idnits 2.17.1 draft-westerlund-avtcore-rtp-topologies-update-02.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 : ---------------------------------------------------------------------------- -- The draft header indicates that this document obsoletes RFC5117, but the abstract doesn't seem to directly say this. It does mention RFC5117 though, so this could be OK. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 25, 2013) is 4071 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-10) exists of draft-ietf-avtcore-rtp-security-options-01 == Outdated reference: A later version (-02) exists of draft-lennox-avtcore-rtp-multi-stream-01 -- Obsolete informational reference (is this intentional?): RFC 5766 (Obsoleted by RFC 8656) Summary: 0 errors (**), 0 flaws (~~), 3 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group M. Westerlund 3 Internet-Draft Ericsson 4 Obsoletes: 5117 (if approved) S. Wenger 5 Intended status: Informational Vidyo 6 Expires: August 29, 2013 February 25, 2013 8 RTP Topologies 9 draft-westerlund-avtcore-rtp-topologies-update-02 11 Abstract 13 This document discusses point to point and multi-endpoint topologies 14 used in Real-time Transport Protocol (RTP)-based environments. In 15 particular, centralized topologies commonly employed in the video 16 conferencing industry are mapped to the RTP terminology. 18 This document is updated with additional topologies and are intended 19 to replace RFC 5117. 21 Status of this Memo 23 This Internet-Draft is submitted in full conformance with the 24 provisions of BCP 78 and BCP 79. 26 Internet-Drafts are working documents of the Internet Engineering 27 Task Force (IETF). Note that other groups may also distribute 28 working documents as Internet-Drafts. The list of current Internet- 29 Drafts is at http://datatracker.ietf.org/drafts/current/. 31 Internet-Drafts are draft documents valid for a maximum of six months 32 and may be updated, replaced, or obsoleted by other documents at any 33 time. It is inappropriate to use Internet-Drafts as reference 34 material or to cite them other than as "work in progress." 36 This Internet-Draft will expire on August 29, 2013. 38 Copyright Notice 40 Copyright (c) 2013 IETF Trust and the persons identified as the 41 document authors. All rights reserved. 43 This document is subject to BCP 78 and the IETF Trust's Legal 44 Provisions Relating to IETF Documents 45 (http://trustee.ietf.org/license-info) in effect on the date of 46 publication of this document. Please review these documents 47 carefully, as they describe your rights and restrictions with respect 48 to this document. Code Components extracted from this document must 49 include Simplified BSD License text as described in Section 4.e of 50 the Trust Legal Provisions and are provided without warranty as 51 described in the Simplified BSD License. 53 Table of Contents 55 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 56 2. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 3 57 2.1. Glossary . . . . . . . . . . . . . . . . . . . . . . . . . 3 58 3. Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . 4 59 3.1. Point to Point . . . . . . . . . . . . . . . . . . . . . . 4 60 3.2. Point to Point via Middlebox . . . . . . . . . . . . . . . 5 61 3.2.1. Translators . . . . . . . . . . . . . . . . . . . . . 5 62 3.2.2. Back to Back RTP sessions . . . . . . . . . . . . . . 8 63 3.3. Point to Multipoint Using Multicast . . . . . . . . . . . 9 64 3.3.1. Any Source Multicast (ASM) . . . . . . . . . . . . . . 9 65 3.3.2. Source Specific Multicast (SSM) . . . . . . . . . . . 11 66 3.3.3. SSM with Local Unicast Resources . . . . . . . . . . . 12 67 3.4. Point to Multipoint Using Mesh . . . . . . . . . . . . . . 14 68 3.5. Point to Multipoint Using the RFC 3550 Translator . . . . 15 69 3.5.1. Relay - Transport Translator . . . . . . . . . . . . . 15 70 3.5.2. Media Translator . . . . . . . . . . . . . . . . . . . 16 71 3.6. Point to Multipoint Using the RFC 3550 Mixer Model . . . . 16 72 3.6.1. Media Mixing . . . . . . . . . . . . . . . . . . . . . 18 73 3.6.2. Media Switching . . . . . . . . . . . . . . . . . . . 21 74 3.7. Source Projecting Middlebox . . . . . . . . . . . . . . . 23 75 3.8. Point to Multipoint Using Video Switching MCUs . . . . . . 26 76 3.9. Point to Multipoint Using RTCP-Terminating MCU . . . . . . 27 77 3.10. De-composite Endpoint . . . . . . . . . . . . . . . . . . 29 78 3.11. Non-Symmetric Mixer/Translators . . . . . . . . . . . . . 30 79 3.12. Combining Topologies . . . . . . . . . . . . . . . . . . . 30 80 4. Comparing Topologies . . . . . . . . . . . . . . . . . . . . . 31 81 4.1. Topology Properties . . . . . . . . . . . . . . . . . . . 31 82 4.1.1. All to All Media Transmission . . . . . . . . . . . . 31 83 4.1.2. Transport or Media Interoperability . . . . . . . . . 32 84 4.1.3. Per Domain Bit-Rate Adaptation . . . . . . . . . . . . 32 85 4.1.4. Aggregation of Media . . . . . . . . . . . . . . . . . 32 86 4.1.5. View of All Session Participants . . . . . . . . . . . 32 87 4.1.6. Loop Detection . . . . . . . . . . . . . . . . . . . . 33 88 4.2. Comparison of Topologies . . . . . . . . . . . . . . . . . 33 89 5. Security Considerations . . . . . . . . . . . . . . . . . . . 33 90 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 35 91 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 36 92 8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 36 93 8.1. Normative References . . . . . . . . . . . . . . . . . . . 36 94 8.2. Informative References . . . . . . . . . . . . . . . . . . 36 95 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 37 97 1. Introduction 99 Real-time Transport Protocol (RTP) [RFC3550] topologies describe 100 methods for interconnecting RTP entities and their processing 101 behavior of RTP and RTCP. This document tries to address past and 102 existing confusion, especially with respect to terms not defined in 103 RTP but in common use in the conversational communication industry, 104 such as MCU. In doing so, this memo provides a common information 105 basis for future discussion and specification work. It attempts to 106 clarify and explain sections of the Real-time Transport Protocol 107 (RTP) spec [RFC3550] in an informal way. It is not intended to 108 update or change what is normatively specified within RFC 3550. 110 When the Audio-Visual Profile with Feedback (AVPF) [RFC4585] was 111 developed the main emphasis lay in the efficient support of point to 112 point and small multipoint scenarios without centralized multipoint 113 control. However, in practice, many small multipoint conferences 114 operate utilizing devices known as Multipoint Control Units (MCUs). 115 MCUs may implement Mixer or Translator (in RTP [RFC3550] terminology) 116 functionality and signalling support. They may also contain 117 additional application functionality. This document focuses on the 118 media transport aspects of the MCU that can be realized using RTP, as 119 discussed below. Further considered are the properties of Mixers and 120 Translators, and how some types of deployed MCUs deviate from these 121 properties. 123 2. Definitions 125 2.1. Glossary 127 ASM: Any Source Multicast 129 AVPF: The Extended RTP Profile for RTCP-based Feedback 131 CSRC: Contributing Source 133 Link: The data transport to the next IP hop 135 MCU: Multipoint Control Unit 137 Path: The concatenation of multiple links, resulting in an end-to- 138 end data transfer. 140 PtM: Point to Multipoint 141 PtP: Point to Point 143 SSM: Source-Specific Multicast 145 SSRC: Synchronization Source 147 3. Topologies 149 This subsection defines several topologies that are relevant for 150 codec control but also RTP usage in other contexts. The section 151 starts with point to point cases, without and with middleboxes. Then 152 follows a number of different methods for establishing point to 153 multipoint communication. These are structure around the most 154 fundamental enabler, i.e. multicast, a mesh of connections, 155 translators, mixers and source projection middlebox, to finally 156 discuss MCUs. The section ends by discussing de-composed endpoints, 157 asymmetric middlebox behaviors and combining topologies. 159 The topologies may be referenced in other documents by a shortcut 160 name, indicated by the prefix "Topo-". 162 For each of the RTP-defined topologies, we discuss how RTP, RTCP, and 163 the carried media are handled. With respect to RTCP, we also discuss 164 the handling of RTCP feedback messages as defined in [RFC4585] and 165 [RFC5104]. Any important differences between the two will be 166 illuminated in the discussion. At this stage we don't intended to 167 discuss in detail how each and every feedback messages should be 168 treated in the various topologies. 170 3.1. Point to Point 172 Shortcut name: Topo-Point-to-Point 174 The Point to Point (PtP) topology (Figure 1) consists of two 175 endpoints, communicating using unicast. Both RTP and RTCP traffic 176 are conveyed endpoint-to-endpoint, using unicast traffic only (even 177 if, in exotic cases, this unicast traffic happens to be conveyed over 178 an IP-multicast address). 180 +---+ +---+ 181 | A |<------->| B | 182 +---+ +---+ 184 Figure 1: Point to Point 186 The main property of this topology is that A sends to B, and only B, 187 while B sends to A, and only A. This avoids all complexities of 188 handling multiple endpoints and combining the requirements from them. 189 Note that an endpoint can still use multiple RTP Synchronization 190 Sources (SSRCs) in an RTP session. The number of RTP sessions in use 191 between A and B can also be of any number. 193 RTCP feedback messages for the indicated SSRCs are communicated 194 directly between the endpoints. Therefore, this topology poses 195 minimal (if any) issues for any feedback messages. For RTP sessions 196 which use multiple SSRC per endpoint it can be relevant to implement 197 support for cross reporting suppression as defined in "Real-Time 198 Transport Protocol (RTP) Considerations for Endpoints Sending 199 Multiple Media Streams" [I-D.lennox-avtcore-rtp-multi-stream]. 201 3.2. Point to Point via Middlebox 203 This section discusses cases where two endpoints communicate but have 204 one or more middlebox involved in the RTP session. 206 3.2.1. Translators 208 Shortcut name: Topo-PtP-Translator 210 Two main categories of Translators can be distinguished; Transport 211 Translators and Media translators. Both Translator types share 212 common attributes that separate them from Mixers. For each media 213 stream that the Translator receives, it generates an individual 214 stream in the other domain. A translator keeps the SSRC for a stream 215 across the translation, whereas a Mixer can select a single media 216 stream, or send out multiple mixed media streams, but always under 217 its own SSRC, possibly using the CSRC field to indicate the source(s) 218 of the content. Mixers are more common in point to multipoint cases 219 than in PtP. The reason is that in PtP use cases the primary focus 220 is interoperability, such as transcoding to a codec the receiver 221 supports, which can be done by a media translator. 223 As specified in Section 7.1 of [RFC3550], the SSRC space is common 224 for all participants in the RTP session, independent of on which side 225 of the Translator the session resides. Therefore, it is the 226 responsibility of the participants to run SSRC collision detection, 227 and the SSRC is thus a field the Translator cannot change. Any SDES 228 information associated with a SSRC or CSRC also needs to be forwarded 229 between the domains for any SSRC/CSRC used in the different domains. 231 A Translator commonly does not use an SSRC of its own, and is not 232 visible as an active participant in the session. One reason to have 233 its own SSRC is when a Translator acts as a quality monitor that 234 sends RTCP reports and therefore is required to have an SSRC. 235 Another example is the case when a Translator is prepared to use RTCP 236 feedback messages. This may, for example, occur in a translator 237 configured to detect packet loss of important video packets and wants 238 to trigger repair by the media sender, by sending feedback messages. 239 While such feedback could use the SSRC of the target for the 240 translator, but this in turn would require translation of the targets 241 RTCP reports to make them consistent. It may be simpler to expose an 242 additional SSRC in the session, the only concern are endpoints 243 failing to support the full RTP specification, thus having issues 244 with multiple SSRCs reporting on the RTP streams sent by that 245 endpoint. 247 In general, a Translator implementation should consider which RTCP 248 feedback messages or codec-control messages it needs to understand in 249 relation to the functionality of the Translator itself. This is 250 completely in line with the requirement to also translate RTCP 251 messages between the domains. 253 3.2.1.1. Transport Relay/Anchoring 255 There exist a number of different types of middleboxes that might be 256 inserted between two RTP endpoints on the transport level, e.g. 257 perform changes on the IP/UDP headers, and are, therefore, basic 258 transport translators. These middleboxes come in many variations 259 including NAT [RFC3022] traversal by pinning the media path to a 260 public address domain relay, network topologies where the media flow 261 is required to pass a particular point for audit by employing 262 relaying, or preserving privacy by hiding each peers transport 263 addresses to the other party. Other protocols or functionalities 264 that provide this behavior are TURN [RFC5766] servers, Session Border 265 Gateways and Media Processing Nodes with media anchoring 266 functionalities. 268 +---+ +---+ +---+ 269 | A |<------>| T |<------->| B | 270 +---+ +---+ +---+ 272 Figure 2: Point to Point with Translator 274 What is common for these functions is that they are normally 275 transparent on RTP level, i.e. they perform no changes on any RTP or 276 RTCP packet fields, only on the lower layers. However, they may 277 effect the path the RTP and RTCP packets are routed between the 278 endpoints in the RTP session, and thereby only indirectly affect the 279 RTP session. For this reason, one could believe that transport 280 translator type middleboxes do not need to included in this document. 281 However, this topology can raise additional requirements the RTP 282 implementation and its interactions with the signalling solution. 283 Both in signalling and in certain RTCP field other network addresses 284 than those of the relay can occur, due to that B has different 285 network address than the relay (T). However, implementation not 286 capable of this will neither not work when endpoints are subject to 287 NAT. 289 3.2.1.2. Transport Translator 291 Transport Translators (Topo-Trn-Translator) do not modify the media 292 stream itself, but are concerned with transport parameters. 293 Transport parameters, in the sense of this section, comprise the 294 transport addresses (to bridge different domains such unicast to 295 multicast) and the media packetization to allow other transport 296 protocols to be interconnected to a session (in gateways). Of the 297 transport Translators, this memo is primarily interested in those 298 that use RTP on both sides, and this is assumed henceforth. 299 Translators that bridge between different protocol worlds need to be 300 concerned about the mapping of the SSRC/CSRC (Contributing Source) 301 concept to the non-RTP protocol. When designing a Translator to a 302 non-RTP-based media transport, one crucial factor lies in how to 303 handle different sources and their identities. This problem space is 304 not discussed henceforth. 306 The most basic transport translators that operate below RTP level was 307 already discussed in Section 3.2.1.1. 309 3.2.1.3. Media Translator 311 Media Translators (Topo-Media-Translator), in contrast, modify the 312 media stream itself. This process is commonly known as transcoding. 313 The modification of the media stream can be as small as removing 314 parts of the stream, and it can go all the way to a full transcoding 315 (down to the sample level or equivalent) utilizing a different media 316 codec. Media Translators are commonly used to connect entities 317 without a common interoperability point in the media encoding. 319 Stand-alone Media Translators are rare. Most commonly, a combination 320 of Transport and Media Translators are used to translate both the 321 media stream and the transport aspects of a stream between two 322 transport domains (or clouds). 324 When media translation occurs, the Translator's task regarding 325 handling of RTCP traffic becomes substantially more complex. In this 326 case, the Translator needs to rewrite B's RTCP Receiver Report before 327 forwarding them to A. The rewriting is needed as the stream received 328 by B is not the same stream as the other participants receive. For 329 example, the number of packets transmitted to B may be lower than 330 what A sends, due to the different media format and data rate. 331 Therefore, if the Receiver Reports were forwarded without changes, 332 the extended highest sequence number would indicate that B were 333 substantially behind in reception, while it most likely it would not 334 be. Therefore, the Translator must translate that number to a 335 corresponding sequence number for the stream the Translator received. 336 Similar arguments can be made for most other fields in the RTCP 337 Receiver Reports. 339 A media Translator may in some cases act on behalf of the "real" 340 source and respond to RTCP feedback messages. This may occur, for 341 example, when a receiver requests a bandwidth reduction, and the 342 media Translator has not detected any congestion or other reasons for 343 bandwidth reduction between the media source and itself. In that 344 case, it is sensible that the media Translator reacts to the codec 345 control messages itself, for example, by transcoding to a lower media 346 rate. 348 A variant of translator behaviour worth pointing out is the one 349 depicted in Figure 3 of an endpoint A sends a media flow to B. On the 350 path there is a device T that on A's behalf does something with the 351 media streams, for example adds an RTP session with FEC information 352 for A's media streams. In this case, T needs to bind the new FEC 353 streams to A's media stream, for example by using the same CNAME as 354 A. 356 +------+ +------+ +------+ 357 | | | | | | 358 | A |------->| T |-------->| B | 359 | | | |---FEC-->| | 360 +------+ +------+ +------+ 362 Figure 3: When De-composition is a Translator 364 This type of functionality where T does something with the media 365 stream on behalf of A is covered under the media translator 366 definition. 368 3.2.2. Back to Back RTP sessions 370 There exist middleboxes that interconnect two endpoints through 371 themselves not by being part of a common RTP session. Instead they 372 establish two different RTP sessions, one between A and the middlebox 373 (MB) and another between the MB and B. 375 |<--Session A-->| |<--Session B-->| 376 +------+ +------+ +------+ 377 | A |------->| MB |-------->| B | 378 +------+ +------+ +------+ 379 Figure 4: When De-composition is a Translator 381 The MB acts as a application level gateway and bridges the two RTP 382 session. This bridging can be as basic as forwarding the RTP 383 payloads between the sessions, or more complex including media 384 transcoding. The difference with the single RTP session context is 385 the handling of the SSRCs and the other session related identifiers, 386 such as CNAMEs. With two different RTP sessions these can be freely 387 changed and it becomes the MB task to maintain the right relations. 389 The signalling or other above-RTP level functionalities referencing 390 RTP media streams may be what is most impacted by using two RTP 391 sessions and changing identifiers. The structure with two RTP 392 sessions also puts a congestion control requirement on the middlebox, 393 because it becomes fully responsible for the media stream it sources 394 into each of the sessions. 396 This can be solved locally or by bridging also statistics from the 397 receiving endpoint. However, from an implementation point this 398 requires the implementation to support dealing with a number of 399 inconsistencies. First, packet loss must be detected for an RTP flow 400 sent from A to the MB, and that loss must be reported through a 401 skipped sequence number in the flow from the MB to B. This coupling 402 and the resulting inconsistencies is conceptually easier to handle 403 when considering the two flows as belonging to a single RTP session. 405 3.3. Point to Multipoint Using Multicast 407 Multicast is a IP layer functionality that is available in some 408 networks. Two main flavors can be distinguished: Any Source 409 Multicast (ASM) where any multicast group participant can send to the 410 group address and expect the packet to reach all group participants; 411 and Source Specific Multicast (SSM), where only a particular IP host 412 sends to the multicast group. Both these models are discussed below 413 in their respective section. 415 3.3.1. Any Source Multicast (ASM) 417 Shortcut name: Topo-ASM (was Topo-Multicast) 418 +-----+ 419 +---+ / \ +---+ 420 | A |----/ \---| B | 421 +---+ / Multi- \ +---+ 422 + Cast + 423 +---+ \ Network / +---+ 424 | C |----\ /---| D | 425 +---+ \ / +---+ 426 +-----+ 428 Figure 5: Point to Multipoint Using Multicast 430 Point to Multipoint (PtM) is defined here as using a multicast 431 topology as a transmission model, in which traffic from any 432 participant reaches all the other participants, except for cases such 433 as: 435 o packet loss, or 437 o when a participant does not wish to receive the traffic for a 438 specific multicast group and, therefore, has not subscribed to the 439 IP-multicast group in question. This scenario can occur, for 440 example, where a multi-media session is distributed using two or 441 more multicast groups and a participant is subscribed only to a 442 subset of these sessions. 444 In the above context, "traffic" encompasses both RTP and RTCP 445 traffic. The number of participants can vary between one and many, 446 as RTP and RTCP scale to very large multicast groups (the theoretical 447 limit of the number of participants in a single RTP session is in the 448 range of billions). The above can be realized using Any Source 449 Multicast (ASM). 451 For feedback usage, it is useful to define a "small multicast group" 452 as a group where the number of participants is so low (and other 453 factors such as the connectivity is so good) that it allows the 454 participants to use early or immediate feedback, as defined in AVPF 455 [RFC4585]. Even when the environment would allow for the use of a 456 small multicast group, some applications may still want to use the 457 more limited options for RTCP feedback available to large multicast 458 groups, for example when there is a likelyhood that the threshold of 459 the small multicast group (in terms of participants) may be exceeded 460 during the lifetime of a session. 462 RTCP feedback messages in multicast reach, like media, every 463 subscriber (subject to packet losses and multicast group 464 subscription). Therefore, the feedback suppression mechanism 465 discussed in [RFC4585] is typically required. Each individual node 466 needs to process every feedback message it receives, not to determine 467 if it is affected or if the feedback message applies only to some 468 other participant, but also to derive timing restriction for the 469 sending of its own feedback messages, if any. 471 3.3.2. Source Specific Multicast (SSM) 473 In Any Source Multicast, any of the participants can send to all the 474 other participants, by sending a packet to the multicast group. In 475 contrast, Source Specific Multicast [RFC4607] refers to scenarios 476 where only a single source (Distribution Source) can send to the 477 multicast group, creating a topology that looks like the one below: 479 +--------+ +-----+ 480 |Media | | | Source-specific 481 |Sender 1|<----->| D S | Multicast 482 +--------+ | I O | +--+----------------> R(1) 483 | S U | | | | 484 +--------+ | T R | | +-----------> R(2) | 485 |Media |<----->| R C |->+ | : | | 486 |Sender 2| | I E | | +------> R(n-1) | | 487 +--------+ | B | | | | | | 488 : | U | +--+--> R(n) | | | 489 : | T +-| | | | | 490 : | I | |<---------+ | | | 491 +--------+ | O |F|<---------------+ | | 492 |Media | | N |T|<--------------------+ | 493 |Sender M|<----->| | |<-------------------------+ 494 +--------+ +-----+ RTCP Unicast 496 FT = Feedback Target 497 Transport from the Feedback Target to the Distribution 498 Source is via unicast or multicast RTCP if they are not 499 co-located. 501 Figure 6: Point to Multipoint using Source Specific Multicast 503 In the SSM topology (Figure 6) a number of RTP sources (1 to M) are 504 allowed to send media to the SSM group. These send media to a 505 dedicated distribution source, which then forwards the media streams 506 to the multicast group on behalf of the original senders. The media 507 streams reach the Receivers (R(1) to R(n)). The Receivers' RTCP 508 cannot be sent to the multicast group, as the SSM multicast group by 509 definition has only a single source. To support RTCP, an RTP 510 extension for SSM [RFC5760] was defined. It uses unicast 511 transmission to send RTCP from each of the receivers to one or more 512 Feedback Targets (FT). The feedback targets relay the RTCP 513 unmodified, or provide summary of the participants RTCP reports 514 towards the whole group by forwarding the RTCP traffic to the 515 distribution source. Figure 6 only shows a single feedback target 516 integrated in the distribution source, but for scalability the FT can 517 be many and have responsibility for sub-groups of the receivers. For 518 summary reports, however, there must be a single feedback aggregating 519 all the summaries to a common message to the whole receiver group. 521 The RTP extension for SSM specifies how feedback (both reception 522 information and specific feedback events) are handled. The more 523 general problems associated with the use of multicast, where everyone 524 receives what the distribution source sends needs to be accounted 525 for. 527 The result of this is some common behaviours for RTP multicast: 529 1. Multicast applications often use a group of RTP sessions, not 530 one. Each endpoint needs to be a member of most or all of these 531 RTP sessions in order to perform well. 533 2. Within each RTP session, the number of media sinks is likely to 534 be much larger than the number of RTP sources. 536 3. Multicast applications need signalling functions to identify the 537 relationships between RTP sessions. 539 4. Multicast applications need signalling functions to identify the 540 relationships between SSRCs in different RTP sessions. 542 All multicast configurations share a signalling requirement: all of 543 the participants need to have the same RTP and payload type 544 configuration. Otherwise, A could, for example, be using payload 545 type 97 to identify the video codec H.264, while B would identify it 546 as MPEG-2. 548 Security solutions for this type of group communications are also 549 challenging. First, the key-management and the security protocol 550 must support group communication. Source authentication becomes more 551 difficult and requires special solutions. For more discussion on 552 this please review Options for Securing RTP Sessions 553 [I-D.ietf-avtcore-rtp-security-options]. 555 3.3.3. SSM with Local Unicast Resources 557 [RFC6285] "Unicast-Based Rapid Acquisition of Multicast RTP Sessions" 558 results in additional extensions to SSM Topology. 560 ----------- -------------- 561 | |------------------------------------>| | 562 | |.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.->| | 563 | | | | 564 | Multicast | ---------------- | | 565 | Source | | Retransmission | | | 566 | |-------->| Server (RS) | | | 567 | |.-.-.-.->| | | | 568 | | | ------------ | | | 569 ----------- | | Feedback | |<.=.=.=.=.| | 570 | | Target (FT)| |<~~~~~~~~~| RTP Receiver | 571 PRIMARY MULTICAST | ------------ | | (RTP_Rx) | 572 RTP SESSION with | | | | 573 UNICAST FEEDBACK | | | | 574 | | | | 575 - - - - - - - - - - - |- - - - - - - - |- - - - - |- - - - - - - |- - 576 | | | | 577 UNICAST BURST | ------------ | | | 578 (or RETRANSMISSION) | | Burst/ | |<~~~~~~~~>| | 579 RTP SESSION | | Retrans. | |.........>| | 580 | |Source (BRS)| |<.=.=.=.=>| | 581 | ------------ | | | 582 | | | | 583 ---------------- -------------- 585 -------> Multicast RTP Flow 586 .-.-.-.> Multicast RTCP Flow 587 .=.=.=.> Unicast RTCP Reports 588 ~~~~~~~> Unicast RTCP Feedback Messages 589 .......> Unicast RTP Flow 591 Figure 7 593 The Rapid acquisition extension allows an endpoint joining an SSM 594 multicast session to request media starting with the last sync-point 595 (from where media can be decoded without prior packets) to be sent at 596 high speed until such time where, after decoding of these bursted 597 media packets, the correct media timing is established, i.e. media 598 packets are received within adequate buffer intervals for this 599 application. This is accomplished by first establishing an unicast 600 PtP RTP session between the BRS (Figure 7) and the RTP Receiver. 601 That session is used to transmit cached packets from the multicast 602 group at higher then nominal speed so to synchronize the receiver to 603 the ongoing multicast packet flow. Once the RTP receiver and its 604 decoder have caught up with the multicast session's current delivery, 605 the receiver switches over to receiving from the multicast group 606 directly. The (still existing) PtP RTP session can be used as a 607 repair channel, i.e. for RTP Retransmission traffic of those packets 608 that were not received from the multicast group. 610 3.4. Point to Multipoint Using Mesh 612 Shortcut name: Topo-Mesh 614 +---+ +---+ 615 | A |<---->| B | 616 +---+ +---+ 617 ^ ^ 618 \ / 619 \ / 620 v v 621 +---+ 622 | C | 623 +---+ 625 Figure 8: Point to Multi-Point using Mesh 627 Based on the RTP session definition, it is clearly possible to have a 628 joint RTP session over multiple unicast transport flows like the 629 above three endpoint joint session. In this case, A needs to send 630 its' media streams and RTCP packets to both B and C over their 631 respective transport flows. As long as all participants do the same, 632 everyone will have a joint view of the RTP session. 634 This doesn't create any additional requirements beyond the need to 635 have multiple transport flows associated with a single RTP session. 636 Note that an endpoint may use a single local port to receive all 637 these transport flows, or it might have separate local reception 638 ports for each of the endpoints. 640 An alternative structure for establishing the above topology is to 641 use independent RTP sessions between each pair of peers, i.e. three 642 different RTP sessions. In some scenarios, the same RTP media stream 643 is being sent from each sending endpoint. In others, some form of 644 local adaptation takes place in one or more of the RTP media streams, 645 rendering them non-identical. From a topologies viewpoint, a 646 difference exists in the behaviours around RTCP. For example, when a 647 single RTP session spans all three endpoints and their connecting 648 flows, a RTCP bandwidth is calculated and used for this single one 649 joint session. In contrast, when there are multiple independent RTP 650 sessions, each has its local RTCP bandwidth allocation. Also, when 651 multiple sessions are used, endpoints not directly involved in these 652 sessions do not have any awareness of the conditions occurring in 653 sessions not involving that endpoint. For example, in case of the 654 three endpoint configuration above, endpoint A has no awareness of 655 the conditions occurring in the session between endpoints B and C 656 (whereas, if a single RTP session were used, it would have such 657 awareness). Loop detection is also affected. With independent RTP 658 sessions, the SSRC/CSRC can't be used to determine when a endpoint 659 receives its own media stream or a mixed media stream including its 660 own media stream a condition known as a loop. The identification of 661 loops and, in most cases, its avoidance, has to be achieved by other 662 means, for example through signaling, or the use of an RTP external 663 name space binding SSRC/CSRC among any communicating RTP sessions in 664 the mesh. 666 3.5. Point to Multipoint Using the RFC 3550 Translator 668 This section discusses some additional usages related to point to 669 multipoint of Translators compared to the point to point only cases 670 in Section 3.2.1. 672 3.5.1. Relay - Transport Translator 674 Shortcut name: Topo-PtM-Trn-Translator 676 This section discusses Transport Translator only usages to enable 677 multipoint sessions. 679 +-----+ 680 +---+ / \ +------------+ +---+ 681 | A |<---/ \ | |<---->| B | 682 +---+ / Multi- \ | | +---+ 683 + Cast +->| Translator | 684 +---+ \ Network / | | +---+ 685 | C |<---\ / | |<---->| D | 686 +---+ \ / +------------+ +---+ 687 +-----+ 689 Figure 9: Point to Multipoint Using Multicast 691 Figure 9 depicts an example of a Transport Translator performing at 692 least IP address translation. It allows the (non-multicast-capable) 693 participants B and D to take part in an any source multicast session 694 by having the Translator forward their unicast traffic to the 695 multicast addresses in use, and vice versa. It must also forward B's 696 traffic to D, and vice versa, to provide each of B and D with a 697 complete view of the session. 699 +---+ +------------+ +---+ 700 | A |<---->| |<---->| B | 701 +---+ | | +---+ 702 | Translator | 703 +---+ | | +---+ 704 | C |<---->| |<---->| D | 705 +---+ +------------+ +---+ 707 Figure 10: RTP Translator (Relay) with Only Unicast Paths 709 Another Translator scenario is depicted in Figure 10. Herein, the 710 Translator connects multiple users of a conference through unicast. 711 This can be implemented using a very simple transport Translator, 712 which in this document is called a relay. The relay forwards all 713 traffic it receives, both RTP and RTCP, to all other participants. 714 In doing so, a multicast network is emulated without relying on a 715 multicast-capable network infrastructure. 717 For RTCP feedback this results in a similar set of considerations 718 those described in the ASM RTP topology. It also puts some 719 additional signalling requirements onto the session establishment; 720 for example, a common configuration of RTP payload types is required. 722 3.5.2. Media Translator 724 In the context of multipoint communications a Media Translator is not 725 providing new mechanisms to establish a multipoint session. It is 726 much more an enabler or facilitator that ensures one or some sub-set 727 of session participants can participate in the session. 729 If B in Figure 9 were behind a limited network path, the Translator 730 may perform media transcoding to allow the traffic received from the 731 other participants to reach B without overloading the path. This 732 transcoding can help the other participants in the Multicast part of 733 the session, by not requiring the quality transmitted by A to be 734 lowered to the nitrates that B is actually capable of receiving. 736 3.6. Point to Multipoint Using the RFC 3550 Mixer Model 738 Shortcut name: Topo-Mixer 740 A Mixer is a middlebox that aggregates multiple RTP streams, which 741 are part of a session, by generating a new RTP stream and, in most 742 cases, by manipulation of the media data. One common application for 743 a Mixer is to allow a participant to receive a session with a reduced 744 amount of resources. 746 +-----+ 747 +---+ / \ +-----------+ +---+ 748 | A |<---/ \ | |<---->| B | 749 +---+ / Multi- \ | | +---+ 750 + Cast +->| Mixer | 751 +---+ \ Network / | | +---+ 752 | C |<---\ / | |<---->| D | 753 +---+ \ / +-----------+ +---+ 754 +-----+ 756 Figure 11: Point to Multipoint Using the RFC 3550 Mixer Model 758 A Mixer can be viewed as a device terminating the media streams 759 received from other session participants. Using the media data from 760 the received media streams, a Mixer generates a media stream that is 761 sent to the session participant. 763 The content that the Mixer provides is the mixed aggregate of what 764 the Mixer receives over the PtP or PtM paths, which are part of the 765 same conference session. 767 The Mixer is the content source, as it mixes the content (often in 768 the uncompressed domain) and then encodes it for transmission to a 769 participant. The CSRC Count (CC) and CSRC fields in the RTP header 770 can be used to indicate the contributors of to the newly generated 771 stream. The SSRCs of the to-be-mixed streams on the Mixer input 772 appear as the CSRCs at the Mixer output. That output stream uses a 773 unique SSRC that identifies the Mixer's stream. The CSRC should be 774 forwarded between the different conference participants to allow for 775 loop detection and identification of sources that are part of the 776 global session. Note that Section 7.1 of RFC 3550 requires the SSRC 777 space to be shared between domains for these reasons. This also 778 implies that any SDES information normally needs to be forwarded 779 across the mixer. 781 The Mixer is responsible for generating RTCP packets in accordance 782 with its role. It is a receiver and should therefore send receiver 783 reports for the media streams it receives. In its role as a media 784 sender, it should also generate sender reports for those media 785 streams it sends. As specified in Section 7.3 of RFC 3550, a Mixer 786 must not forward RTCP unaltered between the two domains. 788 The Mixer depicted in Figure 11 is involved in three domains that 789 need to be separated: the any source multicast network (including 790 participants A and C), participant B, and participant D. Assuming all 791 four participants in the conference are interested in receiving 792 content from each other participant, the Mixer produces different 793 mixed streams for B and D, as the one to B may contain content 794 received from D, and vice versa. However, the Mixer may only need 795 one SSRC per media type in each domain that is the receiving entity 796 and transmitter of mixed content. 798 In the multicast domain, a Mixer still needs to provide a mixed view 799 of the other domains. This makes the Mixer simpler to implement and 800 avoids any issues with advanced RTCP handling or loop detection, 801 which would be problematic if the Mixer were providing non-symmetric 802 behavior. Please see Section 3.11 for more discussion on this topic. 803 However, the mixing operation in each domain could potentially be 804 different. 806 A Mixer is responsible for receiving RTCP feedback messages and 807 handling them appropriately. The definition of "appropriate" depends 808 on the message itself and the context. In some cases, the reception 809 of a codec-control message by the Mixer may result in the generation 810 and transmission of RTCP feedback messages by the Mixer to the 811 participants in the other domain(s). In other cases, a message is 812 handled by the Mixer itself and therefore not forwarded to any other 813 domain. 815 When replacing the multicast network in Figure 11 (to the left of the 816 Mixer) with individual unicast paths as depicted in Figure 12, the 817 Mixer model is very similar to the one discussed in Section 3.9 818 below. Please see the discussion in Section 3.9 about the 819 differences between these two models. 821 +---+ +------------+ +---+ 822 | A |<---->| |<---->| B | 823 +---+ | | +---+ 824 | Mixer | 825 +---+ | | +---+ 826 | C |<---->| |<---->| D | 827 +---+ +------------+ +---+ 829 Figure 12: RTP Mixer with Only Unicast Paths 831 Lets now discuss in more detail different mixing operations that a 832 mixer can perform and how they can affect the RTP and RTCP. 834 3.6.1. Media Mixing 836 The media mixing mixer is likely the one that most think of when they 837 hear the term "mixer". Its basic pattern of operation is that it 838 receives media streams from (typically several) participants. Of 839 those, it selects (either through static configuration or by dynamic, 840 content dependent means such as voice activation) the stream(s) to be 841 included in a media domain mix. Then it creates a single outgoing 842 stream from this mix. 844 The most commonly deployed media mixer is probably the audio mixer, 845 used in voice conferencing, where the output consists of a mixture of 846 all the input streams; this needs minimal signalling to be 847 successfully set up. Audio mixing is relatively straightforward and 848 commonly possible for a reasonable number of participants. Lets 849 assume that you want to mix N streams from different participants. 850 The mixer needs to decode those N streams, typically into the sample 851 domain. Then it needs to produce N or N+1 mixes, the reasons that 852 different mixes are needed being that each contributing source get a 853 mix of all other sources except its own, as this would result in an 854 echo. When N is lower than the number of all participants one may 855 produce a Mix of all N streams for the group that are currently not 856 included in the mix, thus N+1 mixes. These audio streams are then 857 encoded again, RTP packetized and sent out. In many cases, audio 858 level normalization is also required before the actual mixing 859 process. 861 Video can't really be "mixed" and produce something particularly 862 useful for the users, however creating an composition out of the 863 contributed video streams is possible and known as "tiling". For 864 example the reconstructed, appropriately scaled down videos can be 865 spatially arranged in a set of tiles, each tile containing the video 866 from a participant. Tiles can be of different sizes, so that, for 867 example, a particularly important participant, or the loudest 868 speaker, is being shown on in larger tile than other participants. A 869 self-picture can be included in the tiling, which can either be 870 locally produced or be a feedback from a received and reconstructed 871 video image (allowing for confidence monitoring, the participant sees 872 himself/herself just as other participants see him/her). The tiling 873 normally operates on reconstructed video in the sample domain. The 874 tiled image is encoded, packetized, and sent by the mixer. It is 875 possible that a middlebox with media mixing duties contains only a 876 single mixer of the aforementioned type, in which case all 877 participants necessarily see the same tiled video, even if it is 878 being sent over different RTP streams. More common, however, are 879 mixing arrangement where an individual mixer is available for each 880 outgoing port of the middlebox, allowing individual compositions for 881 each participant. 883 One problem with media mixing is that it consumes both large amount 884 of media processing (for the actual mixing process in the 885 uncompressed domain) and encoding resources (for the encoding of the 886 mixed signal). Another problem is the quality degradation created by 887 decoding and re-encoding the media that is encapsulated in the RTP 888 media stream, which is the result of the lossy nature of most, if not 889 all, commonly used media codecs. A third problem is the latency 890 introduced by the media mixing, which can be substantial and 891 annoyingly noticeable in case of video. The advantage of media 892 mixing is that it is quite simplistic for the clients to handle the 893 single media stream (which includes the mixed aggregate of many 894 sources), as they don't need to handle multiple decodings, local 895 mixing and composition. In fact, mixers were introduced in pre-RTP 896 times so that legacy, single stream receiving endpoints can 897 successfully participate in what a user would recognize as a 898 multiparty session. 899 +-A---------+ +-MIXER----------------------+ 900 | +-RTP1----| |-RTP1------+ +-----+ | 901 | | +-Audio-| |-Audio---+ | +---+ | | | 902 | | | AA1|--------->|---------+-+-|DEC|->| | | 903 | | | |<---------|MA1 <----+ | +---+ | | | 904 | | | | |(BA1+CA1)|\| +---+ | | | 905 | | +-------| |---------+ +-|ENC|<-| B+C | | 906 | +---------| |-----------+ +---+ | | | 907 +-----------+ | | | | 908 | | M | | 909 +-B---------+ | | E | | 910 | +-RTP2----| |-RTP2------+ | D | | 911 | | +-Audio-| |-Audio---+ | +---+ | I | | 912 | | | BA1|--------->|---------+-+-|DEC|->| A | | 913 | | | |<---------|MA2 <----+ | +---+ | | | 914 | | +-------| |(BA1+CA1)|\| +---+ | | | 915 | +---------| |---------+ +-|ENC|<-| A+C | | 916 +-----------+ |-----------+ +---+ | | | 917 | | M | | 918 +-C---------+ | | I | | 919 | +-RTP3----| |-RTP3------+ | X | | 920 | | +-Audio-| |-Audio---+ | +---+ | E | | 921 | | | CA1|--------->|---------+-+-|DEC|->| R | | 922 | | | |<---------|MA3 <----+ | +---+ | | | 923 | | +-------| |(BA1+CA1)|\| +---+ | | | 924 | +---------| |---------+ +-|ENC|<-| A+B | | 925 +-----------+ |-----------+ +---+ +-----+ | 926 +----------------------------+ 928 Figure 13: Session and SSRC details for Media Mixer 930 From an RTP perspective media mixing can be very straightforward as 931 can be seen in Figure 13. The mixer presents one SSRC towards the 932 receiving client, e.g. MA1 to Peer A; the associated stream of which 933 is the media mix of the other participants. As, in this example, 934 each peer receives a different version produced by the mixer, there 935 is no actual relation between the different RTP sessions in the 936 actual media or the transport level information. There are, however, 937 common relationships between RTP1-RTP3 namely SSRC space and identity 938 information. When A receives the MA1 stream which is a combination 939 of BA1 and CA1 streams, the mixer may include CSRC information in the 940 MA1 stream to identify the contributing source BA1 and CA1, allowing 941 the receiver to identify the contributing sources even if this were 942 not possible through the media itself or other signaling means. 944 The CSRC has, in turn, utility in RTP extensions, like the Mixer to 945 Client audio levels RTP header extension [RFC6465]. If the SSRC from 946 endpoint to mixer leg are used as CSRC in another RTP session, then 947 RTP1, RTP2 and RTP3 become one joint session as they have a common 948 SSRC space. At this stage, the mixer also need to consider which 949 RTCP information it needs to expose in the different legs. In the 950 above scenario, commonly, a mixer would expose nothing more than the 951 Source Description (SDES) information and RTCP BYE for CSRC leaving 952 the session. The main goal would be to enable the correct binding 953 against the application logic and other information sources. This 954 also enables loop detection in the RTP session. 956 3.6.2. Media Switching 958 Media switching mixers are commonly used in such limited 959 functionality scenarios where no, or only very limited, concurrent 960 presentation of multiple sources is required by the application. An 961 RTP Mixer based on media switching avoids the media decoding and 962 encoding cycle in the mixer, as it conceptually forwards the encoded 963 media stream as it was being sent to the mixer, but not the 964 decryption and re-encryption cycle as it rewrites RTP headers. 965 Forwarding media (in contrast to reconstructing-mixing-encoding 966 media) reduces the amount of computational resources needed in the 967 mixer and increases the media quality (both in terms of fidelity and 968 reduced latency) per transmitted bit. 970 A media switching mixer maintains a pool of SSRCs representing 971 conceptual or functional streams the mixer can produce. These 972 streams are created by selecting media from one of RTP media streams 973 received by the mixer and forwarded to the peer using the mixer's own 974 SSRCs. The mixer can switch between available sources if that is 975 required by the concept for the source, like currently active 976 speaker. Note that the mixer, in most cases, still need to perform a 977 certain amount of media processing, as many media formats do not 978 allow to "tune" into the stream at arbitrary points of their 979 bitstream. 981 To achieve a coherent RTP media stream from the mixer's SSRC, the 982 mixer needs to rewrite the incoming RTP packet's header. First the 983 SSRC field must be set to the value of the Mixer's SSRC. Secondly, 984 the sequence number must be the next in the sequence of outgoing 985 packets it sent. Thirdly the RTP timestamp value needs to be 986 adjusted using an offset that changes each time one switch media 987 source. Finally depending on the negotiation the RTP payload type 988 value representing this particular RTP payload configuration may have 989 to be changed if the different endpoint mixer legs have not arrived 990 on the same numbering for a given configuration. This also requires 991 that the different end-points do support a common set of codecs, 992 otherwise media transcoding for codec compatibility is still 993 required. 995 Lets consider the operation of media switching mixer that supports a 996 video conference with six participants (A-F) where the two latest 997 speakers in the conference are shown to each participants. Thus the 998 mixer has two SSRCs sending video to each peer, and each peer is 999 capable of locally handling two video streams simultaneously. 1000 +-A---------+ +-MIXER----------------------+ 1001 | +-RTP1----| |-RTP1------+ +-----+ | 1002 | | +-Video-| |-Video---+ | | | | 1003 | | | AV1|------------>|---------+-+------->| S | | 1004 | | | |<------------|MV1 <----+-+-BV1----| W | | 1005 | | | |<------------|MV2 <----+-+-EV1----| I | | 1006 | | +-------| |---------+ | | T | | 1007 | +---------| |-----------+ | C | | 1008 +-----------+ | | H | | 1009 | | | | 1010 +-B---------+ | | M | | 1011 | +-RTP2----| |-RTP2------+ | A | | 1012 | | +-Video-| |-Video---+ | | T | | 1013 | | | BV1|------------>|---------+-+------->| R | | 1014 | | | |<------------|MV3 <----+-+-AV1----| I | | 1015 | | | |<------------|MV4 <----+-+-EV1----| X | | 1016 | | +-------| |---------+ | | | | 1017 | +---------| |-----------+ | | | 1018 +-----------+ | | | | 1019 : : : : 1020 : : : : 1021 +-F---------+ | | | | 1022 | +-RTP6----| |-RTP6------+ | | | 1023 | | +-Video-| |-Video---+ | | | | 1024 | | | CV1|------------>|---------+-+------->| | | 1025 | | | |<------------|MV11 <---+-+-AV1----| | | 1026 | | | |<------------|MV12 <---+-+-EV1----| | | 1027 | | +-------| |---------+ | | | | 1028 | +---------| |-----------+ +-----+ | 1029 +-----------+ +----------------------------+ 1031 Figure 14: Media Switching RTP Mixer 1033 The Media Switching RTP mixer can, similarly to the Media Mixing 1034 Mixer, reduce the bit-rate required for media transmission towards 1035 the different peers by selecting and forwarding only a sub-set of RTP 1036 media streams it receives from the conference participants. In many 1037 practical cases, the link capacities of either direction between 1038 peers and mixer are the same, which effectively limits the subset to 1039 a single media stream. 1041 To ensure that a media receiver can correctly decode the RTP media 1042 stream after a switch, a state saving (frame-based) codec needs to 1043 start its decoding from independent refresh points or similar points 1044 in the bitstream. For some codecs, for example frame based speech 1045 and audio codecs, this is easily achieved by starting the decoding at 1046 RTP packet boundaries (proper packetization on the encoder side 1047 assumed), as each packet boundary provides a refresh point. For 1048 other (mostly video-) codecs, refresh points are less common in the 1049 bitstream or may not be present at all without an explicit request to 1050 the respective encoder. For this purpose there exists the Full Intra 1051 Request [RFC5104] RTCP codec control message. 1053 Also in this type of mixer one could consider to terminate the RTP 1054 sessions fully between the different end-point and mixer legs. The 1055 same arguments and considerations as discussed in Section 3.9 need to 1056 be taken into consideration and apply here. 1058 3.7. Source Projecting Middlebox 1060 Another method for handling media in the RTP mixer is to project all 1061 potential RTP sources (SSRCs) into a per end-point independent RTP 1062 session. The middlebox can select which of the potential sources 1063 that are currently actively transmitting media, despite that the 1064 middlebox, in another RTP session, may receive media from that end- 1065 point. This is similar to the media switching Mixer but has some 1066 important differences in RTP details. 1068 +-A---------+ +-Middlebox-----------------+ 1069 | +-RTP1----| |-RTP1------+ +-----+ | 1070 | | +-Video-| |-Video---+ | | | | 1071 | | | AV1|------------>|---------+-+------>| | | 1072 | | | |<------------|BV1 <----+-+-------| S | | 1073 | | | |<------------|CV1 <----+-+-------| W | | 1074 | | | |<------------|DV1 <----+-+-------| I | | 1075 | | | |<------------|EV1 <----+-+-------| T | | 1076 | | | |<------------|FV1 <----+-+-------| C | | 1077 | | +-------| |---------+ | | H | | 1078 | +---------| |-----------+ | | | 1079 +-----------+ | | M | | 1080 | | A | | 1081 +-B---------+ | | T | | 1082 | +-RTP2----| |-RTP2------+ | R | | 1083 | | +-Video-| |-Video---+ | | I | | 1084 | | | BV1|------------>|---------+-+------>| X | | 1085 | | | |<------------|AV1 <----+-+-------| | | 1086 | | | |<------------|CV1 <----+-+-------| | | 1087 | | | | : : : |: : : : : : : : :| | | 1088 | | | |<------------|FV1 <----+-+-------| | | 1089 | | +-------| |---------+ | | | | 1090 | +---------| |-----------+ | | | 1091 +-----------+ | | | | 1092 : : : : 1093 : : : : 1094 +-F---------+ | | | | 1095 | +-RTP6----| |-RTP6------+ | | | 1096 | | +-Video-| |-Video---+ | | | | 1097 | | | FV1|------------>|---------+-+------>| | | 1098 | | | |<------------|AV1 <----+-+-------| | | 1099 | | | | : : : |: : : : : : : : :| | | 1100 | | | |<------------|EV1 <----+-+-------| | | 1101 | | +-------| |---------+ | | | | 1102 | +---------| |-----------+ +-----+ | 1103 +-----------+ +---------------------------+ 1105 Figure 15: Media Projecting Middlebox 1107 In the six participant conference depicted above in (Figure 15) one 1108 can see that end-point A is aware of five incoming SSRCs, BV1-FV1. 1109 If this middlebox intends to have a similar behaviour as in 1110 Section 3.6.2 where the mixer provides the end-points with the two 1111 latest speaking end-points, then only two out of these five SSRCs 1112 need concurrently transmit media to A. As the middlebox selects the 1113 source in the different RTP sessions that transmit media to the end- 1114 points, each RTP media stream requires some rewriting of RTP header 1115 fields when being projected from one session into another. In 1116 particular, the sequence number needs to be consecutively incremented 1117 based on the packet actually being transmitted in each RTP session. 1118 Therefore, the RTP sequence number offset will change each time a 1119 source is turned on in a RTP session. The timestamp (possibly 1120 offset) stays the same. 1122 As the RTP sessions are independent, the SSRC numbers used can also 1123 be handled independently, thereby bypassing the requirement for SSRC 1124 collision detection and avoidance. On the other hand, tools such as 1125 remapping tables between the RTP sessions are required. For example, 1126 the stream that is being sent by endpoint B to the middlebox (BV1) 1127 may use an SSRC value of 12345678. When that media stream is sent to 1128 endpoint F by the middlebox, it can use any SSRC value, e.g. 1129 87654321. As a result, each endpoint may have a different view of 1130 the application usage of a particular SSRC. Any RTP level identity 1131 information, such as SDES items also needs to update the SSRC 1132 referenced, if the included SDES items are intended to be global. 1133 Thus the application must not use SSRC as references to RTP media 1134 streams when communicating with other peers directly. This also 1135 affects loop detection which will fail to work, as there is no common 1136 namespace and identities across the different legs in the 1137 communication session on RTP level. Instead this responsibility 1138 falls onto higher layers. 1140 The middlebox is also responsible to receive any RTCP codec control 1141 requests coming from an end-point, and decide if it can act on the 1142 request locally or needs to translate the request into the RTP 1143 session that contains the media source. Both end-points and the 1144 middlebox need to implement conference related codec control 1145 functionalities to provide a good experience. Commonly used are Full 1146 Intra Request to request from the media source to provide switching 1147 points between the sources, and Temporary Maximum Media Bit-rate 1148 Request (TMMBR) to enable the middlebox to aggregate congestion 1149 control responses towards the media source so to enable it to adjust 1150 its bit-rate (obviously only in case the limitation is not in the 1151 source to middlebox link). 1153 This version of the middlebox also puts different requirements on the 1154 end-point when it comes to decoder instances and handling of the RTP 1155 media streams providing media. As each projected SSRC can, at any 1156 time, provide media, the end-point either needs to be able to handle 1157 as many decoder instances as the middlebox received, or have 1158 efficient switching of decoder contexts in a more limited set of 1159 actual decoder instances to cope with the switches. The application 1160 also gets more responsibility to update how the media provided is to 1161 be presented to the user. 1163 Note, this could potentially be seen as a media translator which 1164 include an on/off logic as part of its media translation. The main 1165 difference would be a common global SSRC space in the case of the 1166 Media Translator and the mapped one used in the above. It also has 1167 mixer aspects, as the streams it provides are not basically 1168 translated version, but instead they have conceptual property 1169 assigned to them. Thus this topology appears to be some hybrid 1170 between the translator and mixer model. 1172 3.8. Point to Multipoint Using Video Switching MCUs 1174 Shortcut name: Topo-Video-switch-MCU 1176 +---+ +------------+ +---+ 1177 | A |------| Multipoint |------| B | 1178 +---+ | Control | +---+ 1179 | Unit | 1180 +---+ | (MCU) | +---+ 1181 | C |------| |------| D | 1182 +---+ +------------+ +---+ 1184 Figure 16: Point to Multipoint Using a Video Switching MCU 1186 This PtM topology was common before, although the RTCP-terminating 1187 MCUs, as discussed in the next section, where perhaps even more 1188 common. This topology, as well as the following one, was a result of 1189 lack of wide availability of IP multicast technologies, as well as 1190 the simplicity of content switching when compared to content mixing. 1191 The technology is commonly implemented in what is known as "Video 1192 Switching MCUs". 1194 A video switching MCU forwards to a participant a single media 1195 stream, selected from the available streams. The criteria for 1196 selection are often based on voice activity in the audio-visual 1197 conference, but other conference management mechanisms (like 1198 presentation mode or explicit floor control) are known to exist as 1199 well. 1201 The video switching MCU may also perform media translation to modify 1202 the content in bit-rate, encoding, or resolution. However, it still 1203 may indicate the original sender of the content through the SSRC. In 1204 this case, the values of the CC and CSRC fields are retained. 1206 If not terminating RTP, the RTCP Sender Reports are forwarded for the 1207 currently selected sender. All RTCP Receiver Reports are freely 1208 forwarded between the participants. In addition, the MCU may also 1209 originate RTCP control traffic in order to control the session and/or 1210 report on status from its viewpoint. 1212 The video switching MCU has most of the attributes of a Translator. 1213 However, its stream selection is a mixing behavior. This behavior 1214 has some RTP and RTCP issues associated with it. The suppression of 1215 all but one media stream results in most participants seeing only a 1216 subset of the sent media streams at any given time, often a single 1217 stream per conference. Therefore, RTCP Receiver Reports only report 1218 on these streams. Consequently, the media senders that are not 1219 currently forwarded receive a view of the session that indicates 1220 their media streams disappear somewhere en route. This makes the use 1221 of RTCP for congestion control, or any type of quality reporting, 1222 very problematic. 1224 To avoid the aforementioned issues, the MCU needs to implement two 1225 features. First, it needs to act as a Mixer (see Section 3.6) and 1226 forward the selected media stream under its own SSRC and with the 1227 appropriate CSRC values. Second, the MCU needs to modify the RTCP 1228 RRs it forwards between the domains. As a result, it is recommended 1229 that one implement a centralized video switching conference using a 1230 Mixer according to RFC 3550, instead of the shortcut implementation 1231 described here. 1233 3.9. Point to Multipoint Using RTCP-Terminating MCU 1235 Shortcut name: Topo-RTCP-terminating-MCU 1237 +---+ +------------+ +---+ 1238 | A |<---->| Multipoint |<---->| B | 1239 +---+ | Control | +---+ 1240 | Unit | 1241 +---+ | (MCU) | +---+ 1242 | C |<---->| |<---->| D | 1243 +---+ +------------+ +---+ 1245 Figure 17: Point to Multipoint Using Content Modifying MCUs 1247 In this PtM scenario, each participant runs an RTP point-to-point 1248 session between itself and the MCU. This is a very commonly deployed 1249 topology in multipoint video conferencing. The content that the MCU 1250 provides to each participant is either: 1252 a. a selection of the content received from the other participants, 1253 or 1255 b. the mixed aggregate of what the MCU receives from the other PtP 1256 paths, which are part of the same conference session. 1258 In case a), the MCU may modify the content in bit-rate, encoding, or 1259 resolution. No explicit RTP mechanism is used to establish the 1260 relationship between the original media sender and the version the 1261 MCU sends. In other words, the outgoing sessions typically use a 1262 different SSRC, and may well use a different payload type (PT), even 1263 if this different PT happens to be mapped to the same media type. 1264 This is a result of the individually negotiated session for each 1265 participant. 1267 In case b), the MCU is the content source as it mixes the content and 1268 then encodes it for transmission to a participant. According to RTP 1269 [RFC3550], the SSRC of the contributors are to be signalled using the 1270 CSRC/CC mechanism. In practice, today, most deployed MCUs do not 1271 implement this feature. Instead, the identification of the 1272 participants whose content is included in the Mixer's output is not 1273 indicated through any explicit RTP mechanism. That is, most deployed 1274 MCUs set the CSRC Count (CC) field in the RTP header to zero, thereby 1275 indicating no available CSRC information, even if they could identify 1276 the content sources as suggested in RTP. 1278 The main feature that sets this topology apart from what RFC 3550 1279 describes is the breaking of the common RTP session across the 1280 centralized device, such as the MCU. This results in the loss of 1281 explicit RTP-level indication of all participants. If one were using 1282 the mechanisms available in RTP and RTCP to signal this explicitly, 1283 the topology would follow the approach of an RTP Mixer. The lack of 1284 explicit indication has at least the following potential problems: 1286 1. Loop detection cannot be performed on the RTP level. When 1287 carelessly connecting two misconfigured MCUs, a loop could be 1288 generated. 1290 2. There is no information about active media senders available in 1291 the RTP packet. As this information is missing, receivers cannot 1292 use it. It also deprives the client of information related to 1293 currently active senders in a machine-usable way, thus preventing 1294 clients from indicating currently active speakers in user 1295 interfaces, etc. 1297 Note that deployed MCUs (and endpoints) rely on signalling layer 1298 mechanisms for the identification of the contributing sources, for 1299 example, a SIP conferencing package [RFC4575]. This alleviates, to 1300 some extent, the aforementioned issues resulting from ignoring RTP's 1301 CSRC mechanism. 1303 As a result of the shortcomings of this topology, it is recommended 1304 to instead implement the Mixer concept as specified by RFC 3550. 1306 3.10. De-composite Endpoint 1308 The implementation of an application may desire to send a subset of 1309 the application's data to each of multiple devices, each with its own 1310 network address. A very basic use case for this would be to separate 1311 audio and video processing for a particular endpoint, like a 1312 conference room, into one device handling the audio and another 1313 handling the video, being interconnected by some control functions 1314 allowing them to behave as a single endpoint in all aspects except 1315 for transport Figure 18. 1317 Which decomposition scheme is possible is highly dependent on the RTP 1318 session usage. It is not really feasible to decompose one logical 1319 end-point into two different transport nodes in one RTP session. A 1320 third party monitor would report such an attempt as two entities 1321 being two different end-points with a CNAME collision. As a result, 1322 a fully RTP conformant de-composited endpoint is one where the 1323 different decomposed parts use separate RTP sessions to send and/or 1324 receive media streams intended for them. 1326 +---------------------+ 1327 | Endpoint A | 1328 | Local Area Network | 1329 | +------------+ | 1330 | +->| Audio |<+-RTP---\ 1331 | | +------------+ | \ +------+ 1332 | | +------------+ | +-->| | 1333 | +->| Video |<+-RTP-------->| B | 1334 | | +------------+ | +-->| | 1335 | | +------------+ | / +------+ 1336 | +->| Control |<+-SIP---/ 1337 | +------------+ | 1338 +---------------------+ 1340 Figure 18: De-composite End-Point 1342 In the above usage, let us assume that the different RTP sessions are 1343 used for audio and video. The audio and video parts, however, use a 1344 common CNAME and also have a common clock to ensure that 1345 synchronization and clock drift handling works, despite the 1346 decomposition. Also, the RTCP handling works correctly as long as 1347 only one part of the de-composite is part of each RTP session. That 1348 way any differences in the path between A's audio entity and B and 1349 A's video and B are related to different SSRCs in different RTP 1350 sessions. 1352 The requirement that can be derived from the above usage is that the 1353 transport flows for each RTP session might be under common control, 1354 but still are addressed to what looks like different endpoints (based 1355 on addresses and ports). This geometry cannot be accomplished using 1356 one RTP session, so in this case, multiple RTP sessions are needed. 1358 3.11. Non-Symmetric Mixer/Translators 1360 Shortcut name: Topo-Asymmetric 1362 It is theoretically possible to construct an MCU that is a Mixer in 1363 one direction and a Translator in another. The main reason to 1364 consider this would be to allow topologies similar to Figure 11, 1365 where the Mixer does not need to mix in the direction from B or D 1366 towards the multicast domains with A and C. Instead, the media 1367 streams from B and D are forwarded without changes. Avoiding this 1368 mixing would save media processing resources that perform the mixing 1369 in cases where it isn't needed. However, there would still be a need 1370 to mix B's stream towards D. Only in the direction B -> multicast 1371 domain or D -> multicast domain would it be possible to work as a 1372 Translator. In all other directions, it would function as a Mixer. 1374 The Mixer/Translator would still need to process and change the RTCP 1375 before forwarding it in the directions of B or D to the multicast 1376 domain. One issue is that A and C do not know about the mixed-media 1377 stream the Mixer sends to either B or D. Therefore, any reports 1378 related to these streams must be removed. Also, receiver reports 1379 related to A and C's media stream would be missing. To avoid A and C 1380 thinking that B and D aren't receiving A and C at all, the Mixer 1381 needs to insert locally generated reports reflecting the situation 1382 for the streams from A and C into B and D's Sender Reports. In the 1383 opposite direction, the Receiver Reports from A and C about B's and 1384 D's stream also need to be aggregated into the Mixer's Receiver 1385 Reports sent to B and D. Since B and D only have the Mixer as source 1386 for the stream, all RTCP from A and C must be suppressed by the 1387 Mixer. 1389 This topology is so problematic and it is so easy to get the RTCP 1390 processing wrong, that it is not recommended to implement this 1391 topology. 1393 3.12. Combining Topologies 1395 Topologies can be combined and linked to each other using Mixers or 1396 Translators. However, care must be taken in handling the SSRC/CSRC 1397 space. A Mixer does not forward RTCP from sources in other domains, 1398 but instead generates its own RTCP packets for each domain it mixes 1399 into, including the necessary Source Description (SDES) information 1400 for both the CSRCs and the SSRCs. Thus, in a mixed domain, the only 1401 SSRCs seen will be the ones present in the domain, while there can be 1402 CSRCs from all the domains connected together with a combination of 1403 Mixers and Translators. The combined SSRC and CSRC space is common 1404 over any Translator or Mixer. This is important to facilitate loop 1405 detection, something that is likely to be even more important in 1406 combined topologies due to the mixed behavior between the domains. 1407 Any hybrid, like the Topo-Video-switch-MCU or Topo-Asymmetric, 1408 requires considerable thought on how RTCP is dealt with. 1410 4. Comparing Topologies 1412 The topologies discussed in Section 3 have different properties. 1413 This section first lists these properties and maps the different 1414 topologies to them. Please note that even if a certain property is 1415 supported within a particular topology concept, the necessary 1416 functionality may, in many cases, be optional to implement. 1418 Note: This section has not yet been updated with the new additions of 1419 topologies. 1421 4.1. Topology Properties 1423 4.1.1. All to All Media Transmission 1425 Multicast, at least Any Source Multicast (ASM), provides the 1426 functionality that everyone may send to, or receive from, everyone 1427 else within the session. MCUs, Mixers, and Translators may all 1428 provide that functionality at least on some basic level. However, 1429 there are some differences in which type of reachability they 1430 provide. 1432 The transport Translator function called "relay", in Section 3.5, is 1433 the one that provides the emulation of ASM that is closest to true 1434 IP-multicast-based, all to all transmission. Media Translators, 1435 Mixers, and the MCU variants do not provide a fully meshed forwarding 1436 on the transport level; instead, they only allow limited forwarding 1437 of content from the other session participants. 1439 The "all to all media transmission" requires that any media 1440 transmitting entity considers the path to the least capable receiver. 1441 Otherwise, the media transmissions may overload that path. 1442 Therefore, a media sender needs to monitor the path from itself to 1443 any of the participants, to detect the currently least capable 1444 receiver, and adapt its sending rate accordingly. As multiple 1445 participants may send simultaneously, the available resources may 1446 vary. RTCP's Receiver Reports help performing this monitoring, at 1447 least on a medium time scale. 1449 The transmission of RTCP automatically adapts to any changes in the 1450 number of participants due to the transmission algorithm, defined in 1451 the RTP specification [RFC3550], and the extensions in AVPF [RFC4585] 1452 (when applicable). That way, the resources utilized for RTCP stay 1453 within the bounds configured for the session. 1455 4.1.2. Transport or Media Interoperability 1457 Translators, Mixers, and RTCP-terminating MCU all allow changing the 1458 media encoding or the transport to other properties of the other 1459 domain, thereby providing extended interoperability in cases where 1460 the participants lack a common set of media codecs and/or transport 1461 protocols. 1463 4.1.3. Per Domain Bit-Rate Adaptation 1465 Participants are most likely to be connected to each other with a 1466 heterogeneous set of paths. This makes congestion control in a Point 1467 to Multipoint set problematic. For the ASM and "relay" scenario, 1468 each individual sender has to adapt to the receiver with the least 1469 capable path. This is no longer necessary when Media Translators, 1470 Mixers, or MCUs are involved, as each participant only needs to adapt 1471 to the slowest path within its own domain. The Translator, Mixer, or 1472 MCU topologies all require their respective outgoing streams to 1473 adjust the bit-rate, packet-rate, etc., to adapt to the least capable 1474 path in each of the other domains. That way one can avoid lowering 1475 the quality to the least-capable participant in all the domains at 1476 the cost (complexity, delay, equipment) of the Mixer or Translator. 1478 4.1.4. Aggregation of Media 1480 In the all to all media property mentioned above and provided by ASM, 1481 all simultaneous media transmissions share the available bit-rate. 1482 For participants with limited reception capabilities, this may result 1483 in a situation where even a minimal acceptable media quality cannot 1484 be accomplished. This is the result of multiple media streams 1485 needing to share the available resources. The solution to this 1486 problem is to provide for a Mixer or MCU to aggregate the multiple 1487 streams into a single one. This aggregation can be performed 1488 according to different methods. Mixing or selection are two common 1489 methods. 1491 4.1.5. View of All Session Participants 1493 The RTP protocol includes functionality to identify the session 1494 participants through the use of the SSRC and CSRC fields. In 1495 addition, it is capable of carrying some further identity information 1496 about these participants using the RTCP Source Descriptors (SDES). 1498 To maintain this functionality, it is necessary that RTCP is handled 1499 correctly in domain bridging function. This is specified for 1500 Translators and Mixers. The MCU described in Section 3.8 does not 1501 entirely fulfill this. The one described in Section 3.9 does not 1502 support this at all. 1504 4.1.6. Loop Detection 1506 In complex topologies with multiple interconnected domains, it is 1507 possible to form media loops. RTP and RTCP support detecting such 1508 loops, as long as the SSRC and CSRC identities are correctly set in 1509 forwarded packets. It is likely that loop detection works for the 1510 MCU, described in Section 3.8, at least as long as it forwards the 1511 RTCP between the participants. However, the MCU in Section 3.9 will 1512 definitely break the loop detection mechanism. 1514 4.2. Comparison of Topologies 1516 The table below attempts to summarize the properties of the different 1517 topologies. The legend to the topology abbreviations are: Topo- 1518 Point-to-Point (PtP), Topo-Multicast (Multic), Topo-Trns-Translator 1519 (TTrn), Topo-Media-Translator (including Transport Translator) 1520 (MTrn), Topo-Mixer (Mixer), Topo-Asymmetric (ASY), Topo-Video-switch- 1521 MCU (MCUs), and Topo-RTCP-terminating-MCU (MCUt). In the table 1522 below, Y indicates Yes or full support, N indicates No support, (Y) 1523 indicates partial support, and N/A indicates not applicable. 1525 Property PtP Multic TTrn MTrn Mixer ASY MCUs MCUt 1526 ------------------------------------------------------------------ 1527 All to All media N Y Y Y (Y) (Y) (Y) (Y) 1528 Interoperability N/A N Y Y Y Y N Y 1529 Per Domain Adaptation N/A N N Y Y Y N Y 1530 Aggregation of media N N N N Y (Y) Y Y 1531 Full Session View Y Y Y Y Y Y (Y) N 1532 Loop Detection Y Y Y Y Y Y (Y) N 1534 Please note that the Media Translator also includes the transport 1535 Translator functionality. 1537 5. Security Considerations 1539 The use of Mixers and Translators has impact on security and the 1540 security functions used. The primary issue is that both Mixers and 1541 Translators modify packets, thus preventing the use of integrity and 1542 source authentication, unless they are trusted devices that take part 1543 in the security context, e.g., the device can send Secure Realtime 1544 Transport Protocol (SRTP) and Secure Realtime Transport Control 1545 Protocol (SRTCP) [RFC3711] packets to session endpoints. If 1546 encryption is employed, the media Translator and Mixer need to be 1547 able to decrypt the media to perform its function. A transport 1548 Translator may be used without access to the encrypted payload in 1549 cases where it translates parts that are not included in the 1550 encryption and integrity protection, for example, IP address and UDP 1551 port numbers in a media stream using SRTP [RFC3711]. However, in 1552 general, the Translator or Mixer needs to be part of the signalling 1553 context and get the necessary security associations (e.g., SRTP 1554 crypto contexts) established with its RTP session participants. 1556 Including the Mixer and Translator in the security context allows the 1557 entity, if subverted or misbehaving, to perform a number of very 1558 serious attacks as it has full access. It can perform all the 1559 attacks possible (see RFC 3550 and any applicable profiles) as if the 1560 media session were not protected at all, while giving the impression 1561 to the session participants that they are protected. 1563 Transport Translators have no interactions with cryptography that 1564 works above the transport layer, such as SRTP, since that sort of 1565 Translator leaves the RTP header and payload unaltered. Media 1566 Translators, on the other hand, have strong interactions with 1567 cryptography, since they alter the RTP payload. A media Translator 1568 in a session that uses cryptographic protection needs to perform 1569 cryptographic processing to both inbound and outbound packets. 1571 A media Translator may need to use different cryptographic keys for 1572 the inbound and outbound processing. For SRTP, different keys are 1573 required, because an RFC 3550 media Translator leaves the SSRC 1574 unchanged during its packet processing, and SRTP key sharing is only 1575 allowed when distinct SSRCs can be used to protect distinct packet 1576 streams. 1578 When the media Translator uses different keys to process inbound and 1579 outbound packets, each session participant needs to be provided with 1580 the appropriate key, depending on whether they are listening to the 1581 Translator or the original source. (Note that there is an 1582 architectural difference between RTP media translation, in which 1583 participants can rely on the RTP Payload Type field of a packet to 1584 determine appropriate processing, and cryptographically protected 1585 media translation, in which participants must use information that is 1586 not carried in the packet.) 1588 When using security mechanisms with Translators and Mixers, it is 1589 possible that the Translator or Mixer could create different security 1590 associations for the different domains they are working in. Doing so 1591 has some implications: 1593 First, it might weaken security if the Mixer/Translator accepts a 1594 weaker algorithm or key in one domain than in another. Therefore, 1595 care should be taken that appropriately strong security parameters 1596 are negotiated in all domains. In many cases, "appropriate" 1597 translates to "similar" strength. If a key management system does 1598 allow the negotiation of security parameters resulting in a different 1599 strength of the security, then this system should notify the 1600 participants in the other domains about this. 1602 Second, the number of crypto contexts (keys and security related 1603 state) needed (for example, in SRTP [RFC3711]) may vary between 1604 Mixers and Translators. A Mixer normally needs to represent only a 1605 single SSRC per domain and therefore needs to create only one 1606 security association (SRTP crypto context) per domain. In contrast, 1607 a Translator needs one security association per participant it 1608 translates towards, in the opposite domain. Considering Figure 9, 1609 the Translator needs two security associations towards the multicast 1610 domain, one for B and one for D. It may be forced to maintain a set 1611 of totally independent security associations between itself and B and 1612 D respectively, so as to avoid two-time pad occurrences. These 1613 contexts must also be capable of handling all the sources present in 1614 the other domains. Hence, using completely independent security 1615 associations (for certain keying mechanisms) may force a Translator 1616 to handle N*DM keys and related state; where N is the total number of 1617 SSRCs used over all domains and DM is the total number of domains. 1619 There exist a number of different mechanisms to provide keys to the 1620 different participants. One example is the choice between group keys 1621 and unique keys per SSRC. The appropriate keying model is impacted 1622 by the topologies one intends to use. The final security properties 1623 are dependent on both the topologies in use and the keying 1624 mechanisms' properties, and need to be considered by the application. 1625 Exactly which mechanisms are used is outside of the scope of this 1626 document. Please review RTP Security Options 1627 [I-D.ietf-avtcore-rtp-security-options] to get a better understanding 1628 of most of the available options. 1630 6. IANA Considerations 1632 This document makes no request of IANA. 1634 Note to RFC Editor: this section may be removed on publication as an 1635 RFC. 1637 7. Acknowledgements 1639 The authors would like to thank Bo Burman, Umesh Chandra, Roni Even, 1640 Keith Lantz, Ladan Gharai, Geoff Hunt, and Mark Baugher for their 1641 help in reviewing this document. 1643 8. References 1645 8.1. Normative References 1647 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 1648 Jacobson, "RTP: A Transport Protocol for Real-Time 1649 Applications", STD 64, RFC 3550, July 2003. 1651 [RFC3711] Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K. 1652 Norrman, "The Secure Real-time Transport Protocol (SRTP)", 1653 RFC 3711, March 2004. 1655 [RFC4575] Rosenberg, J., Schulzrinne, H., and O. Levin, "A Session 1656 Initiation Protocol (SIP) Event Package for Conference 1657 State", RFC 4575, August 2006. 1659 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 1660 "Extended RTP Profile for Real-time Transport Control 1661 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, 1662 July 2006. 1664 8.2. Informative References 1666 [I-D.ietf-avtcore-rtp-security-options] 1667 Westerlund, M. and C. Perkins, "Options for Securing RTP 1668 Sessions", draft-ietf-avtcore-rtp-security-options-01 1669 (work in progress), October 2012. 1671 [I-D.lennox-avtcore-rtp-multi-stream] 1672 Lennox, J. and M. Westerlund, "Real-Time Transport 1673 Protocol (RTP) Considerations for Endpoints Sending 1674 Multiple Media Streams", 1675 draft-lennox-avtcore-rtp-multi-stream-01 (work in 1676 progress), October 2012. 1678 [RFC3022] Srisuresh, P. and K. Egevang, "Traditional IP Network 1679 Address Translator (Traditional NAT)", RFC 3022, 1680 January 2001. 1682 [RFC4607] Holbrook, H. and B. Cain, "Source-Specific Multicast for 1683 IP", RFC 4607, August 2006. 1685 [RFC5104] Wenger, S., Chandra, U., Westerlund, M., and B. Burman, 1686 "Codec Control Messages in the RTP Audio-Visual Profile 1687 with Feedback (AVPF)", RFC 5104, February 2008. 1689 [RFC5760] Ott, J., Chesterfield, J., and E. Schooler, "RTP Control 1690 Protocol (RTCP) Extensions for Single-Source Multicast 1691 Sessions with Unicast Feedback", RFC 5760, February 2010. 1693 [RFC5766] Mahy, R., Matthews, P., and J. Rosenberg, "Traversal Using 1694 Relays around NAT (TURN): Relay Extensions to Session 1695 Traversal Utilities for NAT (STUN)", RFC 5766, April 2010. 1697 [RFC6285] Ver Steeg, B., Begen, A., Van Caenegem, T., and Z. Vax, 1698 "Unicast-Based Rapid Acquisition of Multicast RTP 1699 Sessions", RFC 6285, June 2011. 1701 [RFC6465] Ivov, E., Marocco, E., and J. Lennox, "A Real-time 1702 Transport Protocol (RTP) Header Extension for Mixer-to- 1703 Client Audio Level Indication", RFC 6465, December 2011. 1705 Authors' Addresses 1707 Magnus Westerlund 1708 Ericsson 1709 Farogatan 6 1710 SE-164 80 Kista 1711 Sweden 1713 Phone: +46 10 714 82 87 1714 Email: magnus.westerlund@ericsson.com 1716 Stephan Wenger 1717 Vidyo 1718 433 Hackensack Ave 1719 Hackensack, NJ 07601 1720 USA 1722 Email: stewe@stewe.org