idnits 2.17.1 draft-ietf-rtcweb-rtp-usage-09.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (September 05, 2013) is 3885 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Unused Reference: 'I-D.ietf-avt-srtp-ekt' is defined on line 1760, but no explicit reference was found in the text == Outdated reference: A later version (-13) exists of draft-ietf-avtcore-multi-media-rtp-session-03 == Outdated reference: A later version (-18) exists of draft-ietf-avtcore-rtp-circuit-breakers-03 == Outdated reference: A later version (-12) exists of draft-ietf-avtcore-rtp-multi-stream-optimisation-00 == Outdated reference: A later version (-11) exists of draft-ietf-avtcore-rtp-multi-stream-01 == Outdated reference: A later version (-11) exists of draft-ietf-avtext-multiple-clock-rates-09 == Outdated reference: A later version (-54) exists of draft-ietf-mmusic-sdp-bundle-negotiation-04 == Outdated reference: A later version (-20) exists of draft-ietf-rtcweb-security-arch-07 == Outdated reference: A later version (-12) exists of draft-ietf-rtcweb-security-05 ** Obsolete normative reference: RFC 5285 (Obsoleted by RFC 8285) == Outdated reference: A later version (-10) exists of draft-ietf-avtcore-rtp-topologies-update-00 == Outdated reference: A later version (-19) exists of draft-ietf-rtcweb-overview-08 == Outdated reference: A later version (-16) exists of draft-ietf-rtcweb-use-cases-and-requirements-11 == Outdated reference: A later version (-07) exists of draft-westerlund-avtcore-transport-multiplexing-06 Summary: 1 error (**), 0 flaws (~~), 14 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 RTCWEB Working Group C. S. Perkins 3 Internet-Draft University of Glasgow 4 Intended status: Standards Track M. Westerlund 5 Expires: March 09, 2014 Ericsson 6 J. Ott 7 Aalto University 8 September 05, 2013 10 Web Real-Time Communication (WebRTC): Media Transport and Use of RTP 11 draft-ietf-rtcweb-rtp-usage-09 13 Abstract 15 The Web Real-Time Communication (WebRTC) framework provides support 16 for direct interactive rich communication using audio, video, text, 17 collaboration, games, etc. between two peers' web-browsers. This 18 memo describes the media transport aspects of the WebRTC framework. 19 It specifies how the Real-time Transport Protocol (RTP) is used in 20 the WebRTC context, and gives requirements for which RTP features, 21 profiles, and extensions need to be supported. 23 Status of This Memo 25 This Internet-Draft is submitted in full conformance with the 26 provisions of BCP 78 and BCP 79. 28 Internet-Drafts are working documents of the Internet Engineering 29 Task Force (IETF). Note that other groups may also distribute 30 working documents as Internet-Drafts. The list of current Internet- 31 Drafts is at http://datatracker.ietf.org/drafts/current/. 33 Internet-Drafts are draft documents valid for a maximum of six months 34 and may be updated, replaced, or obsoleted by other documents at any 35 time. It is inappropriate to use Internet-Drafts as reference 36 material or to cite them other than as "work in progress." 38 This Internet-Draft will expire on March 09, 2014. 40 Copyright Notice 42 Copyright (c) 2013 IETF Trust and the persons identified as the 43 document authors. All rights reserved. 45 This document is subject to BCP 78 and the IETF Trust's Legal 46 Provisions Relating to IETF Documents 47 (http://trustee.ietf.org/license-info) in effect on the date of 48 publication of this document. Please review these documents 49 carefully, as they describe your rights and restrictions with respect 50 to this document. Code Components extracted from this document must 51 include Simplified BSD License text as described in Section 4.e of 52 the Trust Legal Provisions and are provided without warranty as 53 described in the Simplified BSD License. 55 Table of Contents 57 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 58 2. Rationale . . . . . . . . . . . . . . . . . . . . . . . . . . 4 59 3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 60 4. WebRTC Use of RTP: Core Protocols . . . . . . . . . . . . . . 5 61 4.1. RTP and RTCP . . . . . . . . . . . . . . . . . . . . . . 5 62 4.2. Choice of the RTP Profile . . . . . . . . . . . . . . . . 6 63 4.3. Choice of RTP Payload Formats . . . . . . . . . . . . . . 7 64 4.4. Use of RTP Sessions . . . . . . . . . . . . . . . . . . . 8 65 4.5. RTP and RTCP Multiplexing . . . . . . . . . . . . . . . . 9 66 4.6. Reduced Size RTCP . . . . . . . . . . . . . . . . . . . . 10 67 4.7. Symmetric RTP/RTCP . . . . . . . . . . . . . . . . . . . 10 68 4.8. Choice of RTP Synchronisation Source (SSRC) . . . . . . . 10 69 4.9. Generation of the RTCP Canonical Name (CNAME) . . . . . . 11 70 5. WebRTC Use of RTP: Extensions . . . . . . . . . . . . . . . . 11 71 5.1. Conferencing Extensions . . . . . . . . . . . . . . . . . 12 72 5.1.1. Full Intra Request (FIR) . . . . . . . . . . . . . . 13 73 5.1.2. Picture Loss Indication (PLI) . . . . . . . . . . . . 13 74 5.1.3. Slice Loss Indication (SLI) . . . . . . . . . . . . . 13 75 5.1.4. Reference Picture Selection Indication (RPSI) . . . . 13 76 5.1.5. Temporal-Spatial Trade-off Request (TSTR) . . . . . . 14 77 5.1.6. Temporary Maximum Media Stream Bit Rate Request 78 (TMMBR) . . . . . . . . . . . . . . . . . . . . . . . 14 79 5.2. Header Extensions . . . . . . . . . . . . . . . . . . . . 14 80 5.2.1. Rapid Synchronisation . . . . . . . . . . . . . . . . 14 81 5.2.2. Client-to-Mixer Audio Level . . . . . . . . . . . . . 15 82 5.2.3. Mixer-to-Client Audio Level . . . . . . . . . . . . . 15 83 5.2.4. Associating RTP Media Streams and Signalling Contexts 15 84 6. WebRTC Use of RTP: Improving Transport Robustness . . . . . . 15 85 6.1. Negative Acknowledgements and RTP Retransmission . . . . 16 86 6.2. Forward Error Correction (FEC) . . . . . . . . . . . . . 17 87 7. WebRTC Use of RTP: Rate Control and Media Adaptation . . . . 17 88 7.1. Boundary Conditions and Circuit Breakers . . . . . . . . 18 89 7.2. RTCP Limitations for Congestion Control . . . . . . . . . 19 90 7.3. Congestion Control Interoperability and Legacy Systems . 19 91 8. WebRTC Use of RTP: Performance Monitoring . . . . . . . . . . 20 92 9. WebRTC Use of RTP: Future Extensions . . . . . . . . . . . . 21 93 10. Signalling Considerations . . . . . . . . . . . . . . . . . . 21 94 11. WebRTC API Considerations . . . . . . . . . . . . . . . . . . 23 95 12. RTP Implementation Considerations . . . . . . . . . . . . . . 23 96 12.1. Configuration and Use of RTP Sessions . . . . . . . . . 24 97 12.1.1. Use of Multiple Media Flows Within an RTP Session . 24 98 12.1.2. Use of Multiple RTP Sessions . . . . . . . . . . . . 25 99 12.1.3. Differentiated Treatment of Flows . . . . . . . . . 30 100 12.2. Source, Flow, and Participant Identification . . . . . . 31 101 12.2.1. Media Streams . . . . . . . . . . . . . . . . . . . 31 102 12.2.2. Media Streams: SSRC Collision Detection . . . . . . 32 103 12.2.3. Media Synchronisation Context . . . . . . . . . . . 33 104 12.2.4. Correlation of Media Streams . . . . . . . . . . . . 34 105 13. Security Considerations . . . . . . . . . . . . . . . . . . . 34 106 14. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 34 107 15. Open Issues . . . . . . . . . . . . . . . . . . . . . . . . . 35 108 16. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 35 109 17. References . . . . . . . . . . . . . . . . . . . . . . . . . 35 110 17.1. Normative References . . . . . . . . . . . . . . . . . . 35 111 17.2. Informative References . . . . . . . . . . . . . . . . . 38 112 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 40 114 1. Introduction 116 The Real-time Transport Protocol (RTP) [RFC3550] provides a framework 117 for delivery of audio and video teleconferencing data and other real- 118 time media applications. Previous work has defined the RTP protocol, 119 along with numerous profiles, payload formats, and other extensions. 120 When combined with appropriate signalling, these form the basis for 121 many teleconferencing systems. 123 The Web Real-Time communication (WebRTC) framework provides the 124 protocol building blocks to support direct, interactive, real-time 125 communication using audio, video, collaboration, games, etc., between 126 two peers' web-browsers. This memo describes how the RTP framework 127 is to be used in the WebRTC context. It proposes a baseline set of 128 RTP features that are to be implemented by all WebRTC-aware end- 129 points, along with suggested extensions for enhanced functionality. 131 This memo specifies a protocol intended for use within the WebRTC 132 framework, but is not restricted to that context. An overview of the 133 WebRTC framework is given in [I-D.ietf-rtcweb-overview]. 135 The structure of this memo is as follows. Section 2 outlines our 136 rationale in preparing this memo and choosing these RTP features. 137 Section 3 defines terminology. Requirements for core RTP protocols 138 are described in Section 4 and suggested RTP extensions are described 139 in Section 5. Section 6 outlines mechanisms that can increase 140 robustness to network problems, while Section 7 describes congestion 141 control and rate adaptation mechanisms. The discussion of mandated 142 RTP mechanisms concludes in Section 8 with a review of performance 143 monitoring and network management tools that can be used in the 144 WebRTC context. Section 9 gives some guidelines for future 145 incorporation of other RTP and RTP Control Protocol (RTCP) extensions 146 into this framework. Section 10 describes requirements placed on the 147 signalling channel. Section 11 discusses the relationship between 148 features of the RTP framework and the WebRTC application programming 149 interface (API), and Section 12 discusses RTP implementation 150 considerations. This memo concludes with an appendix discussing 151 several different RTP Topologies, and how they affect the RTP 152 session(s) and various implementation details of possible realization 153 of central nodes. 155 2. Rationale 157 The RTP framework comprises the RTP data transfer protocol, the RTP 158 control protocol, and numerous RTP payload formats, profiles, and 159 extensions. This range of add-ons has allowed RTP to meet various 160 needs that were not envisaged by the original protocol designers, and 161 to support many new media encodings, but raises the question of what 162 extensions are to be supported by new implementations. The 163 development of the WebRTC framework provides an opportunity for us to 164 review the available RTP features and extensions, and to define a 165 common baseline feature set for all WebRTC implementations of RTP. 166 This builds on the past 20 years development of RTP to mandate the 167 use of extensions that have shown widespread utility, while still 168 remaining compatible with the wide installed base of RTP 169 implementations where possible. 171 Other RTP and RTCP extensions not discussed in this document can be 172 implemented by WebRTC end-points if they are beneficial for new use 173 cases. However, they are not necessary to address the WebRTC use 174 cases and requirements identified to date 175 [I-D.ietf-rtcweb-use-cases-and-requirements]. 177 While the baseline set of RTP features and extensions defined in this 178 memo is targeted at the requirements of the WebRTC framework, it is 179 expected to be broadly useful for other conferencing-related uses of 180 RTP. In particular, it is likely that this set of RTP features and 181 extensions will be appropriate for other desktop or mobile video 182 conferencing systems, or for room-based high-quality telepresence 183 applications. 185 3. Terminology 187 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 188 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 189 document are to be interpreted as described in [RFC2119]. The RFC 190 2119 interpretation of these key words applies only when written in 191 ALL CAPS. Lower- or mixed-case uses of these key words are not to be 192 interpreted as carrying special significance in this memo. 194 We define the following terms: 196 RTP Media Stream: A sequence of RTP packets, and associated RTCP 197 packets, using a single synchronisation source (SSRC) that 198 together carries part or all of the content of a specific Media 199 Type from a specific sender source within a given RTP session. 201 RTP Session: As defined by [RFC3550], the endpoints belonging to the 202 same RTP Session are those that share a single SSRC space. That 203 is, those endpoints can see an SSRC identifier transmitted by any 204 one of the other endpoints. An endpoint can see an SSRC either 205 directly in RTP and RTCP packets, or as a contributing source 206 (CSRC) in RTP packets from a mixer. The RTP Session scope is 207 hence decided by the endpoints' network interconnection topology, 208 in combination with RTP and RTCP forwarding strategies deployed by 209 endpoints and any interconnecting middle nodes. 211 WebRTC MediaStream: The MediaStream concept defined by the W3C in 212 the API. 214 Other terms are used according to their definitions from the RTP 215 Specification [RFC3550]. 217 4. WebRTC Use of RTP: Core Protocols 219 The following sections describe the core features of RTP and RTCP 220 that need to be implemented, along with the mandated RTP profiles and 221 payload formats. Also described are the core extensions providing 222 essential features that all WebRTC implementations need to implement 223 to function effectively on today's networks. 225 4.1. RTP and RTCP 227 The Real-time Transport Protocol (RTP) [RFC3550] is REQUIRED to be 228 implemented as the media transport protocol for WebRTC. RTP itself 229 comprises two parts: the RTP data transfer protocol, and the RTP 230 control protocol (RTCP). RTCP is a fundamental and integral part of 231 RTP, and MUST be implemented in all WebRTC applications. 233 The following RTP and RTCP features are sometimes omitted in limited 234 functionality implementations of RTP, but are REQUIRED in all WebRTC 235 implementations: 237 o Support for use of multiple simultaneous SSRC values in a single 238 RTP session, including support for RTP end-points that send many 239 SSRC values simultaneously, following [RFC3550] and 240 [I-D.ietf-avtcore-rtp-multi-stream]. Support for the RTCP 241 optimisations for multi-SSRC sessions defined in 242 [I-D.ietf-avtcore-rtp-multi-stream-optimisation] is RECOMMENDED. 244 * (tbd: do endpoints need to signal the maximum number of SSRCs 245 that they support (e.g., draft-westerlund-mmusic-max-ssrc-01) 246 and/or some constraint on the maximum number of simultaneous 247 streams of various kinds that can be decoded?) 249 o Random choice of SSRC on joining a session; collision detection 250 and resolution for SSRC values (see also Section 4.8). 252 o Support for reception of RTP data packets containing CSRC lists, 253 as generated by RTP mixers, and RTCP packets relating to CSRCs. 255 o Support for sending correct synchronization information in the 256 RTCP Sender Reports, to allow a receiver to implement lip-sync, 257 with RECOMMENDED support for the rapid RTP synchronisation 258 extensions (see Section 5.2.1). 260 o Support for sending and receiving RTCP SR, RR, SDES, and BYE 261 packet types, with OPTIONAL support for other RTCP packet types; 262 implementations MUST ignore unknown RTCP packet types. 264 o Support for multiple end-points in a single RTP session, and for 265 scaling the RTCP transmission interval according to the number of 266 participants in the session; support for randomised RTCP 267 transmission intervals to avoid synchronisation of RTCP reports; 268 support for RTCP timer reconsideration. 270 o Support for configuring the RTCP bandwidth as a fraction of the 271 media bandwidth, and for configuring the fraction of the RTCP 272 bandwidth allocated to senders, e.g., using the SDP "b=" line. 274 It is known that a significant number of legacy RTP implementations, 275 especially those targeted at VoIP-only systems, do not support all of 276 the above features, and in some cases do not support RTCP at all. 277 Implementers are advised to consider the requirements for graceful 278 degradation when interoperating with legacy implementations. 280 Other implementation considerations are discussed in Section 12. 282 4.2. Choice of the RTP Profile 283 The complete specification of RTP for a particular application domain 284 requires the choice of an RTP Profile. For WebRTC use, the Extended 285 Secure RTP Profile for RTCP-Based Feedback (RTP/SAVPF) [RFC5124], as 286 extended by [RFC7007], MUST be implemented. This builds on the basic 287 RTP/AVP profile [RFC3551], the RTP profile for RTCP-based feedback 288 (RTP/AVPF) [RFC4585], and the secure RTP profile (RTP/SAVP) 289 [RFC3711]. 291 The RTCP-based feedback extensions [RFC4585] are needed for the 292 improved RTCP timer model, that allows more flexible transmission of 293 RTCP packets in response to events, rather than strictly according to 294 bandwidth. This is vital for being able to report congestion events. 295 These extensions also save RTCP bandwidth, and will commonly only use 296 the full RTCP bandwidth allocation if there are many events that 297 require feedback. They are also needed to make use of the RTP 298 conferencing extensions discussed in Section 5.1. 300 Note: The enhanced RTCP timer model defined in the RTP/AVPF 301 profile is backwards compatible with legacy systems that implement 302 only the base RTP/AVP profile, given some constraints on parameter 303 configuration such as the RTCP bandwidth value and "trr-int" (the 304 most important factor for interworking with RTP/AVP end-points via 305 a gateway is to set the trr-int parameter to a value representing 306 4 seconds). 308 The secure RTP profile [RFC3711] is needed to provide media 309 encryption, integrity protection, replay protection and a limited 310 form of source authentication. WebRTC implementations MUST NOT send 311 packets using the basic RTP/AVP profile or the RTP/AVPF profile; they 312 MUST employ the full RTP/SAVPF profile to protect all RTP and RTCP 313 packets that are generated. The default and mandatory to implement 314 transforms listed in Section 5 of [RFC3711] SHALL apply. 316 Implementations MUST support DTLS-SRTP [RFC5764] for key-management. 317 Other key management schemes MAY be supported. 319 4.3. Choice of RTP Payload Formats 321 The set of mandatory to implement codecs and RTP payload formats for 322 WebRTC is not specified in this memo. Implementations can support 323 any codec for which an RTP payload format and associated signalling 324 is defined. Implementation cannot assume that the other participants 325 in an RTP session understand any RTP payload format, no matter how 326 common; the mapping between RTP payload type numbers and specific 327 configurations of particular RTP payload formats MUST be agreed 328 before those payload types/formats can be used. In an SDP context, 329 this can be done using the "a=rtpmap:" and "a=fmtp:" attributes 330 associated with an "m=" line. 332 Endpoints can signal support for multiple RTP payload formats, or 333 multiple configurations of a single RTP payload format, as long as 334 each unique RTP payload format configuration uses a different RTP 335 payload type number. As outlined in Section 4.8, the RTP payload 336 type number is sometimes used to associate an RTP media stream with a 337 signalling context. This association is possible provided unique RTP 338 payload type numbers are used in each context. For example, an RTP 339 media stream can be associated with an SDP "m=" line by comparing the 340 RTP payload type numbers used by the media stream with payload types 341 signalled in the "a=rtpmap:" lines in the media sections of the SDP. 342 If RTP media streams are being associated with signalling contexts 343 based on the RTP payload type, then the assignment of RTP payload 344 type numbers MUST be unique across signalling contexts; if the same 345 RTP payload format configuration is used in multiple contexts, then a 346 different RTP payload type number has to be assigned in each context 347 to ensure uniqueness. If the RTP payload type number is not being 348 used to associated RTP media streams with a signalling context, then 349 the same RTP payload type number can be used to indicate the exact 350 same RTP payload format configuration in multiple contexts. 352 An endpoint that has signalled support for multiple RTP payload 353 formats SHOULD accept data in any of those payload formats at any 354 time, unless it has previously signalled limitations on its decoding 355 capability. This requirement is constrained if several types of 356 media (e.g., audio and video) are sent in the same RTP session. In 357 such a case, a source (SSRC) is restricted to switching only between 358 the RTP payload formats signalled for the type of media that is being 359 sent by that source; see Section 4.4. To support rapid rate 360 adaptation by changing codec, RTP does not require advance signalling 361 for changes between RTP payload formats that were signalled during 362 session set-up. 364 An RTP sender that changes between two RTP payload types that use 365 different RTP clock rates MUST follow the recommendations in 366 Section 4.1 of [I-D.ietf-avtext-multiple-clock-rates]. RTP receivers 367 MUST follow the recommendations in Section 4.3 of 368 [I-D.ietf-avtext-multiple-clock-rates], in order to support sources 369 that switch between clock rates in an RTP session (these 370 recommendations for receivers are backwards compatible with the case 371 where senders use only a single clock rate). 373 4.4. Use of RTP Sessions 375 An association amongst a set of participants communicating using RTP 376 is known as an RTP session. A participant can be involved in several 377 RTP sessions at the same time. In a multimedia session, each type of 378 media has typically been carried in a separate RTP session (e.g., 379 using one RTP session for the audio, and a separate RTP session using 380 different transport addresses for the video). WebRTC implementations 381 of RTP are REQUIRED to implement support for multimedia sessions in 382 this way, separating each session using different transport-layer 383 addresses (e.g., different UDP ports) for compatibility with legacy 384 systems. 386 In modern day networks, however, with the widespread use of network 387 address/port translators (NAT/NAPT) and firewalls, it is desirable to 388 reduce the number of transport-layer flows used by RTP applications. 389 This can be done by sending all the RTP media streams in a single RTP 390 session, which will comprise a single transport-layer flow (this will 391 prevent the use of some quality-of-service mechanisms, as discussed 392 in Section 12.1.3). Implementations are REQUIRED to support 393 transport of all RTP media streams, independent of media type, in a 394 single RTP session according to 395 [I-D.ietf-avtcore-multi-media-rtp-session]. If multiple types of 396 media are to be used in a single RTP session, all participants in 397 that session MUST agree to this usage. In an SDP context, 398 [I-D.ietf-mmusic-sdp-bundle-negotiation] can be used to signal this. 400 It is also possible to use a shim-based approach to run multiple RTP 401 sessions on a single transport-layer flow. This gives advantages in 402 some gateway scenarios, and makes it easy to distinguish groups of 403 RTP media streams that might need distinct processing. One way of 404 doing this is described in 405 [I-D.westerlund-avtcore-transport-multiplexing]. At the time of this 406 writing, there is no consensus to use a shim-based approach in WebRTC 407 implementations. 409 Further discussion about when different RTP session structures and 410 multiplexing methods are suitable can be found in 411 [I-D.westerlund-avtcore-multiplex-architecture]. 413 4.5. RTP and RTCP Multiplexing 415 Historically, RTP and RTCP have been run on separate transport layer 416 addresses (e.g., two UDP ports for each RTP session, one port for RTP 417 and one port for RTCP). With the increased use of Network Address/ 418 Port Translation (NAPT) this has become problematic, since 419 maintaining multiple NAT bindings can be costly. It also complicates 420 firewall administration, since multiple ports need to be opened to 421 allow RTP traffic. To reduce these costs and session set-up times, 422 support for multiplexing RTP data packets and RTCP control packets on 423 a single port for each RTP session is REQUIRED, as specified in 424 [RFC5761]. For backwards compatibility, implementations are also 425 REQUIRED to support RTP and RTCP sent on separate transport-layer 426 addresses. 428 Note that the use of RTP and RTCP multiplexed onto a single transport 429 port ensures that there is occasional traffic sent on that port, even 430 if there is no active media traffic. This can be useful to keep NAT 431 bindings alive, and is the recommend method for application level 432 keep-alives of RTP sessions [RFC6263]. 434 4.6. Reduced Size RTCP 436 RTCP packets are usually sent as compound RTCP packets, and [RFC3550] 437 requires that those compound packets start with an Sender Report (SR) 438 or Receiver Report (RR) packet. When using frequent RTCP feedback 439 messages under the RTP/AVPF Profile [RFC4585] these statistics are 440 not needed in every packet, and unnecessarily increase the mean RTCP 441 packet size. This can limit the frequency at which RTCP packets can 442 be sent within the RTCP bandwidth share. 444 To avoid this problem, [RFC5506] specifies how to reduce the mean 445 RTCP message size and allow for more frequent feedback. Frequent 446 feedback, in turn, is essential to make real-time applications 447 quickly aware of changing network conditions, and to allow them to 448 adapt their transmission and encoding behaviour. Support for non- 449 compound RTCP feedback packets [RFC5506] is REQUIRED, but MUST be 450 negotiated using the signalling channel before use. For backwards 451 compatibility, implementations are also REQUIRED to support the use 452 of compound RTCP feedback packets if the remote endpoint does not 453 agree to the use of non-compound RTCP in the signalling exchange. 455 4.7. Symmetric RTP/RTCP 457 To ease traversal of NAT and firewall devices, implementations are 458 REQUIRED to implement and use Symmetric RTP [RFC4961]. The reasons 459 for using symmetric RTP is primarily to avoid issues with NAT and 460 Firewalls by ensuring that the flow is actually bi-directional and 461 thus kept alive and registered as flow the intended recipient 462 actually wants. In addition, it saves resources, specifically ports 463 at the end-points, but also in the network as NAT mappings or 464 firewall state is not unnecessary bloated. Also the amount of QoS 465 state is reduced. 467 4.8. Choice of RTP Synchronisation Source (SSRC) 469 Implementations are REQUIRED to support signalled RTP synchronisation 470 source (SSRC) identifiers, using the "a=ssrc:" SDP attribute defined 471 in Section 4.1 and Section 5 of [RFC5576]. Implementations MUST also 472 support the "previous-ssrc" source attribute defined in Section 6.2 473 of [RFC5576]. Other per-SSRC attributes defined in [RFC5576] MAY be 474 supported. 476 Use of the "a=ssrc:" attribute to signal SSRC identifiers in an RTP 477 session is OPTIONAL. Implementations MUST be prepared to accept RTP 478 and RTCP packets using SSRCs that have not been explicitly signalled 479 ahead of time. Implementations MUST support random SSRC assignment, 480 and MUST support SSRC collision detection and resolution, according 481 to [RFC3550]. When using signalled SSRC values, collision detection 482 MUST be performed as described in Section 5 of [RFC5576]. 484 It is often desirable to associate an RTP media stream with a non-RTP 485 context (e.g., to associate an RTP media stream with an "m=" line in 486 a session description formatted using SDP). If SSRCs are signalled 487 this is straightforward (in SDP the "a=ssrc:" line will be at the 488 media level, allowing a direct association with an "m=" line). If 489 SSRCs are not signalled, the RTP payload type numbers used in an RTP 490 media stream are often sufficient to associate that media stream with 491 a signalling context (e.g., if RTP payload type numbers are assigned 492 as described in Section 4.3 of this memo, the RTP payload types used 493 by an RTP media stream can be compared with values in SDP "a=rtpmap:" 494 lines, which are at the media level in SDP, and so map to an "m=" 495 line). 497 4.9. Generation of the RTCP Canonical Name (CNAME) 499 The RTCP Canonical Name (CNAME) provides a persistent transport-level 500 identifier for an RTP endpoint. While the Synchronisation Source 501 (SSRC) identifier for an RTP endpoint can change if a collision is 502 detected, or when the RTP application is restarted, its RTCP CNAME is 503 meant to stay unchanged, so that RTP endpoints can be uniquely 504 identified and associated with their RTP media streams within a set 505 of related RTP sessions. For proper functionality, each RTP endpoint 506 needs to have a unique RTCP CNAME value. 508 The RTP specification [RFC3550] includes guidelines for choosing a 509 unique RTP CNAME, but these are not sufficient in the presence of NAT 510 devices. In addition, long-term persistent identifiers can be 511 problematic from a privacy viewpoint. Accordingly, support for 512 generating a short-term persistent RTCP CNAMEs following [RFC7022] is 513 RECOMMENDED. 515 An WebRTC end-point MUST support reception of any CNAME that matches 516 the syntax limitations specified by the RTP specification [RFC3550] 517 and cannot assume that any CNAME will be chosen according to the form 518 suggested above. 520 5. WebRTC Use of RTP: Extensions 522 There are a number of RTP extensions that are either needed to obtain 523 full functionality, or extremely useful to improve on the baseline 524 performance, in the WebRTC application context. One set of these 525 extensions is related to conferencing, while others are more generic 526 in nature. The following subsections describe the various RTP 527 extensions mandated or suggested for use within the WebRTC context. 529 5.1. Conferencing Extensions 531 RTP is inherently a group communication protocol. Groups can be 532 implemented using a centralised server, multi-unicast, or using IP 533 multicast. While IP multicast is popular in IPTV systems, overlay- 534 based topologies dominate in interactive conferencing environments. 535 Such overlay-based topologies typically use one or more central 536 servers to connect end-points in a star or flat tree topology. These 537 central servers can be implemented in a number of ways as discussed 538 in the memo on RTP Topologies 539 [I-D.ietf-avtcore-rtp-topologies-update]. 541 Not all of the possible the overlay-based topologies are suitable for 542 use in the WebRTC environment. Specifically: 544 o The use of video switching MCUs makes the use of RTCP for 545 congestion control and quality of service reports problematic (see 546 Section 3.6.2 of [I-D.ietf-avtcore-rtp-topologies-update]). 548 o The use of content modifying MCUs with RTCP termination breaks RTP 549 loop detection, and prevents receivers from identifying active 550 senders (see section 3.8 of 551 [I-D.ietf-avtcore-rtp-topologies-update]). 553 Accordingly, only Point to Point (Topo-Point-to-Point), Multiple 554 concurrent Point to Point (Mesh) and RTP Mixers (Topo-Mixer) 555 topologies are needed to achieve the use-cases to be supported in 556 WebRTC initially. These RECOMMENDED topologies are expected to be 557 supported by all WebRTC end-points (these topologies require no 558 special RTP-layer support in the end-point if the RTP features 559 mandated in this memo are implemented). 561 The RTP extensions described in Section 5.1.1 to Section 5.1.6 are 562 designed to be used with centralised conferencing, where an RTP 563 middlebox (e.g., a conference bridge) receives a participant's RTP 564 media streams and distributes them to the other participants. These 565 extensions are not necessary for interoperability; an RTP endpoint 566 that does not implement these extensions will work correctly, but 567 might offer poor performance. Support for the listed extensions will 568 greatly improve the quality of experience and, to provide a 569 reasonable baseline quality, some these extensions are mandatory to 570 be supported by WebRTC end-points. 572 The RTCP conferencing extensions are defined in Extended RTP Profile 573 for Real-time Transport Control Protocol (RTCP)-Based Feedback (RTP/ 574 AVPF) [RFC4585] and the "Codec Control Messages in the RTP Audio- 575 Visual Profile with Feedback (AVPF)" (CCM) [RFC5104] and are fully 576 usable by the Secure variant of this profile (RTP/SAVPF) [RFC5124]. 578 5.1.1. Full Intra Request (FIR) 580 The Full Intra Request is defined in Sections 3.5.1 and 4.3.1 of the 581 Codec Control Messages [RFC5104]. This message is used to make the 582 mixer request a new Intra picture from a participant in the session. 583 This is used when switching between sources to ensure that the 584 receivers can decode the video or other predictive media encoding 585 with long prediction chains. WebRTC senders MUST understand and 586 react to the FIR feedback message since it greatly improves the user 587 experience when using centralised mixer-based conferencing; support 588 for sending the FIR message is OPTIONAL. 590 5.1.2. Picture Loss Indication (PLI) 592 The Picture Loss Indication is defined in Section 6.3.1 of the RTP/ 593 AVPF profile [RFC4585]. It is used by a receiver to tell the sending 594 encoder that it lost the decoder context and would like to have it 595 repaired somehow. This is semantically different from the Full Intra 596 Request above as there could be multiple ways to fulfil the request. 597 WebRTC senders MUST understand and react to this feedback message as 598 a loss tolerance mechanism; receivers MAY send PLI messages. 600 5.1.3. Slice Loss Indication (SLI) 602 The Slice Loss Indicator is defined in Section 6.3.2 of the RTP/AVPF 603 profile [RFC4585]. It is used by a receiver to tell the encoder that 604 it has detected the loss or corruption of one or more consecutive 605 macro blocks, and would like to have these repaired somehow. Support 606 for this feedback message is OPTIONAL as a loss tolerance mechanism. 608 5.1.4. Reference Picture Selection Indication (RPSI) 610 Reference Picture Selection Indication (RPSI) is defined in 611 Section 6.3.3 of the RTP/AVPF profile [RFC4585]. Some video coding 612 standards allow the use of older reference pictures than the most 613 recent one for predictive coding. If such a codec is in used, and if 614 the encoder has learned about a loss of encoder-decoder 615 synchronisation, a known-as-correct reference picture can be used for 616 future coding. The RPSI message allows this to be signalled. 617 Support for RPSI messages is OPTIONAL. 619 5.1.5. Temporal-Spatial Trade-off Request (TSTR) 621 The temporal-spatial trade-off request and notification are defined 622 in Sections 3.5.2 and 4.3.2 of [RFC5104]. This request can be used 623 to ask the video encoder to change the trade-off it makes between 624 temporal and spatial resolution, for example to prefer high spatial 625 image quality but low frame rate. Support for TSTR requests and 626 notifications is OPTIONAL. 628 5.1.6. Temporary Maximum Media Stream Bit Rate Request (TMMBR) 630 This feedback message is defined in Sections 3.5.4 and 4.2.1 of the 631 Codec Control Messages [RFC5104]. This message and its notification 632 message are used by a media receiver to inform the sending party that 633 there is a current limitation on the amount of bandwidth available to 634 this receiver. This can be various reasons for this: for example, an 635 RTP mixer can use this message to limit the media rate of the sender 636 being forwarded by the mixer (without doing media transcoding) to fit 637 the bottlenecks existing towards the other session participants. 638 WebRTC senders are REQUIRED to implement support for TMMBR messages, 639 and MUST follow bandwidth limitations set by a TMMBR message received 640 for their SSRC. The sending of TMMBR requests is OPTIONAL. 642 5.2. Header Extensions 644 The RTP specification [RFC3550] provides the capability to include 645 RTP header extensions containing in-band data, but the format and 646 semantics of the extensions are poorly specified. The use of header 647 extensions is OPTIONAL in the WebRTC context, but if they are used, 648 they MUST be formatted and signalled following the general mechanism 649 for RTP header extensions defined in [RFC5285], since this gives 650 well-defined semantics to RTP header extensions. 652 As noted in [RFC5285], the requirement from the RTP specification 653 that header extensions are "designed so that the header extension may 654 be ignored" [RFC3550] stands. To be specific, header extensions MUST 655 only be used for data that can safely be ignored by the recipient 656 without affecting interoperability, and MUST NOT be used when the 657 presence of the extension has changed the form or nature of the rest 658 of the packet in a way that is not compatible with the way the stream 659 is signalled (e.g., as defined by the payload type). Valid examples 660 might include metadata that is additional to the usual RTP 661 information. 663 5.2.1. Rapid Synchronisation 665 Many RTP sessions require synchronisation between audio, video, and 666 other content. This synchronisation is performed by receivers, using 667 information contained in RTCP SR packets, as described in the RTP 668 specification [RFC3550]. This basic mechanism can be slow, however, 669 so it is RECOMMENDED that the rapid RTP synchronisation extensions 670 described in [RFC6051] be implemented in addition to RTCP SR-based 671 synchronisation. The rapid synchronisation extensions use the 672 general RTP header extension mechanism [RFC5285], which requires 673 signalling, but are otherwise backwards compatible. 675 5.2.2. Client-to-Mixer Audio Level 677 The Client to Mixer Audio Level extension [RFC6464] is an RTP header 678 extension used by a client to inform a mixer about the level of audio 679 activity in the packet to which the header is attached. This enables 680 a central node to make mixing or selection decisions without decoding 681 or detailed inspection of the payload, reducing the complexity in 682 some types of central RTP nodes. It can also save decoding resources 683 in receivers, which can choose to decode only the most relevant RTP 684 media streams based on audio activity levels. 686 The Client-to-Mixer Audio Level [RFC6464] extension is RECOMMENDED to 687 be implemented. If it is implemented, it is REQUIRED that the header 688 extensions are encrypted according to [RFC6904] since the information 689 contained in these header extensions can be considered sensitive. 691 5.2.3. Mixer-to-Client Audio Level 693 The Mixer to Client Audio Level header extension [RFC6465] provides 694 the client with the audio level of the different sources mixed into a 695 common mix by a RTP mixer. This enables a user interface to indicate 696 the relative activity level of each session participant, rather than 697 just being included or not based on the CSRC field. This is a pure 698 optimisations of non critical functions, and is hence OPTIONAL to 699 implement. If it is implemented, it is REQUIRED that the header 700 extensions are encrypted according to [RFC6904] since the information 701 contained in these header extensions can be considered sensitive. 703 5.2.4. Associating RTP Media Streams and Signalling Contexts 705 (tbd: it seems likely that we need a mechanism to associate RTP media 706 streams with signalling contexts. The mechanism by which this is 707 done will likely be some combination of an RTP header extension, 708 periodic transmission of a new RTCP SDES item, and some signalling 709 extension. The semantics of those items are not yet settled; see 710 draft-westerlund-avtext-rtcp-sdes-srcname, draft-ietf-mmusic-msid, 711 and draft-even-mmusic-application-token for discussion). 713 6. WebRTC Use of RTP: Improving Transport Robustness 714 There are tools that can make RTP media streams robust against packet 715 loss and reduce the impact of loss on media quality. However, they 716 all add extra bits compared to a non-robust stream. The overhead of 717 these extra bits needs to be considered, and the aggregate bit-rate 718 MUST be rate controlled to avoid causing network congestion (see 719 Section 7). As a result, improving robustness might require a lower 720 base encoding quality, but has the potential to deliver that quality 721 with fewer errors. The mechanisms described in the following sub- 722 sections can be used to improve tolerance to packet loss. 724 6.1. Negative Acknowledgements and RTP Retransmission 726 As a consequence of supporting the RTP/SAVPF profile, implementations 727 will support negative acknowledgements (NACKs) for RTP data packets 728 [RFC4585]. This feedback can be used to inform a sender of the loss 729 of particular RTP packets, subject to the capacity limitations of the 730 RTCP feedback channel. A sender can use this information to optimise 731 the user experience by adapting the media encoding to compensate for 732 known lost packets, for example. 734 Senders are REQUIRED to understand the Generic NACK message defined 735 in Section 6.2.1 of [RFC4585], but MAY choose to ignore this feedback 736 (following Section 4.2 of [RFC4585]). Receivers MAY send NACKs for 737 missing RTP packets; [RFC4585] provides some guidelines on when to 738 send NACKs. It is not expected that a receiver will send a NACK for 739 every lost RTP packet, rather it needs to consider the cost of 740 sending NACK feedback, and the importance of the lost packet, to make 741 an informed decision on whether it is worth telling the sender about 742 a packet loss event. 744 The RTP Retransmission Payload Format [RFC4588] offers the ability to 745 retransmit lost packets based on NACK feedback. Retransmission needs 746 to be used with care in interactive real-time applications to ensure 747 that the retransmitted packet arrives in time to be useful, but can 748 be effective in environments with relatively low network RTT (an RTP 749 sender can estimate the RTT to the receivers using the information in 750 RTCP SR and RR packets, as described at the end of Section 6.4.1 of 751 [RFC3550]). The use of retransmissions can also increase the forward 752 RTP bandwidth, and can potentially worsen the problem if the packet 753 loss was caused by network congestion. We note, however, that 754 retransmission of an important lost packet to repair decoder state 755 can have lower cost than sending a full intra frame. It is not 756 appropriate to blindly retransmit RTP packets in response to a NACK. 757 The importance of lost packets and the likelihood of them arriving in 758 time to be useful needs to be considered before RTP retransmission is 759 used. 761 Receivers are REQUIRED to implement support for RTP retransmission 762 packets [RFC4588]. Senders MAY send RTP retransmission packets in 763 response to NACKs if the RTP retransmission payload format has been 764 negotiated for the session, and if the sender believes it is useful 765 to send a retransmission of the packet(s) referenced in the NACK. An 766 RTP sender is not expected to retransmit every NACKed packet. 768 6.2. Forward Error Correction (FEC) 770 The use of Forward Error Correction (FEC) can provide an effective 771 protection against some degree of packet loss, at the cost of steady 772 bandwidth overhead. There are several FEC schemes that are defined 773 for use with RTP. Some of these schemes are specific to a particular 774 RTP payload format, others operate across RTP packets and can be used 775 with any payload format. It needs to be noted that using redundant 776 encoding or FEC will lead to increased play out delay, which needs to 777 be considered when choosing the redundancy or FEC formats and their 778 respective parameters. 780 If an RTP payload format negotiated for use in a WebRTC session 781 supports redundant transmission or FEC as a standard feature of that 782 payload format, then that support MAY be used in the WebRTC session, 783 subject to any appropriate signalling. 785 There are several block-based FEC schemes that are designed for use 786 with RTP independent of the chosen RTP payload format. At the time 787 of this writing there is no consensus on which, if any, of these FEC 788 schemes is appropriate for use in the WebRTC context. Accordingly, 789 this memo makes no recommendation on the choice of block-based FEC 790 for WebRTC use. 792 7. WebRTC Use of RTP: Rate Control and Media Adaptation 794 WebRTC will be used in heterogeneous network environments using a 795 variety set of link technologies, including both wired and wireless 796 links, to interconnect potentially large groups of users around the 797 world. As a result, the network paths between users can have widely 798 varying one-way delays, available bit-rates, load levels, and traffic 799 mixtures. Individual end-points can send one or more RTP media 800 streams to each participant in a WebRTC conference, and there can be 801 several participants. Each of these RTP media streams can contain 802 different types of media, and the type of media, bit rate, and number 803 of flows can be highly asymmetric. Non-RTP traffic can share the 804 network paths RTP flows. Since the network environment is not 805 predictable or stable, WebRTC endpoints MUST ensure that the RTP 806 traffic they generate can adapt to match changes in the available 807 network capacity. 809 The quality of experience for users of WebRTC implementation is very 810 dependent on effective adaptation of the media to the limitations of 811 the network. End-points have to be designed so they do not transmit 812 significantly more data than the network path can support, except for 813 very short time periods, otherwise high levels of network packet loss 814 or delay spikes will occur, causing media quality degradation. The 815 limiting factor on the capacity of the network path might be the link 816 bandwidth, or it might be competition with other traffic on the link 817 (this can be non-WebRTC traffic, traffic due to other WebRTC flows, 818 or even competition with other WebRTC flows in the same session). 820 An effective media congestion control algorithm is therefore an 821 essential part of the WebRTC framework. However, at the time of this 822 writing, there is no standard congestion control algorithm that can 823 be used for interactive media applications such as WebRTC flows. 824 Some requirements for congestion control algorithms for WebRTC 825 sessions are discussed in [I-D.jesup-rtp-congestion-reqs], and it is 826 expected that a future version of this memo will mandate the use of a 827 congestion control algorithm that satisfies these requirements. 829 7.1. Boundary Conditions and Circuit Breakers 831 In the absence of a concrete congestion control algorithm, all WebRTC 832 implementations MUST implement the RTP circuit breaker algorithm that 833 is in described [I-D.ietf-avtcore-rtp-circuit-breakers]. The circuit 834 breaker defines a conservative boundary condition for safe operation, 835 chosen such that applications that trigger the circuit breaker will 836 almost certainly be causing severe network congestion. Any future 837 RTP congestion control algorithms are expected to operate within the 838 envelope allowed by the circuit breaker. 840 The session establishment signalling will also necessarily establish 841 boundaries to which the media bit-rate will conform. The choice of 842 media codecs provides upper- and lower-bounds on the supported bit- 843 rates that the application can utilise to provide useful quality, and 844 the packetization choices that exist. In addition, the signalling 845 channel can establish maximum media bit-rate boundaries using the SDP 846 "b=AS:" or "b=CT:" lines, and the RTP/AVPF Temporary Maximum Media 847 Stream Bit Rate (TMMBR) Requests (see Section 5.1.6 of this memo). 848 The combination of media codec choice and signalled bandwidth limits 849 SHOULD be used to limit traffic based on known bandwidth limitations, 850 for example the capacity of the edge links, to the extent possible. 852 7.2. RTCP Limitations for Congestion Control 854 Experience with the congestion control algorithms of TCP [RFC5681], 855 TFRC [RFC5348], and DCCP [RFC4341], [RFC4342], [RFC4828], has shown 856 that feedback on packet arrivals needs to be sent roughly once per 857 round trip time. We note that the real-time media traffic might not 858 have to adapt to changing path conditions as rapidly as needed for 859 the elastic applications TCP was designed for, but frequent feedback 860 is still needed to allow the congestion control algorithm to track 861 the path dynamics. 863 The total RTCP bandwidth is limited in its transmission rate to a 864 fraction of the RTP traffic (by default 5%). RTCP packets are larger 865 than, e.g., TCP ACKs (even when non-compound RTCP packets are used). 866 The RTP media stream bit rate thus limits the maximum feedback rate 867 as a function of the mean RTCP packet size. 869 Interactive communication might not be able to afford waiting for 870 packet losses to occur to indicate congestion, because an increase in 871 play out delay due to queuing (most prominent in wireless networks) 872 can easily lead to packets being dropped due to late arrival at the 873 receiver. Therefore, more sophisticated cues might need to be 874 reported -- to be defined in a suitable congestion control framework 875 as noted above -- which, in turn, increase the report size again. 876 For example, different RTCP XR report blocks (jointly) provide the 877 necessary details to implement a variety of congestion control 878 algorithms, but the (compound) report size grows quickly. 880 In group communication, the share of RTCP bandwidth needs to be 881 shared by all group members, reducing the capacity and thus the 882 reporting frequency per node. 884 Example: assuming 512 kbit/s video yields 3200 bytes/s RTCP 885 bandwidth, split across two entities in a point-to-point session. An 886 endpoint could thus send a report of 100 bytes about every 70ms or 887 for every other frame in a 30 fps video. 889 7.3. Congestion Control Interoperability and Legacy Systems 891 There are legacy implementations that do not implement RTCP, and 892 hence do not provide any congestion feedback. Congestion control 893 cannot be performed with these end-points. WebRTC implementations 894 that need to interwork with such end-points MUST limit their 895 transmission to a low rate, equivalent to a VoIP call using a low 896 bandwidth codec, that is unlikely to cause any significant 897 congestion. 899 When interworking with legacy implementations that support RTCP using 900 the RTP/AVP profile [RFC3551], congestion feedback is provided in 901 RTCP RR packets every few seconds. Implementations that have to 902 interwork with such end-points MUST ensure that they keep within the 903 RTP circuit breaker [I-D.ietf-avtcore-rtp-circuit-breakers] 904 constraints to limit the congestion they can cause. 906 If a legacy end-point supports RTP/AVPF, this enables negotiation of 907 important parameters for frequent reporting, such as the "trr-int" 908 parameter, and the possibility that the end-point supports some 909 useful feedback format for congestion control purpose such as TMMBR 910 [RFC5104]. Implementations that have to interwork with such end- 911 points MUST ensure that they stay within the RTP circuit breaker 912 [I-D.ietf-avtcore-rtp-circuit-breakers] constraints to limit the 913 congestion they can cause, but might find that they can achieve 914 better congestion response depending on the amount of feedback that 915 is available. 917 With proprietary congestion control algorithms issues can arise when 918 different algorithms and implementations interact in a communication 919 session. If the different implementations have made different 920 choices in regards to the type of adaptation, for example one sender 921 based, and one receiver based, then one could end up in situation 922 where one direction is dual controlled, when the other direction is 923 not controlled. This memo cannot mandate behaviour for proprietary 924 congestion control algorithms, but implementations that use such 925 algorithms ought to be aware of this issue, and try to ensure that 926 both effective congestion control is negotiated for media flowing in 927 both directions. If the IETF were to standardise both sender- and 928 receiver-based congestion control algorithms for WebRTC traffic in 929 the future, the issues of interoperability, control, and ensuring 930 that both directions of media flow are congestion controlled would 931 also need to be considered. 933 8. WebRTC Use of RTP: Performance Monitoring 935 As described in Section 4.1, implementations are REQUIRED to generate 936 RTCP Sender Report (SR) and Reception Report (RR) packets relating to 937 the RTP media streams they send and receive. These RTCP reports can 938 be used for performance monitoring purposes, since they include basic 939 packet loss and jitter statistics. 941 A large number of additional performance metrics are supported by the 942 RTCP Extended Reports (XR) framework [RFC3611]. It is not yet clear 943 what extended metrics are appropriate for use in the WebRTC context, 944 so implementations are not expected to generate any RTCP XR packets. 945 However, implementations that can use detailed performance monitoring 946 data MAY generate RTCP XR packets as appropriate; the use of such 947 packets SHOULD be signalled in advance. 949 All WebRTC implementations MUST be prepared to receive RTP XR report 950 packets, whether or not they were signalled. There is no requirement 951 that the data contained in such reports be used, or exposed to the 952 Javascript application, however. 954 9. WebRTC Use of RTP: Future Extensions 956 It is possible that the core set of RTP protocols and RTP extensions 957 specified in this memo will prove insufficient for the future needs 958 of WebRTC applications. In this case, future updates to this memo 959 MUST be made following the Guidelines for Writers of RTP Payload 960 Format Specifications [RFC2736] and Guidelines for Extending the RTP 961 Control Protocol [RFC5968], and SHOULD take into account any future 962 guidelines for extending RTP and related protocols that have been 963 developed. 965 Authors of future extensions are urged to consider the wide range of 966 environments in which RTP is used when recommending extensions, since 967 extensions that are applicable in some scenarios can be problematic 968 in others. Where possible, the WebRTC framework will adopt RTP 969 extensions that are of general utility, to enable easy implementation 970 of a gateway to other applications using RTP, rather than adopt 971 mechanisms that are narrowly targeted at specific WebRTC use cases. 973 10. Signalling Considerations 975 RTP is built with the assumption that an external signalling channel 976 exists, and can be used to configure RTP sessions and their features. 977 The basic configuration of an RTP session consists of the following 978 parameters: 980 RTP Profile: The name of the RTP profile to be used in session. The 981 RTP/AVP [RFC3551] and RTP/AVPF [RFC4585] profiles can interoperate 982 on basic level, as can their secure variants RTP/SAVP [RFC3711] 983 and RTP/SAVPF [RFC5124]. The secure variants of the profiles do 984 not directly interoperate with the non-secure variants, due to the 985 presence of additional header fields for authentication in SRTP 986 packets and cryptographic transformation of the payload. WebRTC 987 requires the use of the RTP/SAVPF profile, and this MUST be 988 signalled if SDP is used. Interworking functions might transform 989 this into the RTP/SAVP profile for a legacy use case, by 990 indicating to the WebRTC end-point that the RTP/SAVPF is used, and 991 limiting the usage of the "a=rtcp:" attribute to indicate a trr- 992 int value of 4 seconds. 994 Transport Information: Source and destination IP address(s) and 995 ports for RTP and RTCP MUST be signalled for each RTP session. In 996 WebRTC these transport addresses will be provided by ICE that 997 signals candidates and arrives at nominated candidate address 998 pairs. If RTP and RTCP multiplexing [RFC5761] is to be used, such 999 that a single port is used for RTP and RTCP flows, this MUST be 1000 signalled (see Section 4.5). If several RTP sessions are to be 1001 multiplexed onto a single transport layer flow, this MUST also be 1002 signalled (see Section 4.4). 1004 RTP Payload Types, media formats, and format parameters: The mapping 1005 between media type names (and hence the RTP payload formats to be 1006 used), and the RTP payload type numbers MUST be signalled. Each 1007 media type MAY also have a number of media type parameters that 1008 MUST also be signalled to configure the codec and RTP payload 1009 format (the "a=fmtp:" line from SDP). Section 4.3 of this memo 1010 discusses requirements for uniqueness of payload types. 1012 RTP Extensions: The RTP extensions to be used SHOULD be agreed upon, 1013 including any parameters for each respective extension. At the 1014 very least, this will help avoiding using bandwidth for features 1015 that the other end-point will ignore. But for certain mechanisms 1016 there is requirement for this to happen as interoperability 1017 failure otherwise happens. 1019 RTCP Bandwidth: Support for exchanging RTCP Bandwidth values to the 1020 end-points will be necessary. This SHALL be done as described in 1021 "Session Description Protocol (SDP) Bandwidth Modifiers for RTP 1022 Control Protocol (RTCP) Bandwidth" [RFC3556], or something 1023 semantically equivalent. This also ensures that the end-points 1024 have a common view of the RTCP bandwidth, this is important as too 1025 different view of the bandwidths can lead to failure to 1026 interoperate. 1028 These parameters are often expressed in SDP messages conveyed within 1029 an offer/answer exchange. RTP does not depend on SDP or on the offer 1030 /answer model, but does require all the necessary parameters to be 1031 agreed upon, and provided to the RTP implementation. We note that in 1032 the WebRTC context it will depend on the signalling model and API how 1033 these parameters need to be configured but they will be need to 1034 either set in the API or explicitly signalled between the peers. 1036 11. WebRTC API Considerations 1038 The WebRTC API and its media function have the concept of a WebRTC 1039 MediaStream that consists of zero or more tracks. A track is an 1040 individual stream of media from any type of media source like a 1041 microphone or a camera, but also conceptual sources, like a audio mix 1042 or a video composition, are possible. The tracks within a WebRTC 1043 MediaStream are expected to be synchronized. 1045 A track correspond to the media received with one particular SSRC. 1046 There might be additional SSRCs associated with that SSRC, like for 1047 RTP retransmission or Forward Error Correction. However, one SSRC 1048 will identify an RTP media stream and its timing. 1050 As a result, a WebRTC MediaStream is a collection of SSRCs carrying 1051 the different media included in the synchronised aggregate. 1052 Therefore, also the synchronization state associated with the 1053 included SSRCs are part of concept. It is important to consider that 1054 there can be multiple different WebRTC MediaStreams containing a 1055 given Track (SSRC). To avoid unnecessary duplication of media at the 1056 transport level in such cases, a need arises for a binding defining 1057 which WebRTC MediaStreams a given SSRC is associated with at the 1058 signalling level. 1060 The API also needs to be capable of handling when new SSRCs are 1061 received but not previously signalled by signalling in some fashion. 1062 Note, that not all SSRCs carries media directly associated with a 1063 media source, instead they can be repair or redundancy information 1064 for one or a set of SSRCs. 1066 A proposal for how the binding between WebRTC MediaStreams and SSRC 1067 can be done is specified in "Cross Session Stream Identification in 1068 the Session Description Protocol" [I-D.alvestrand-rtcweb-msid]. 1070 (tbd: This text needs to be improved and achieved consensus on. 1071 Interim meeting in June 2012 shows large differences in opinions.) 1073 (tbd: It is an open question whether these considerations are best 1074 discussed in this draft, in the W3C WebRTC API spec, or elsewhere. 1076 12. RTP Implementation Considerations 1078 The following discussion provides some guidance on the implementation 1079 of the RTP features described in this memo. The focus is on a WebRTC 1080 end-point implementation perspective, and while some mention is made 1081 of the behaviour of middleboxes, that is not the focus of this memo. 1083 12.1. Configuration and Use of RTP Sessions 1085 A WebRTC end-point will be a simultaneous participant in one or more 1086 RTP sessions. Each RTP session can convey multiple media flows, and 1087 can include media data from multiple end-points. In the following, 1088 we outline some ways in which WebRTC end-points can configure and use 1089 RTP sessions. 1091 12.1.1. Use of Multiple Media Flows Within an RTP Session 1093 RTP is a group communication protocol, and in a WebRTC context every 1094 RTP session can potentially contain multiple media flows. There are 1095 several reasons why this might be desirable: 1097 Multiple media types: Outside of WebRTC, it is common to use one RTP 1098 session for each type of media (e.g., one RTP session for audio 1099 and one for video, each sent on a different UDP port). However, 1100 to reduce the number of UDP ports used, the default in WebRTC is 1101 to send all types of media in a single RTP session, as described 1102 in Section 4.4, using RTP and RTCP multiplexing (Section 4.5) to 1103 further reduce the number of UDP ports needed. This RTP session 1104 then uses only one UDP flow, but will contain multiple RTP media 1105 streams, each containing a different type of media. A common 1106 example might be an end-point with a camera and microphone that 1107 sends two RTP streams, one video and one audio, into a single RTP 1108 session. 1110 Multiple Capture Devices: A WebRTC end-point might have multiple 1111 cameras, microphones, or other media capture devices, and so might 1112 want to generate several RTP media streams of the same media type. 1113 Alternatively, it might want to send media from a single capture 1114 device in several different formats or quality settings at once. 1115 Both can result in a single end-point sending multiple RTP media 1116 streams of the same media type into a single RTP session at the 1117 same time. 1119 Associated Repair Data: An end-point might send a media stream that 1120 is somehow associated with another stream. For example, it might 1121 send an RTP stream that contains FEC or retransmission data 1122 relating to another stream. Some RTP payload formats send this 1123 sort of associated repair data as part of the original media 1124 stream, while others send it as a separate stream. 1126 Layered or Multiple Description Coding: An end-point can use a 1127 layered media codec, for example H.264 SVC, or a multiple 1128 description codec, that generates multiple media flows, each with 1129 a distinct RTP SSRC, within a single RTP session. 1131 RTP Mixers, Translators, and Other Middleboxes: An RTP session, in 1132 the WebRTC context, is a point-to-point association between an 1133 end-point and some other peer device, where those devices share a 1134 common SSRC space. The peer device might be another WebRTC end- 1135 point, or it might be an RTP mixer, translator, or some other form 1136 of media processing middlebox. In the latter cases, the middlebox 1137 might send mixed or relayed RTP streams from several participants, 1138 that the WebRTC end-point will need to render. Thus, even though 1139 a WebRTC end-point might only be a member of a single RTP session, 1140 the peer device might be extending that RTP session to incorporate 1141 other end-points. WebRTC is a group communication environment and 1142 end-points need to be capable of receiving, decoding, and playing 1143 out multiple RTP media streams at once, even in a single RTP 1144 session. 1146 (tbd: Are any mechanism needed to signal limitations in the number 1147 of active SSRC that an end-point can handle?) 1149 (tbd: need to discuss signalling for the above here, preferably by 1150 referring to a separate document that describes SDP use for WebRTC) 1152 12.1.2. Use of Multiple RTP Sessions 1154 In addition to sending and receiving multiple media streams within a 1155 single RTP session, a WebRTC end-point might participate in multiple 1156 RTP sessions. There are several reasons why a WebRTC end-point might 1157 choose to do this: 1159 To interoperate with legacy devices: The common practice in the non- 1160 WebRTC world is to send different types of media in separate RTP 1161 sessions, for example using one RTP session for audio and another 1162 RTP session, on a different UDP port, for video. All WebRTC end- 1163 points need to support the option of sending different types of 1164 media on different RTP sessions, so they can interwork with such 1165 legacy devices. This is discussed further in Section 4.4. 1167 To provide enhanced quality of service: Some network-based quality 1168 of service mechanisms operate on the granularity of UDP 5-tuples. 1169 If it is desired to use these mechanisms to provide differentiated 1170 quality of service for some RTP flows, then those RTP flows need 1171 to be sent in a separate RTP session using a different UDP port 1172 number, and with appropriate quality of service marking. This is 1173 discussed further in Section 12.1.3. 1175 To separate media with different purposes: An end-point might want 1176 to send media streams that have different purposes on different 1177 RTP sessions, to make it easy for the peer device to distinguish 1178 them. For example, some centralised multiparty conferencing 1179 systems display the active speaker in high resolution, but show 1180 low resolution "thumbnails" of other participants. Such systems 1181 might configure the end-points to send simulcast high- and low- 1182 resolution versions of their video using separate RTP sessions, to 1183 simplify the operation of the central mixer In the WebRTC context 1184 this appears to be most easily accomplished by establishing 1185 multiple PeerConnection all being feed the same set of WebRTC 1186 MediaStreams. Each PeerConnection is then configured to deliver a 1187 particular media quality and thus media bit-rate, and will produce 1188 an independently encoded version with the codec parameters agreed 1189 specifically in the context of that PeerConnection. The central 1190 mixer can always distinguish packets corresponding to the low- and 1191 high-resolution streams by inspecting their SSRC, RTP payload 1192 type, or some other information contained in RTP header extensions 1193 or RTCP packets, but it can be easier to distinguish the flows if 1194 they arrive on separate RTP sessions on separate UDP ports. 1196 To directly connect with multiple peers: A multi-party conference 1197 does not need to use a central mixer. Rather, a multi-unicast 1198 mesh can be created, comprising several distinct RTP sessions, 1199 with each participant sending RTP traffic over a separate RTP 1200 session (that is, using an independent an PeerConnection object) 1201 to every other participant, as shown in Figure 1. This topology 1202 has the benefit of not requiring a central mixer node that is 1203 trusted to access and manipulate the media data. The downside is 1204 that it increases the used bandwidth at each sender by requiring 1205 one copy of the RTP media streams for each participant that are 1206 part of the same session beyond the sender itself. 1208 The multi-unicast topology could also be implemented as a single 1209 RTP session, spanning multiple peer-to-peer transport layer 1210 connections, or as several pairwise RTP sessions, one between each 1211 pair of peers. To maintain a coherent mapping between the 1212 relation between RTP sessions and PeerConnection objects we 1213 recommend that this is implemented as several individual RTP 1214 sessions. The only downside is that end-point A will not learn of 1215 the quality of any transmission happening between B and C, since 1216 it will not see RTCP reports for the RTP session between B and C, 1217 whereas it would it all three participants were part of a single 1218 RTP session. Experience with the Mbone tools (experimental RTP- 1219 based multicast conferencing tools from the late 1990s) has showed 1220 that RTCP reception quality reports for third parties can usefully 1221 be presented to the users in a way that helps them understand 1222 asymmetric network problems, and the approach of using separate 1223 RTP sessions prevents this. However, an advantage of using 1224 separate RTP sessions is that it enables using different media 1225 bit-rates and RTP session configurations between the different 1226 peers, thus not forcing B to endure the same quality reductions if 1227 there are limitations in the transport from A to C as C will. It 1228 it believed that these advantages outweigh the limitations in 1229 debugging power. 1231 To indirectly connect with multiple peers: A common scenario in 1232 multi-party conferencing is to create indirect connections to 1233 multiple peers, using an RTP mixer, translator, or some other type 1234 of RTP middlebox. Figure 2 outlines a simple topology that might 1235 be used in a four-person centralised conference. The middlebox 1236 acts to optimise the transmission of RTP media streams from 1237 certain perspectives, either by only sending some of the received 1238 RTP media stream to any given receiver, or by providing a combined 1239 RTP media stream out of a set of contributing streams. 1241 There are various methods of implementation for the middlebox. If 1242 implemented as a standard RTP mixer or translator, a single RTP 1243 session will extend across the middlebox and encompass all the 1244 end-points in one multi-party session. Other types of middlebox 1245 might use separate RTP sessions between each end-point and the 1246 middlebox. A common aspect is that these central nodes can use a 1247 number of tools to control the media encoding provided by a WebRTC 1248 end-point. This includes functions like requesting breaking the 1249 encoding chain and have the encoder produce a so called Intra 1250 frame. Another is limiting the bit-rate of a given stream to 1251 better suit the mixer view of the multiple down-streams. Others 1252 are controlling the most suitable frame-rate, picture resolution, 1253 the trade-off between frame-rate and spatial quality. The 1254 middlebox gets the significant responsibility to correctly perform 1255 congestion control, source identification, manage synchronization 1256 while providing the application with suitable media optimizations. 1257 The middlebox is also has to be a trusted node when it comes to 1258 security, since it manipulates either the RTP header or the media 1259 itself (or both) received from one end-point, before sending it on 1260 towards the end-point(s), thus they need to be able to decrypt and 1261 then encrypt it before sending it out. 1263 RTP Mixers can create a situation where an end-point experiences a 1264 situation in-between a session with only two end-points and 1265 multiple RTP sessions. Mixers are expected to not forward RTCP 1266 reports regarding RTP media streams across themselves. This is 1267 due to the difference in the RTP media streams provided to the 1268 different end-points. The original media source lacks information 1269 about a mixer's manipulations prior to sending it the different 1270 receivers. This scenario also results in that an end-point's 1271 feedback or requests goes to the mixer. When the mixer can't act 1272 on this by itself, it is forced to go to the original media source 1273 to fulfil the receivers request. This will not necessarily be 1274 explicitly visible any RTP and RTCP traffic, but the interactions 1275 and the time to complete them will indicate such dependencies. 1277 Providing source authentication in multi-party scenarios is a 1278 challenge. In the mixer-based topologies, end-points source 1279 authentication is based on, firstly, verifying that media comes 1280 from the mixer by cryptographic verification and, secondly, trust 1281 in the mixer to correctly identify any source towards the end- 1282 point. In RTP sessions where multiple end-points are directly 1283 visible to an end-point, all end-points will have knowledge about 1284 each others' master keys, and can thus inject packets claimed to 1285 come from another end-point in the session. Any node performing 1286 relay can perform non-cryptographic mitigation by preventing 1287 forwarding of packets that have SSRC fields that came from other 1288 end-points before. For cryptographic verification of the source 1289 SRTP would require additional security mechanisms, for example 1290 TESLA for SRTP [RFC4383], that are not part of the base WebRTC 1291 standards. 1293 To forward media between multiple peers: It might be desirable for 1294 an end-point that receives an RTP media stream to be able to 1295 forward that media stream to a third party. The are obvious 1296 security and privacy implications in this, but also potential 1297 uses. If it is to be allowed, there are two implementation 1298 strategies: either the browser can relay the flow at the RTP 1299 layer, or it transcode and forward the media at the application 1300 layer. 1302 A relay approach will result in the RTP session be extended beyond 1303 the PeerConnection, making both the original end-point and the 1304 destination to which the media is forwarded part of the RTP 1305 session. These end-points can have different path 1306 characteristics, and hence different reception quality. Thus 1307 sender's congestion control needs to be capable of handling this. 1308 The security solution can either support mechanism that the sender 1309 informs both receivers of the key; alternatively the end-point 1310 that is forwarding the media needs to decrypt and then re-encrypt 1311 using a new key. The relay based approach has the advantage that 1312 the forwarding end-point does not need to transcode the media, 1313 thus maintaining the quality of the encoding and reducing the 1314 computational complexity requirements. If the right security 1315 solutions are supported then the end-point that receives the 1316 forwarded media will be able to verify the authenticity of the 1317 media coming from the original sender. A downside is that the 1318 original sender is forced to take both receivers into 1319 consideration when delivering content. 1321 The media transcoder approach is similar to having the forwarding 1322 end-point act as Mixer, terminating the RTP session, combined with 1323 a transcoder. The original sender will only see a single receiver 1324 of its media. The receiving end-point will responsible to produce 1325 a RTP media stream suitable for onwards transmission. This might 1326 require media transcoding for congestion control purpose to 1327 produce a suitable bit-rate. Thus loosing media quality in the 1328 transcoding and forcing the forwarding end-point to spend the 1329 resource on the transcoding. The media transcoding does result in 1330 a separation of the two different legs removing almost all 1331 dependencies, and allowing the forwarding end-point to optimize 1332 its media transcoding operation. It also allows forwarding 1333 without the original sender being aware of the forwarding. The 1334 cost is greatly increased computational complexity on the 1335 forwarding node. 1337 (tbd: ought media forwarding be allowed?) 1339 +---+ +---+ 1340 | A |<--->| B | 1341 +---+ +---+ 1342 ^ ^ 1343 \ / 1344 \ / 1345 v v 1346 +---+ 1347 | C | 1348 +---+ 1350 Figure 1: Multi-unicast using several RTP sessions 1352 +---+ +-------------+ +---+ 1353 | A |<---->| |<---->| B | 1354 +---+ | RTP mixer, | +---+ 1355 | translator, | 1356 | or other | 1357 +---+ | middlebox | +---+ 1358 | C |<---->| |<---->| D | 1359 +---+ +-------------+ +---+ 1361 Figure 2: RTP mixer with only unicast paths 1363 12.1.3. Differentiated Treatment of Flows 1365 There are use cases for differentiated treatment of RTP media 1366 streams. Such differentiation can happen at several places in the 1367 system. First of all is the prioritization within the end-point 1368 sending the media, which controls, both which RTP media streams that 1369 will be sent, and their allocation of bit-rate out of the current 1370 available aggregate as determined by the congestion control. 1372 It is expected that the WebRTC API will allow the application to 1373 indicate relative priorities for different MediaStreamTracks. These 1374 priorities can then be used to influence the local RTP processing, 1375 especially when it comes to congestion control response in how to 1376 divide the available bandwidth between the RTP flows. Any changes in 1377 relative priority will also need to be considered for RTP flows that 1378 are associated with the main RTP flows, such as RTP retransmission 1379 streams and FEC. The importance of such associated RTP traffic flows 1380 is dependent on the media type and codec used, in regards to how 1381 robust that codec is to packet loss. However, a default policy might 1382 to be to use the same priority for associated RTP flows as for the 1383 primary RTP flow. 1385 Secondly, the network can prioritize packet flows, including RTP 1386 media streams. Typically, differential treatment includes two steps, 1387 the first being identifying whether an IP packet belongs to a class 1388 that has to be treated differently, the second the actual mechanism 1389 to prioritize packets. This is done according to three methods: 1391 DiffServ: The end-point marks a packet with a DiffServ code point to 1392 indicate to the network that the packet belongs to a particular 1393 class. 1395 Flow based: Packets that need to be given a particular treatment are 1396 identified using a combination of IP and port address. 1398 Deep Packet Inspection: A network classifier (DPI) inspects the 1399 packet and tries to determine if the packet represents a 1400 particular application and type that is to be prioritized. 1402 Flow-based differentiation will provide the same treatment to all 1403 packets within a flow, i.e., relative prioritization is not possible. 1404 Moreover, if the resources are limited it might not be possible to 1405 provide differential treatment compared to best-effort for all the 1406 flows in a WebRTC application. When flow-based differentiation is 1407 available the WebRTC application needs to know about it so that it 1408 can provide the separation of the RTP media streams onto different 1409 UDP flows to enable a more granular usage of flow based 1410 differentiation. That way at least providing different 1411 prioritization of audio and video if desired by application. 1413 DiffServ assumes that either the end-point or a classifier can mark 1414 the packets with an appropriate DSCP so that the packets are treated 1415 according to that marking. If the end-point is to mark the traffic 1416 two requirements arise in the WebRTC context: 1) The WebRTC 1417 application or browser has to know which DSCP to use and that it can 1418 use them on some set of RTP media streams. 2) The information needs 1419 to be propagated to the operating system when transmitting the 1420 packet. These issues are discussed in DSCP and other packet markings 1421 for RTCWeb QoS [I-D.ietf-rtcweb-qos]. 1423 For packet based marking schemes it would be possible in the context 1424 to mark individual RTP packets differently based on the relative 1425 priority of the RTP payload. For example video codecs that has I,P 1426 and B pictures could prioritise any payloads carrying only B frames 1427 less, as these are less damaging to loose. But as default policy all 1428 RTP packets related to a media stream ought to be provided with the 1429 same prioritization. 1431 It is also important to consider how RTCP packets associated with a 1432 particular RTP media flow need to be marked. RTCP compound packets 1433 with Sender Reports (SR), ought to be marked with the same priority 1434 as the RTP media flow itself, so the RTCP-based round-trip time (RTT) 1435 measurements are done using the same flow priority as the media flow 1436 experiences. RTCP compound packets containing RR packet ought to be 1437 sent with the priority used by the majority of the RTP media flows 1438 reported on. RTCP packets containing time-critical feedback packets 1439 can use higher priority to improve the timeliness and likelihood of 1440 delivery of such feedback. 1442 12.2. Source, Flow, and Participant Identification 1444 12.2.1. Media Streams 1446 Each RTP media stream is identified by a unique synchronisation 1447 source (SSRC) identifier. The SSRC identifier is carried in the RTP 1448 data packets comprising a media stream, and is also used to identify 1449 that stream in the corresponding RTCP reports. The SSRC is chosen as 1450 discussed in Section 4.8. The first stage in demultiplexing RTP and 1451 RTCP packets received at a WebRTC end-point is to separate the media 1452 streams based on their SSRC value; once that is done, additional 1453 demultiplexing steps can determine how and where to render the media. 1455 RTP allows a mixer, or other RTP-layer middlebox, to combine media 1456 flows from multiple sources to form a new media flow. The RTP data 1457 packets in that new flow can include a Contributing Source (CSRC) 1458 list, indicating which original SSRCs contributed to the combined 1459 packet. As described in Section 4.1, implementations need to support 1460 reception of RTP data packets containing a CSRC list and RTCP packets 1461 that relate to sources present in the CSRC list. The CSRC list can 1462 change on a packet-by-packet basis, depending on the mixing operation 1463 being performed. Knowledge of what sources contributed to a 1464 particular RTP packet can be important if the user interface 1465 indicates which participants are active in the session. Changes in 1466 the CSRC list included in packets needs to be exposed to the WebRTC 1467 application using some API, if the application is to be able to track 1468 changes in session participation. It is desirable to map CSRC values 1469 back into WebRTC MediaStream identities as they cross this API, to 1470 avoid exposing the SSRC/CSRC name space to JavaScript applications. 1472 If the mixer-to-client audio level extension [RFC6465] is being used 1473 in the session (see Section 5.2.3), the information in the CSRC list 1474 is augmented by audio level information for each contributing source. 1475 This information can usefully be exposed in the user interface. 1477 12.2.2. Media Streams: SSRC Collision Detection 1479 The RTP standard [RFC3550] requires any RTP implementation to have 1480 support for detecting and handling SSRC collisions, i.e., resolve the 1481 conflict when two different end-points use the same SSRC value. This 1482 requirement also applies to WebRTC end-points. There are several 1483 scenarios where SSRC collisions can occur. 1485 In a point-to-point session where each SSRC is associated with either 1486 of the two end-points and where the main media carrying SSRC 1487 identifier will be announced in the signalling channel, a collision 1488 is less likely to occur due to the information about used SSRCs 1489 provided by Source-Specific SDP Attributes [RFC5576]. Still if both 1490 end-points start uses an new SSRC identifier prior to having 1491 signalled it to the peer and received acknowledgement on the 1492 signalling message, there can be collisions. The Source-Specific SDP 1493 Attributes [RFC5576] contains no mechanism to resolve SSRC collisions 1494 or reject a end-points usage of an SSRC. 1496 There could also appear SSRC values that are not signalled. This is 1497 more likely than it appears as certain RTP functions need extra SSRCs 1498 to provide functionality related to another (the "main") SSRC, for 1499 example, SSRC multiplexed RTP retransmission [RFC4588]. In those 1500 cases, an end-point can create a new SSRC that strictly doesn't need 1501 to be announced over the signalling channel to function correctly on 1502 both RTP and PeerConnection level. 1504 The more likely case for SSRC collision is that multiple end-points 1505 in a multiparty conference create new sources and signals those 1506 towards the central server. In cases where the SSRC/CSRC are 1507 propagated between the different end-points from the central node 1508 collisions can occur. 1510 Another scenario is when the central node manages to connect an end- 1511 point's PeerConnection to another PeerConnection the end-point 1512 already has, thus forming a loop where the end-point will receive its 1513 own traffic. While is is clearly considered a bug, it is important 1514 that the end-point is able to recognise and handle the case when it 1515 occurs. This case becomes even more problematic when media mixers, 1516 and so on, are involved, where the stream received is a different 1517 stream but still contains this client's input. 1519 These SSRC/CSRC collisions can only be handled on RTP level as long 1520 as the same RTP session is extended across multiple PeerConnections 1521 by a RTP middlebox. To resolve the more generic case where multiple 1522 PeerConnections are interconnected, then identification of the media 1523 source(s) part of a MediaStreamTrack being propagated across multiple 1524 interconnected PeerConnection needs to be preserved across these 1525 interconnections. 1527 12.2.3. Media Synchronisation Context 1529 When an end-point sends media from more than one media source, it 1530 needs to consider if (and which of) these media sources are to be 1531 synchronized. In RTP/RTCP, synchronisation is provided by having a 1532 set of RTP media streams be indicated as coming from the same 1533 synchronisation context and logical end-point by using the same RTCP 1534 CNAME identifier. 1536 The next provision is that the internal clocks of all media sources, 1537 i.e., what drives the RTP timestamp, can be correlated to a system 1538 clock that is provided in RTCP Sender Reports encoded in an NTP 1539 format. By correlating all RTP timestamps to a common system clock 1540 for all sources, the timing relation of the different RTP media 1541 streams, also across multiple RTP sessions can be derived at the 1542 receiver and, if desired, the streams can be synchronized. The 1543 requirement is for the media sender to provide the correlation 1544 information; it is up to the receiver to use it or not. 1546 12.2.4. Correlation of Media Streams 1548 (tbd: this need to outline the approach to mapping media streams to 1549 the signalling context defined in the unified plan) 1551 (tbd: need to discuss correlation between associated RTP streams, for 1552 example between a media stream and its associated FEC stream) 1554 13. Security Considerations 1556 The overall security architecture for WebRTC is described in 1557 [I-D.ietf-rtcweb-security-arch], and security considerations for the 1558 WebRTC framework are described in [I-D.ietf-rtcweb-security]. These 1559 considerations apply to this memo also. 1561 The security considerations of the RTP specification, the RTP/SAVPF 1562 profile, and the various RTP/RTCP extensions and RTP payload formats 1563 that form the complete protocol suite described in this memo apply. 1564 We do not believe there are any new security considerations resulting 1565 from the combination of these various protocol extensions. 1567 The Extended Secure RTP Profile for Real-time Transport Control 1568 Protocol (RTCP)-Based Feedback [RFC5124] (RTP/SAVPF) provides 1569 handling of fundamental issues by offering confidentiality, integrity 1570 and partial source authentication. A mandatory to implement media 1571 security solution is (tbd). 1573 RTCP packets convey a Canonical Name (CNAME) identifier that is used 1574 to associate media flows that need to be synchronised across related 1575 RTP sessions. Inappropriate choice of CNAME values can be a privacy 1576 concern, since long-term persistent CNAME identifiers can be used to 1577 track users across multiple WebRTC calls. Section 4.9 of this memo 1578 provides guidelines for generation of untraceable CNAME values that 1579 alleviate this risk. 1581 The guidelines in [RFC6562] apply when using variable bit rate (VBR) 1582 audio codecs such as Opus (see Section 4.3 for discussion of mandated 1583 audio codecs). These guidelines in [RFC6562] also apply, but are of 1584 lesser importance, when using the client-to-mixer audio level header 1585 extensions (Section 5.2.2) or the mixer-to-client audio level header 1586 extensions (Section 5.2.3). 1588 14. IANA Considerations 1590 This memo makes no request of IANA. 1592 Note to RFC Editor: this section is to be removed on publication as 1593 an RFC. 1595 15. Open Issues 1597 This section contains a summary of the open issues or to be done 1598 things noted in the document: 1600 1. tbd: The API mapping to RTP level concepts has to be agreed and 1601 documented in Section 11. 1603 2. tbd: An open question if any requirements are needed to agree and 1604 limit the number of simultaneously used media sources (SSRCs) 1605 within an RTP session. See Section 4.1. 1607 3. tbd: The method for achieving simulcast of a media source has to 1608 be decided. 1610 4. tbd: Possible documentation of what support for differentiated 1611 treatment that are needed on RTP level as the API and the network 1612 level specification matures as discussed in Section 12.1.3. 1614 16. Acknowledgements 1616 The authors would like to thank Harald Alvestrand, Cary Bran, Charles 1617 Eckel, Cullen Jennings, Bernard Aboba, and the other members of the 1618 IETF RTCWEB working group for their valuable feedback. 1620 17. References 1622 17.1. Normative References 1624 [I-D.ietf-avtcore-multi-media-rtp-session] 1625 Westerlund, M., Perkins, C., and J. Lennox, "Sending 1626 Multiple Types of Media in a Single RTP Session", draft- 1627 ietf-avtcore-multi-media-rtp-session-03 (work in 1628 progress), July 2013. 1630 [I-D.ietf-avtcore-rtp-circuit-breakers] 1631 Perkins, C. and V. Singh, "Multimedia Congestion Control: 1632 Circuit Breakers for Unicast RTP Sessions", draft-ietf- 1633 avtcore-rtp-circuit-breakers-03 (work in progress), July 1634 2013. 1636 [I-D.ietf-avtcore-rtp-multi-stream-optimisation] 1637 Lennox, J., Westerlund, M., Wu, Q., and C. Perkins, 1638 "Sending Multiple Media Streams in a Single RTP Session: 1639 Grouping RTCP Reception Statistics and Other Feedback ", 1640 draft-ietf-avtcore-rtp-multi-stream-optimisation-00 (work 1641 in progress), July 2013. 1643 [I-D.ietf-avtcore-rtp-multi-stream] 1644 Lennox, J., Westerlund, M., Wu, W., and C. Perkins, 1645 "Sending Multiple Media Streams in a Single RTP Session", 1646 draft-ietf-avtcore-rtp-multi-stream-01 (work in progress), 1647 July 2013. 1649 [I-D.ietf-avtext-multiple-clock-rates] 1650 Petit-Huguenin, M. and G. Zorn, "Support for Multiple 1651 Clock Rates in an RTP Session", draft-ietf-avtext- 1652 multiple-clock-rates-09 (work in progress), April 2013. 1654 [I-D.ietf-mmusic-sdp-bundle-negotiation] 1655 Holmberg, C., Alvestrand, H., and C. Jennings, 1656 "Multiplexing Negotiation Using Session Description 1657 Protocol (SDP) Port Numbers", draft-ietf-mmusic-sdp- 1658 bundle-negotiation-04 (work in progress), June 2013. 1660 [I-D.ietf-rtcweb-security-arch] 1661 Rescorla, E., "WebRTC Security Architecture", draft-ietf- 1662 rtcweb-security-arch-07 (work in progress), July 2013. 1664 [I-D.ietf-rtcweb-security] 1665 Rescorla, E., "Security Considerations for WebRTC", draft- 1666 ietf-rtcweb-security-05 (work in progress), July 2013. 1668 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1669 Requirement Levels", BCP 14, RFC 2119, March 1997. 1671 [RFC2736] Handley, M. and C. Perkins, "Guidelines for Writers of RTP 1672 Payload Format Specifications", BCP 36, RFC 2736, December 1673 1999. 1675 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 1676 Jacobson, "RTP: A Transport Protocol for Real-Time 1677 Applications", STD 64, RFC 3550, July 2003. 1679 [RFC3551] Schulzrinne, H. and S. Casner, "RTP Profile for Audio and 1680 Video Conferences with Minimal Control", STD 65, RFC 3551, 1681 July 2003. 1683 [RFC3556] Casner, S., "Session Description Protocol (SDP) Bandwidth 1684 Modifiers for RTP Control Protocol (RTCP) Bandwidth", RFC 1685 3556, July 2003. 1687 [RFC3711] Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K. 1688 Norrman, "The Secure Real-time Transport Protocol (SRTP)", 1689 RFC 3711, March 2004. 1691 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 1692 "Extended RTP Profile for Real-time Transport Control 1693 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, July 1694 2006. 1696 [RFC4588] Rey, J., Leon, D., Miyazaki, A., Varsa, V., and R. 1697 Hakenberg, "RTP Retransmission Payload Format", RFC 4588, 1698 July 2006. 1700 [RFC4961] Wing, D., "Symmetric RTP / RTP Control Protocol (RTCP)", 1701 BCP 131, RFC 4961, July 2007. 1703 [RFC5104] Wenger, S., Chandra, U., Westerlund, M., and B. Burman, 1704 "Codec Control Messages in the RTP Audio-Visual Profile 1705 with Feedback (AVPF)", RFC 5104, February 2008. 1707 [RFC5124] Ott, J. and E. Carrara, "Extended Secure RTP Profile for 1708 Real-time Transport Control Protocol (RTCP)-Based Feedback 1709 (RTP/SAVPF)", RFC 5124, February 2008. 1711 [RFC5285] Singer, D. and H. Desineni, "A General Mechanism for RTP 1712 Header Extensions", RFC 5285, July 2008. 1714 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 1715 Real-Time Transport Control Protocol (RTCP): Opportunities 1716 and Consequences", RFC 5506, April 2009. 1718 [RFC5761] Perkins, C. and M. Westerlund, "Multiplexing RTP Data and 1719 Control Packets on a Single Port", RFC 5761, April 2010. 1721 [RFC5764] McGrew, D. and E. Rescorla, "Datagram Transport Layer 1722 Security (DTLS) Extension to Establish Keys for the Secure 1723 Real-time Transport Protocol (SRTP)", RFC 5764, May 2010. 1725 [RFC6051] Perkins, C. and T. Schierl, "Rapid Synchronisation of RTP 1726 Flows", RFC 6051, November 2010. 1728 [RFC6464] Lennox, J., Ivov, E., and E. Marocco, "A Real-time 1729 Transport Protocol (RTP) Header Extension for Client-to- 1730 Mixer Audio Level Indication", RFC 6464, December 2011. 1732 [RFC6465] Ivov, E., Marocco, E., and J. Lennox, "A Real-time 1733 Transport Protocol (RTP) Header Extension for Mixer-to- 1734 Client Audio Level Indication", RFC 6465, December 2011. 1736 [RFC6562] Perkins, C. and JM. Valin, "Guidelines for the Use of 1737 Variable Bit Rate Audio with Secure RTP", RFC 6562, March 1738 2012. 1740 [RFC6904] Lennox, J., "Encryption of Header Extensions in the Secure 1741 Real-time Transport Protocol (SRTP)", RFC 6904, April 1742 2013. 1744 [RFC7007] Terriberry, T., "Update to Remove DVI4 from the 1745 Recommended Codecs for the RTP Profile for Audio and Video 1746 Conferences with Minimal Control (RTP/AVP)", RFC 7007, 1747 August 2013. 1749 [RFC7022] Begen, A., Perkins, C., Wing, D., and E. Rescorla, 1750 "Guidelines for Choosing RTP Control Protocol (RTCP) 1751 Canonical Names (CNAMEs)", RFC 7022, September 2013. 1753 17.2. Informative References 1755 [I-D.alvestrand-rtcweb-msid] 1756 Alvestrand, H., "Cross Session Stream Identification in 1757 the Session Description Protocol", draft-alvestrand- 1758 rtcweb-msid-02 (work in progress), May 2012. 1760 [I-D.ietf-avt-srtp-ekt] 1761 Wing, D., McGrew, D., and K. Fischer, "Encrypted Key 1762 Transport for Secure RTP", draft-ietf-avt-srtp-ekt-03 1763 (work in progress), October 2011. 1765 [I-D.ietf-avtcore-rtp-topologies-update] 1766 Westerlund, M. and S. Wenger, "RTP Topologies", draft- 1767 ietf-avtcore-rtp-topologies-update-00 (work in progress), 1768 April 2013. 1770 [I-D.ietf-rtcweb-overview] 1771 Alvestrand, H., "Overview: Real Time Protocols for Brower- 1772 based Applications", draft-ietf-rtcweb-overview-08 (work 1773 in progress), September 2013. 1775 [I-D.ietf-rtcweb-qos] 1776 Dhesikan, S., Druta, D., Jones, P., and J. Polk, "DSCP and 1777 other packet markings for RTCWeb QoS", draft-ietf-rtcweb- 1778 qos-00 (work in progress), October 2012. 1780 [I-D.ietf-rtcweb-use-cases-and-requirements] 1781 Holmberg, C., Hakansson, S., and G. Eriksson, "Web Real- 1782 Time Communication Use-cases and Requirements", draft- 1783 ietf-rtcweb-use-cases-and-requirements-11 (work in 1784 progress), June 2013. 1786 [I-D.jesup-rtp-congestion-reqs] 1787 Jesup, R. and H. Alvestrand, "Congestion Control 1788 Requirements For Real Time Media", draft-jesup-rtp- 1789 congestion-reqs-00 (work in progress), March 2012. 1791 [I-D.westerlund-avtcore-multiplex-architecture] 1792 Westerlund, M., Perkins, C., and H. Alvestrand, 1793 "Guidelines for using the Multiplexing Features of RTP", 1794 draft-westerlund-avtcore-multiplex-architecture-03 (work 1795 in progress), February 2013. 1797 [I-D.westerlund-avtcore-transport-multiplexing] 1798 Westerlund, M. and C. Perkins, "Multiple RTP Sessions on a 1799 Single Lower-Layer Transport", draft-westerlund-avtcore- 1800 transport-multiplexing-06 (work in progress), August 2013. 1802 [RFC3611] Friedman, T., Caceres, R., and A. Clark, "RTP Control 1803 Protocol Extended Reports (RTCP XR)", RFC 3611, November 1804 2003. 1806 [RFC4341] Floyd, S. and E. Kohler, "Profile for Datagram Congestion 1807 Control Protocol (DCCP) Congestion Control ID 2: TCP-like 1808 Congestion Control", RFC 4341, March 2006. 1810 [RFC4342] Floyd, S., Kohler, E., and J. Padhye, "Profile for 1811 Datagram Congestion Control Protocol (DCCP) Congestion 1812 Control ID 3: TCP-Friendly Rate Control (TFRC)", RFC 4342, 1813 March 2006. 1815 [RFC4383] Baugher, M. and E. Carrara, "The Use of Timed Efficient 1816 Stream Loss-Tolerant Authentication (TESLA) in the Secure 1817 Real-time Transport Protocol (SRTP)", RFC 4383, February 1818 2006. 1820 [RFC4828] Floyd, S. and E. Kohler, "TCP Friendly Rate Control 1821 (TFRC): The Small-Packet (SP) Variant", RFC 4828, April 1822 2007. 1824 [RFC5348] Floyd, S., Handley, M., Padhye, J., and J. Widmer, "TCP 1825 Friendly Rate Control (TFRC): Protocol Specification", RFC 1826 5348, September 2008. 1828 [RFC5576] Lennox, J., Ott, J., and T. Schierl, "Source-Specific 1829 Media Attributes in the Session Description Protocol 1830 (SDP)", RFC 5576, June 2009. 1832 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 1833 Control", RFC 5681, September 2009. 1835 [RFC5968] Ott, J. and C. Perkins, "Guidelines for Extending the RTP 1836 Control Protocol (RTCP)", RFC 5968, September 2010. 1838 [RFC6263] Marjou, X. and A. Sollaud, "Application Mechanism for 1839 Keeping Alive the NAT Mappings Associated with RTP / RTP 1840 Control Protocol (RTCP) Flows", RFC 6263, June 2011. 1842 Authors' Addresses 1844 Colin Perkins 1845 University of Glasgow 1846 School of Computing Science 1847 Glasgow G12 8QQ 1848 United Kingdom 1850 Email: csp@csperkins.org 1852 Magnus Westerlund 1853 Ericsson 1854 Farogatan 6 1855 SE-164 80 Kista 1856 Sweden 1858 Phone: +46 10 714 82 87 1859 Email: magnus.westerlund@ericsson.com 1861 Joerg Ott 1862 Aalto University 1863 School of Electrical Engineering 1864 Espoo 02150 1865 Finland 1867 Email: jorg.ott@aalto.fi