a single RTP session, spanning multiple peer-to-peer
transport layer connections, or as several pairwise RTP sessions, one
between each pair of peers. To maintain a coherent mapping between the
relation between RTP sessions and PeerConnections we recommend that
one implements this as individual RTP sessions. The only downside is
that end-point A will not learn of the quality of any transmission
happening between B and C based on RTCP. This has not been seen as a
significant downside as now one has yet seen a need for why A would
need to know about the B's and C's communication. An advantage of
using separate RTP sessions is that it enables using different media
bit-rates to the differnt peers, thus not forcing B to endure the same
quality reductions if there are limiations in the transport from A to
C as C will.Lets review how the RTP sessions looks from A's perspective by
considering both how the media is a handled and what PeerConnections
and RTP sessions that are setup in .
A's microphone is captured and the digital audio can then be feed into
two different encoder instances each beeing associated with two
different PeerConnections (PeerC1 and PeerC2) each containing
independent RTP sessions (RTP1 and RTP2). The SSRCs in each RTP
session will be completely independent and the media bit-rate produced
by the encoder can also be tuned to address any congestion control
requirements between A and B differently then for the path A to C.For media encodings which are more resource consuming, like video,
one could expect that it will be common that end-points that are
resource costrained will use a different implementation strategy where
the encoder is shared between the different PeerConnections as shown
below .This will clearly save resources consumed by encoding but does
introduce the need for the end-point A to make decisions on how it
encodes the media so it suites delivery to both B and C. This is not
limited to congestion control, also prefered resolution to receive
based on dispaly area available is another aspect requiring
consideration. The need for this type of descion logic does arise in
several different topologies and implementation.An mixer is a centralised point
that selects or mixes content in a conference to optimise the RTP
session so that each end-point only needs connect to one entity, the
mixer. The mixer can also reduce the bit-rate needed from the mixer
down to a conference participants as the media sent from the mixer to
the end-point can be optimised in different ways. These optimisations
include methods like only choosing media from the currently most
active speaker or mixing together audio so that only one audio stream
is required in stead of 3 in the depicted scenario.Mixers has two downsides, the first is that the mixer must be a
trusted node as they either performs media operations or at least
repacketize the media. Both type of operations requires when using
SRTP that the mixer verifies integrity, decrypts the content, perform
its operation and form new RTP packets, encrypts and integegrity
protect them. This applies to all types of mixers described below.The second downside is that all these operations and optimization
of the session requires processing. How much depends on the
implementation as will become evident below.The implementation of an mixer can take several different forms and
we will discuss the main themes available that doesn't break RTP.Please note that a Mixer could also contain translator
functionalities, like a media transcoder to adjust the media bit-rate
or codec used on a particular RTP media stream.This type of mixer is one which clearly can be called RTP mixer
is likely the one that most thinks of when they hear the term mixer.
Its basic patter of operation is that it will receive the different
participants RTP media stream. Select which that are to be included
in a media domain mix of the incomming RTP media streams. Then
create a single outgoing stream from this mix.Audio mixing is straight forward and commonly possible to do for
a number of participants. Lets assume that you want to mix N number
of streams from different participants. Then the mixer need to
perform N decodings. Then it needs to produce N or N+1 mixes, the
reasons that different mixes are needed are so that each
contributing source get a mix which don't contain themselves, 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 curently not included in the mix, thus N+1 mixes. These
audio streams are then encoded again, RTP packetized and sent
out.Video can't really be "mixed" and produce something particular
useful for the users, however creating an composition out of the
contributed video streams can be done. In fact it can be done in a
number of ways, tiling the different streams creating a chessboard,
selecting someone as more important and showing them large and a
number of other sources as smaller is another. Also here one
commonly need to produce a number of different compositions so that
the contributing part doesn't need to see themselves. Then the mixer
re-encodes the created video stream, RTP packetize it and send it
outThe problem with media mixing is that it both consume large
amount of media processing and encoding resources. The second is the
quality degradation created by decoding and re-encoding the RTP
media stream. Its advantage is that it is quite simplistic for the
clients to handle as they don't need to handle local mixing and
composition.From an RTP perspective media mixing can be very straight forward
as can be seen in . The mixer
present one SSRC towards the peer client, e.g. MA1 to Peer A, which
is the media mix of the other particpants. As each peer receives a
different version produced by the mixer there are no actual relation
between the different RTP sessions in the actual media or the
transport level information. There is however one connection between
RTP1-RTP3 in this figure. It has to do with the SSRC space and the
identity information. When A receives the MA1 stream which is a
combination of BA1 and CA1 streams in the other PeerConnections RTP
could enable the mixer to include CSRC information in the MA1 stream
to identify the contributing source BA1 and CA1.The CSRC has in its turn utility in RTP extensions, like the in
discussed Mixer to Client audio levels RTP header
extension. If the SSRC from one PeerConnection are used as
CSRC in another PeerConnection then RTP1, RTP2 and RTP3 becomes one
joint session as they have a common SSRC space. At this stage one
also need to consider which RTCP information one need to expose in
the different legs. For the above situation commonly nothing more
than the Source Description (SDES) information and RTCP BYE for CSRC
need to be exposed. 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.There exist an possible implementation choice to have the RTP
sessions being separated between the different legs in the
multi-party communication session and only generate RTP media
streams in each without carrying on RTP/RTCP level any identity
information about the contributing sources. This removes both the
functionaltiy that CSRC can provide and the possibility to use any
extensions that build on CSRC and the loop detection. It may
appear a simplification if SSRC collision would occur between two
different end-points as they can be avoide to be resolved and
instead remapped between the independent sessions if at all
exposed. However, SSRC/CSRC remapping requiresthat SSRC/CSRC are
never exposed to the WebRTC javascript client to use as reference.
This as they only have local importance if they are used on a
multi-party session scope the result would be missreferencing.
Also SSRC collision handling will still be needed as it may occur
between the mixer and the end-point.Session termination may appear to resolve some issues, it
however creates other issues that needs resolving, like loop
detection, identification of contributing sources and the need to
handle mapped identities and ensure that the right one is used
towards the right identities and never used directly between
multiple end-points.An RTP Mixer based on media switching avoids the media decoding
and encoding cycle in the mixer, but not the decryption and
re-encryption cycle as one rewrites RTP headers. This both reduces
the amount of computational resources needed in the mixer and
increases the media quality per transmitted bit. This is achieve by
letting the mixer have a number of SSRCs that represents conceptual
or functional streams the mixer produces. These streams are created
by selecting media from one of the by the mixer received RTP media
streams and forward the media using the mixers own SSRCs. The mixer
can then switch between available sources if that is required by the
concept for the source, like currently active speaker.To achieve a coherent RTP media stream from the mixer's SSRC the
mixer is forced to rewrite the incoming RTP packet's header. First
the SSRC field must be set to the value of the Mixer's SSRC.
Secondly, the sequence number must be the next in the sequence of
outgoing packets it sent. Thirdly the RTP timestamp value needs to
be adjusted using an offset that changes each time one switch media
source. Finally depending on the negotiation the RTP payload type
value representing this particular RTP payload configuration may
have to be changed if the different PeerConnections have not arrived
on the same numbering for a given configuration. This also requires
that the different end-points do support a common set of codecs,
otherwise media transcoding for codec compatibility is still
required.Lets consider the operation of media switching mixer that
supports a video conference with six participants (A-F) where the
two latest speakers in the conference are shown to each
participants. Thus the mixer has two SSRCs sending video to each
peer.The Media Switching RTP mixer can similar to the Media Mixing one
reduce the bit-rate needed towards the different peers by selecting
and switching in a sub-set of RTP media streams out of the ones it
receives from the conference participations.To ensure that a media receiver can correctly decode the RTP
media stream after a switch, it becomes necessary to ensure for
state saving codecs that they start from default state at the point
of switching. Thus one common tool for video is to request that the
encoding creates an intra picture, something that isn't dependent on
earlier state. This can be done using Full Intra Request RTCP codec
control message as discussed in .Also in this type of mixer one could consider to terminate the
RTP sessions fully between the different PeerConnection. The same
arguments and conisderations as discussed in applies here.Another method for handling media in the RTP mixer is to project
all potential sources (SSRCs) into a per end-point independent RTP
session. The mixer can then select which of the potential sources
that are currently actively transmitting media, despite that the
mixer in another RTP session recieves media from that end-point.
This is similar to the media switching Mixer but have some important
differences in RTP details.So in this six participant conference depicted above in one can see that end-point A will
in this case be aware of 5 incoming SSRCs, BV1-FV1. If this mixer
intend to have the same behavior as in where the mixer provides the
end-points with the two latest speaking end-points, then only two
out of these five SSRCs will concurrently transmitt media to A. As
the mixer selects which source in the different RTP sessions that
transmit media to the end-points each RTP media stream will require
some rewriting when being projected from one session into another.
The main thing is that the sequence number will need to be
consequitvely incremented based on the packet actually being
transmitted in each RTP session. Thus the RTP sequence number offset
will change each time a source is turned on in RTP session.As the RTP sessions are independent the SSRC numbers used can be
handled indepdentently also thus working around any SSRC collisions
by having remapping tables between the RTP sessions. However the
related WebRTC MediaStream signalling must be correspondlingly
changed to ensure consistent WebRTC MediaStream to SSRC mappings
between the different PeerConnections and the same comment that
higher functions must not use SSRC as references to RTP media
streams applies also here.The mixer will also be responsible to act on any RTCP codec
control requests comming from an end-point and decide if it can act
on it locally or needs to translate the request into the RTP session
that contains the media source. Both end-points and the mixer will
need to implement conference related codec control functionalities
to provide a good experience. Full Intra Request to request from the
media source to provide switching points between the sources,
Temporary Maximum Media Bit-rate Request (TMMBR) to enable the mixer
to aggregate congestion control response towards the media source
and have it adjust its bit-rate in case the limitation is not in the
source to mixer link.This version of the mixer also puts different requirements on the
end-point 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 end-point either needs to handle having thus
many allocated decoder instances or have efficient switching of
decoder contexts in a more limited set of actual decoder instances
to cope with the switches. The WebRTC application also gets more
responsibility to update how the media provides is to be presented
to the user.There is also a variety of translators. The core commonality is
that they do not need to make themselves visible in the RTP level by
having an SSRC themselves. Instead they sit between one or more
end-point and perform translation at some level. It can be media
transcoding, protocol translation or covering missing functionality
for a legacy end-point or simply relay packets between transport
domains or to realize multi-party. We will go in details below.A transcoder operates on media level and really used for two
purposes, the first is to allow two end-points that doesn't have a
common set of media codecs to communicate by translating from one
codec to another. The second is to change the bit-rate to a lower
one. For WebRTC end-points communicating with each other only the
first one should at all be relevant. In certain legacy deployment
media transcoder will be necessary to ensure both codecs and
bit-rate falls within the envelope the legacy end-point
supports.As transcoding requires access to the media the transcoder must
within the security context and access any media encryption and
integrity keys. On the RTP plane a media transcoder will in practice
fork the RTP session into two different domains that are highly
decoupled when it comes to media parameters and reporting, but not
identities. To maintain signalling bindings to SSRCs a transcoder is
likely needing to use the SSRC of one end-point to represent the
transcoded RTP media stream to the other end-point(s). The
congestion control loop can be terminated in the transcoder as the
media bit-rate being sent by the transcoder can be adjusted
independently of the incoming bit-rate. However, for optimizing
performance and resource consumption the translator needs to
consider what signals or bit-rate reductions it should send towards
the source end-point. For example receving a 2.5 mbps video stream
and then send out a 250 kbps video stream after transcoding is a
vaste of resources. In most cases a 500 kbps video stream from the
source in the right resolution is likely to provide equal quality
after transcoding as the 2.5 mbps source stream. At the same time
increasing media bit-rate futher than what is needed to represent
the incoming quality accurate is also wasted resources. exposes some important
details. First of all you can see the SSRC identifiers used by the
translator are the corresponding end-points. Secondly, there is a
relation between the RTP sessions in the two different
PeerConnections that are represtented by having both parts be
identified by the same level and they need to share certain
contexts. Also certain type of RTCP messages will need to be bridged
between the two parts. Certain RTCP feedback messages are likely
needed to be soruced by the translator in response to actions by the
translator and its media encoder.Gateways are used when some protocol feature that is required is
not supported by an end-point wants to participate in session. This
RTP translator in takes on the role
of ensuring that from the perspective of participant A, participant
B appears as a fully compliant WebRTC end-point (that is, it is the
combination of the Translator and participant B that looks like a
WebRTC end point).For WebRTC there are a number of requirements that could force
the need for a gateway if a WebRTC end-point is to communicate with
a legacy end-point, such as support of ICE and DTLS-SRTP for
keymanagement. On RTP level the main functions that may be missing
in a legacy implementation that otherswise support RTP are RTCP in
general, SRTP implementation, congestion control and feedback
messages required to make it work.The legacy gateway may be implemented in several ways and what it
need to change is higly dependent on what functions it need to proxy
for the legacy end-point. One possibility is depicted in where the RTP media streams are
compatible and forward without changes. However, their RTP header
values are captured to enable the RTCP translator to create RTCP
reception information related to the leg between the end-point and
the translator. This can then be combined with the more basic RTCP
reports that the legacy endpoint (B) provides to give compatible and
expected RTCP reporting to A. Thus enabling at least full congestion
control on the path between A and the translator. If B has limited
possibilities for congestion response for the media then the
translator may need the capabilities to perform media transcoding to
address cases where it otherwise would need to terminate media
transmission.As the translator are generating RTP/RTCP traffic on behalf of B
to A it will need to be able to correctly protect these packets that
it translates or generates. Thus security context information are
required in this type of translator if it operates on the RTP/RTCP
packet content or media. In fact one of the more likley scenario is
that the translator (gateway) will need to have two different
security contexts one towards A and one towards B and for each
RTP/RTCP packet do a authenticity verification, decryption followed
by a encryption and integirty protection operation to resolve
missmatch in security systems.There exist a class of translators that operates on transport
level below RTP and thus do not effect RTP/RTCP packets directly.
They come in two distinct flavors, the one used to bridge between
two different transport or address domains to more function as a
gateway and the second one which is to to provide a group
communication feature as depicted below in .The first kind is straight forward and is likely to exist in
WebRTC context when an legacy end-point is compatible with the
exception for ICE, and thus needs a gateway that terminates the ICE
and then forwards all the RTP/RTCP traffic and keymanagment to the
end-point only rewriting the IP/UDP to forward the packet to the
legacy node.The second type is useful if one wants a less complex central
node or a central node that is outside of the security context and
thus do not have access to the media. This relay takes on the role
of forwarding the media (RTP and RTCP) packets to the other
end-points but doesn't perform any RTP or media processing. Such a
device simply forwards the media from each sender to all of the
other particpants, and is sometimes called a transport-layer
translator. In , participant A will only
need to send a media once to the relay, which will redistribute it
by sending a copy of the stream to participants B, C, and D.
Participant A will still receive three RTP streams with the media
from B, C and D if they transmit simultaneously. This is from an RTP
perspective resulting in an RTP session that behaves equivalent to
one transporter over an IP Any Source Multicast (ASM).This results in one common RTP session between all participants
despite that there will be independent PeerConnections created to
the translator as depicted below .As the Relay RTP and RTCP packets between the UDP flows as
indicated by the arrows for the media flow a given WebRTC end-point,
like A will see the remote sources BV1 and CV1. There will be also
two different network paths between A, and B or C. This results in
that the client A must be capable of handlilng that when determining
congestion state that there might exist multiple destinations on the
far side of a PeerConnection and that these paths shall be treated
differently. It also results in a requirement to combine the
different congestion states into a decision to transmit a particular
RTP media stream suitable to all participants.It is also important to note that the relay can not perform
selective relaying of some sources and not others. The reason is
that the RTCP reporting in that case becomes incosistent and without
explicit information about it being blocked must be interpret as
severe congestion.In this usage it is also necessary that the session management
has configured a common set of RTP configuration including RTP
payload formats as when A sends a packet with pt=97 it will arrive
at both B and C carrying pt=97 and having the same packetization and
encoding, no entity will have manipulated the packet.When it comes to security there exist some additional
requirements to ensure that the property that the relay can't read
the media traffic is enforced. First of all the key to be used must
be agreed such so that the relay doesn't get it, e.g. no DTLS-SRTP
handshake with the relay, instead some other method must be used.
Secondly, the keying structure must be capable of handling multiple
end-points in the same RTP session.The second problem can basically be solved in two ways. Either a
common master key from which all derive their per source key for
SRTP. The second alternative which might be more practical is that
each end-point has its own key used to protects all RTP/RTCP packets
it sends. Each participants key are then distributed to the other
participants. This second method could be implemented using
DTLS-SRTP to a special key server and then use Encrypted Key Transport to
distribute the actual used key to the other participants in the RTP
session . The first one could be
achieved using MIKEY messages in SDP.The relay can still verify that a given SSRC isn't used or
spoofed by another participant within the multi-party session by
binding SSRCs on their first usage to a given source address and
port pair. Packets carrying that source SSRC from other addresses
can be suppressed to prevent spoofing. This is possible as long as
SRTP is used which leaves the SSRC of the packet originator in RTP
and RTCP packets in the clear. If such packet level method for
enforcing source authentication within the group, then there exist
cryptographic methods such as TESLA
that could be used for true source authentication.An WebRTC end-point (B in ) will
receive a WebRTC MediaStream (set of SSRCs) over a PeerConnection
(from A). For the moment is not decided if the end-point is allowed or
not to in its turn send that WebRTC MediaStream over another
PeerConnection to C. This section discusses the RTP and end-point
implications of allowing such functionality, which on the API level is
extremely simplistic to perform.There exist two main approaches to how B forwards the media from A
to C. The first one is to simply relay the RTP media stream. The
second one is for B to act as a transcoder. Lets consider both
approaches.A relay approache will result in that the WebRTC end-points will
have to have the same capabilities as being discussed in Relay. Thus A will see an RTP session that
is extended beyond the PeerConnection and see two different receiving
end-points with different path characteristics (B and C). Thus A's
congestion control needs to be capable of handling this. The security
solution can either support mechanism that allows A to inform C about
the key A is using despite B and C having agreed on another set of
keys. Alternatively B will decrypt and then re-encrypt using a new
key. The relay based approach has the advantage that B does not need
to transcode the media thus both maintaining the quality of the
encoding and reducing B's complexity requirements. If the right
security solutions are supported then also C will be able to verify
the authenticity of the media comming from A. As downside A are forced
to take both B and C into consideration when delivering content.The media transcoder approach is similar to having B act as Mixer
terminating the RTP session combined with the transcoder as discussed
in . A will only see B as receiver of
its media. B will responsible to produce a RTP media stream suitable
for the B to C PeerConnection. This may require media transcoding for
congestion control purpose to produce a suitable bit-rate. Thus
loosing media quality in the transcoding and forcing B to spend the
resource on the transcoding. The media transcoding does result in a
separation of the two different legs removing almost all dependencies.
B could choice to implement logic to optimize its media transcoding
operation, by for example requesting media properties that are
suitable for C also, thus trying to avoid it having to transcode the
content and only forward the media payloads between the two sides. For
that optimization to be practical WebRTC end-points must support
sufficiently good tools for codec control.This section discusses simulcast in the meaning of providing a
node, for example a stream switching Mixer, with multiple different
encoded version of the same media source. In the WebRTC context that
appears to be most easily accomplished by establishing mutliple
PeerConnection all being feed the same set of WebRTC MediaStreams.
Each PeerConnection is then configured to deliver a particular media
quality and thus media bit-rate. This will work well as long as the
end-point implements media encoding according to . Then each PeerConnection will receive an
independently encoded version and the codec parameters can be agreed
specifically in the context of this PeerConnection.For simulcast to work one needs to prevent that the end-point
deliver content encoded as depicted in . If a single encoder instance is feed
to multiple PeerConnections the intention of performing simulcast will
fail.Thus it should be considered to explicitly signal which of the two
implementation strategies that are desired and which will be done. At
least making the application and possible the central node interested
in receiving simulcast of an end-points RTP media streams to be aware
if it will function or not.