idnits 2.17.1 draft-ietf-avtcore-ecn-for-rtp-04.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 : ---------------------------------------------------------------------------- == There are 3 instances of lines with private range IPv4 addresses in the document. If these are generic example addresses, they should be changed to use any of the ranges defined in RFC 6890 (or successor): 192.0.2.x, 198.51.100.x or 203.0.113.x. -- The document has examples using IPv4 documentation addresses according to RFC6890, but does not use any IPv6 documentation addresses. Maybe there should be IPv6 examples, too? Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (July 11, 2011) is 4672 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) == Missing Reference: 'TBA1' is mentioned on line 682, but not defined == Missing Reference: 'TBA2' is mentioned on line 842, but not defined ** Downref: Normative reference to an Experimental RFC: RFC 2762 ** Obsolete normative reference: RFC 5245 (Obsoleted by RFC 8445, RFC 8839) ** Obsolete normative reference: RFC 5389 (Obsoleted by RFC 8489) -- Obsolete informational reference (is this intentional?): RFC 4347 (Obsoleted by RFC 6347) -- Obsolete informational reference (is this intentional?): RFC 4566 (Obsoleted by RFC 8866) -- Obsolete informational reference (is this intentional?): RFC 4960 (Obsoleted by RFC 9260) -- Obsolete informational reference (is this intentional?): RFC 5751 (Obsoleted by RFC 8551) Summary: 3 errors (**), 0 flaws (~~), 4 warnings (==), 6 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group M. Westerlund 3 Internet-Draft I. Johansson 4 Intended status: Standards Track Ericsson 5 Expires: January 12, 2012 C. Perkins 6 University of Glasgow 7 P. O'Hanlon 8 UCL 9 K. Carlberg 10 G11 11 July 11, 2011 13 Explicit Congestion Notification (ECN) for RTP over UDP 14 draft-ietf-avtcore-ecn-for-rtp-04 16 Abstract 18 This memo specifies how Explicit Congestion Notification (ECN) can be 19 used with the Real-time Transport Protocol (RTP) running over UDP, 20 using RTP Control Protocol (RTCP) as a feedback mechanism. It 21 defines a new RTCP Extended Report (XR) block for periodic ECN 22 feedback, a new RTCP transport feedback message for timely reporting 23 of congestion events, and a Session Traversal Utilities for NAT 24 (STUN) extension used in the optional initialization method using 25 Interactive Connectivity Establishment (ICE). Signalling and 26 procedures for negotiation of capabilities and initialization methods 27 are also defined. 29 Status of this Memo 31 This Internet-Draft is submitted in full conformance with the 32 provisions of BCP 78 and BCP 79. 34 Internet-Drafts are working documents of the Internet Engineering 35 Task Force (IETF). Note that other groups may also distribute 36 working documents as Internet-Drafts. The list of current Internet- 37 Drafts is at http://datatracker.ietf.org/drafts/current/. 39 Internet-Drafts are draft documents valid for a maximum of six months 40 and may be updated, replaced, or obsoleted by other documents at any 41 time. It is inappropriate to use Internet-Drafts as reference 42 material or to cite them other than as "work in progress." 44 This Internet-Draft will expire on January 12, 2012. 46 Copyright Notice 48 Copyright (c) 2011 IETF Trust and the persons identified as the 49 document authors. All rights reserved. 51 This document is subject to BCP 78 and the IETF Trust's Legal 52 Provisions Relating to IETF Documents 53 (http://trustee.ietf.org/license-info) in effect on the date of 54 publication of this document. Please review these documents 55 carefully, as they describe your rights and restrictions with respect 56 to this document. Code Components extracted from this document must 57 include Simplified BSD License text as described in Section 4.e of 58 the Trust Legal Provisions and are provided without warranty as 59 described in the Simplified BSD License. 61 Table of Contents 63 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 64 2. Conventions, Definitions and Acronyms . . . . . . . . . . . . 5 65 3. Discussion, Requirements, and Design Rationale . . . . . . . . 6 66 3.1. Requirements . . . . . . . . . . . . . . . . . . . . . . . 7 67 3.2. Applicability . . . . . . . . . . . . . . . . . . . . . . 8 68 3.3. Interoperability . . . . . . . . . . . . . . . . . . . . . 12 69 4. Overview of Use of ECN with RTP/UDP/IP . . . . . . . . . . . . 12 70 5. RTCP Extensions for ECN feedback . . . . . . . . . . . . . . . 15 71 5.1. RTP/AVPF Transport Layer ECN Feedback packet . . . . . . . 15 72 5.2. RTCP XR Report block for ECN summary information . . . . . 18 73 6. SDP Signalling Extensions for ECN . . . . . . . . . . . . . . 20 74 6.1. Signalling ECN Capability using SDP . . . . . . . . . . . 20 75 6.2. RTCP ECN Feedback SDP Parameter . . . . . . . . . . . . . 25 76 6.3. XR Block ECN SDP Parameter . . . . . . . . . . . . . . . . 25 77 6.4. ICE Parameter to Signal ECN Capability . . . . . . . . . . 25 78 7. Use of ECN with RTP/UDP/IP . . . . . . . . . . . . . . . . . . 26 79 7.1. Negotiation of ECN Capability . . . . . . . . . . . . . . 26 80 7.2. Initiation of ECN Use in an RTP Session . . . . . . . . . 26 81 7.3. Ongoing Use of ECN Within an RTP Session . . . . . . . . . 33 82 7.4. Detecting Failures . . . . . . . . . . . . . . . . . . . . 36 83 8. Processing ECN in RTP Translators and Mixers . . . . . . . . . 40 84 8.1. Transport Translators . . . . . . . . . . . . . . . . . . 40 85 8.2. Fragmentation and Reassembly in Translators . . . . . . . 40 86 8.3. Generating RTCP ECN Feedback in Media Transcoders . . . . 42 87 8.4. Generating RTCP ECN Feedback in Mixers . . . . . . . . . . 43 88 9. Implementation considerations . . . . . . . . . . . . . . . . 44 89 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 44 90 10.1. SDP Attribute Registration . . . . . . . . . . . . . . . . 44 91 10.2. RTP/AVPF Transport Layer Feedback Message . . . . . . . . 45 92 10.3. RTCP Feedback SDP Parameter . . . . . . . . . . . . . . . 45 93 10.4. RTCP XR Report blocks . . . . . . . . . . . . . . . . . . 45 94 10.5. RTCP XR SDP Parameter . . . . . . . . . . . . . . . . . . 45 95 10.6. STUN attribute . . . . . . . . . . . . . . . . . . . . . . 45 96 10.7. ICE Option . . . . . . . . . . . . . . . . . . . . . . . . 46 97 11. Security Considerations . . . . . . . . . . . . . . . . . . . 46 98 12. Examples of SDP Signalling . . . . . . . . . . . . . . . . . . 48 99 12.1. Basic SDP Offer/Answer . . . . . . . . . . . . . . . . . . 48 100 12.2. Declarative Multicast SDP . . . . . . . . . . . . . . . . 50 101 13. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 51 102 14. References . . . . . . . . . . . . . . . . . . . . . . . . . . 51 103 14.1. Normative References . . . . . . . . . . . . . . . . . . . 51 104 14.2. Informative References . . . . . . . . . . . . . . . . . . 52 105 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 54 107 1. Introduction 109 This memo outlines how Explicit Congestion Notification (ECN) 110 [RFC3168] can be used for Real-time Transport Protocol (RTP) 111 [RFC3550] flows running over UDP/IP which use RTP Control Protocol 112 (RTCP) as a feedback mechanism. The solution consists of feedback of 113 ECN congestion experienced markings to the sender using RTCP, 114 verification of ECN functionality end-to-end, and procedures for how 115 to initiate ECN usage. Since the initiation process has some 116 dependencies on the signalling mechanism used to establish the RTP 117 session, a specification for signalling mechanisms using Session 118 Description Protocol (SDP) [RFC4566] is included. 120 ECN is getting attention as a method to minimise the impact of 121 congestion on real-time multimedia traffic. The use of ECN provides 122 a way for the network to send a congestion control signal to a media 123 transport without having to impair the media. Unlike packet loss, 124 ECN signals unambiguously indicate congestion to the transport as 125 quickly as feedback delays allow, and without confusing congestion 126 with losses that might have occurred for other reasons such as 127 transmission errors, packet-size errors, routing errors, badly 128 implemented middleboxes, policy violations and so forth. 130 The introduction of ECN into the Internet requires changes to both 131 the network and transport layers. At the network layer, IP 132 forwarding has to be updated to allow routers to mark packets, rather 133 than discarding them in times of congestion [RFC3168]. In addition, 134 transport protocols have to be modified to inform the sender that ECN 135 marked packets are being received, so it can respond to the 136 congestion. The Transmission Control Protocol (TCP) [RFC3168], 137 Stream Control Transmission Protocol (SCTP) [RFC4960] and Datagram 138 Congestion Control Protocol (DCCP) [RFC4340] have been updated to 139 support ECN, but to date there is no specification how UDP-based 140 transports, such as RTP [RFC3550], can use ECN. This is due to the 141 lack of feedback mechanisms directly in UDP. Instead the signaling 142 control protocol on top of UDP needs to provide that feedback. For 143 RTP that feedback is provided by RTCP. 145 The remainder of this memo is structured as follows. We start by 146 describing the conventions, definitions and acronyms used in this 147 memo in Section 2, and the design rationale and applicability in 148 Section 3. Section 4 gives an overview of how ECN is used with RTP 149 over UDP. RTCP extensions for ECN feedback are defined in Section 5, 150 and SDP signalling extensions in Section 6. The details of how ECN 151 is used with RTP over UDP are defined in Section 7. In Section 8 we 152 describe how ECN is handled in RTP translators and mixers. Section 9 153 discusses some implementation considerations, Section 10 lists IANA 154 considerations, and Section 11 discusses security considerations. 156 2. Conventions, Definitions and Acronyms 158 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 159 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 160 "OPTIONAL" in this document are to be interpreted as described in RFC 161 2119 [RFC2119]. 163 Abbreviations and Definitions: 165 Sender: A sender of RTP packets carrying an encoded media stream. 166 The sender can change how the media transmission is performed by 167 varying the media coding or packetisation. It is one end-point of 168 the ECN control loop. 170 Receiver: A receiver of RTP packets with the intention to consume 171 the media stream. It sends RTCP feedback on the received stream. 172 It is the other end-point of the ECN control loop. 174 ECN Capable Host: A sender or receiver of a media stream that is 175 capable of setting and/or processing ECN marks. 177 ECN Capable Transport (ECT): A transport flow where both sender and 178 receiver are ECN capable hosts. Packets sent by an ECN Capable 179 Transport will be marked as ECT(0) or ECT(1) on transmission. See 180 [RFC3168] for the definition of the ECT(0) and ECT(1) marks. 182 ECN-CE: ECN Congestion Experienced mark (see [RFC3168]). 184 ECN Capable Packets: Packets with ECN mark set to either ECT(0), 185 ECT(1) or ECN-CE. 187 Not-ECT packets: Packets that are not sent by an ECN capable 188 transport, and are not ECN-CE marked. 190 ECN Oblivious Relay: A router or middlebox that treats ECN Capable 191 Packets no differently from Not-ECT packets. 193 ECN Capable Queue: A queue that supports ECN-CE marking of ECN- 194 Capable Packets to indicate congestion. 196 ECN Blocking Middlebox: A middlebox that discards ECN-Capable 197 Packets. 199 ECN Reverting Middlebox: A middlebox that changes ECN-Capable 200 Packets to Not-ECT packets by removing the ECN mark. 202 Note that RTP mixers or translators that operate in such a manner 203 that they terminate or split the ECN control loop will take on the 204 role of receivers or senders. This is further discussed in 205 Section 3.2. 207 3. Discussion, Requirements, and Design Rationale 209 ECN has been specified for use with TCP [RFC3168], SCTP [RFC4960], 210 and DCCP [RFC4340] transports. These are all unicast protocols which 211 negotiate the use of ECN during the initial connection establishment 212 handshake (supporting incremental deployment, and checking if ECN 213 marked packets pass all middleboxes on the path). ECN-CE marks are 214 immediately echoed back to the sender by the receiving end-point 215 using an additional bit in feedback messages, and the sender then 216 interprets the mark as equivalent to a packet loss for congestion 217 control purposes. 219 If RTP is run over TCP, SCTP, or DCCP, it can use the native ECN 220 support provided by those protocols. This memo does not concern 221 itself further with these use cases. However, RTP is more commonly 222 run over UDP. This combination does not currently support ECN, and 223 we observe that it has significant differences from the other 224 transport protocols for which ECN has been specified. These include: 226 Signalling: RTP relies on separate signalling protocols to negotiate 227 parameters before a session can be created, and doesn't include an 228 in-band handshake or negotiation at session set-up time (i.e., 229 there is no equivalent to the TCP three-way handshake in RTP). 231 Feedback: RTP does not explicitly acknowledge receipt of datagrams. 232 Instead, the RTP Control Protocol (RTCP) provides reception 233 quality feedback, and other back channel communication, for RTP 234 sessions. The feedback interval is generally on the order of 235 seconds, rather than once per network RTT (although the RTP/AVPF 236 profile [RFC4585] allows more rapid feedback in most cases). RTCP 237 is also very much oriented around counting packets, which makes 238 byte counting congestion algorithms difficult to utilize. 240 Congestion Response: While it is possible to adapt the transmission 241 of many audio/visual streams in response to network congestion, 242 and such adaptation is required by [RFC3550], the dynamics of the 243 congestion response may be quite different to those of TCP or 244 other transport protocols. 246 Middleboxes: The RTP framework explicitly supports the concept of 247 mixers and translators, which are middleboxes that are involved in 248 media transport functions. 250 Multicast: RTP is explicitly a group communication protocol, and was 251 designed from the start to support IP multicast (primarily Any 252 Source Multicast (ASM) [RFC1112], although a recent extension 253 supports Source Specific Multicast (SSM) [RFC3569] with unicast 254 feedback [RFC5760]). 256 Application Awareness: When ECN support is provided within the 257 transport protocol, the ability of the application to react to 258 congestion is limited, since it has little visibility into the 259 transport layer. By adding support of ECN to RTP using RTCP 260 feedback, the application is made aware of congestion, allowing a 261 wider range of reactions in response to that loss. 263 Counting vs Detecting Congestion: TCP and the protocols derived from 264 it are mainly designed to respond the same whether they experience 265 a burst of congestion indications within one RTT or just one. 266 Whereas real-time applications may be concerned with the amount of 267 congestion experienced, whether it is distributed smoothly or in 268 bursts. When feedback of ECN was added to TCP [RFC3168], the 269 receiver was designed to flip the echo congestion experienced 270 (ECE) flag to 1 for a whole RTT then flop it back to zero. 271 Whereas ECN feedback in RTCP will need to report a count of how 272 much congestion has been experienced within an RTCP reporting 273 period, irrespective of round trip times. 275 These differences will significantly alter the shape of ECN support 276 in RTP-over-UDP compared to ECN support in TCP, SCTP, and DCCP, but 277 do not invalidate the need for ECN support. 279 ECN support is more important for RTP sessions than, for instance, is 280 the case for TCP. This is because the impact of packet loss in real- 281 time audio-visual media flows is highly visible to users. Effective 282 ECN support for RTP flows running over UDP will allow real-time 283 audio-visual applications to respond to the onset of congestion 284 before routers are forced to drop packets, allowing those 285 applications to control how they reduce their transmission rate, and 286 hence media quality, rather than responding to, and trying to conceal 287 the effects of unpredictable packet loss. Furthermore, widespread 288 deployment for ECN and active queue management in routers, should it 289 occur, can potentially reduce unnecessary queueing delays in routers, 290 lowering the round-trip time and benefiting interactive applications 291 of RTP, such as voice telephony. 293 3.1. Requirements 295 Considering ECN, transport protocols supporting ECN, and RTP based 296 applications one can create a set of requirements that must be 297 satisfied to at least some degree if ECN is to used by RTP over UDP. 299 o REQ 1: A mechanism MUST exist to negotiate and initiate the use of 300 ECN for RTP/UDP/IP sessions so that an RTP sender will not send 301 packets with ECT in the IP header unless it knows that all 302 potential receivers will understand any ECN-CE indications they 303 might receive. 305 o REQ 2: A mechanism MUST exist to feed back the reception of any 306 packets that are ECN-CE marked to the packet sender. 308 o REQ 3: The provided mechanism SHOULD minimise the possibility of 309 cheating (either by the sender or receiver). 311 o REQ 4: Some detection and fallback mechanism SHOULD exist to avoid 312 loss of communication due to the attempted usage of ECN in case an 313 intermediate node clears ECT or drops packets that are ECT marked. 315 o REQ 5: Negotiation of ECN SHOULD NOT significantly increase the 316 time taken to negotiate and set-up the RTP session (an extra RTT 317 before the media can flow is unlikely to be acceptable for some 318 use cases). 320 o REQ 6: Negotiation of ECN SHOULD NOT cause media clipping at the 321 start of a session. 323 The following sections describes how these requirements can be met 324 for RTP over UDP. 326 3.2. Applicability 328 The use of ECN with RTP over UDP is dependent on negotiation of ECN 329 capability between the sender and receiver(s), and validation of ECN 330 support in all elements of the network path(s) traversed. RTP is 331 used in a heterogeneous range of network environments and topologies, 332 with various different signalling protocols. The mechanisms defined 333 here make it possible to verify support for ECN in each of these 334 environments, and irrespective of the topology. 336 Due to the need for each RTP sender that intends to use ECN with RTP 337 to track all participants in the RTP session, the sub-sampling of the 338 group membership as specified by "Sampling of the Group Membership in 339 RTP" [RFC2762] MUST NOT be used. 341 The use of ECN is further dependent on a capability of the RTP media 342 flow to react to congestion signalled by ECN marked packets. 343 Depending on the application, media codec, and network topology, this 344 adaptation can occur in various forms and at various nodes. As an 345 example, the sender can change the media encoding, or the receiver 346 can change the subscription to a layered encoding, or either reaction 347 can be accomplished by a transcoding middlebox. RFC 5117 identifies 348 seven topologies in which RTP sessions may be configured, and which 349 may affect the ability to use ECN: 351 Topo-Point-to-Point: This is a standard unicast flow. ECN may be 352 used with RTP in this topology in an analogous manner to its use 353 with other unicast transport protocols, with RTCP conveying ECN 354 feedback messages. 356 Topo-Multicast: This is either an any source multicast (ASM) group 357 [RFC3569] with potentially several active senders and multicast 358 RTCP feedback, or a source specific multicast (SSM) group 359 [RFC4607] with a single sender and unicast RTCP feedback from 360 receivers. RTCP is designed to scale to large group sizes while 361 avoiding feedback implosion (see Section 6.2 of [RFC3550], 362 [RFC4585], and [RFC5760]), and can be used by a sender to 363 determine if all its receivers, and the network paths to those 364 receivers, support ECN (see Section 7.2). It is somewhat more 365 difficult to determine if all network paths from all senders to 366 all receivers support ECN. Accordingly, we allow ECN to be used 367 by an RTP sender using multicast UDP provided the sender has 368 verified that the paths to all its known receivers support ECN, 369 and irrespective of whether the paths from other senders to their 370 receivers support ECN ("all its known receivers" are all the SSRCs 371 that the RTP sender has received RTP or RTCP from the last five 372 reporting intervals, i.e., they have not timed out). Note that 373 group membership may change during the lifetime of a multicast RTP 374 session, potentially introducing new receivers that are not ECN 375 capable or have a path that doesn't support ECN. Senders must use 376 the mechanisms described in Section 7.4 to check that all 377 receivers, and the network paths traversed to reach those 378 receivers, continue to support ECN, and they need to fallback to 379 non-ECN use if any receivers join that do not. 381 Topo-Translator: An RTP translator is an RTP-level middlebox that is 382 invisible to the other participants in the RTP session (although 383 it is usually visible in the associated signalling session). 384 There are two types of RTP translator: those that do not modify 385 the media stream, and are concerned with transport parameters, for 386 example a multicast to unicast gateway; and those that do modify 387 the media stream, for example transcoding between different media 388 codecs. A single RTP session traverses the translator, and the 389 translator must rewrite RTCP messages passing through it to match 390 the changes it makes to the RTP data packets. A legacy, ECN- 391 unaware, RTP translator is expected to ignore the ECN bits on 392 received packets, and to set the ECN bits to not-ECT when sending 393 packets, so causing ECN negotiation on the path containing the 394 translator to fail (any new RTP translator that does not wish to 395 support ECN may do so similarly). An ECN aware RTP translator may 396 act in one of three ways: 398 * If the translator does not modify the media stream, it should 399 copy the ECN bits unchanged from the incoming to the outgoing 400 datagrams, unless it is overloaded and experiencing congestion, 401 in which case it may mark the outgoing datagrams with an ECN-CE 402 mark. Such a translator passes RTCP feedback unchanged. See 403 Section 8.1. 405 * If the translator modifies the media stream to combine or split 406 RTP packets, but does not otherwise transcode the media, it 407 must manage the ECN bits in a way analogous to that described 408 in Section 5.3 of [RFC3168], see Section 8.2 for details. 410 * If the translator is a media transcoder, or otherwise modifies 411 the content of the media stream, the output RTP media stream 412 may have radically different characteristics than the input RTP 413 media stream. Each side of the translator must then be 414 considered as a separate transport connection, with its own ECN 415 processing. This requires the translator interpose itself into 416 the ECN negotiation process, effectively splitting the 417 connection into two parts with their own negotiation. Once 418 negotiation has been completed, the translator must generate 419 RTCP ECN feedback back to the source based on its own 420 reception, and must respond to RTCP ECN feedback received from 421 the receiver(s) (see Section 8.3). 423 It is recognised that ECN and RTCP processing in an RTP translator 424 that modifies the media stream is non-trivial. 426 Topo-Mixer: A mixer is an RTP-level middlebox that aggregates 427 multiple RTP streams, mixing them together to generate a new RTP 428 stream. The mixer is visible to the other participants in the RTP 429 session, and is also usually visible in the associated signalling 430 session. The RTP flows on each side of the mixer are treated 431 independently for ECN purposes, with the mixer generating its own 432 RTCP ECN feedback, and responding to ECN feedback for data it 433 sends. Since unicast transport between the mixer and any end- 434 point are treated independently, it would seem reasonable to allow 435 the transport on one side of the mixer to use ECN, while the 436 transport on the other side of the mixer is not ECN capable, if 437 this is desired. See Section 8.4 for details in how mixers should 438 process ECN. 440 Topo-Video-switch-MCU: A video switching MCU receives several RTP 441 flows, but forwards only one of those flows onwards to the other 442 participants at a time. The flow that is forwarded changes during 443 the session, often based on voice activity. Since only a subset 444 of the RTP packets generated by a sender are forwarded to the 445 receivers, a video switching MCU can break ECN negotiation (the 446 success of the ECN negotiation may depend on the voice activity of 447 the participant at the instant the negotiation takes place - shout 448 if you want ECN). It also breaks congestion feedback and 449 response, since RTP packets are dropped by the MCU depending on 450 voice activity rather than network congestion. This topology is 451 widely used in legacy products, but is NOT RECOMMENDED for new 452 implementations and SHALL NOT be used with ECN. 454 Topo-RTCP-terminating-MCU: In this scenario, each participant runs 455 an RTP point-to-point session between itself and the MCU. Each of 456 these sessions is treated independently for the purposes of ECN 457 and RTCP feedback, potentially with some using ECN and some not. 459 Topo-Asymmetric: It is theoretically possible to build a middlebox 460 that is a combination of an RTP mixer in one direction and an RTP 461 translator in the other. To quote RFC 5117 "This topology is so 462 problematic and it is so easy to get the RTCP processing wrong, 463 that it is NOT RECOMMENDED to implement this topology." 465 These topologies may be combined within a single RTP session. 467 The ECN mechanism defined in this memo is applicable to both sender 468 and receiver controlled congestion algorithms. The mechanism ensures 469 that both senders and receivers will know about ECN-CE markings and 470 any packet losses. Thus the actual decision point for the congestion 471 control is not relevant. This is a great benefit as the rate of an 472 RTP session can be varied in a number of ways, for example a unicast 473 media sender might use TFRC [RFC5348] or some other algorithm, while 474 a multicast session could use a sender based scheme adapting to the 475 lowest common supported rate, or a receiver driven mechanism using 476 layered coding to support more heterogeneous paths. 478 To ensure timely feedback of ECN-CE marked packets when needed, this 479 mechanism requires support for the RTP/AVPF profile [RFC4585] or any 480 of its derivatives, such as RTP/SAVPF [RFC5124]. The standard RTP/ 481 AVP profile [RFC3551] does not allow any early or immediate 482 transmission of RTCP feedback, and has a minimal RTCP interval whose 483 default value (5 seconds) is many times the normal RTT between sender 484 and receiver. 486 3.3. Interoperability 488 The interoperability requirements for this specification are that 489 there is at least one common interoperability point for all 490 implementations. Since initialization using RTP and RTCP 491 (Section 7.2.1) is the one method that works in all cases, although 492 is not optimal for all uses, it is selected as mandatory to implement 493 this initialisation method. This method requires both the RTCP XR 494 extension and the ECN feedback format, which require the RTP/AVPF 495 profile to ensure timely feedback. 497 When one considers all the uses of ECN for RTP it is clear that there 498 exist congestion control mechanisms that are receiver driven only 499 (Section 7.3.3). These congestion control mechanism do not require 500 timely feedback of congestion events to the sender. If such a 501 congestion control mechanism is combined with an initialization 502 method that also doesn't require timely feedback using RTCP, like the 503 leap of faith or the ICE based method then neither the ECN feedback 504 format nor the RTP/AVPF profile would appear to be needed. However, 505 fault detection can be greatly improved by using receiver side 506 detection (Section 7.4.1) and early reporting of such cases using the 507 ECN feedback mechanism. 509 For interoperability we mandate the implementation of the RTP/AVPF 510 profile, with both RTCP extensions and the necessary signalling to 511 support a common operations mode. This specification recommends the 512 use of RTP/AVPF in all cases as negotiation of the common 513 interoperability point requires RTP/AVPF, mixed negotiation of RTP/ 514 AVP and RTP/AVPF depending on other SDP attributes in the same media 515 block is difficult, and the fact that fault detection can be improved 516 when using RTP/AVPF. 518 The use of the ECN feedback format is also recommended, but cases 519 exist where its use is not required due to no need for timely 520 feedback. These will be explicitly noted using the term "no timely 521 feedback required", and generally occur in combination with receiver 522 driven congestion control, and with the leap-of-faith and ICE-based 523 initialization methods. We also note that any receiver driven 524 congestion control solution that still requires RTCP for signalling 525 of any adaptation information to the sender will still require RTP/ 526 AVPF for timeliness. 528 4. Overview of Use of ECN with RTP/UDP/IP 530 The solution for using ECN with RTP over UDP/IP consists of four 531 different pieces that together make the solution work: 533 1. Negotiation of the capability to use ECN with RTP/UDP/IP 535 2. Initiation and initial verification of ECN capable transport 537 3. Ongoing use of ECN within an RTP session 539 4. Handling of dynamic behavior through failure detection, 540 verification and fallback 542 Before an RTP session can be created, a signalling protocol is used 543 to discover the other participants and negotiate or configure session 544 parameters (see Section 7.1). One of the parameters that must be 545 agreed is the capability of a participant to support ECN. Note that 546 all participants having the capability of supporting ECN does not 547 necessarily imply that ECN is usable in an RTP session, since there 548 may be middleboxes on the path between the participants which don't 549 pass ECN-marked packets (for example, a firewall that blocks traffic 550 with the ECN bits set). This document defines the information that 551 needs to be negotiated, and provides a mapping to SDP for use in both 552 declarative and offer/answer contexts. 554 When a sender joins a session for which all participants claim to 555 support ECN, it must verify if that support is usable. There are 556 three ways in which this verification can be done: 558 o The sender may generate a (small) subset of its RTP data packets 559 with the ECN field set to ECT(0) or ECT(1). Each receiver will 560 then send an RTCP feedback packet indicating the reception of the 561 ECT marked RTP packets. Upon reception of this feedback from each 562 receiver it knows of, the sender can consider ECN functional for 563 its traffic. Each sender does this verification independently. 564 When a new receiver joins an existing RTP session, it will send 565 RTCP reports in the usual manner. If those RTCP reports include 566 ECN information, verification will have succeeded and sources can 567 continue to send ECT packets. If not, verification fails and each 568 sender MUST stop using ECN (see Section 7.2.1 for details). 570 o Alternatively, ECN support can be verified during an initial end- 571 to-end STUN exchange (for example, as part of ICE connection 572 establishment). After having verified connectivity without ECN 573 capability an extra STUN exchange, this time with the ECN field 574 set to ECT(0) or ECT(1), is performed on the candidate path that 575 is about to be used. If successful the path's capability to 576 convey ECN marked packets is verified. A new STUN attribute is 577 defined to convey feedback that the ECT marked STUN request was 578 received (see Section 7.2.2), along with an ICE signalling option 579 (Section 6.4) to indicate that the check is to be performed. 581 o Thirdly, the sender may make a leap of faith that ECN will work. 582 This is only recommended for applications that know they are 583 running in controlled environments where ECN functionality has 584 been verified through other means. In this mode it is assumed 585 that ECN works, and the system reacts to failure indicators if the 586 assumption proved wrong. The use of this method relies on a high 587 confidence that ECN operation will be successful, or an 588 application where failure is not serious. The impact on the 589 network and other users must be considered when making a leap of 590 faith, so there are limitations on when this method is allowed 591 (see Section 7.2.3). 593 The first mechanism, using RTP with RTCP feedback, has the advantage 594 of working for all RTP sessions, but the disadvantages of potential 595 clipping if ECN marked RTP packets are discarded by middleboxes, and 596 slow verification of ECN support. The STUN-based mechanism is faster 597 to verify ECN support, but only works in those scenarios supported by 598 end-to-end STUN, such as within an ICE exchange. The third one, 599 leap-of-faith, has the advantage of avoiding additional tests or 600 complexities and enabling ECN usage from the first media packet. The 601 downside is that if the end-to-end path contains middleboxes that do 602 not pass ECN, the impact on the application can be severe: in the 603 worst case, all media could be lost if a middlebox that discards ECN 604 marked packets is present. A less severe effect, but still requiring 605 reaction, is the presence of a middlebox that re-marks ECT marked 606 packets to non-ECT, possibly marking packets with an ECN-CE mark as 607 non-ECT. This could result in increased levels of congestion due to 608 non-responsiveness, and impact media quality as applications end up 609 relying on packet loss as an indication of congestion. 611 Once ECN support has been verified (or assumed) to work for all 612 receivers, a sender marks all its RTP packets as ECT packets, while 613 receivers rapidly feed back reports on any ECN-CE marks to the sender 614 using RTCP in RTP/AVPF immediate or early feedback mode, unless no 615 timely feedback is required. Each feedback report indicates the 616 receipt of new ECN-CE marks since the last ECN feedback packet, and 617 also counts the total number of ECN-CE marked packets as a cumulative 618 sum. This is the mechanism to provide the fastest possible feedback 619 to senders about ECN-CE marks. On receipt of an ECN-CE marked 620 packet, the system must react to congestion as-if packet loss has 621 been reported. Section 7.3 describes the ongoing use of ECN within 622 an RTP session. 624 This rapid feedback is not optimised for reliability, so another 625 mechanism, RTCP XR ECN summary reports, is used to ensure more 626 reliable, but less timely, reporting of the ECN information. The ECN 627 summary report contains the same information as the ECN feedback 628 format, only packed differently for better efficiency with reports 629 for many sources. It is sent in a compound RTCP packet, along with 630 regular RTCP reception reports. By using cumulative counters for 631 observed ECN-CE, ECT, not-ECT, packet duplication, and packet loss 632 the sender can determine what events have happened since the last 633 report, independently of any RTCP packets having been lost. 635 RTCP reports MUST NOT be ECT marked, since ECT marked traffic may be 636 dropped if the path is not ECN compliant. RTCP is used to provide 637 feedback about what has been transmitted and what ECN markings that 638 are received, so it is important that it is received in cases when 639 ECT marked traffic is not getting through. 641 There are numerous reasons why the path the RTP packets take from the 642 sender to the receiver may change, e.g., mobility, link failure 643 followed by re-routing around it. Such an event may result in the 644 packet being sent through a node that is ECN non-compliant, thus re- 645 marking or dropping packets with ECT set. To prevent this from 646 impacting the application for longer than necessary, the operation of 647 ECN is constantly monitored by all senders (Section 7.4). Both the 648 RTCP XR ECN summary reports and the ECN feedback packets allow the 649 sender to compare the number of ECT(0), ECT(1), and non-ECT marked 650 packets received with the number that were sent, while also reporting 651 ECN-CE marked and lost packets. If these numbers do not agree, it 652 can be inferred that the path does not reliably pass ECN-marked 653 packets. A sender detecting a possible ECN non-compliance issue 654 should then stop sending ECT marked packets to determine if that 655 allows the packets to be correctly delivered. If the issues can be 656 connected to ECN, then ECN usage is suspended. 658 5. RTCP Extensions for ECN feedback 660 This memo defines two new RTCP extensions: one RTP/AVPF [RFC4585] 661 transport layer feedback format for urgent ECN information, and one 662 RTCP XR [RFC3611] ECN summary report block type for regular reporting 663 of the ECN marking information. 665 5.1. RTP/AVPF Transport Layer ECN Feedback packet 667 This RTP/AVPF transport layer feedback format is intended for use in 668 RTP/AVPF early or immediate feedback modes when information needs to 669 urgently reach the sender. Thus its main use is to report reception 670 of an ECN-CE marked RTP packet so that the sender may perform 671 congestion control, or to speed up the initiation procedures by 672 rapidly reporting that the path can support ECN-marked traffic. The 673 feedback format is also defined with reduced size RTCP [RFC5506] in 674 mind, where RTCP feedback packets may be sent without accompanying 675 Sender or Receiver Reports that would contain the Extended Highest 676 Sequence number and the accumulated number of packet losses. Both 677 are important for ECN to verify functionality and keep track of when 678 CE marking does occur. 680 The RTP/AVPF transport layer feedback packet starts with the common 681 header defined by the RTP/AVPF profile [RFC4585] which is reproduced 682 in Figure 1. The FMT field takes the value [TBA1] to indicate that 683 the Feedback Control Information (FCI) contains ECN Feedback report, 684 as defined in Figure 2. 686 0 1 2 3 687 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 688 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 689 |V=2|P| FMT=TBA1| PT=RTPFB=205 | length | 690 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 691 | SSRC of packet sender | 692 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 693 | SSRC of media source | 694 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 695 : Feedback Control Information (FCI) : 696 : : 698 Figure 1: RTP/AVPF Common Packet Format for Feedback Messages 700 0 1 2 3 701 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 702 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 703 | Extended Highest Sequence Number | 704 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 705 | ECT (0) Counter | 706 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 707 | ECT (1) Counter | 708 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 709 | ECN-CE Counter | not-ECT Counter | 710 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 711 | Loss Packet Counter | Duplication Counter | 712 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 714 Figure 2: ECN Feedback Report Format 716 The ECN Feedback Report contains the following fields: 718 Extended Highest Sequence Number: The 32-bit Extended highest 719 sequence number received, as defined by [RFC3550]. Indicates the 720 highest RTP sequence number to which this report relates. 722 ECT(0) Counter: The 32-bit cumulative number of RTP packets with 723 ECT(0) received from this SSRC. 725 ECT(1) Counter: The 32-bit cumulative number of RTP packets with 726 ECT(1) received from this SSRC. 728 ECN-CE Counter: The cumulative number of RTP packets received from 729 this SSRC since the receiver joined the RTP session that were 730 ECN-CE marked, including ECN-CE marks in any duplicate packets. 731 The receiver should keep track of this value using a local 732 representation that is at least 32-bits, and only include the 16- 733 bits with least significance. In other words, the field will wrap 734 if more than 65535 ECN-CE marked packets have been received. 736 not-ECT Counter: The cumulative number of RTP packets received from 737 this SSRC since the receiver joined the RTP session that had an 738 ECN field value of not-ECT. The receiver should keep track of 739 this value using a local representation that is at least 32-bits, 740 and only include the 16-bits with least significance. In other 741 words, the field will wrap if more than 65535 not-ECT packets have 742 been received. 744 Lost Packets Counter: The cumulative number of RTP packets that the 745 receiver expected to receive minus the number of packets it 746 actually received that are not a duplicate of an already received 747 packet, from this SSRC since the receiver joined the RTP session. 748 Note that packets that arrive late are not counted as lost. The 749 receiver should keep track of this value using a local 750 representation that is at least 32-bits, and only include the 16- 751 bits with least significance. In other words, the field will wrap 752 if more than 65535 packets are lost. 754 Duplication Counter: The cumulative number of RTP packets received 755 that are a duplicate of an already received packet from this SSRC 756 since the receiver joined the RTP session. The receiver should 757 keep track of this value using a local representation that is at 758 least 32-bits, and only include the 16-bits with least 759 significance. In other words, the field will wrap if more than 760 65535 duplicate packets have been received. 762 All fields in the ECN Feedback Report are unsigned integers in 763 network byte order. Each ECN Feedback Report corresponds to a single 764 RTP source (SSRC). Multiple sources can be reported by including 765 multiple ECN Feedback Reports packets in an compound RTCP packet. 767 The counters SHALL be initiated to 0 for each new SSRC received. 768 This to enable detection of ECN-CE marks or Packet loss on the 769 initial report from a specific participant. 771 The use of at least 32-bit counters allows even extremely high packet 772 volume applications to not have wrapping of counters within any 773 timescale close to the RTCP reporting intervals. However, 32-bits 774 are not sufficiently large to disregard the fact that wrappings may 775 happen during the life time of a long-lived RTP session. Thus 776 handling of wrapping of these counters MUST be supported. It is 777 recommended that implementations uses local representation of these 778 counters that are longer than 32-bits to enable easy handling of 779 wraps. 781 There is a difference in packet duplication reports between the 782 packet loss counter that is defined in the Receiver Report Block 783 [RFC3550] and that defined here. To avoid holding state for what RTP 784 sequence numbers have been received, [RFC3550] specifies that one can 785 count packet loss by counting the number of received packets and 786 comparing it to the number of packets expected. As a result a packet 787 duplication can hide a packet loss. However, when populating the ECN 788 Feedback report, a receiver needs to track the sequence numbers 789 actually received and count duplicates and packet loss separately to 790 provide a more reliable indication. Reordering may however still 791 result in that packet loss is reported in one report and then removed 792 in the next. 794 The ECN-CE counter is robust for packet duplication. Adding each 795 received ECN-CE marked packet to the counter is not an issue, in fact 796 it is required to ensure complete tracking of the ECN state. If one 797 of the clones was ECN-CE marked that is still an indication of 798 congestion. Packet duplication has potential impact on the ECN 799 verification and thus there is a need to count the duplicates. 801 5.2. RTCP XR Report block for ECN summary information 803 This unilateral XR report block combined with RTCP SR or RR report 804 blocks carries the same information as the ECN Feedback Report and is 805 be based on the same underlying information. However, the ECN 806 Feedback Report is intended to report on an ECN-CE mark as soon as 807 possible, while this extended report is for the regular RTCP 808 reporting and continuous verification of the ECN functionality end- 809 to-end. 811 The ECN Summary report block consists of one RTCP XR report block 812 header, shown in Figure 3 followed by one or more ECN summary report 813 data blocks, as defined in Figure 4. 815 0 1 2 3 816 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 817 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 818 | BT=[TBA2] | Reserved | Block Length | 819 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 821 Figure 3: RTCP XR Report Header 823 0 1 2 3 824 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 825 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 826 | SSRC of Media Sender | 827 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 828 | ECT (0) Counter | 829 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 830 | ECT (1) Counter | 831 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 832 | ECN-CE Counter | not-ECT Counter | 833 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 834 | Loss Packet Counter | Duplication Counter | 835 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 837 Figure 4: RTCP XR ECN Summary Report 839 The RTCP XR ECN Summary Report contains the following fields: 841 BT: Block Type identifying the ECN summary report block. Value is 842 [TBA2]. 844 Reserved: All bits SHALL be set to 0 on transmission and ignored on 845 reception. 847 Block Length: The length of the report block. Used to indicate the 848 number of report data blocks present in the ECN summary report. 849 This length will be 5*n, where n is the number of ECN summary 850 report blocks, since blocks are a fixed size. The block length 851 MAY be zero if there is nothing to report. Receivers MUST discard 852 reports where the block length is not a multiple of five octets, 853 since these cannot be valid. 855 SSRC of Media Sender: The SSRC identifying the media sender this 856 report is for. 858 ECT(0) Counter: as in Section 5.1. 860 ECT(1) Counter: as in Section 5.1. 862 ECN-CE Counter: as in Section 5.1. 864 not-ECT Counter: as in Section 5.1. 866 Loss Packet Counter: as in Section 5.1. 868 Duplication Counter: as in Section 5.1. 870 The Extended Highest Sequence number counter for each SSRC is not 871 present in RTCP XR report, in contrast to the feedback version. The 872 reason is that this summary report will rely on the information sent 873 in the Sender Report (SR) or Receiver Report (RR) blocks part of the 874 same RTCP compound packet. The Extended Highest Sequence number is 875 available from the SR or RR. 877 All the SSRCs that are present in the SR or RR SHOULD also be 878 included in the RTCP XR ECN summary report. In cases where the 879 number of senders are so large that the combination of SR/RR and the 880 ECN summary for all the senders exceed the MTU, then only a subset of 881 the senders SHOULD be included so that the reports for the subset 882 fits within the MTU. The subsets SHOULD be selected round-robin 883 across multiple intervals so that all sources are periodically 884 reported. In case there are no SSRCs that currently are counted as 885 senders in the session, the report block SHALL still be sent with no 886 report block entry and a zero report block length to continuously 887 indicate to the other participants the receiver capability to report 888 ECN information. 890 6. SDP Signalling Extensions for ECN 892 This section defines a number of SDP signalling extensions used in 893 the negotiation of the ECN for RTP support when using SDP. This 894 includes one SDP attribute "ecn-capable-rtp" that negotiates the 895 actual operation of ECN for RTP. Two SDP signalling parameters are 896 defined to indicate the use of the RTCP XR ECN summary block and the 897 RTP/AVPF feedback format for ECN. One ICE option SDP representation 898 is also defined. 900 6.1. Signalling ECN Capability using SDP 902 One new SDP attribute, "a=ecn-capable-rtp", is defined. This is a 903 media level attribute, and MUST NOT be used at the session level. It 904 is not subject to the character set chosen. The aim of this 905 signalling is to indicate the capability of the sender and receivers 906 to support ECN, and to negotiate the method of ECN initiation to be 907 used in the session. The attribute takes a list of initiation 908 methods, ordered in decreasing preference. The defined values for 909 the initiation method are: 911 rtp: Using RTP and RTCP as defined in Section 7.2.1. 913 ice: Using STUN within ICE as defined in Section 7.2.2. 915 leap: Using the leap of faith method as defined in Section 7.2.3. 917 Further methods may be specified in the future, so unknown methods 918 MUST be ignored upon reception. 920 In addition, a number of OPTIONAL parameters may be included in the 921 "a=ecn-capable-rtp" attribute as follows: 923 mode: This parameter signals the endpoint's capability to set and 924 read ECN marks in UDP packets. An examination of various 925 operating systems has shown that end-system support for ECN 926 marking of UDP packets may be symmetric or asymmetric. By this we 927 mean that some systems may allow end points to set the ECN bits in 928 an outgoing UDP packet but not read them, while others may allow 929 applications to read the ECN bits but not set them. This 930 either/or case may produce an asymmetric support for ECN and thus 931 should be conveyed in the SDP signalling. The "mode=setread" 932 state is the ideal condition where an endpoint can both set and 933 read ECN bits in UDP packets. The "mode=setonly" state indicates 934 that an endpoint can set the ECT bit, but cannot read the ECN bits 935 from received UDP packets to determine if upstream congestion 936 occurred. The "mode=readonly" state indicates that the endpoint 937 can read the ECN bits to determine if congestion has occurred for 938 incoming packets, but it cannot set the ECT bits in outgoing UDP 939 packets. When the "mode=" parameter is omitted it is assumed that 940 the node has "setread" capabilities. This option can provide for 941 an early indication that ECN cannot be used in a session. This 942 would be case when both the offerer and answerer set the "mode=" 943 parameter to "setonly" or both set it to "readonly". 945 ect: This parameter makes it possible to express the preferred ECT 946 marking. This is either "random", "0", or "1", with "0" being 947 implied if not specified. The "ect" parameter describes a 948 receiver preference, and is useful in the case where the receiver 949 knows it is behind a link using IP header compression, the 950 efficiency of which would be seriously disrupted if it were to 951 receive packets with randomly chosen ECT marks. It is RECOMMENDED 952 that ECT(0) marking be used. 954 The ABNF [RFC5234] grammar for the "a=ecn-capable-rtp" attribute is 955 shown in Figure 5. 957 ecn-attribute = "a=ecn-capable-rtp:" SP init-list [SP parm-list] 958 init-list = init-value *("," init-value) 959 init-value = "rtp" / "ice" / "leap" / init-ext 960 init-ext = token 961 parm-list = parm-value *(";" SP parm-value) 962 parm-value = mode / ect / parm-ext 963 mode = "mode=" ("setonly" / "setread" / "readonly") 964 ect = "ect=" ("0" / "1" / "random") 965 parm-ext = parm-name "=" parm-value-ext 966 parm-name = token 967 parm-value-ext = token / quoted-string 968 quoted-string = DQUOTE *qdtext DQUOTE 969 qdtext = %x20-21 / %x23-7E / %x80-FF 970 ; any 8-bit ASCII except <"> 972 ; external references: 973 ; token: from RFC 4566 974 ; SP and DQUOTE from RFC 5234 976 Figure 5: ABNF Grammar for the "a=ecn-capable-rtp" attribute 978 6.1.1. Use of "a=ecn-capable-rtp:" with the Offer/Answer Model 980 When SDP is used with the offer/answer model [RFC3264], the party 981 generating the SDP offer MUST insert an "a=ecn-capable-rtp" attribute 982 into the media section of the SDP offer of each RTP session for which 983 it wishes to use ECN. The attribute includes one or more ECN 984 initiation methods in a comma separated list in decreasing order of 985 preference, with any number of optional parameters following. The 986 answering party compares the list of initiation methods in the offer 987 with those it supports in order of preference. If there is a match, 988 and if the receiver wishes to attempt to use ECN in the session, it 989 includes an "a=ecn-capable-rtp" attribute containing its single 990 preferred choice of initiation method, and any optional parameters, 991 in the media sections of the answer. If there is no matching 992 initiation method capability, or if the receiver does not wish to 993 attempt to use ECN in the session, it does not include an "a=ecn- 994 capable-rtp" attribute in its answer. If the attribute is removed in 995 the answer then ECN MUST NOT be used in any direction for that media 996 flow. If there are initialization methods that are unknown, they 997 MUST be ignored on reception and MUST NOT be included in an answer. 999 The endpoints' capability to set and read ECN marks, as expressed by 1000 the optional "mode=" parameter, determines whether ECN support can be 1001 negotiated for flows in one or both directions: 1003 o If the "mode=setonly" parameter is present in the "a=ecn-capable- 1004 rtp" attribute of the offer and the answering party is also 1005 "mode=setonly", then there is no common ECN capability, and the 1006 answer MUST NOT include the "a=ecn-capable-rtp" attribute. 1007 Otherwise, if the offer is "mode=setonly" then ECN may only be 1008 initiated in the direction from the offering party to the 1009 answering party. 1011 o If the "mode=readonly" parameter is present in the "a=ecn-capable- 1012 rtp" attribute of the offer and the answering party is 1013 "mode=readonly", then there is no common ECN capability, and the 1014 answer MUST NOT include the "a=ecn-capable-rtp" attribute. 1015 Otherwise, if the offer is "mode=readonly" then ECN may only be 1016 initiated in the direction from the answering party to the 1017 offering party. 1019 o If the "mode=setread" parameter is present in the "a=ecn-capable- 1020 rtp" attribute of the offer and the answering party is "setonly", 1021 then ECN may only be initiated in the direction from the answering 1022 party to the offering party. If the offering party is 1023 "mode=setread" but the answering party is "mode=readonly", then 1024 ECN may only be initiated in the direction from the offering party 1025 to the answering party. If both offer and answer are 1026 "mode=setread", then ECN may be initiated in both directions. 1027 Note that "mode=setread" is implied by the absence of a "mode=" 1028 parameter in the offer or the answer. 1030 o An offer that does not include a "mode=" parameter MUST be treated 1031 as-if a "mode=setread" parameter had been included. 1033 In an RTP session using multicast and ECN, participants that intend 1034 to send RTP packets SHOULD support setting ECT marks in RTP packets 1035 (i.e., should be "mode=setonly" or "mode=setread"). Participants 1036 receiving data need the capability to read ECN marks on incoming 1037 packets. It is important that receivers can read ECN marks (are 1038 "mode=readonly" or "mode=setread"), since otherwise no sender in the 1039 multicast session will be able to enable ECN. Accordingly, receivers 1040 that are "mode=setonly" SHOULD NOT join multicast RTP sessions that 1041 use ECN. If session participants that are not aware of the ECN for 1042 RTP signalling are invited to a multicast session, and simply ignore 1043 the signalling attribute, the other party in the offer/answer 1044 exchange SHOULD terminate the SDP dialogue so that the participant 1045 leaves the session. 1047 The "ect=" parameter in the "a=ecn-capable-rtp" attribute is set 1048 independently in the offer and the answer. Its value in the offer 1049 indicates a preference for the sending behaviour of the answering 1050 party, and its value in the answer indicates a sending preference for 1051 the behaviour of the offering party. It will be the senders choice 1052 to honour the receivers preference for what to receive or not. In 1053 multicast sessions, all senders SHOULD set the ECT marks using the 1054 value declared in the "ect=" parameter. 1056 Unknown optional parameters MUST be ignored on reception, and MUST 1057 NOT be included in the answer. That way new parameters may be 1058 introduced and verified to be supported by the other end-point by 1059 having them include it in any answer. 1061 6.1.2. Use of "a=ecn-capable-rtp:" with Declarative SDP 1063 When SDP is used in a declarative manner, for example in a multicast 1064 session using the Session Announcement Protocol (SAP, [RFC2974]), 1065 negotiation of session description parameters is not possible. The 1066 "a=ecn-capable-rtp" attribute MAY be added to the session description 1067 to indicate that the sender will use ECN in the RTP session. The 1068 attribute MUST include a single method of initiation. Participants 1069 MUST NOT join such a session unless they have the capability to 1070 receive ECN-marked UDP packets, implement the method of initiation, 1071 and can generate RTCP ECN feedback. The mode parameter MAY also be 1072 included in declarative usage, to indicate the minimal capability is 1073 required by the consumer of the SDP. So for example in a SSM session 1074 the participants configured with a particular SDP will all be in a 1075 media receive only mode, thus mode=readonly will work as the 1076 capability of reporting on the ECN markings in the received is what 1077 is required. However, using "mode=readonly" also in ASM sessions is 1078 reasonable, unless all senders are required to attempt to use ECN for 1079 their outgoing RTP data traffic, in which case the mode needs to be 1080 set to "setread". 1082 6.1.3. General Use of the "a=ecn-capable-rtp:" Attribute 1084 The "a=ecn-capable-rtp" attribute MAY be used with RTP media sessions 1085 using UDP/IP transport. It MUST NOT be used for RTP sessions using 1086 TCP, SCTP, or DCCP transport, or for non-RTP sessions. 1088 As described in Section 7.3.3, RTP sessions using ECN require rapid 1089 RTCP ECN feedback, unless timely feedback is not required due to a 1090 receiver driven congestion control. To ensure that the sender can 1091 react to ECN-CE marked packets timely feedback is usually required. 1092 Thus, the use of the Extended RTP Profile for RTCP-Based Feedback 1093 (RTP/AVPF) [RFC4585] or other profile that inherits RTP/AVPF's 1094 signalling rules, MUST be signalled unless timely feedback is not 1095 required. If timely feedback is not required it is still RECOMMENDED 1096 to used RTP/AVPF. The signalling of an RTP/AVPF based profile is 1097 likely to be required even if the preferred method of initialization 1098 and the congestion control does not require timely feedback, as the 1099 common interoperable method is likely to be signalled or the improved 1100 fault reaction is desired. 1102 6.2. RTCP ECN Feedback SDP Parameter 1104 A new "nack" feedback parameter "ecn" is defined to indicate the 1105 usage of the RTCP ECN feedback packet format (Section 5.1). The ABNF 1106 [RFC5234] definition of the SDP parameter extension is: 1108 rtcp-fb-nack-param = 1109 rtcp-fb-nack-param /= ecn-fb-par 1110 ecn-fb-par = SP "ecn" 1112 The offer/answer rules for this SDP feedback parameters are specified 1113 in the RTP/AVPF profile [RFC4585]. 1115 6.3. XR Block ECN SDP Parameter 1117 A new unilateral RTCP XR block for ECN summary information is 1118 specified, thus the XR block SDP signalling also needs to be extended 1119 with a parameter. This is done in the same way as for the other XR 1120 blocks. The XR block SDP attribute as defined in Section 5.1 of the 1121 RTCP XR specification [RFC3611] is defined to be extensible. As no 1122 parameter values are needed for this ECN summary block, this 1123 parameter extension consists of a simple parameter name used to 1124 indicate support and intent to use the XR block. 1126 xr-format = 1127 xr-format /= ecn-summary-par 1128 ecn-summary-par = "ecn-sum" 1130 For SDP declarative and offer/answer usage, see the RTCP XR 1131 specification [RFC3611] and its description of how to handle 1132 unilateral parameters. 1134 6.4. ICE Parameter to Signal ECN Capability 1136 One new ICE [RFC5245] option, "rtp+ecn", is defined. This is used 1137 with the SDP session level "a=ice-options" attribute in an SDP offer 1138 to indicate that the initiator of the ICE exchange has the capability 1139 to support ECN for RTP-over-UDP flows (via "a=ice-options: rtp+ecn"). 1140 The answering party includes this same attribute at the session level 1141 in the SDP answer if it also has the capability, and removes the 1142 attribute if it does not wish to use ECN, or doesn't have the 1143 capability to use ECN. If the ICE initiation method (Section 7.2.2) 1144 is actually going to be used, it is also needs to be explicitly 1145 negotiated using the "a=ecn-capable-rtp" attribute. This ICE option 1146 SHALL be included when the ICE initiation method is offered or 1147 declared in the SDP. 1149 Note: This signalling mechanism is not strictly needed as long as 1150 the STUN ECN testing capability is used within the context of this 1151 document. It may however be useful if the ECN verification 1152 capability is used in additional contexts. 1154 7. Use of ECN with RTP/UDP/IP 1156 In the detailed specification of the behaviour below, the different 1157 functions in the general case will first be discussed. In case 1158 special considerations are needed for middleboxes, multicast usage 1159 etc, those will be specially discussed in related subsections. 1161 7.1. Negotiation of ECN Capability 1163 The first stage of ECN negotiation for RTP-over-UDP is to signal the 1164 capability to use ECN. An RTP system that supports ECN and uses SDP 1165 for its signalling MUST implement the SDP extension to signal ECN 1166 capability as described in Section 6.1, the RTCP ECN feedback SDP 1167 parameter defined in Section 6.2, and the XR Block ECN SDP parameter 1168 defined in Section 6.3. It MAY also implement alternative ECN 1169 capability negotiation schemes, such as the ICE extension described 1170 in Section 6.4. Other signalling systems will need to define 1171 signalling parameters corresponding to those defined for SDP. 1173 The "ecn-capable-rtp" SDP attribute MUST be used when employing ECN 1174 for RTP according to this specification in systems using SDP. As the 1175 RTCP XR ECN summary report is required independently of the 1176 initialization method or congestion control scheme, the "rtcp-xr" 1177 attribute with the "ecn-sum" parameter MUST also be used. The 1178 "rtcp-fb" attribute with the "nack" parameter "ecn" MUST be used 1179 whenever the initialization method or a congestion control algorithm 1180 requiring timely sender side knowledge of received CE markings. If 1181 the congestion control scheme requires additional signalling, this 1182 should be indicated as appropriate. 1184 7.2. Initiation of ECN Use in an RTP Session 1186 Once the sender and the receiver(s) have agreed that they have the 1187 capability to use ECN within a session, they may attempt to initiate 1188 ECN use. All session participants connected over the same transport 1189 will need to use the same initiation method. RTP mixers or 1190 translators can use different initiation methods to different 1191 participants that are connected over different underlying transports. 1192 The mixer or translator will need to do individual signalling with 1193 each participant to ensure it is consistent with the ECN support in 1194 those cases where it does not function as one end-point for the ECN 1195 control loop. 1197 At the start of the RTP session, when the first packets with ECT are 1198 sent, it is important to verify that IP packets with ECN field values 1199 of ECT or ECN-CE will reach their destination(s). There is some risk 1200 that the use of ECN will result in either reset of the ECN field, or 1201 loss of all packets with ECT or ECN-CE markings. If the path between 1202 the sender and the receivers exhibits either of these behaviours one 1203 needs to stop using ECN immediately to protect both the network and 1204 the application. 1206 The RTP senders and receivers SHALL NOT ECT mark their RTCP traffic 1207 at any time. This is to ensure that packet loss due to ECN marking 1208 will not effect the RTCP traffic and the necessary feedback 1209 information it carries. 1211 An RTP system that supports ECN MUST implement the initiation of ECN 1212 using in-band RTP and RTCP described in Section 7.2.1. It MAY also 1213 implement other mechanisms to initiate ECN support, for example the 1214 STUN-based mechanism described in Section 7.2.2, or use the leap of 1215 faith option if the session supports the limitations provided in 1216 Section 7.2.3. If support for both in-band and out-of-band 1217 mechanisms is signalled, the sender SHOULD try ECN negotiation using 1218 STUN with ICE first, and if it fails, fallback to negotiation using 1219 RTP and RTCP ECN feedback. 1221 No matter how ECN usage is initiated, the sender MUST continually 1222 monitor the ability of the network, and all its receivers, to support 1223 ECN, following the mechanisms described in Section 7.4. This is 1224 necessary because path changes or changes in the receiver population 1225 may invalidate the ability of the system to use ECN. 1227 7.2.1. Detection of ECT using RTP and RTCP 1229 The ECN initiation phase using RTP and RTCP to detect if the network 1230 path supports ECN comprises three stages. Firstly, the RTP sender 1231 generates some small fraction of its traffic with ECT marks to act as 1232 probe for ECN support. Then, on receipt of these ECT-marked packets, 1233 the receivers send RTCP ECN feedback packets and RTCP ECN summary 1234 reports to inform the sender that their path supports ECN. Finally, 1235 the RTP sender makes the decision to use ECN or not, based on whether 1236 the paths to all RTP receivers have been verified to support ECN. 1238 Generating ECN Probe Packets: During the ECN initiation phase, an 1239 RTP sender SHALL mark a small fraction of its RTP traffic as ECT, 1240 while leaving the reminder of the packets unmarked. The main 1241 reason for only marking some packets is to maintain usable media 1242 delivery during the ECN initiation phase in those cases where ECN 1243 is not supported by the network path. A secondary reason to send 1244 some not-ECT packets are to ensure that the receivers will send 1245 RTCP reports on this sender, even if all ECT marked packets are 1246 lost in transit. The not-ECT packets also provide a base-line to 1247 compare performance parameters against. A fourth reason for only 1248 probing with a small number of packets is to reduce the risk that 1249 significant numbers of congestion markings might be lost if ECT is 1250 cleared to Not-ECT by an ECN-Reverting Middlebox. Then any 1251 resulting lack of congestion response is likely to have little 1252 damaging affect on others. An RTP sender is RECOMMENDED to send a 1253 minimum of two packets with ECT markings per RTCP reporting 1254 interval. In case a random ECT pattern is intended to be used, at 1255 least one packet with ECT(0) and one with ECT(1) should be sent 1256 per reporting interval; in case a single ECT marking is to be 1257 used, only that ECT value SHOULD be sent. The RTP sender SHALL 1258 continue to send some ECT marked traffic as long as the ECN 1259 initiation phase continues. The sender SHOULD NOT mark all RTP 1260 packets as ECT during the ECN initiation phase. 1262 This memo does not mandate which RTP packets are marked with ECT 1263 during the ECN initiation phase. An implementation should insert 1264 ECT marks in RTP packets in a way that minimises the impact on 1265 media quality if those packets are lost. The choice of packets to 1266 mark is clearly very media dependent, but the use of RTP NO-OP 1267 payloads [I-D.ietf-avt-rtp-no-op], if supported, would be an 1268 appropriate choice. For audio formats, if would make sense for 1269 the sender to mark comfort noise packets or similar. For video 1270 formats, packets containing P- or B-frames, rather than I-frames, 1271 would be an appropriate choice. No matter which RTP packets are 1272 marked, those packets MUST NOT be sent in duplicate, with and 1273 without ECT, since the RTP sequence number is used to identify 1274 packets that are received with ECN markings. 1276 Generating RTCP ECN Feedback: If ECN capability has been negotiated 1277 in an RTP session, the receivers in the session MUST listen for 1278 ECT or ECN-CE marked RTP packets, and generate RTCP ECN feedback 1279 packets (Section 5.1) to mark their receipt. An immediate or 1280 early (depending on the RTP/AVPF mode) ECN feedback packet SHOULD 1281 be generated on receipt of the first ECT or ECN-CE marked packet 1282 from a sender that has not previously sent any ECT traffic. Each 1283 regular RTCP report MUST also contain an ECN summary report 1284 (Section 5.2). Reception of subsequent ECN-CE marked packets MUST 1285 result in additional early or immediate ECN feedback packets being 1286 sent unless no timely feedback is required. 1288 Determination of ECN Support: RTP is a group communication protocol, 1289 where members can join and leave the group at any time. This 1290 complicates the ECN initiation phase, since the sender must wait 1291 until it believes the group membership has stabilised before it 1292 can determine if the paths to all receivers support ECN (group 1293 membership changes after the ECN initiation phase has completed 1294 are discussed in Section 7.3). 1296 An RTP sender shall consider the group membership to be stable 1297 after it has been in the session and sending ECT-marked probe 1298 packets for at least three RTCP reporting intervals (i.e., after 1299 sending its third regularly scheduled RTCP packet), and when a 1300 complete RTCP reporting interval has passed without changes to the 1301 group membership. ECN initiation is considered successful when 1302 the group membership is stable, and all known participants have 1303 sent one or more RTCP ECN feedback packets or RTCP XR ECN summary 1304 reports indicating correct receipt of the ECT-marked RTP packets 1305 generated by the sender. 1307 As an optimisation, if an RTP sender is initiating ECN usage 1308 towards a unicast address, then it MAY treat the ECN initiation as 1309 provisionally successful if it receives an RTCP ECN feedback 1310 report or an RTCP XR ECN summary report indicating successful 1311 receipt of the ECT-marked packets, with no negative indications, 1312 from a single RTP receiver (where a single RTP receiver is 1313 considered as all SSRCs used by a single RTCP CNAME). After 1314 declaring provisional success, the sender MAY generate ECT-marked 1315 packets as described in Section 7.3, provided it continues to 1316 monitor the RTCP reports for a period of three RTCP reporting 1317 intervals from the time the ECN initiation started, to check if 1318 there are any other participants in the session. Thus as long as 1319 any additional SSRC that report on the ECN usage are using the 1320 same RTCP CNAME as the previous reports and they are all 1321 indicating functional ECN the sender may continue. If other 1322 participants are detected, i.e., other RTCP CNAMEs, the sender 1323 MUST fallback to only ECT-marking a small fraction of its RTP 1324 packets, while it determines if ECN can be supported following the 1325 full procedure described above. Different RTCP CNAMEs received 1326 over an unicast transport may occur when using translators in a 1327 multi-party RTP session (e.g., when using a centralised conference 1328 bridge). 1330 Note: The above optimization supports peer to peer unicast 1331 transport with several SSRCs multiplexed onto the same flow 1332 (e.g., a single participant with two video cameras, or SSRC 1333 multiplexed RTP retransmission [RFC4588]). It is desirable to 1334 be able to rapidly negotiate ECN support for such a session, 1335 but the optimisation above can fail if there are 1336 implementations that use the same CNAME for different parts of 1337 a distributed implementation that have different transport 1338 characteristics (e.g., if a single logical endpoint is split 1339 across multiple hosts). 1341 ECN initiation is considered to have failed at the instant when 1342 any RTP session participant sends an RTCP packet that doesn't 1343 contain an RTCP ECN feedback report or ECN summary report, but has 1344 an RTCP RR with an extended RTP sequence number field that 1345 indicates that it should have received multiple (>3) ECT marked 1346 RTP packets. This can be due to failure to support the ECN 1347 feedback format by the receiver or some middlebox, or the loss of 1348 all ECT marked packets. Both indicate a lack of ECN support. 1350 If the ECN negotiation succeeds, this indicates that the path can 1351 pass some ECN-marked traffic, and that the receivers support ECN 1352 feedback. This does not necessarily imply that the path can robustly 1353 convey ECN feedback; Section 7.3 describes the ongoing monitoring 1354 that must be performed to ensure the path continues to robustly 1355 support ECN. 1357 When a sender or receiver detects ECN failures on paths they should 1358 log these to enable follow up and statistics gathering regarding 1359 broken paths. The logging mechanism used is implementation 1360 dependent. 1362 7.2.2. Detection of ECT using STUN with ICE 1364 This section describes an OPTIONAL method that can be used to avoid 1365 media impact and also ensure an ECN capable path prior to media 1366 transmission. This method is considered in the context where the 1367 session participants are using ICE [RFC5245] to find working 1368 connectivity. We need to use ICE rather than STUN only, as the 1369 verification needs to happen from the media sender to the address and 1370 port on which the receiver is listening. 1372 Note that this method is only applicable to sessions when the remote 1373 destinations are unicast addresses. In addition, transport 1374 translators that do not terminate the ECN control loop and may 1375 distribute received packets to more than one other receiver must 1376 either disallow this method (and use the RTP/RTCP method instead), or 1377 implement additional handling as discussed below. This is because 1378 the ICE initialization method verifies the underlying transport to 1379 one particular address and port. If the receiver at that address and 1380 port intends to use the received packets in a multi-point session 1381 then the tested capabilities and the actual session behavior are not 1382 matched. 1384 To minimise the impact of set-up delay, and to prioritise the fact 1385 that one has a working connectivity rather than necessarily finding 1386 the best ECN capable network path, this procedure is applied after 1387 having performed a successful connectivity check for a candidate, 1388 which is nominated for usage. At that point an additional 1389 connectivity check is performed, sending the "ECN Check" attribute in 1390 a STUN packet that is ECT marked. On reception of the packet, a STUN 1391 server supporting this extension will note the received ECN field 1392 value, and send a STUN/UDP/IP packet in reply with the ECN field set 1393 to not-ECT and including an ECN check attribute. A STUN server that 1394 doesn't understand the extension, or is incapable of reading the ECN 1395 values on incoming STUN packets, should follow the rule in the STUN 1396 specification for unknown comprehension-optional attributes, and 1397 ignore the attribute, resulting in the sender receiving a STUN 1398 response without the ECN Check STUN attribute. 1400 The STUN ECN check attribute contains one field and a flag, as shown 1401 in Figure 6. The flag indicates whether the echo field contains a 1402 valid value or not. The field is the ECN echo field, and when valid 1403 contains the two ECN bits from the packet it echoes back. The ECN 1404 check attribute is a comprehension optional attribute. 1406 0 1 2 3 1407 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1408 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1409 | Type | Length | 1410 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1411 | Reserved |ECF|V| 1412 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1414 Figure 6: ECN Check STUN Attribute 1416 V: Valid (1 bit) ECN Echo value field is valid when set to 1, and 1417 invalid when set 0. 1419 ECF: ECN Echo value field (2 bits) contains the ECN field value of 1420 the STUN packet it echoes back when field is valid. If invalid 1421 the content is arbitrary. 1423 Reserved: Reserved bits (29 bits) SHALL be set to 0 on transmission, 1424 and SHALL be ignored on reception. 1426 This attribute MAY be included in any STUN request to request the ECN 1427 field to be echoed back. In STUN requests the V bit SHALL be set to 1428 0. A compliant STUN server receiving a request with the ECN Check 1429 attribute SHALL read the ECN field value of the IP/UDP packet the 1430 request was received in. Upon forming the response the server SHALL 1431 include the ECN Check attribute setting the V bit to valid and 1432 include the read value of the ECN field into the ECF field. If the 1433 STUN responder was unable to ascertain, due to temporary errors, the 1434 ECN value of the STUN request, it SHALL set the V bit in the response 1435 to 0. The STUN client may retry immediately. 1437 The ICE based initialization method does require some special 1438 consideration when used by a translator. This is especially for 1439 transport translators and translators that fragment or reassemble 1440 packets, since they do not separate the ECN control loops between the 1441 end-points and the translator. When using ICE-based initiation, such 1442 a translator must ensure that any participants joining an RTP session 1443 for which ECN has been negotiated are successfully verified in the 1444 direction from the translator to the joining participant. 1445 Alternatively, it must correctly handle remarking of ECT RTP packets 1446 towards that participant. When a new participant joins the session, 1447 the translator will perform a check towards the new participant. If 1448 that is successfully completed the ECT properties of the session are 1449 maintained for the other senders in the session. If the check fails 1450 then the existing senders will now see a participant that fails to 1451 receive ECT. Thus the failure detection in those senders will 1452 eventually detect this. However to avoid misusing the network on the 1453 path from the translator to the new participant, the translator SHALL 1454 remark the traffic intended to be forwarded from ECT to non-ECT. Any 1455 packet intended to be forward that are ECN-CE marked SHALL be discard 1456 and not sent. In cases where the path from a new participant to the 1457 translator fails the ECT check then only that sender will not 1458 contribute any ECT marked traffic towards the translator. 1460 7.2.3. Leap of Faith ECT initiation method 1462 This method for initiating ECN usage is a leap of faith that assumes 1463 that ECN will work on the used path(s). The method is to go directly 1464 to "ongoing use of ECN" as defined in Section 7.3. Thus all RTP 1465 packets MAY be marked as ECT and the failure detection MUST be used 1466 to detect any case when the assumption that the path was ECT capable 1467 is wrong. This method is only recommended for controlled 1468 environments where the whole path(s) between sender and receiver(s) 1469 has been built and verified to be ECT. 1471 If the sender marks all packets as ECT while transmitting on a path 1472 that contains an ECN-blocking middlebox, then receivers downstream of 1473 that middlebox will not receive any RTP data packets from the sender, 1474 and hence will not consider it to be an active RTP SSRC. The sender 1475 can detect this and revert to sending packets without ECT marks, 1476 since RTCP SR/RR packets from such receivers will either not include 1477 a report for sender's SSRC, or will report that no packets have been 1478 received, but this takes at least one RTCP reporting interval. It 1479 should be noted that a receiver might generate its first RTCP packet 1480 immediately on joining a unicast session, or very shortly after 1481 joining a RTP/AVPF session, before it has had chance to receive any 1482 data packets. A sender that receives RTCP SR/RR packet indicating 1483 lack of reception by a receiver SHOULD therefore wait for a second 1484 RTCP report from that receiver to be sure that the lack of reception 1485 is due to ECT-marking. Since this recovery process can take several 1486 tens of seconds, during which time the RTP session is unusable for 1487 media, it is NOT RECOMMENDED that the leap-of-faith ECT initiation 1488 method be used in environments where ECN-blocking middleboxes are 1489 likely to be present. 1491 7.3. Ongoing Use of ECN Within an RTP Session 1493 Once ECN has been successfully initiated for an RTP sender, that 1494 sender begins sending all RTP data packets as ECT-marked, and its 1495 receivers send ECN feedback information via RTCP packets. This 1496 section describes procedures for sending ECT-marked data, providing 1497 ECN feedback information via RTCP, responding to ECN feedback 1498 information, and detecting failures and misbehaving receivers. 1500 7.3.1. Transmission of ECT-marked RTP Packets 1502 After a sender has successfully initiated ECN use, it SHOULD mark all 1503 the RTP data packets it sends as ECT. The sender SHOULD mark packets 1504 as ECT(0) unless the receiver expresses a preference for ECT(1) or 1505 random using the "ect" parameter in the "a=ecn-capable-rtp" 1506 attribute. 1508 The sender SHALL NOT include ECT marks on outgoing RTCP packets, and 1509 SHOULD NOT include ECT marks on any other outgoing control messages 1510 (e.g., STUN [RFC5389] packets, DTLS [RFC4347] handshake packets, or 1511 ZRTP [RFC6189] control packets) that are multiplexed on the same UDP 1512 port. For control packets there might be exceptions, like the STUN 1513 based ECN check defined in Section 7.2.2. 1515 7.3.2. Reporting ECN Feedback via RTCP 1517 An RTP receiver that receives a packet with an ECN-CE mark, or that 1518 detects a packet loss, MUST schedule the transmission of an RTCP ECN 1519 feedback packet as soon as possible (subject to the constraints of 1520 [RFC4585] and [RFC3550]) to report this back to the sender unless no 1521 timely feedback is required. The feedback RTCP packet SHALL consist 1522 of at least one ECN feedback packet (Section 5.1) reporting on the 1523 packets received since the last ECN feedback packet, and will contain 1524 (at least) an RTCP SR/RR packet and an SDES packet, unless reduced 1525 size RTCP [RFC5506] is used. The RTP/AVPF profile in early or 1526 immediate feedback mode SHOULD be used where possible, to reduce the 1527 interval before feedback can be sent. To reduce the size of the 1528 feedback message, reduced size RTCP [RFC5506] MAY be used if 1529 supported by the end-points. Both RTP/AVPF and reduced size RTCP 1530 MUST be negotiated in the session set-up signalling before they can 1531 be used. 1533 Every time a regular compound RTCP packet is to be transmitted, an 1534 ECN-capable RTP receiver MUST include an RTCP XR ECN summary report 1535 as described in Section 5.2 as part of the compound packet. 1537 The multicast feedback implosion problem, that occurs when many 1538 receivers simultaneously send feedback to a single sender, must be 1539 considered. The RTP/AVPF transmission rules will limit the amount of 1540 feedback that can be sent, avoiding the implosion problem but also 1541 delaying feedback by varying degrees from nothing up to a full RTCP 1542 reporting interval. As a result, the full extent of a congestion 1543 situation may take some time to reach the sender, although some 1544 feedback should arrive in a reasonably timely manner, allowing the 1545 sender to react on a single or a few reports. 1547 7.3.3. Response to Congestion Notifications 1549 The reception of RTP packets with ECN-CE marks in the IP header is a 1550 notification that congestion is being experienced. The default 1551 reaction on the reception of these ECN-CE marked packets MUST be to 1552 provide the congestion control algorithm with a congestion 1553 notification that triggers the algorithm to react as if packet loss 1554 had occurred. There should be no difference in congestion response 1555 if ECN-CE marks or packet drops are detected. 1557 We note that there MAY be other reactions to ECN-CE specified in the 1558 future. Such an alternative reaction MUST be specified and 1559 considered to be safe for deployment under any restrictions 1560 specified. A potential example for an alternative reaction could be 1561 emergency communications (such as that generated by first responders, 1562 as opposed to the general public) in networks where the user has been 1563 authorized. A more detailed description of these other reactions, as 1564 well as the types of congestion control algorithms used by end-nodes, 1565 is outside of the scope of this document. 1567 Depending on the media format, type of session, and RTP topology 1568 used, there are several different types of congestion control that 1569 can be used: 1571 Sender-Driven Congestion Control: The sender is responsible for 1572 adapting the transmitted bit-rate in response to RTCP ECN 1573 feedback. When the sender receives the ECN feedback data it feeds 1574 this information into its congestion control or bit-rate 1575 adaptation mechanism so that it can react as if packet loss was 1576 reported. The congestion control algorithm to be used is not 1577 specified here, although TFRC [RFC5348] is one example that might 1578 be used. 1580 Receiver-Driven Congestion Control: If a receiver driven congestion 1581 control mechanism is used, the receiver can react to the ECN-CE 1582 marks without contacting the sender. This may allow faster 1583 response than sender-driven congestion control in some 1584 circumstances. Receiver-driven congestion control is usually 1585 implemented by providing the content in a layered way, with each 1586 layer providing improved media quality but also increased 1587 bandwidth usage. The receiver locally monitors the ECN-CE marks 1588 on received packets to check if it experiences congestion with the 1589 current number of layers. If congestion is experienced, the 1590 receiver drops one layer, so reducing the resource consumption on 1591 the path towards itself. For example, if a layered media encoding 1592 scheme such as H.264 SVC is used, the receiver may change its 1593 layer subscription, and so reduce the bit rate it receives. The 1594 receiver MUST still send RTCP XR ECN Summary to the sender, even 1595 if it can adapt without contact with the sender, so that the 1596 sender can determine if ECN is supported on the network path. The 1597 timeliness of RTCP feedback is less of a concern with receiver 1598 driven congestion control, and regular RTCP reporting of ECN 1599 summary information is sufficient (without using RTP/AVPF 1600 immediate or early feedback). 1602 Hybrid: There might be mechanisms that utilize both some receiver 1603 behaviors and some sender side monitoring, thus requiring both 1604 feedback of congestion events to the sender and taking receiver 1605 decisions and possible signalling to the sender. In this case the 1606 congestion control algorithm needs to use the signalling to 1607 indicate which features of ECN for RTP are required. 1609 Responding to congestion indication in the case of multicast traffic 1610 is a more complex problem than for unicast traffic. The fundamental 1611 problem is diverse paths, i.e., when different receivers don't see 1612 the same path, and thus have different bottlenecks, so the receivers 1613 may get ECN-CE marked packets due to congestion at different points 1614 in the network. This is problematic for sender driven congestion 1615 control, since when receivers are heterogeneous in regards to 1616 capacity, the sender is limited to transmitting at the rate the 1617 slowest receiver can support. This often becomes a significant 1618 limitation as group size grows. Also, as group size increases the 1619 frequency of reports from each receiver decreases, which further 1620 reduces the responsiveness of the mechanism. Receiver-driven 1621 congestion control has the advantage that each receiver can choose 1622 the appropriate rate for its network path, rather than all receivers 1623 having to settle for the lowest common rate. 1625 We note that ECN support is not a silver bullet to improving 1626 performance. The use of ECN gives the chance to respond to 1627 congestion before packets are dropped in the network, improving the 1628 user experience by allowing the RTP application to control how the 1629 quality is reduced. An application which ignores ECN Congestion 1630 Experienced feedback is not immune to congestion: the network will 1631 eventually begin to discard packets if traffic doesn't respond. It 1632 is in the best interest of an application to respond to ECN 1633 congestion feedback promptly, to avoid packet loss. 1635 7.4. Detecting Failures 1637 Senders and receivers can deliberately ignore ECN-CE and thus get a 1638 benefit over behaving flows (cheating). Th ECN Nonce [RFC3540] is an 1639 addition to TCP that attempts to solve this issue as long as the 1640 sender acts on behalf of the network. The assumption about the 1641 senders acting on the behalf of the network may be reduced due to the 1642 nature of peer-to-peer use of RTP. Still a significant portion of 1643 RTP senders are infrastructure devices (for example, streaming media 1644 servers) that do have an interest in protecting both service quality 1645 and the network. Even though there may be cases where nonce can be 1646 applicable also for RTP, it is not included in this specification. 1647 This as a receiver interested in cheating would simple claim to not 1648 support nonce, or even ECN itself. It is, however, worth mentioning 1649 that, as real-time media is commonly sensitive to increased delay and 1650 packet loss, it will be in both the media sender and receivers 1651 interest to minimise the number and duration of any congestion events 1652 as they will adversely affect media quality. 1654 RTP sessions can also suffer from path changes resulting in a non-ECN 1655 compliant node becoming part of the path. That node may perform 1656 either of two actions that has effect on the ECN and application 1657 functionality. The gravest is if the node drops packets with the ECN 1658 field set to ECT(0), ECT(1), or ECN-CE. This can be detected by the 1659 receiver when it receives an RTCP SR packet indicating that a sender 1660 has sent a number of packets that it has not received. The sender 1661 may also detect such a middlebox based on the receiver's RTCP RR 1662 packet, when the extended sequence number is not advanced due to the 1663 failure to receive packets. If the packet loss is less than 100%, 1664 then packet loss reporting in either the ECN feedback information or 1665 RTCP RR will indicate the situation. The other action is to re-mark 1666 a packet from ECT or ECN-CE to not-ECT. That has less dire results, 1667 however it should be detected so that ECN usage can be suspended to 1668 prevent misusing the network. 1670 The RTCP XR ECN summary packet and the ECN feedback packet allow the 1671 sender to compare the number of ECT marked packets of different types 1672 received with the number it actually sent. The number of ECT packets 1673 received, plus the number of ECN-CE marked and lost packets, should 1674 correspond to the number of sent ECT marked packets plus the number 1675 of received duplicates. If these numbers doesn't agree there are two 1676 likely reasons, a translator changing the stream or not carrying the 1677 ECN markings forward, or that some node re-marks the packets. In 1678 both cases the usage of ECN is broken on the path. By tracking all 1679 the different possible ECN field values a sender can quickly detect 1680 if some non-compliant behavior is happening on the path. 1682 Thus packet losses and non-matching ECN field value statistics are 1683 possible indication of issues with using ECN over the path. The next 1684 section defines both sender and receiver reactions to these cases. 1686 7.4.1. Fallback mechanisms 1688 Upon the detection of a potential failure, both the sender and the 1689 receiver can react to mitigate the situation. 1691 A receiver that detects a packet loss burst MAY schedule an early 1692 feedback packet that includes at least the RTCP RR and the ECN 1693 feedback message to report this to the sender. This will speed up 1694 the detection of the loss at the sender, thus triggering sender side 1695 mitigation. 1697 A sender that detects high packet loss rates for ECT-marked packets 1698 SHOULD immediately switch to sending packets as not-ECT to determine 1699 if the losses are potentially due to the ECT markings. If the losses 1700 disappear when the ECT-marking is discontinued, the RTP sender should 1701 go back to initiation procedures to attempt to verify the apparent 1702 loss of ECN capability of the used path. If a re-initiation fails 1703 then the two possible actions exist: 1705 1. Periodically retry the ECN initiation to detect if a path change 1706 occurs to a path that is ECN capable. 1708 2. Renegotiate the session to disable ECN support. This is a choice 1709 that is suitable if the impact of ECT probing on the media 1710 quality is noticeable. If multiple initiations have been 1711 successful, but the following full usage of ECN has resulted in 1712 the fallback procedures, then disabling of the ECN support is 1713 RECOMMENDED. 1715 We foresee the possibility of flapping ECN capability due to several 1716 reasons: video switching MCU or similar middleboxes that selects to 1717 deliver media from the sender only intermittently; load balancing 1718 devices may in worst case result in that some packets take a 1719 different network path then the others; mobility solutions that 1720 switch underlying network path in a transparent way for the sender or 1721 receiver; and membership changes in a multicast group. It is however 1722 appropriate to mention that there are also issues such as re-routing 1723 of traffic due to a flappy route table or excessive reordering and 1724 other issues that are not directly ECN related but nevertheless may 1725 cause problems for ECN. 1727 7.4.2. Interpretation of ECN Summary information 1729 This section contains discussion on how the ECN summary report 1730 information can be used to detect various types of ECN path issues. 1731 We first review the information the RTCP reports provide on a per 1732 source (SSRC) basis: 1734 ECN-CE Counter: The number of RTP packets received so far in the 1735 session with an ECN field set to CE. 1737 ECT (0/1) Counters: The number of RTP packets received so far in the 1738 session with an ECN field set to ECT (0) and ECT (1) respectively. 1740 not-ECT Counter: The number of RTP packets received so far in the 1741 session with an ECN field set to not-ECT. 1743 Lost Packets counter: The number of RTP packets that where expected 1744 based on sequence numbers but never received. 1746 Duplication Counter: The number of received RTP packets that are 1747 duplicates of already received ones. 1749 Extended Highest Sequence number: The highest sequence number seen 1750 when sending this report, but with additional bits, to handle 1751 disambiguation when wrapping the RTP sequence number field. 1753 The counters will be initialised to zero to provide value for the RTP 1754 stream sender from the first report. After the first report, the 1755 changes between the last received report and the previous report are 1756 determined by simply taking the values of the latest minus the 1757 previous, taking wrapping into account. This definition is also 1758 robust to packet losses, since if one report is missing, the 1759 reporting interval becomes longer, but is otherwise equally valid. 1761 In a perfect world, the number of not-ECT packets received should be 1762 equal to the number sent minus the lost packets counter, and the sum 1763 of the ECT(0), ECT(1), and ECN-CE counters should be equal to the 1764 number of ECT marked packet sent. Two issues may cause a mismatch in 1765 these statistics: severe network congestion or unresponsive 1766 congestion control might cause some ECT-marked packets to be lost, 1767 and packet duplication might result in some packets being received, 1768 and counted in the statistics, multiple times (potentially with a 1769 different ECN-mark on each copy of the duplicate). 1771 The rate of packet duplication is tracked, allowing one to take the 1772 duplication into account. The value of the ECN field for duplicates 1773 will also be counted and when comparing the figures one needs to take 1774 some fraction of packet duplicates that are non-ECT and some fraction 1775 of packet duplicates being ECT into account into the calculation. 1776 Thus when only sending non-ECT then the number of sent packets plus 1777 reported duplicates equals the number of received non-ECT. When 1778 sending only ECT then number of sent ECT packets plus duplicates will 1779 equal ECT(0), ECT(1), ECN-CE and packet loss. When sending a mix of 1780 non-ECT and ECT then there is an uncertainty if any duplicate or 1781 packet loss was an non-ECT or ECT. If the packet duplication is 1782 completely independent of the usage of ECN, then the fraction of 1783 packet duplicates should be in relation to the number of non-ECT vs 1784 ECT packet sent during the period of comparison. This relation does 1785 not hold for packet loss, where higher rates of packet loss for non- 1786 ECT is expected than for ECT traffic. 1788 Detecting clearing of ECN field: If the ratio between ECT and not-ECT 1789 transmitted in the reports has become all not-ECT, or has 1790 substantially changed towards not-ECT, then this is clearly an 1791 indication that the path results in clearing of the ECT field. 1793 Dropping of ECT packets: To determine if the packet drop ratio is 1794 different between not-ECT and ECT marked transmission requires a mix 1795 of transmitted traffic. The sender should compare if the delivery 1796 percentage (delivered / transmitted) between ECT and not-ECT is 1797 significantly different. Care must be taken if the number of packets 1798 are low in either of the categories. One must also take into account 1799 the level of CE marking. A CE marked packet would have been dropped 1800 unless it was ECT marked. Thus, the packet loss level for not-ECT 1801 should be approximately equal to the loss rate for ECT when counting 1802 the CE marked packets as lost ones. A sender performing this 1803 calculation needs to ensure that the difference is statistically 1804 significant. 1806 If erroneous behavior is detected, it should be logged to enable 1807 follow up and statistics gathering. 1809 8. Processing ECN in RTP Translators and Mixers 1811 RTP translators and mixers that support ECN for RTP are required to 1812 process, and potentially modify or generate ECN marking in RTP 1813 packets. They also need to process, and potentially modify or 1814 generate RTCP ECN feedback packets for the translated and/or mixed 1815 streams. This includes both downstream RTCP reports generated by the 1816 media sender, and also reports generated by the receivers, flowing 1817 upstream back towards the sender. 1819 8.1. Transport Translators 1821 Some translators only perform transport level translations, like 1822 copying packets from one address domain, like unicast to multicast. 1823 It may also perform relaying like copying an incoming packet to a 1824 number of unicast receivers. This section details the ECN related 1825 actions for RTP and RTCP. 1827 For the RTP data packets the translator, which does not modify the 1828 media stream, SHOULD copy the ECN bits unchanged from the incoming to 1829 the outgoing datagrams, unless the translator itself is overloaded 1830 and experiencing congestion, in which case it may mark the outgoing 1831 datagrams with an ECN-CE mark. 1833 A Transport translator does not modify RTCP packets. It however MUST 1834 perform the corresponding transport translation of the RTCP packets 1835 as it does with RTP packets being sent from the same source/ 1836 end-point. 1838 8.2. Fragmentation and Reassembly in Translators 1840 An RTP translator may fragment or reassemble RTP data packets without 1841 changing the media encoding, and without reference to the congestion 1842 state of the networks it bridges. An example of this might be to 1843 combine packets of a voice-over-IP stream coded with one 20ms frame 1844 per RTP packet into new RTP packets with two 20ms frames per packet, 1845 thereby reducing the header overheads and so stream bandwidth, at the 1846 expense of an increase in latency. If multiple data packets are re- 1847 encoded into one, or vice versa, the RTP translator MUST assign new 1848 sequence numbers to the outgoing packets. Losses in the incoming RTP 1849 packet stream may also induce corresponding gaps in the outgoing RTP 1850 sequence numbers. An RTP translator MUST rewrite RTCP packets to 1851 make the corresponding changes to their sequence numbers, and to 1852 reflect the impact of the fragmentation or reassembly. This section 1853 describes how that rewriting is to be done for RTCP ECN feedback 1854 packets. Section 7.2 of [RFC3550] describes general procedures for 1855 other RTCP packet types. 1857 The processing of arriving RTP packets for this case is as follows. 1858 If an ECN marked packet is split into two, then both the outgoing 1859 packets MUST be ECN marked identically to the original; if several 1860 ECN marked packets are combined into one, the outgoing packet MUST be 1861 either ECN-CE marked or dropped if any of the incoming packets are 1862 ECN-CE marked. If the outgoing combined packet is not ECN-CE marked, 1863 then it MUST be ECT marked if any of the incoming packets were ECT 1864 marked. 1866 RTCP ECN feedback packets (Section 5.1) contain seven fields that are 1867 rewritten in an RTP translator that fragments or reassembles packets: 1868 the extended highest sequence number, the duplication counter, the 1869 lost packets counter, the ECN-CE counter, and not-ECT counter, the 1870 ECT(0) counter, and the ECT(1) counter. The RTCP XR report block for 1871 ECN summary information (Section 5.2) includes all of these fields 1872 except the extended highest sequence number which is present in the 1873 report block in an SR or RR packet. The procedures for rewriting 1874 these fields are the same for both RTCP ECN feedback packet and the 1875 RTCP XR ECN summary packet. 1877 When receiving an RTCP ECN feedback packet for the translated stream, 1878 an RTP translator first determines the range of packets to which the 1879 report corresponds. The extended highest sequence number in the RTCP 1880 ECN feedback packet (or in the RTCP SR/RR packet contained within the 1881 compound packet, in the case of RTCP XR ECN summary reports) 1882 specifies the end sequence number of the range. For the first RTCP 1883 ECN feedback packet received, the initial extended sequence number of 1884 the range may be determined by subtracting the sum of the lost 1885 packets counter, the ECN-CE counter, the not-ECT counter, the ECT(0) 1886 counter and the ECT(1) counter minus the duplication counter, from 1887 the extended highest sequence number. For subsequent RTCP ECN 1888 feedback packets, the starting sequence number may be determined as 1889 being one after the extended highest sequence number of the previous 1890 RTCP ECN feedback packet received from the same SSRC. These values 1891 are in the sequence number space of the translated packets. 1893 Based on its knowledge of the translation process, the translator 1894 determines the sequence number range for the corresponding original, 1895 pre-translation, packets. The extended highest sequence number in 1896 the RTCP ECN feedback packet is rewritten to match the final sequence 1897 number in the pre-translation sequence number range. 1899 The translator then determines the ratio, R, of the number of packets 1900 in the translated sequence number space (numTrans) to the number of 1901 packets in the pre-translation sequence number space (numOrig) such 1902 that R = numTrans / numOrig. The counter values in the RTCP ECN 1903 feedback report are then scaled by dividing each of them by R. For 1904 example, if the translation process combines two RTP packets into 1905 one, then numOrig will be twice numTrans, giving R=0.5, and the 1906 counters in the translated RTCP ECN feedback packet will be twice 1907 those in the original. 1909 The ratio, R, may have a value that leads to non-integer multiples of 1910 the counters when translating the RTCP packet. For example, a VoIP 1911 translator that combines two adjacent RTP packets into one if they 1912 contain active speech data, but passes comfort noise packets 1913 unchanged, would have an R values of between 0.5 and 1.0 depending on 1914 the amount of active speech. Since the counter values in the 1915 translated RTCP report are integer values, rounding will be necessary 1916 in this case. 1918 When rounding counter values in the translated RTCP packet, the 1919 translator should try to ensure that they sum to the number of RTP 1920 packets in the pre-translation sequence number space (numOrig). The 1921 translator should also try to ensure that no non-zero counter is 1922 rounded to a zero value, since that will lose information that a 1923 particular type of event has occurred. It is recognised that it may 1924 be impossible to satisfy both of these constraints; in such cases, it 1925 is better to ensure that no non-zero counter is mapped to a zero 1926 value, since this preserves congestion adaptation and helps the RTCP- 1927 based ECN initiation process. 1929 One should be aware of the impact this type of translators have on 1930 the measurement of packet duplication. A translator performing 1931 aggregation and most likely also an fragmenting translator will 1932 suppress any duplication happening prior to itself. Thus the reports 1933 and what is being scaled will only represent packet duplication 1934 happening from the translator to the receiver reporting on the flow. 1936 It should be noted that scaling the RTCP counter values in this way 1937 is meaningful only on the assumption that the level of congestion in 1938 the network is related to the number of packets being sent. This is 1939 likely to be a reasonable assumption in the type of environment where 1940 RTP translators that fragment or reassemble packets are deployed, as 1941 their entire purpose is to change the number of packets being sent to 1942 adapt to known limitations of the network, but is not necessarily 1943 valid in general. 1945 The rewritten RTCP ECN feedback report is sent from the other side of 1946 the translator to that which it arrived (as part of a compound RTCP 1947 packet containing other translated RTCP packets, where appropriate). 1949 8.3. Generating RTCP ECN Feedback in Media Transcoders 1951 An RTP translator that acts as a media transcoder cannot directly 1952 forward RTCP packets corresponding to the transcoded stream, since 1953 those packets will relate to the non-transcoded stream, and will not 1954 be useful in relation to the transcoded RTP flow. Such a transcoder 1955 will need to interpose itself into the RTCP flow, acting as a proxy 1956 for the receiver to generate RTCP feedback in the direction of the 1957 sender relating to the pre-transcoded stream, and acting in place of 1958 the sender to generate RTCP relating to the transcoded stream, to be 1959 sent towards the receiver. This section describes how this proxying 1960 is to be done for RTCP ECN feedback packets. Section 7.2 of 1961 [RFC3550] describes general procedures for other RTCP packet types. 1963 An RTP translator acting as a media transcoder in this manner does 1964 not have its own SSRC, and hence is not visible to other entities at 1965 the RTP layer. RTCP ECN feedback packets and RTCP XR report blocks 1966 for ECN summary information that are received from downstream relate 1967 to the translated stream, and so must be processed by the translator 1968 as if it were the original media source. These reports drive the 1969 congestion control loop and media adaptation between the translator 1970 and the downstream receiver. If there are multiple downstream 1971 receivers, a logically separate transcoder instance must be used for 1972 each receiver, and must process RTCP ECN feedback and summary reports 1973 independently to the other transcoder instances. An RTP translator 1974 acting as a media transcoder in this manner MUST NOT forward RTCP ECN 1975 feedback packets or RTCP XR ECN summary reports from downstream 1976 receivers in the upstream direction. 1978 An RTP translator acting as a media transcoder will generate RTCP 1979 reports upstream towards the original media sender, based on the 1980 reception quality of the original media stream at the translator. 1981 The translator will run a separate congestion control loop and media 1982 adaptation between itself and the media sender for each of its 1983 downstream receivers, and must generate RTCP ECN feedback packets and 1984 RTCP XR ECN summary reports for that congestion control loop using 1985 the SSRC of that downstream receiver. 1987 8.4. Generating RTCP ECN Feedback in Mixers 1989 An RTP mixer terminates one-or-more RTP flows, combines them into a 1990 single outgoing media stream, and transmits that new stream as a 1991 separate RTP flow. A mixer has its own SSRC, and is visible to other 1992 participants in the session at the RTP layer. 1994 An ECN-aware RTP mixer must generate RTCP ECN feedback packets and 1995 RTCP XR report blocks for ECN summary information relating to the RTP 1996 flows it terminates, in exactly the same way it would if it were an 1997 RTP receiver. These reports form part of the congestion control loop 1998 between the mixer and the media senders generating the streams it is 1999 mixing. A separate control loop runs between each sender and the 2000 mixer. 2002 An ECN-aware RTP mixer will negotiate and initiate the use of ECN on 2003 the mixed RTP flows it generates, and will accept and process RTCP 2004 ECN feedback reports and RTCP XR report blocks for ECN relating to 2005 those mixed flows as if it were a standard media sender. A 2006 congestion control loop runs between the mixer and its receivers, 2007 driven in part by the ECN reports received. 2009 An RTP mixer MUST NOT forward RTCP ECN feedback packets or RTCP XR 2010 ECN summary reports from downstream receivers in the upstream 2011 direction. 2013 9. Implementation considerations 2015 To allow the use of ECN with RTP over UDP, the RTP implementation 2016 should be able to set the ECT bits in outgoing UDP datagrams, and 2017 should be able to read the value of the ECT bits on received UDP 2018 datagrams. The standard Berkeley sockets API pre-dates the 2019 specification of ECN, and does not provide the functionality which is 2020 required for this mechanism to be used with UDP flows, making this 2021 specification difficult to implement portably. 2023 10. IANA Considerations 2025 Note to RFC Editor: please replace "RFC XXXX" below with the RFC 2026 number of this memo, and remove this note. 2028 10.1. SDP Attribute Registration 2030 Following the guidelines in [RFC4566], the IANA is requested to 2031 register one new SDP attribute: 2033 o Contact name, email address and telephone number: Authors of 2034 RFCXXXX 2036 o Attribute-name: ecn-capable-rtp 2038 o Type of attribute: media-level 2040 o Subject to charset: no 2042 This attribute defines the ability to negotiate the use of ECT (ECN 2043 capable transport) for RTP flows running over UDP/IP. This attribute 2044 should be put in the SDP offer if the offering party wishes to 2045 receive an ECT flow. The answering party should include the 2046 attribute in the answer if it wish to receive an ECT flow. If the 2047 answerer does not include the attribute then ECT MUST be disabled in 2048 both directions. 2050 10.2. RTP/AVPF Transport Layer Feedback Message 2052 The IANA is requested to register one new RTP/AVPF Transport Layer 2053 Feedback Message in the table of FMT values for RTPFB Payload Types 2054 [RFC4585] as defined in Section 5.1: 2056 Name: RTCP-ECN-FB 2057 Long name: RTCP ECN Feedback 2058 Value: TBA1 2059 Reference: RFC XXXX 2061 10.3. RTCP Feedback SDP Parameter 2063 The IANA is requested to register one new SDP "rtcp-fb" attribute 2064 "nack" parameter "ecn" in the SDP ("ack" and "nack" Attribute Values) 2065 registry. 2067 Value name: ecn 2068 Long name: Explicit Congestion Notification 2069 Usable with: nack 2070 Reference: RFC XXXX 2072 10.4. RTCP XR Report blocks 2074 The IANA is requested to register one new RTCP XR Block Type as 2075 defined in Section 5.2: 2077 Block Type: TBA2 2078 Name: ECN Summary Report 2079 Reference: RFC XXXX 2081 10.5. RTCP XR SDP Parameter 2083 The IANA is requested to register one new RTCP XR SDP Parameter "ecn- 2084 sum" in the "RTCP XR SDP Parameters" registry. 2086 Parameter name XR block (block type and name) 2087 -------------- ------------------------------------ 2088 ecn-sum TBA2 ECN Summary Report Block 2090 10.6. STUN attribute 2092 A new STUN [RFC5389] attribute in the Comprehension-optional range 2093 under IETF Review (0x0000 - 0x3FFF) is request to be assigned to the 2094 STUN attribute defined in Section 7.2.2. The STUN attribute registry 2095 can currently be found at: http://www.iana.org/assignments/ 2096 stun-parameters/stun-parameters.xhtml. 2098 10.7. ICE Option 2100 A new ICE option "rtp+ecn" is registered in the registry that "IANA 2101 Registry for Interactive Connectivity Establishment (ICE) Options" 2102 [I-D.ietf-mmusic-ice-options-registry] creates. 2104 11. Security Considerations 2106 The use of ECN with RTP over UDP as specified in this document has 2107 the following known security issues that need to be considered. 2109 External threats to the RTP and RTCP traffic: 2111 Denial of Service affecting RTCP: An attacker that can modify the 2112 traffic between the media sender and a receiver can achieve either 2113 of two things: 1) Report a lot of packets as being Congestion 2114 Experience marked, thus forcing the sender into a congestion 2115 response; or 2) Ensure that the sender disable the usage of ECN by 2116 reporting failures to receive ECN by changing the counter fields. 2117 This can also be accomplished by injecting false RTCP packets to 2118 the media sender. Reporting a lot of ECN-CE marked traffic is 2119 likely the more efficient denial of service tool as that may 2120 likely force the application to use lowest possible bit-rates. 2121 The prevention against an external threat is to integrity protect 2122 the RTCP feedback information and authenticate the sender. 2124 Information leakage: The ECN feedback mechanism exposes the 2125 receivers perceived packet loss, what packets it considers to be 2126 ECN-CE marked and its calculation of the ECN-none. This is mostly 2127 not considered as sensitive information. If it is considered 2128 sensitive the RTCP feedback should be encrypted. 2130 Changing the ECN bits: An on-path attacker that sees the RTP packet 2131 flow from sender to receiver and who has the capability to change 2132 the packets can rewrite ECT into ECN-CE thus forcing the sender or 2133 receiver to take congestion control response. This denial of 2134 service against the media quality in the RTP session is impossible 2135 for an end-point to protect itself against. Only network 2136 infrastructure nodes can detect this illicit re-marking. It will 2137 be mitigated by turning off ECN, however, if the attacker can 2138 modify its response to drop packets the same vulnerability exist. 2140 Denial of Service affecting the session set-up signalling: If an 2141 attacker can modify the session signalling it can prevent the 2142 usage of ECN by removing the signalling attributes used to 2143 indicate that the initiator is capable and willing to use ECN with 2144 RTP/UDP. This attack can be prevented by authentication and 2145 integrity protection of the signalling. We do note that any 2146 attacker that can modify the signalling has more interesting 2147 attacks they can perform than prevent the usage of ECN, like 2148 inserting itself as a middleman in the media flows enabling wire- 2149 tapping also for an off-path attacker. 2151 The following are threats that exist from misbehaving senders or 2152 receivers: 2154 Receivers cheating: A receiver may attempt to cheat and fail to 2155 report reception of ECN-CE marked packets. The benefit for a 2156 receiver cheating in its reporting would be to get an unfair bit- 2157 rate share across the resource bottleneck. It is far from certain 2158 that a receiver would be able to get a significant larger share of 2159 the resources. That assumes a high enough level of aggregation 2160 that there are flows to acquire shares from. The risk of cheating 2161 is that failure to react to congestion results in packet loss and 2162 increased path delay. 2164 Receivers misbehaving: A receiver may prevent the usage of ECN in an 2165 RTP session by reporting itself as non ECN capable, forcing the 2166 sender to turn off usage of ECN. In a point-to-point scenario 2167 there is little incentive to do this as it will only affect the 2168 receiver. Thus failing to utilise an optimisation. For multi- 2169 party session there exist some motivation why a receiver would 2170 misbehave as it can prevent also the other receivers from using 2171 ECN. As an insider into the session it is difficult to determine 2172 if a receiver is misbehaving or simply incapable, making it 2173 basically impossible in the incremental deployment phase of ECN 2174 for RTP usage to determine this. If additional information about 2175 the receivers and the network is known it might be possible to 2176 deduce that a receiver is misbehaving. If it can be determined 2177 that a receiver is misbehaving, the only response is to exclude it 2178 from the RTP session and ensure that is does not any longer have 2179 any valid security context to affect the session. 2181 Misbehaving Senders: The enabling of ECN gives the media packets a 2182 higher degree of probability to reach the receiver compared to 2183 not-ECT marked ones on a ECN capable path. However, this is no 2184 magic bullet and failure to react to congestion will most likely 2185 only slightly delay a network buffer over-run, in which its 2186 session also will experience packet loss and increased delay. 2187 There is some possibility that the media senders traffic will push 2188 other traffic out of the way without being affected too 2189 negatively. However, we do note that a media sender still needs 2190 to implement congestion control functions to prevent the media 2191 from being badly affected by congestion events. Thus the 2192 misbehaving sender is getting a unfair share. This can only be 2193 detected and potentially prevented by network monitoring and 2194 administrative entities. See Section 7 of [RFC3168] for more 2195 discussion of this issue. 2197 We note that the end-point security functions needed to prevent an 2198 external attacker from inferring with the signalling are source 2199 authentication and integrity protection. To prevent information 2200 leakage from the feedback packets encryption of the RTCP is also 2201 needed. For RTP there exist multiple solutions possible depending on 2202 the application context. Secure RTP (SRTP) [RFC3711] does satisfy 2203 the requirement to protect this mechanism despite only providing 2204 authentication if a entity is within the security context or not. 2205 IPsec [RFC4301] and DTLS [RFC4347] can also provide the necessary 2206 security functions. 2208 The signalling protocols used to initiate an RTP session also need to 2209 be source authenticated and integrity protected to prevent an 2210 external attacker from modifying any signalling. Here an appropriate 2211 mechanism to protect the used signalling needs to be used. For SIP/ 2212 SDP ideally S/MIME [RFC5751] would be used. However, with the 2213 limited deployment a minimal mitigation strategy is to require use of 2214 SIPS (SIP over TLS) [RFC3261] [RFC5630] to at least accomplish hop- 2215 by-hop protection. 2217 We do note that certain mitigation methods will require network 2218 functions. 2220 12. Examples of SDP Signalling 2222 This section contain a few different examples of the signalling 2223 mechanism defined in this specification in an SDP context. If there 2224 are discrepancies between these examples and the specification text, 2225 the specification text is definitive. 2227 12.1. Basic SDP Offer/Answer 2229 This example is a basic offer/answer SDP exchange, assumed done by 2230 SIP (not shown). The intention is to establish a basic audio session 2231 point to point between two users. 2233 The Offer: 2235 v=0 2236 o=jdoe 3502844782 3502844782 IN IP4 10.0.1.4 2237 s=VoIP call 2238 i=SDP offer for VoIP call with ICE and ECN for RTP 2239 b=AS:128 2240 b=RR:2000 2241 b=RS:2500 2242 a=ice-pwd:YH75Fviy6338Vbrhrlp8Yh 2243 a=ice-ufrag:9uB6 2244 a=ice-options:rtp+ecn 2245 t=0 0 2246 m=audio 45664 RTP/AVPF 97 98 99 2247 c=IN IP4 192.0.2.3 2248 a=rtpmap:97 G719/48000/1 2249 a=fmtp:97 maxred=160 2250 a=rtpmap:98 AMR-WB/16000/1 2251 a=fmtp:98 octet-align=1; mode-change-capability=2 2252 a=rtpmap:99 PCMA/8000/1 2253 a=maxptime:160 2254 a=ptime:20 2255 a=ecn-capable-rtp: ice rtp ect=0 mode=setread 2256 a=rtcp-fb:* nack ecn 2257 a=rtcp-fb:* trr-int 1000 2258 a=rtcp-xr:ecn-sum 2259 a=rtcp-rsize 2260 a=candidate:1 1 UDP 2130706431 10.0.1.4 8998 typ host 2261 a=candidate:2 1 UDP 1694498815 192.0.2.3 45664 typ srflx raddr 2262 10.0.1.4 rport 8998 2264 This SDP offer offers a single media stream with 3 media payload 2265 types. It proposes to use ECN with RTP, with the ICE based 2266 initialization as being preferred over the RTP/RTCP one. Leap of 2267 faith is not suggested to be used. The offerer is capable of both 2268 setting and reading the ECN bits. In addition the RTCP ECN feedback 2269 packet is configured and the RTCP XR ECN summary report. ICE is also 2270 proposed with two candidates. It also supports reduced size RTCP and 2271 are willing to use it. 2273 The Answer: 2275 v=0 2276 o=jdoe 3502844783 3502844783 IN IP4 198.51.100.235 2277 s=VoIP call 2278 i=SDP offer for VoIP call with ICE and ECN for RTP 2279 b=AS:128 2280 b=RR:2000 2281 b=RS:2500 2282 a=ice-pwd:asd88fgpdd777uzjYhagZg 2283 a=ice-ufrag:8hhY 2284 a=ice-options:rtp+ecn 2285 t=0 0 2286 m=audio 53879 RTP/AVPF 97 99 2287 c=IN IP4 198.51.100.235 2288 a=rtpmap:97 G719/48000/1 2289 a=fmtp:97 maxred=160 2290 a=rtpmap:99 PCMA/8000/1 2291 a=maxptime:160 2292 a=ptime:20 2293 a=ecn-capable-rtp: ice ect=0 mode=readonly 2294 a=rtcp-fb:* nack ecn 2295 a=rtcp-fb:* trr-int 1000 2296 a=rtcp-xr:ecn-sum 2297 a=candidate:1 1 UDP 2130706431 198.51.100.235 53879 typ host 2299 The answer confirms that only one media stream will be used. One RTP 2300 Payload type was removed. ECN capability was confirmed, and the 2301 initialization method will be ICE. However, the answerer is only 2302 capable of reading the ECN bits, which means that ECN can only be 2303 used for RTP flowing from the offerer to the answerer. ECT always 2304 set to 0 will be used in both directions. Both the RTCP ECN feedback 2305 packet and the RTCP XR ECN summary report will be used. Reduced size 2306 RTCP will not be used as the answerer has not indicated support for 2307 it in the answer. 2309 12.2. Declarative Multicast SDP 2311 The below session describes an any source multicast using session 2312 with a single media stream. 2314 v=0 2315 o=jdoe 3502844782 3502844782 IN IP4 198.51.100.235 2316 s=Multicast SDP session using ECN for RTP 2317 i=Multicasted audio chat using ECN for RTP 2318 b=AS:128 2319 t=3502892703 3502910700 2320 m=audio 56144 RTP/AVPF 97 2321 c=IN IP4 233.252.0.212/127 2322 a=rtpmap:97 g719/48000/1 2323 a=fmtp:97 maxred=160 2324 a=maxptime:160 2325 a=ptime:20 2326 a=ecn-capable-rtp: rtp mode=readonly; ect=0 2327 a=rtcp-fb:* nack ecn 2328 a=rtcp-fb:* trr-int 1500 2329 a=rtcp-xr:ecn-sum 2331 In the above example, as this is declarative we need to require 2332 certain functionality. As it is ASM the initialization method that 2333 can work here is the RTP/RTCP based one. So that is indicated. The 2334 ECN setting and reading capability to take part of this session is at 2335 least read. If one is capable of setting that is good, but not 2336 required as one can skip using ECN for anything one sends oneself. 2337 The ECT value is recommended to be set to 0 always. The ECN usage in 2338 this session requires both ECN feedback and the XR ECN summary 2339 report, so their use is also indicated. 2341 13. Acknowledgments 2343 The authors wish to thank the following persons for their reviews and 2344 comments: Thomas Belling, Bob Briscoe, Roni Even, Thomas Frankkila, 2345 Christian Groves, Cullen Jennings Tom Van Caenegem, Simo 2346 Veikkolainen, Lei Zhu, and Christer Holmgren. 2348 14. References 2350 14.1. Normative References 2352 [I-D.ietf-mmusic-ice-options-registry] 2353 Westerlund, M. and C. Perkins, "IANA Registry for 2354 Interactive Connectivity Establishment (ICE) Options", 2355 draft-ietf-mmusic-ice-options-registry-02 (work in 2356 progress), May 2011. 2358 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 2359 Requirement Levels", BCP 14, RFC 2119, March 1997. 2361 [RFC2762] Rosenberg, J. and H. Schulzrinne, "Sampling of the Group 2362 Membership in RTP", RFC 2762, February 2000. 2364 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 2365 of Explicit Congestion Notification (ECN) to IP", 2366 RFC 3168, September 2001. 2368 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 2369 Jacobson, "RTP: A Transport Protocol for Real-Time 2370 Applications", STD 64, RFC 3550, July 2003. 2372 [RFC3611] Friedman, T., Caceres, R., and A. Clark, "RTP Control 2373 Protocol Extended Reports (RTCP XR)", RFC 3611, 2374 November 2003. 2376 [RFC5234] Crocker, D. and P. Overell, "Augmented BNF for Syntax 2377 Specifications: ABNF", STD 68, RFC 5234, January 2008. 2379 [RFC5245] Rosenberg, J., "Interactive Connectivity Establishment 2380 (ICE): A Protocol for Network Address Translator (NAT) 2381 Traversal for Offer/Answer Protocols", RFC 5245, 2382 April 2010. 2384 [RFC5348] Floyd, S., Handley, M., Padhye, J., and J. Widmer, "TCP 2385 Friendly Rate Control (TFRC): Protocol Specification", 2386 RFC 5348, September 2008. 2388 [RFC5389] Rosenberg, J., Mahy, R., Matthews, P., and D. Wing, 2389 "Session Traversal Utilities for NAT (STUN)", RFC 5389, 2390 October 2008. 2392 14.2. Informative References 2394 [I-D.ietf-avt-rtp-no-op] 2395 Andreasen, F., "A No-Op Payload Format for RTP", 2396 draft-ietf-avt-rtp-no-op-04 (work in progress), May 2007. 2398 [RFC1112] Deering, S., "Host extensions for IP multicasting", STD 5, 2399 RFC 1112, August 1989. 2401 [RFC2974] Handley, M., Perkins, C., and E. Whelan, "Session 2402 Announcement Protocol", RFC 2974, October 2000. 2404 [RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, 2405 A., Peterson, J., Sparks, R., Handley, M., and E. 2406 Schooler, "SIP: Session Initiation Protocol", RFC 3261, 2407 June 2002. 2409 [RFC3264] Rosenberg, J. and H. Schulzrinne, "An Offer/Answer Model 2410 with Session Description Protocol (SDP)", RFC 3264, 2411 June 2002. 2413 [RFC3540] Spring, N., Wetherall, D., and D. Ely, "Robust Explicit 2414 Congestion Notification (ECN) Signaling with Nonces", 2415 RFC 3540, June 2003. 2417 [RFC3551] Schulzrinne, H. and S. Casner, "RTP Profile for Audio and 2418 Video Conferences with Minimal Control", STD 65, RFC 3551, 2419 July 2003. 2421 [RFC3569] Bhattacharyya, S., "An Overview of Source-Specific 2422 Multicast (SSM)", RFC 3569, July 2003. 2424 [RFC3711] Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K. 2425 Norrman, "The Secure Real-time Transport Protocol (SRTP)", 2426 RFC 3711, March 2004. 2428 [RFC4301] Kent, S. and K. Seo, "Security Architecture for the 2429 Internet Protocol", RFC 4301, December 2005. 2431 [RFC4340] Kohler, E., Handley, M., and S. Floyd, "Datagram 2432 Congestion Control Protocol (DCCP)", RFC 4340, March 2006. 2434 [RFC4347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 2435 Security", RFC 4347, April 2006. 2437 [RFC4566] Handley, M., Jacobson, V., and C. Perkins, "SDP: Session 2438 Description Protocol", RFC 4566, July 2006. 2440 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 2441 "Extended RTP Profile for Real-time Transport Control 2442 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, 2443 July 2006. 2445 [RFC4588] Rey, J., Leon, D., Miyazaki, A., Varsa, V., and R. 2446 Hakenberg, "RTP Retransmission Payload Format", RFC 4588, 2447 July 2006. 2449 [RFC4607] Holbrook, H. and B. Cain, "Source-Specific Multicast for 2450 IP", RFC 4607, August 2006. 2452 [RFC4960] Stewart, R., "Stream Control Transmission Protocol", 2453 RFC 4960, September 2007. 2455 [RFC5124] Ott, J. and E. Carrara, "Extended Secure RTP Profile for 2456 Real-time Transport Control Protocol (RTCP)-Based Feedback 2457 (RTP/SAVPF)", RFC 5124, February 2008. 2459 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 2460 Real-Time Transport Control Protocol (RTCP): Opportunities 2461 and Consequences", RFC 5506, April 2009. 2463 [RFC5630] Audet, F., "The Use of the SIPS URI Scheme in the Session 2464 Initiation Protocol (SIP)", RFC 5630, October 2009. 2466 [RFC5751] Ramsdell, B. and S. Turner, "Secure/Multipurpose Internet 2467 Mail Extensions (S/MIME) Version 3.2 Message 2468 Specification", RFC 5751, January 2010. 2470 [RFC5760] Ott, J., Chesterfield, J., and E. Schooler, "RTP Control 2471 Protocol (RTCP) Extensions for Single-Source Multicast 2472 Sessions with Unicast Feedback", RFC 5760, February 2010. 2474 [RFC6189] Zimmermann, P., Johnston, A., and J. Callas, "ZRTP: Media 2475 Path Key Agreement for Unicast Secure RTP", RFC 6189, 2476 April 2011. 2478 Authors' Addresses 2480 Magnus Westerlund 2481 Ericsson 2482 Farogatan 6 2483 SE-164 80 Kista 2484 Sweden 2486 Phone: +46 10 714 82 87 2487 Email: magnus.westerlund@ericsson.com 2489 Ingemar Johansson 2490 Ericsson 2491 Laboratoriegrand 11 2492 SE-971 28 Lulea 2493 SWEDEN 2495 Phone: +46 73 0783289 2496 Email: ingemar.s.johansson@ericsson.com 2497 Colin Perkins 2498 University of Glasgow 2499 School of Computing Science 2500 Glasgow G12 8QQ 2501 United Kingdom 2503 Email: csp@csperkins.org 2505 Piers O'Hanlon 2506 University College London 2507 Computer Science Department 2508 Gower Street 2509 London WC1E 6BT 2510 United Kingdom 2512 Email: p.ohanlon@cs.ucl.ac.uk 2514 Ken Carlberg 2515 G11 2516 1600 Clarendon Blvd 2517 Arlington VA 2518 USA 2520 Email: carlberg@g11.org.uk