idnits 2.17.1 draft-ietf-avtcore-rtp-circuit-breakers-11.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- -- The draft header indicates that this document updates RFC3550, but the abstract doesn't seem to mention this, which it should. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year (Using the creation date from RFC3550, updated by this document, for RFC5378 checks: 1998-04-07) -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (October 16, 2015) is 3087 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references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 3448 (Obsoleted by RFC 5348) == Outdated reference: A later version (-11) exists of draft-ietf-avtcore-rtp-multi-stream-09 == Outdated reference: A later version (-15) exists of draft-ietf-tsvwg-circuit-breaker-05 -- Obsolete informational reference (is this intentional?): RFC 5405 (Obsoleted by RFC 8085) Summary: 1 error (**), 0 flaws (~~), 3 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 AVTCORE Working Group C. Perkins 3 Internet-Draft University of Glasgow 4 Updates: 3550 (if approved) V. Singh 5 Intended status: Standards Track Aalto University 6 Expires: April 18, 2016 October 16, 2015 8 Multimedia Congestion Control: Circuit Breakers for Unicast RTP Sessions 9 draft-ietf-avtcore-rtp-circuit-breakers-11 11 Abstract 13 The Real-time Transport Protocol (RTP) is widely used in telephony, 14 video conferencing, and telepresence applications. Such applications 15 are often run on best-effort UDP/IP networks. If congestion control 16 is not implemented in the applications, then network congestion will 17 deteriorate the user's multimedia experience. This document does not 18 propose a congestion control algorithm; instead, it defines a minimal 19 set of RTP circuit-breakers. Circuit-breakers are conditions under 20 which an RTP sender needs to stop transmitting media data in order to 21 protect the network from excessive congestion. It is expected that, 22 in the absence of severe congestion, all RTP applications running on 23 best-effort IP networks will be able to run without triggering these 24 circuit breakers. Any future RTP congestion control specification 25 will be expected to operate within the constraints defined by these 26 circuit breakers. 28 Status of This Memo 30 This Internet-Draft is submitted in full conformance with the 31 provisions of BCP 78 and BCP 79. 33 Internet-Drafts are working documents of the Internet Engineering 34 Task Force (IETF). Note that other groups may also distribute 35 working documents as Internet-Drafts. The list of current Internet- 36 Drafts is at http://datatracker.ietf.org/drafts/current/. 38 Internet-Drafts are draft documents valid for a maximum of six months 39 and may be updated, replaced, or obsoleted by other documents at any 40 time. It is inappropriate to use Internet-Drafts as reference 41 material or to cite them other than as "work in progress." 43 This Internet-Draft will expire on April 18, 2016. 45 Copyright Notice 47 Copyright (c) 2015 IETF Trust and the persons identified as the 48 document authors. All rights reserved. 50 This document is subject to BCP 78 and the IETF Trust's Legal 51 Provisions Relating to IETF Documents 52 (http://trustee.ietf.org/license-info) in effect on the date of 53 publication of this document. Please review these documents 54 carefully, as they describe your rights and restrictions with respect 55 to this document. Code Components extracted from this document must 56 include Simplified BSD License text as described in Section 4.e of 57 the Trust Legal Provisions and are provided without warranty as 58 described in the Simplified BSD License. 60 Table of Contents 62 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 63 2. Background . . . . . . . . . . . . . . . . . . . . . . . . . 3 64 3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 6 65 4. RTP Circuit Breakers for Systems Using the RTP/AVP Profile . 7 66 4.1. RTP/AVP Circuit Breaker #1: RTCP Timeout . . . . . . . . 9 67 4.2. RTP/AVP Circuit Breaker #2: Media Timeout . . . . . . . . 10 68 4.3. RTP/AVP Circuit Breaker #3: Congestion . . . . . . . . . 11 69 4.4. RTP/AVP Circuit Breaker #4: Media Usability . . . . . . . 15 70 4.5. Ceasing Transmission . . . . . . . . . . . . . . . . . . 16 71 5. RTP Circuit Breakers and the RTP/AVPF and RTP/SAVPF Profiles 17 72 6. Impact of RTCP Extended Reports (XR) . . . . . . . . . . . . 18 73 7. Impact of Explicit Congestion Notification (ECN) . . . . . . 18 74 8. Impact of Bundled Media and Layered Coding . . . . . . . . . 18 75 9. Security Considerations . . . . . . . . . . . . . . . . . . . 19 76 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 77 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 20 78 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 20 79 12.1. Normative References . . . . . . . . . . . . . . . . . . 20 80 12.2. Informative References . . . . . . . . . . . . . . . . . 20 81 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 23 83 1. Introduction 85 The Real-time Transport Protocol (RTP) [RFC3550] is widely used in 86 voice-over-IP, video teleconferencing, and telepresence systems. 87 Many of these systems run over best-effort UDP/IP networks, and can 88 suffer from packet loss and increased latency if network congestion 89 occurs. Designing effective RTP congestion control algorithms, to 90 adapt the transmission of RTP-based media to match the available 91 network capacity, while also maintaining the user experience, is a 92 difficult but important problem. Many such congestion control and 93 media adaptation algorithms have been proposed, but to date there is 94 no consensus on the correct approach, or even that a single standard 95 algorithm is desirable. 97 This memo does not attempt to propose a new RTP congestion control 98 algorithm. Instead, we propose a small set of RTP circuit breakers. 99 These are conditions under which there is general agreement that an 100 RTP flow is causing serious congestion, and hence ought to cease 101 transmission. The RTP circuit breakers proposed in this memo are a 102 specific instance of the general class of network transport circuit 103 breakers [I-D.ietf-tsvwg-circuit-breaker], designed to act as a 104 protection mechanism of last resort to avoid persistent congestion. 105 It is expected that future standards-track congestion control 106 algorithms for RTP will operate within the envelope defined by this 107 memo. 109 2. Background 111 We consider congestion control for unicast RTP traffic flows. This 112 is the problem of adapting the transmission of an audio/visual data 113 flow, encapsulated within an RTP transport session, from one sender 114 to one receiver, so that it matches the available network bandwidth. 115 Such adaptation needs to be done in a way that limits the disruption 116 to the user experience caused by both packet loss and excessive rate 117 changes. Congestion control for multicast flows is outside the scope 118 of this memo. Multicast traffic needs different solutions, since the 119 available bandwidth estimator for a group of receivers will differ 120 from that for a single receiver, and because multicast congestion 121 control has to consider issues of fairness across groups of receivers 122 that do not apply to unicast flows. 124 Congestion control for unicast RTP traffic can be implemented in one 125 of two places in the protocol stack. One approach is to run the RTP 126 traffic over a congestion controlled transport protocol, for example 127 over TCP, and to adapt the media encoding to match the dictates of 128 the transport-layer congestion control algorithm. This is safe for 129 the network, but can be suboptimal for the media quality unless the 130 transport protocol is designed to support real-time media flows. We 131 do not consider this class of applications further in this memo, as 132 their network safety is guaranteed by the underlying transport. 134 Alternatively, RTP flows can be run over a non-congestion controlled 135 transport protocol, for example UDP, performing rate adaptation at 136 the application layer based on RTP Control Protocol (RTCP) feedback. 137 With a well-designed, network-aware, application, this allows highly 138 effective media quality adaptation, but there is potential to disrupt 139 the network's operation if the application does not adapt its sending 140 rate in a timely and effective manner. We consider this class of 141 applications in this memo. 143 Congestion control relies on monitoring the delivery of a media flow, 144 and responding to adapt the transmission of that flow when there are 145 signs that the network path is congested. Network congestion can be 146 detected in one of three ways: 1) a receiver can infer the onset of 147 congestion by observing an increase in one-way delay caused by queue 148 build-up within the network; 2) if Explicit Congestion Notification 149 (ECN) [RFC3168] is supported, the network can signal the presence of 150 congestion by marking packets using ECN Congestion Experienced (CE) 151 marks; or 3) in the extreme case, congestion will cause packet loss 152 that can be detected by observing a gap in the received RTP sequence 153 numbers. 155 Once the onset of congestion is observed, the receiver has to send 156 feedback to the sender to indicate that the transmission rate needs 157 to be reduced. How the sender reduces the transmission rate is 158 highly dependent on the media codec being used, and is outside the 159 scope of this memo. 161 There are several ways in which a receiver can send feedback to a 162 media sender within the RTP framework: 164 o The base RTP specification [RFC3550] defines RTCP Reception Report 165 (RR) packets to convey reception quality feedback information, and 166 Sender Report (SR) packets to convey information about the media 167 transmission. RTCP SR packets contain data that can be used to 168 reconstruct media timing at a receiver, along with a count of the 169 total number of octets and packets sent. RTCP RR packets report 170 on the fraction of packets lost in the last reporting interval, 171 the cumulative number of packets lost, the highest sequence number 172 received, and the inter-arrival jitter. The RTCP RR packets also 173 contain timing information that allows the sender to estimate the 174 network round trip time (RTT) to the receivers. RTCP reports are 175 sent periodically, with the reporting interval being determined by 176 the number of SSRCs used in the session and a configured session 177 bandwidth estimate (the number of SSRCs used is usually two in a 178 unicast session, one for each participant, but can be greater if 179 the participants send multiple media streams). The interval 180 between reports sent from each receiver tends to be on the order 181 of a few seconds on average, although it varies with the session 182 bandwidth, and sub-second reporting intervals are possible in high 183 bandwidth sessions, and it is randomised to avoid synchronisation 184 of reports from multiple receivers. RTCP RR packets allow a 185 receiver to report ongoing network congestion to the sender. 186 However, if a receiver detects the onset of congestion part way 187 through a reporting interval, the base RTP specification contains 188 no provision for sending the RTCP RR packet early, and the 189 receiver has to wait until the next scheduled reporting interval. 191 o The RTCP Extended Reports (XR) [RFC3611] allow reporting of more 192 complex and sophisticated reception quality metrics, but do not 193 change the RTCP timing rules. RTCP extended reports of potential 194 interest for congestion control purposes are the extended packet 195 loss, discard, and burst metrics [RFC3611], [RFC7002], [RFC7097], 196 [RFC7003], [RFC6958]; and the extended delay metrics [RFC6843], 197 [RFC6798]. Other RTCP Extended Reports that could be helpful for 198 congestion control purposes might be developed in future. 200 o Rapid feedback about the occurrence of congestion events can be 201 achieved using the Extended RTP Profile for RTCP-Based Feedback 202 (RTP/AVPF) [RFC4585] (or its secure variant, RTP/SAVPF [RFC5124]) 203 in place of the RTP/AVP profile [RFC3551]. This modifies the RTCP 204 timing rules to allow RTCP reports to be sent early, in some cases 205 immediately, provided the RTCP transmission rate keeps within its 206 bandwidth allocation. It also defines transport-layer feedback 207 messages, including negative acknowledgements (NACKs), that can be 208 used to report on specific congestion events. RTP Codec Control 209 Messages [RFC5104] extend the RTP/AVPF profile with additional 210 feedback messages that can be used to influence that way in which 211 rate adaptation occurs, but do not further change the dynamics of 212 how rapidly feedback can be sent. Use of the RTP/AVPF profile is 213 dependent on signalling. 215 o Finally, Explicit Congestion Notification (ECN) for RTP over UDP 216 [RFC6679] can be used to provide feedback on the number of packets 217 that received an ECN Congestion Experienced (CE) mark. This RTCP 218 extension builds on the RTP/AVPF profile to allow rapid congestion 219 feedback when ECN is supported. 221 In addition to these mechanisms for providing feedback, the sender 222 can include an RTP header extension in each packet to record packet 223 transmission times [RFC5450]. Accurate transmission timestamps can 224 be helpful for estimating queuing delays, to get an early indication 225 of the onset of congestion. 227 Taken together, these various mechanisms allow receivers to provide 228 feedback on the senders when congestion events occur, with varying 229 degrees of timeliness and accuracy. The key distinction is between 230 systems that use only the basic RTCP mechanisms, without RTP/AVPF 231 rapid feedback, and those that use the RTP/AVPF extensions to respond 232 to congestion more rapidly. 234 3. Terminology 236 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 237 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 238 document are to be interpreted as described in RFC 2119 [RFC2119]. 239 This interpretation of these key words applies only when written in 240 ALL CAPS. Mixed- or lower-case uses of these key words are not to be 241 interpreted as carrying special significance in this memo. 243 The definition of the RTP circuit breaker is specified in terms of 244 the following variables: 246 o Td is the deterministic RTCP reporting interval, as defined in 247 Section 6.3.1 of [RFC3550]. 249 o Tdr is the sender's estimate of the deterministic RTCP reporting 250 interval, Td, calculated by a receiver of the data it is sending. 251 Tdr is not known at the sender, but can be estimated by executing 252 the algorithm in Section 6.2 of [RFC3550] using the average RTCP 253 packet size seen at the sender, the number of members reported in 254 the receiver's SR/RR report blocks, and whether the receiver is 255 sending SR or RR packets. Tdr is recalculated when each new RTCP 256 SR/RR report is received, but the media timeout circuit breaker 257 (see Section 4.2) is only reconsidered when Tdr increases. 259 o Tr is the network round-trip time, calculated by the sender using 260 the algorithm in Section 6.4.1 of [RFC3550] and smoothed using an 261 exponentially weighted moving average as Tr = (0.8 * Tr) + (0.2 * 262 Tr_new) where Tr_new is the latest RTT estimate obtained from an 263 RTCP report. The weight is chosen so old estimates decay over k 264 intervals. 266 o k is the non-reporting threshold (see Section 4.2). 268 o Tf is the media framing interval at the sender. For applications 269 sending at a constant frame rate, Tf is the inter-frame interval. 270 For applications that switch between a small set of possible frame 271 rates, for example when sending speech with comfort noise, where 272 comfort noise frames are sent less often than speech frames, Tf is 273 set to the longest of the inter-frame intervals of the different 274 frame rates. For applications that send periodic frames but 275 dynamically vary their frame rate, Tf is set to the largest inter- 276 frame interval used in the last 10 seconds. For applications that 277 send less than one frame every 10 seconds, or that have no concept 278 of periodic frames (e.g., text conversation [RFC4103], or pointer 279 events [RFC2862]), Tf is set to the time interval since the 280 previous frame when each frame is sent. 282 o G is the frame group size. That is, the number of frames that are 283 coded together based on a particular sending rate setting. If the 284 codec used by the sender can change its rate on each frame, G = 1; 285 otherwise G is set to the number of frames before the codec can 286 adjust to the new rate. For codecs that have the concept of a 287 group-of-pictures (GoP), G is likely the GoP length. 289 o T_rr_interval is the minimal interval between RTCP reports, as 290 defined in Section 3.4 of [RFC4585]; it is only meaningful for 291 implementations of RTP/AVPF profile [RFC4585] or the RTP/SAVPF 292 profile [RFC5124]. 294 o X is estimated throughput a TCP connection would achieve over a 295 path, in bytes per second. 297 o s is the size of RTP packets being sent, in bytes. If the RTP 298 packets being sent vary in size, then the average size over the 299 packet comprising the last 4 * G frames MUST be used (this is 300 intended to be comparable to the four loss intervals used in 301 [RFC5348]). 303 o p is the loss event rate, between 0.0 and 1.0, that would be seen 304 by a TCP connection over a particular path. When used in the RTP 305 congestion circuit breaker, this is approximated as described in 306 Section 4.3. 308 o t_RTO is the retransmission timeout value that would be used by a 309 TCP connection over a particular path, in seconds. This MUST be 310 approximated using t_RTO = 4 * Tr when used as part of the RTP 311 congestion circuit breaker. 313 o b is the number of packets that are acknowledged by a single TCP 314 acknowledgement. Following [RFC3448], it is RECOMMENDED that the 315 value b = 1 is used as part of the RTP congestion circuit breaker. 317 4. RTP Circuit Breakers for Systems Using the RTP/AVP Profile 319 The feedback mechanisms defined in [RFC3550] and available under the 320 RTP/AVP profile [RFC3551] are the minimum that can be assumed for a 321 baseline circuit breaker mechanism that is suitable for all unicast 322 applications of RTP. Accordingly, for an RTP circuit breaker to be 323 useful, it needs to be able to detect that an RTP flow is causing 324 excessive congestion using only basic RTCP features, without needing 325 RTCP XR feedback or the RTP/AVPF profile for rapid RTCP reports. 327 RTCP is a fundamental part of the RTP protocol, and the mechanisms 328 described here rely on the implementation of RTCP. Implementations 329 that claim to support RTP, but that do not implement RTCP, cannot use 330 the circuit breaker mechanisms described in this memo. Such 331 implementations SHOULD NOT be used on networks that might be subject 332 to congestion unless equivalent mechanisms are defined using some 333 non-RTCP feedback channel to report congestion and signal circuit 334 breaker conditions. The RTCP timeout circuit breaker (Section 4.1) 335 will trigger if an implementation of this memo attempts to interwork 336 with an endpoint that does not support RTCP. Implementations that 337 sometimes need to interwork with endpoints that do not support RTCP 338 need to disable the RTP circuit breakers if they don't receive some 339 confirmation via signalling that the remote endpoint implements RTCP 340 (the presence of an SDP "a=rtcp:" attribute in an answer might be 341 such an indication). 343 Three potential congestion signals are available from the basic RTCP 344 SR/RR packets and are reported for each synchronisation source (SSRC) 345 in the RTP session: 347 1. The sender can estimate the network round-trip time once per RTCP 348 reporting interval, based on the contents and timing of RTCP SR 349 and RR packets. 351 2. Receivers report a jitter estimate (the statistical variance of 352 the RTP data packet inter-arrival time) calculated over the RTCP 353 reporting interval. Due to the nature of the jitter calculation 354 ([RFC3550], section 6.4.4), the jitter is only meaningful for RTP 355 flows that send a single data packet for each RTP timestamp value 356 (i.e., audio flows, or video flows where each packet comprises 357 one video frame). 359 3. Receivers report the fraction of RTP data packets lost during the 360 RTCP reporting interval, and the cumulative number of RTP packets 361 lost over the entire RTP session. 363 These congestion signals limit the possible circuit breakers, since 364 they give only limited visibility into the behaviour of the network. 366 RTT estimates are widely used in congestion control algorithms, as a 367 proxy for queuing delay measures in delay-based congestion control or 368 to determine connection timeouts. RTT estimates derived from RTCP SR 369 and RR packets sent according to the RTP/AVP timing rules are too 370 infrequent to be useful for congestion control, and don't give enough 371 information to distinguish a delay change due to routing updates from 372 queuing delay caused by congestion. Accordingly, we cannot use the 373 RTT estimate alone as an RTP circuit breaker. 375 Increased jitter can be a signal of transient network congestion, but 376 in the highly aggregated form reported in RTCP RR packets, it offers 377 insufficient information to estimate the extent or persistence of 378 congestion. Jitter reports are a useful early warning of potential 379 network congestion, but provide an insufficiently strong signal to be 380 used as a circuit breaker. 382 The remaining congestion signals are the packet loss fraction and the 383 cumulative number of packets lost. If considered carefully, these 384 can be effective indicators that congestion is occurring in networks 385 where packet loss is primarily due to queue overflows, although loss 386 caused by non-congestive packet corruption can distort the result in 387 some networks. TCP congestion control [RFC5681] intentionally tries 388 to fill the router queues, and uses the resulting packet loss as 389 congestion feedback. An RTP flow competing with TCP traffic will 390 therefore expect to see a non-zero packet loss fraction that has to 391 be related to TCP dynamics to estimate available capacity. This 392 behaviour of TCP is reflected in the congestion circuit breaker 393 below, and will affect the design of any RTP congestion control 394 protocol. 396 Two packet loss regimes can be observed: 1) RTCP RR packets show a 397 non-zero packet loss fraction, while the extended highest sequence 398 number received continues to increment; and 2) RR packets show a loss 399 fraction of zero, but the extended highest sequence number received 400 does not increment even though the sender has been transmitting RTP 401 data packets. The former corresponds to the TCP congestion avoidance 402 state, and indicates a congested path that is still delivering data; 403 the latter corresponds to a TCP timeout, and is most likely due to a 404 path failure. A third condition is that data is being sent but no 405 RTCP feedback is received at all, corresponding to a failure of the 406 reverse path. We derive circuit breaker conditions for these loss 407 regimes in the following. 409 4.1. RTP/AVP Circuit Breaker #1: RTCP Timeout 411 An RTCP timeout can occur when RTP data packets are being sent, but 412 there are no RTCP reports returned from the receiver. This is either 413 due to a failure of the receiver to send RTCP reports, or a failure 414 of the return path that is preventing those RTCP reporting from being 415 delivered. In either case, it is not safe to continue transmission, 416 since the sender has no way of knowing if it is causing congestion. 418 An RTP sender that has not received any RTCP SR or RTCP RR packets 419 reporting on the SSRC it is using, for a time period of at least 420 three times its deterministic RTCP reporting interval, Td, without 421 the randomization factor, and using the fixed minimum interval of 422 Tmin=5 seconds, SHOULD cease transmission (see Section 4.5). The 423 rationale for this choice of timeout is as described in Section 6.2 424 of [RFC3550] ("so that implementations which do not use the reduced 425 value for transmitting RTCP packets are not timed out by other 426 participants prematurely"), as updated by Section 6.1.4 of 427 [I-D.ietf-avtcore-rtp-multi-stream] to account for the use of the 428 RTP/AVPF profile [RFC4585] or the RTP/SAVPF profile [RFC5124]. 430 To reduce the risk of premature timeout, implementations SHOULD NOT 431 configure the RTCP bandwidth such that Td is larger than 5 seconds. 432 Similarly, implementations that use the RTP/AVPF profile [RFC4585] or 433 the RTP/SAVPF profile [RFC5124] SHOULD NOT configure T_rr_interval to 434 values larger than 4 seconds (the reduced limit for T_rr_interval 435 follows Section 6.1.3 of [I-D.ietf-avtcore-rtp-multi-stream]). 437 The choice of three RTCP reporting intervals as the timeout is made 438 following Section 6.3.5 of RFC 3550 [RFC3550]. This specifies that 439 participants in an RTP session will timeout and remove an RTP sender 440 from the list of active RTP senders if no RTP data packets have been 441 received from that RTP sender within the last two RTCP reporting 442 intervals. Using a timeout of three RTCP reporting intervals is 443 therefore large enough that the other participants will have timed 444 out the sender if a network problem stops the data packets it is 445 sending from reaching the receivers, even allowing for loss of some 446 RTCP packets. 448 If a sender is transmitting a large number of RTP media streams, such 449 that the corresponding RTCP SR or RR packets are too large to fit 450 into the network MTU, the receiver will generate RTCP SR or RR 451 packets in a round-robin manner. In this case, the sender SHOULD 452 treat receipt of an RTCP SR or RR packet corresponding to any SSRC it 453 sent on the same 5-tuple of source and destination IP address, port, 454 and protocol, as an indication that the receiver and return path are 455 working, preventing the RTCP timeout circuit breaker from triggering. 457 4.2. RTP/AVP Circuit Breaker #2: Media Timeout 459 If RTP data packets are being sent, but the RTCP SR or RR packets 460 reporting on that SSRC indicate a non-increasing extended highest 461 sequence number received, this is an indication that those RTP data 462 packets are not reaching the receiver. This could be a short-term 463 issue affecting only a few RTP packets, perhaps caused by a slow to 464 open firewall or a transient connectivity problem, but if the issue 465 persists, it is a sign of a more ongoing and significant problem (a 466 "media timeout"). 468 The time needed to declare a media timeout depends on the parameters 469 Tdr, Tr, Tf, and on the non-reporting threshold k. The value of k is 470 chosen so that when Tdr is large compared to Tr and Tf, receipt of at 471 least k RTCP reports with non-increasing extended highest sequence 472 number received gives reasonable assurance that the forward path has 473 failed, and that the RTP data packets have not been lost by chance. 474 The RECOMMENDED value for k is 5 reports. 476 When Tdr < Tf, then RTP data packets are being sent at a rate less 477 than one per RTCP reporting interval of the receiver, so the extended 478 highest sequence number received can be expected to be non-increasing 479 for some receiver RTCP reporting intervals. Similarly, when Tdr < 480 Tr, some receiver RTCP reporting intervals might pass before the RTP 481 data packets arrive at the receiver, also leading to reports where 482 the extended highest sequence number received is non-increasing. 483 Both issues require the media timeout interval to be scaled relative 484 to the threshold, k. 486 The media timeout RTP circuit breaker is therefore as follows. When 487 starting sending, calculate MEDIA_TIMEOUT using: 489 MEDIA_TIMEOUT = ceil(k * max(Tf, Tr, Tdr) / Tdr) 491 When a sender receives an RTCP packet indicating that the media it's 492 sending is being received, then it cancels the media timeout circuit 493 breaker. If it is still sending, then it MUST calculate a new value 494 for MEDIA_TIMEOUT, and set a new media timeout circuit breaker. 496 If a sender receives an RTCP packet indicating that its media was not 497 received, it MUST calculate a new value for MEDIA_TIMEOUT. If the 498 new value is larger than the previous, is replaces MEDIA_TIMEOUT with 499 the new value, extending the media timeout circuit breaker; otherwise 500 it keeps the original value of MEDIA_TIMEOUT. This process is known 501 as reconsidering the media timeout circuit breaker. 503 If MEDIA_TIMEOUT consecutive RTCP packets are received indicating 504 that the media being sent was not received, and the media timeout 505 circuit breaker has not been cancelled, then the media timeout 506 circuit breaker triggers. When the media timeout circuit breaker 507 triggers, the sender SHOULD cease transmission (see Section 4.5). 509 When stopping sending an RTP stream, a sender MUST cancel the 510 corresponding media timeout circuit breaker. 512 4.3. RTP/AVP Circuit Breaker #3: Congestion 514 If RTP data packets are being sent, and the corresponding RTCP SR or 515 RR packets show non-zero packet loss fraction and increasing extended 516 highest sequence number received, then those RTP data packets are 517 arriving at the receiver, but some degree of congestion is occurring. 518 The RTP/AVP profile [RFC3551] states that: 520 If best-effort service is being used, RTP receivers SHOULD monitor 521 packet loss to ensure that the packet loss rate is within 522 acceptable parameters. Packet loss is considered acceptable if a 523 TCP flow across the same network path and experiencing the same 524 network conditions would achieve an average throughput, measured 525 on a reasonable time scale, that is not less than the RTP flow is 526 achieving. This condition can be satisfied by implementing 527 congestion control mechanisms to adapt the transmission rate (or 528 the number of layers subscribed for a layered multicast session), 529 or by arranging for a receiver to leave the session if the loss 530 rate is unacceptably high. 532 The comparison to TCP cannot be specified exactly, but is intended 533 as an "order-of-magnitude" comparison in time scale and 534 throughput. The time scale on which TCP throughput is measured is 535 the round-trip time of the connection. In essence, this 536 requirement states that it is not acceptable to deploy an 537 application (using RTP or any other transport protocol) on the 538 best-effort Internet which consumes bandwidth arbitrarily and does 539 not compete fairly with TCP within an order of magnitude. 541 The phase "order of magnitude" in the above means within a factor of 542 ten, approximately. In order to implement this, it is necessary to 543 estimate the throughput a TCP connection would achieve over the path. 544 For a long-lived TCP Reno connection, it has been shown that the TCP 545 throughput, X, in bytes per second, can be estimated using [Padhye]: 547 s 548 X = ------------------------------------------------------------- 549 Tr*sqrt(2*b*p/3)+(t_RTO * (3*sqrt(3*b*p/8) * p * (1+32*p*p))) 551 This is the same approach to estimated TCP throughput that is used in 552 [RFC3448]. Under conditions of low packet loss the second term on 553 the denominator is small, so this formula can be approximated with 554 reasonable accuracy as follows [Mathis]: 556 s 557 X = ---------------- 558 Tr*sqrt(2*b*p/3) 560 It is RECOMMENDED that this simplified throughout equation be used, 561 since the reduction in accuracy is small, and it is much simpler to 562 calculate than the full equation. Measurements have shown that the 563 simplified TCP throughput equation is effective as an RTP circuit 564 breaker for multimedia flows sent to hosts on residential networks 565 using ADSL and cable modem links [Singh]. The data shows that the 566 full TCP throughput equation tends to be more sensitive to packet 567 loss and triggers the RTP circuit breaker earlier than the simplified 568 equation. Implementations that desire this extra sensitivity MAY use 569 the full TCP throughput equation in the RTP circuit breaker. Initial 570 measurements in LTE networks have shown that the extra sensitivity is 571 helpful in that environment, with the full TCP throughput equation 572 giving a more balanced circuit breaker response than the simplified 573 TCP equation [Sarker]; other networks might see similar behaviour. 575 No matter what TCP throughput equation is chosen, two parameters need 576 to be estimated and reported to the sender in order to calculate the 577 throughput: the round trip time, Tr, and the loss event rate, p (the 578 packet size, s, is known to the sender). The round trip time can be 579 estimated from RTCP SR and RR packets. This is done too infrequently 580 for accurate statistics, but is the best that can be done with the 581 standard RTCP mechanisms. 583 Report blocks in RTCP SR or RR packets contain the packet loss 584 fraction, rather than the loss event rate, so p cannot be reported 585 (TCP typically treats the loss of multiple packets within a single 586 RTT as one loss event, but RTCP RR packets report the overall 587 fraction of packets lost, and does not report when the packet losses 588 occurred). Using the loss fraction in place of the loss event rate 589 can overestimate the loss. We believe that this overestimate will 590 not be significant, given that we are only interested in order of 591 magnitude comparison ([Floyd] section 3.2.1 shows that the difference 592 is small for steady-state conditions and random loss, but using the 593 loss fraction is more conservative in the case of bursty loss). 595 The congestion circuit breaker is therefore: when a sender that is 596 transmitting at least one RTP packet every max(Tdr, Tr) seconds 597 receives an RTCP SR or RR packet that contains a report block for an 598 SSRC it is using, the sender MUST record the value of the fraction 599 lost field in the report block and the time since the last report 600 block was received for that SSRC. If more than CB_INTERVAL (see 601 below) report blocks have been received for that SSRC, the sender 602 MUST calculate the average fraction lost over the last CB_INTERVAL 603 reporting intervals, and then estimate the TCP throughput that would 604 be achieved over the path using the chosen TCP throughput equation 605 and the measured values of the round-trip time, Tr, the loss event 606 rate, p (approximated by the average fraction lost, as is described 607 below), and the packet size, s. The estimate of the TCP throughput, 608 X, is then compared with the actual sending rate of the RTP stream. 609 If the actual sending rate of the RTP stream is more than 10 * X, 610 then the congestion circuit breaker is triggered. 612 The average fraction lost is calculated based on the sum, over the 613 last CB_INTERVAL reporting intervals, of the fraction lost in each 614 reporting interval multiplied by the duration of the corresponding 615 reporting interval, divided by the total duration of the last 616 CB_INTERVAL reporting intervals. The CB_INTERVAL parameter is set 617 to: 619 CB_INTERVAL = 620 ceil(3*min(max(10*G*Tf, 10*Tr, 3*Tdr), max(15, 3*Td))/(3*Tdr)) 622 The parameters that feed into CB_INTERVAL are chosen to give the 623 congestion control algorithm time to react to congestion. They give 624 at least three RTCP reports, ten round trip times, and ten groups of 625 frames to adjust the rate to reduce the congestion to a reasonable 626 level. It is expected that a responsive congestion control algorithm 627 will begin to respond with the next group of frames after it receives 628 indication of congestion, so CB_INTERVAL ought to be a much longer 629 interval than the congestion response. 631 If the RTP/AVPF profile [RFC4585] or the RTP/SAVPF [RFC5124] is used, 632 and the T_rr_interval parameter is used to reduce the frequency of 633 regular RTCP reports, then the value Tdr in the above expression for 634 the CB_INTERVAL parameter MUST be replaced by max(T_rr_interval, 635 Tdr). 637 The CB_INTERVAL parameter is calculated on joining the session, and 638 recalculated on receipt of each RTCP packet, after checking whether 639 the media timeout circuit breaker or the congestion circuit breaker 640 has been triggered. 642 To ensure a timely response to persistent congestion, implementations 643 SHOULD NOT configure the RTCP bandwidth such that Tdr is larger than 644 5 seconds. Similarly, implementations that use the RTP/AVPF profile 645 [RFC4585] or the RTP/SAVPF profile [RFC5124] SHOULD NOT configure 646 T_rr_interval to values larger than 4 seconds (the reduced limit for 647 T_rr_interval follows Section 6.1.3 of 648 [I-D.ietf-avtcore-rtp-multi-stream]). 650 The rationale for enforcing a minimum sending rate below which the 651 congestion circuit breaker will not trigger is to avoid spurious 652 circuit breaker triggers when the number of packets sent per RTCP 653 reporting interval is small, and hence the fraction lost samples are 654 subject to measurement artefacts. The bound of at least one packet 655 every max(Tdr, Tr) seconds is derived from the one packet per RTT 656 minimum sending rate of TCP [RFC5405], adapted for use with RTP where 657 the RTCP reporting interval is decoupled from the network RTT. 659 When the congestion circuit breaker is triggered, the sender SHOULD 660 cease transmission (see Section 4.5). However, if the sender is able 661 to reduce its sending rate by a factor of (approximately) ten, then 662 it MAY first reduce its sending rate by this factor (or some larger 663 amount) to see if that resolves the congestion. If the sending rate 664 is reduced in this way and the congestion circuit breaker triggers 665 again after the next CB_INTERVAL RTCP reporting intervals, the sender 666 MUST then cease transmission. An example of such a rate reduction 667 might be a video conferencing system that backs off to sending audio 668 only, before completely dropping the call. If such a reduction in 669 sending rate resolves the congestion problem, the sender MAY 670 gradually increase the rate at which it sends data after a reasonable 671 amount of time has passed, provided it takes care not to cause the 672 problem to recur ("reasonable" is intentionally not defined here). 674 The RTCP reporting interval of the media sender does not affect how 675 quickly congestion circuit breaker can trigger. The timing is based 676 on the RTCP reporting interval of the receiver that generates the SR/ 677 RR packets from which the loss rate and RTT estimate are derived 678 (note that RTCP requires all participants in a session to have 679 similar reporting intervals, else the participant timeout rules in 680 [RFC3550] will not work, so this interval is likely similar to that 681 of the sender). If the incoming RTCP SR or RR packets are using a 682 reduced minimum RTCP reporting interval (as specified in Section 6.2 683 of RFC 3550 [RFC3550] or the RTP/AVPF profile [RFC4585]), then that 684 reduced RTCP reporting interval is used when determining if the 685 circuit breaker is triggered. 687 If there are more media streams that can be reported in a single RTCP 688 SR or RR packet, or if the size of a complete RTCP SR or RR packet 689 exceeds the network MTU, then the receiver will report on a subset of 690 sources in each reporting interval, with the subsets selected round- 691 robin across multiple intervals so that all sources are eventually 692 reported [RFC3550]. When generating such round-robin RTCP reports, 693 priority SHOULD be given to reports on sources that have high packet 694 loss rates, to ensure that senders are aware of network congestion 695 they are causing (this is an update to [RFC3550]). 697 4.4. RTP/AVP Circuit Breaker #4: Media Usability 699 Applications that use RTP are generally tolerant to some amount of 700 packet loss. How much packet loss can be tolerated will depend on 701 the application, media codec, and the amount of error correction and 702 packet loss concealment that is applied. There is an upper bound on 703 the amount of loss can be corrected, however, beyond which the media 704 becomes unusable. Similarly, many applications have some upper bound 705 on the media capture to play-out latency that can be tolerated before 706 the application becomes unusable. The latency bound will depend on 707 the application, but typical values can range from the order of a few 708 hundred milliseconds for voice telephony and interactive conferencing 709 applications, up to several seconds for some video-on-demand systems. 711 As a final circuit breaker, RTP senders SHOULD monitor the reported 712 packet loss and delay to estimate whether the media is likely to be 713 suitable for the intended purpose. If the packet loss rate and/or 714 latency is such that the media has become unusable, and has remained 715 unusable for a significant time period, then the application SHOULD 716 cease transmission. Similarly, receivers SHOULD monitor the quality 717 of the media they receive, and if the quality is unusable for a 718 significant time period, they SHOULD terminate the session. This 719 memo intentionally does not define a bound on the packet loss rate or 720 latency that will result in unusable media, nor does it specify what 721 time period is deemed significant, as these are highly application 722 dependent. 724 Sending media that suffers from such high packet loss or latency that 725 it is unusable at the receiver is both wasteful of resources, and of 726 no benefit to the user of the application. It also is highly likely 727 to be congesting the network, and disrupting other applications. As 728 such, the congestion circuit breaker will almost certainly trigger to 729 stop flows where the media would be unusable due to high packet loss 730 or latency. However, in pathological scenarios where the congestion 731 circuit breaker does not stop the flow, it is desirable that the RTP 732 application cease sending useless traffic. The role of the media 733 usability circuit breaker is to protect the network in such cases. 735 4.5. Ceasing Transmission 737 What it means to cease transmission depends on the application, but 738 the intention is that the application will stop sending RTP data 739 packets to a particular destination 3-tuple (transport protocol, 740 destination port, IP address), until the user makes an explicit 741 attempt to restart the call. It is important that a human user is 742 involved in the decision to try to restart the call, since that user 743 will eventually give up if the calls repeatedly trigger the circuit 744 breaker. This will help avoid problems with automatic redial systems 745 from congesting the network. Accordingly, RTP flows halted by the 746 circuit breaker SHOULD NOT be restarted automatically unless the 747 sender has received information that the congestion has dissipated. 749 It is recognised that the RTP implementation in some systems might 750 not be able to determine if a call set-up request was initiated by a 751 human user, or automatically by some scripted higher-level component 752 of the system. These implementations SHOULD rate limit attempts to 753 restart a call to the same destination 3-tuple as used by a previous 754 call that was recently halted by the circuit breaker. The chosen 755 rate limit ought to not exceed the rate at which an annoyed human 756 caller might redial a misbehaving phone. 758 5. RTP Circuit Breakers and the RTP/AVPF and RTP/SAVPF Profiles 760 Use of the Extended RTP Profile for RTCP-based Feedback (RTP/AVPF) 761 [RFC4585] allows receivers to send early RTCP reports in some cases, 762 to inform the sender about particular events in the media stream. 763 There are several use cases for such early RTCP reports, including 764 providing rapid feedback to a sender about the onset of congestion. 765 The RTP/SAVPF Profile [RFC5124] is a secure variant of the RTP/AVPF 766 profile, that is treated the same in the context of the RTP circuit 767 breaker. These feedback profiles are often used with non-compound 768 RTCP reports [RFC5506] to reduce the reporting overhead. 770 Receiving rapid feedback about congestion events potentially allows 771 congestion control algorithms to be more responsive, and to better 772 adapt the media transmission to the limitations of the network. It 773 is expected that many RTP congestion control algorithms will adopt 774 the RTP/AVPF profile or the RTP/SAVPF profile for this reason, 775 defining new transport layer feedback reports that suit their 776 requirements. Since these reports are not yet defined, and likely 777 very specific to the details of the congestion control algorithm 778 chosen, they cannot be used as part of the generic RTP circuit 779 breaker. 781 Reduced-size RTCP reports sent under the RTP/AVPF early feedback 782 rules that do not contain an RTCP SR or RR packet MUST be ignored by 783 the congestion circuit breaker (they do not contain the information 784 needed by the congestion circuit breaker algorithm), but MUST be 785 counted as received packets for the RTCP timeout circuit breaker. 786 Reduced-size RTCP reports sent under the RTP/AVPF early feedback 787 rules that contain RTCP SR or RR packets MUST be processed by the 788 congestion circuit breaker as if they were sent as regular RTCP 789 reports, and counted towards the circuit breaker conditions specified 790 in Section 4 of this memo. This will potentially make the RTP 791 circuit breaker trigger earlier than it would if the RTP/AVPF profile 792 was not used. 794 When using ECN with RTP (see Section 7), early RTCP feedback packets 795 can contain ECN feedback reports. The count of ECN-CE marked packets 796 contained in those ECN feedback reports is counted towards the number 797 of lost packets reported if the ECN Feedback Report report is sent in 798 an compound RTCP packet along with an RTCP SR/RR report packet. 799 Reports of ECN-CE packets sent as reduced-size RTCP ECN feedback 800 packets without an RTCP SR/RR packet MUST be ignored. 802 These rules are intended to allow the use of low-overhead RTP/AVPF 803 feedback for generic NACK messages without triggering the RTP circuit 804 breaker. This is expected to make such feedback suitable for RTP 805 congestion control algorithms that need to quickly report loss events 806 in between regular RTCP reports. The reaction to reduced-size RTCP 807 SR/RR packets is to allow such algorithms to send feedback that can 808 trigger the circuit breaker, when desired. 810 The RTP/AVPF and RTP/SAVPF profiles include the T_rr_interval 811 parameter that can be used to adjust the regular RTCP reporting 812 interval. The use of the T_rr_interval parameter changes the 813 behaviour of the RTP circuit breaker, as described in Section 4. 815 6. Impact of RTCP Extended Reports (XR) 817 RTCP Extended Report (XR) blocks provide additional reception quality 818 metrics, but do not change the RTCP timing rules. Some of the RTCP 819 XR blocks provide information that might be useful for congestion 820 control purposes, others provided non-congestion-related metrics. 821 With the exception of RTCP XR ECN Summary Reports (see Section 7), 822 the presence of RTCP XR blocks in a compound RTCP packet does not 823 affect the RTP circuit breaker algorithm. For consistency and ease 824 of implementation, only the reception report blocks contained in RTCP 825 SR packets, RTCP RR packets, or RTCP XR ECN Summary Report packets, 826 are used by the RTP circuit breaker algorithm. 828 7. Impact of Explicit Congestion Notification (ECN) 830 The use of ECN for RTP flows does not affect the media timeout RTP 831 circuit breaker (Section 4.2) or the RTCP timeout circuit breaker 832 (Section 4.1), since these are both connectivity checks that simply 833 determinate if any packets are being received. 835 ECN-CE marked packets SHOULD be treated as if it were lost for the 836 purposes of congestion control, when determining the optimal media 837 sending rate for an RTP flow. If an RTP sender has negotiated ECN 838 support for an RTP session, and has successfully initiated ECN use on 839 the path to the receiver [RFC6679], then ECN-CE marked packets SHOULD 840 be treated as if they were lost when calculating if the congestion- 841 based RTP circuit breaker (Section 4.3) has been met. The count of 842 ECN-CE marked RTP packets is returned in RTCP XR ECN summary report 843 packets if support for ECN has been initiated for an RTP session. 845 8. Impact of Bundled Media and Layered Coding 847 The RTP circuit breaker operates on a per-RTP session basis. An RTP 848 sender that participates in several RTP sessions MUST treat each RTP 849 session independently with regards to the RTP circuit breaker. 851 An RTP sender can generate several media streams within a single RTP 852 session, with each stream using a different SSRC. This can happen if 853 bundled media are in use, when using simulcast, or when using layered 854 media coding. By default, each SSRC will be treated independently by 855 the RTP circuit breaker. However, the sender MAY choose to treat the 856 flows (or a subset thereof) as a group, such that a circuit breaker 857 trigger for one flow applies to the group of flows as a whole, and 858 either causes the entire group to cease transmission, or the sending 859 rate of the group to reduce by a factor of ten, depending on the RTP 860 circuit breaker triggered. Grouping flows in this way is expected to 861 be especially useful for layered flows sent using multiple SSRCs, as 862 it allows the layered flow to react as a whole, ceasing transmission 863 on the enhancement layers first to reduce sending rate if necessary, 864 rather than treating each layer independently. 866 9. Security Considerations 868 The security considerations of [RFC3550] apply. 870 If the RTP/AVPF profile is used to provide rapid RTCP feedback, the 871 security considerations of [RFC4585] apply. If ECN feedback for RTP 872 over UDP/IP is used, the security considerations of [RFC6679] apply. 874 If non-authenticated RTCP reports are used, an on-path attacker can 875 trivially generate fake RTCP packets that indicate high packet loss 876 rates, causing the circuit breaker to trigger and disrupting an RTP 877 session. This is somewhat more difficult for an off-path attacker, 878 due to the need to guess the randomly chosen RTP SSRC value and the 879 RTP sequence number. This attack can be avoided if RTCP packets are 880 authenticated; authentication options are discussed in [RFC7201]. 882 Timely operation of the RTP circuit breaker depends on the choice of 883 RTCP reporting interval. If the receiver has a reporting interval 884 that is overly long, then the responsiveness of the circuit breaker 885 decreases. In the limit, the RTP circuit breaker can be disabled for 886 all practical purposes by configuring an RTCP reporting interval that 887 is many minutes duration. This issue is not specific to the circuit 888 breaker: long RTCP reporting intervals also prevent reception quality 889 reports, feedback messages, codec control messages, etc., from being 890 used. Implementations are expected to impose an upper limit on the 891 RTCP reporting interval they are willing to negotiate (based on the 892 session bandwidth and RTCP bandwidth fraction) when using the RTP 893 circuit breaker, as discussed in Section 4.3. 895 10. IANA Considerations 897 There are no actions for IANA. 899 11. Acknowledgements 901 The authors would like to thank Bernard Aboba, Harald Alvestrand, 902 Gorry Fairhurst, Nazila Fough, Kevin Gross, Cullen Jennings, Randell 903 Jesup, Jonathan Lennox, Matt Mathis, Stephen McQuistin, Simon 904 Perreault, Eric Rescorla, Abheek Saha, Fabio Verdicchio, and Magnus 905 Westerlund for their valuable feedback. 907 12. References 909 12.1. Normative References 911 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 912 Requirement Levels", BCP 14, RFC 2119, 913 DOI 10.17487/RFC2119, March 1997, 914 . 916 [RFC3448] Handley, M., Floyd, S., Padhye, J., and J. Widmer, "TCP 917 Friendly Rate Control (TFRC): Protocol Specification", 918 RFC 3448, DOI 10.17487/RFC3448, January 2003, 919 . 921 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 922 Jacobson, "RTP: A Transport Protocol for Real-Time 923 Applications", STD 64, RFC 3550, DOI 10.17487/RFC3550, 924 July 2003, . 926 [RFC3551] Schulzrinne, H. and S. Casner, "RTP Profile for Audio and 927 Video Conferences with Minimal Control", STD 65, RFC 3551, 928 DOI 10.17487/RFC3551, July 2003, 929 . 931 [RFC3611] Friedman, T., Ed., Caceres, R., Ed., and A. Clark, Ed., 932 "RTP Control Protocol Extended Reports (RTCP XR)", 933 RFC 3611, DOI 10.17487/RFC3611, November 2003, 934 . 936 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 937 "Extended RTP Profile for Real-time Transport Control 938 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, 939 DOI 10.17487/RFC4585, July 2006, 940 . 942 12.2. Informative References 944 [Floyd] Floyd, S., Handley, M., Padhye, J., and J. Widmer, 945 "Equation-Based Congestion Control for Unicast 946 Applications", Proceedings of the ACM SIGCOMM 947 conference, 2000, DOI 10.1145/347059.347397, August 2000. 949 [I-D.ietf-avtcore-rtp-multi-stream] 950 Lennox, J., Westerlund, M., Wu, W., and C. Perkins, 951 "Sending Multiple Media Streams in a Single RTP Session", 952 draft-ietf-avtcore-rtp-multi-stream-09 (work in progress), 953 September 2015. 955 [I-D.ietf-tsvwg-circuit-breaker] 956 Fairhurst, G., "Network Transport Circuit Breakers", 957 draft-ietf-tsvwg-circuit-breaker-05 (work in progress), 958 October 2015. 960 [Mathis] Mathis, M., Semke, J., Mahdavi, J., and T. Ott, "The 961 macroscopic behavior of the TCP congestion avoidance 962 algorithm", ACM SIGCOMM Computer Communication 963 Review 27(3), DOI 10.1145/263932.264023, July 1997. 965 [Padhye] Padhye, J., Firoiu, V., Towsley, D., and J. Kurose, 966 "Modeling TCP Throughput: A Simple Model and its Empirical 967 Validation", Proceedings of the ACM SIGCOMM 968 conference, 1998, DOI 10.1145/285237.285291, August 1998. 970 [RFC2862] Civanlar, M. and G. Cash, "RTP Payload Format for Real- 971 Time Pointers", RFC 2862, DOI 10.17487/RFC2862, June 2000, 972 . 974 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 975 of Explicit Congestion Notification (ECN) to IP", 976 RFC 3168, DOI 10.17487/RFC3168, September 2001, 977 . 979 [RFC4103] Hellstrom, G. and P. Jones, "RTP Payload for Text 980 Conversation", RFC 4103, DOI 10.17487/RFC4103, June 2005, 981 . 983 [RFC5104] Wenger, S., Chandra, U., Westerlund, M., and B. Burman, 984 "Codec Control Messages in the RTP Audio-Visual Profile 985 with Feedback (AVPF)", RFC 5104, DOI 10.17487/RFC5104, 986 February 2008, . 988 [RFC5124] Ott, J. and E. Carrara, "Extended Secure RTP Profile for 989 Real-time Transport Control Protocol (RTCP)-Based Feedback 990 (RTP/SAVPF)", RFC 5124, DOI 10.17487/RFC5124, February 991 2008, . 993 [RFC5348] Floyd, S., Handley, M., Padhye, J., and J. Widmer, "TCP 994 Friendly Rate Control (TFRC): Protocol Specification", 995 RFC 5348, DOI 10.17487/RFC5348, September 2008, 996 . 998 [RFC5405] Eggert, L. and G. Fairhurst, "Unicast UDP Usage Guidelines 999 for Application Designers", BCP 145, RFC 5405, 1000 DOI 10.17487/RFC5405, November 2008, 1001 . 1003 [RFC5450] Singer, D. and H. Desineni, "Transmission Time Offsets in 1004 RTP Streams", RFC 5450, DOI 10.17487/RFC5450, March 2009, 1005 . 1007 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 1008 Real-Time Transport Control Protocol (RTCP): Opportunities 1009 and Consequences", RFC 5506, DOI 10.17487/RFC5506, April 1010 2009, . 1012 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 1013 Control", RFC 5681, DOI 10.17487/RFC5681, September 2009, 1014 . 1016 [RFC6679] Westerlund, M., Johansson, I., Perkins, C., O'Hanlon, P., 1017 and K. Carlberg, "Explicit Congestion Notification (ECN) 1018 for RTP over UDP", RFC 6679, DOI 10.17487/RFC6679, August 1019 2012, . 1021 [RFC6798] Clark, A. and Q. Wu, "RTP Control Protocol (RTCP) Extended 1022 Report (XR) Block for Packet Delay Variation Metric 1023 Reporting", RFC 6798, DOI 10.17487/RFC6798, November 2012, 1024 . 1026 [RFC6843] Clark, A., Gross, K., and Q. Wu, "RTP Control Protocol 1027 (RTCP) Extended Report (XR) Block for Delay Metric 1028 Reporting", RFC 6843, DOI 10.17487/RFC6843, January 2013, 1029 . 1031 [RFC6958] Clark, A., Zhang, S., Zhao, J., and Q. Wu, Ed., "RTP 1032 Control Protocol (RTCP) Extended Report (XR) Block for 1033 Burst/Gap Loss Metric Reporting", RFC 6958, 1034 DOI 10.17487/RFC6958, May 2013, 1035 . 1037 [RFC7002] Clark, A., Zorn, G., and Q. Wu, "RTP Control Protocol 1038 (RTCP) Extended Report (XR) Block for Discard Count Metric 1039 Reporting", RFC 7002, DOI 10.17487/RFC7002, September 1040 2013, . 1042 [RFC7003] Clark, A., Huang, R., and Q. Wu, Ed., "RTP Control 1043 Protocol (RTCP) Extended Report (XR) Block for Burst/Gap 1044 Discard Metric Reporting", RFC 7003, DOI 10.17487/RFC7003, 1045 September 2013, . 1047 [RFC7097] Ott, J., Singh, V., Ed., and I. Curcio, "RTP Control 1048 Protocol (RTCP) Extended Report (XR) for RLE of Discarded 1049 Packets", RFC 7097, DOI 10.17487/RFC7097, January 2014, 1050 . 1052 [RFC7201] Westerlund, M. and C. Perkins, "Options for Securing RTP 1053 Sessions", RFC 7201, DOI 10.17487/RFC7201, April 2014, 1054 . 1056 [Sarker] Sarker, Z., Singh, V., and C. Perkins, "An Evaluation of 1057 RTP Circuit Breaker Performance on LTE Networks", 1058 Proceedings of the IEEE Infocom workshop on Communication 1059 and Networking Techniques for Contemporary Video, 2014, 1060 April 2014. 1062 [Singh] Singh, V., McQuistin, S., Ellis, M., and C. Perkins, 1063 "Circuit Breakers for Multimedia Congestion Control", 1064 Proceedings of the International Packet Video 1065 Workshop, 2013, DOI 10.1109/PV.2013.6691439, December 1066 2013. 1068 Authors' Addresses 1070 Colin Perkins 1071 University of Glasgow 1072 School of Computing Science 1073 Glasgow G12 8QQ 1074 United Kingdom 1076 Email: csp@csperkins.org 1078 Varun Singh 1079 Aalto University 1080 School of Electrical Engineering 1081 Otakaari 5 A 1082 Espoo, FIN 02150 1083 Finland 1085 Email: varun@comnet.tkk.fi 1086 URI: http://www.netlab.tkk.fi/~varun/