AVTCORE Working Group                                         C. S. Perkins
Internet-Draft                                     University of Glasgow
Updates: 3550 (if approved)                                     V. Singh
Intended status: Standards Track                        Aalto University
Expires: September 24, 2015                               March 23, April 18, 2016                                 October 16, 2015

Multimedia Congestion Control: Circuit Breakers for Unicast RTP Sessions
               draft-ietf-avtcore-rtp-circuit-breakers-10
               draft-ietf-avtcore-rtp-circuit-breakers-11

Abstract

   The Real-time Transport Protocol (RTP) is widely used in telephony,
   video conferencing, and telepresence applications.  Such applications
   are often run on best-effort UDP/IP networks.  If congestion control
   is not implemented in the applications, then network congestion will
   deteriorate the user's multimedia experience.  This document does not
   propose a congestion control algorithm; instead, it defines a minimal
   set of RTP "circuit-breakers". circuit-breakers.  Circuit-breakers are conditions under
   which an RTP sender needs to stop transmitting media data in order to
   protect the network from excessive congestion.  It is expected that,
   in the absence of severe congestion, all RTP applications running on
   best-effort IP networks will be able to run without triggering these
   circuit breakers.  Any future RTP congestion control specification
   will be expected to operate within the constraints defined by these
   circuit breakers.

Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at http://datatracker.ietf.org/drafts/current/.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   This Internet-Draft will expire on September 24, 2015. April 18, 2016.

Copyright Notice

   Copyright (c) 2015 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   2
   2.  Terminology  Background  . . . . . . . . . . . . . . . . . . . . . . . . .   3
   3.  Background  Terminology . . . . . . . . . . . . . . . . . . . . . . . . .   3   6
   4.  RTP Circuit Breakers for Systems Using the RTP/AVP Profile  .   6   7
     4.1.  RTP/AVP Circuit Breaker #1: Media RTCP Timeout  . . . . . . . .   7   9
     4.2.  RTP/AVP Circuit Breaker #2: RTCP Media Timeout . . . . . . . .   8  10
     4.3.  RTP/AVP Circuit Breaker #3: Congestion  . . . . . . . . .   9  11
     4.4.  RTP/AVP Circuit Breaker #4: Media Usability . . . . . . .  13  15
     4.5.  Choice of Circuit Breaker Interval  . . . . . . . . . . .  14
     4.6.  Ceasing Transmission  . . . . . . . . . . . . . . . . . .  15  16
   5.  RTP Circuit Breakers and the RTP/AVPF and RTP/SAVPF Profiles   16   17
   6.  Impact of RTCP Extended Reports (XR)  . . . . . . . . . . . .  17  18
   7.  Impact of RTCP Reporting Groups . . . . . . . . . . . . . . .  17
   8.  Impact of Explicit Congestion Notification (ECN)  . . . . . .  18
   9.
   8.  Impact of Bundled Media and Layered Coding  . . . . . . . . .  18
   10.
   9.  Security Considerations . . . . . . . . . . . . . . . . . . .  18
   11.  19
   10. IANA Considerations . . . . . . . . . . . . . . . . . . . . .  19
   12.
   11. Acknowledgements  . . . . . . . . . . . . . . . . . . . . . .  19
   13.  20
   12. References  . . . . . . . . . . . . . . . . . . . . . . . . .  19
     13.1.  20
     12.1.  Normative References . . . . . . . . . . . . . . . . . .  19
     13.2.  20
     12.2.  Informative References . . . . . . . . . . . . . . . . .  20
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  22  23

1.  Introduction

   The Real-time Transport Protocol (RTP) [RFC3550] is widely used in
   voice-over-IP, video teleconferencing, and telepresence systems.
   Many of these systems run over best-effort UDP/IP networks, and can
   suffer from packet loss and increased latency if network congestion
   occurs.  Designing effective RTP congestion control algorithms, to
   adapt the transmission of RTP-based media to match the available
   network capacity, while also maintaining the user experience, is a
   difficult but important problem.  Many such congestion control and
   media adaptation algorithms have been proposed, but to date there is
   no consensus on the correct approach, or even that a single standard
   algorithm is desirable.

   This memo does not attempt to propose a new RTP congestion control
   algorithm.  Rather, it proposes  Instead, we propose a minimal small set of "RTP RTP circuit
   breakers"; breakers.
   These are conditions under which there is general agreement that an
   RTP flow is causing serious congestion, and hence ought to cease
   transmission.  The RTP circuit breakers proposed in this memo are a
   specific instance of the general class of network transport circuit
   breakers [I-D.ietf-tsvwg-circuit-breaker], designed to act as a
   protection mechanism of last resort to avoid persistent congestion.
   It is expected that future standards-track congestion control
   algorithms for RTP will operate within the envelope defined by this
   memo.

2.  Terminology

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in RFC 2119 [RFC2119].
   This interpretation of these key words applies only when written in
   ALL CAPS.  Mixed- or lower-case uses of these key words are not to be
   interpreted as carrying special significance in this memo.

3.  Background

   We consider congestion control for unicast RTP traffic flows.  This
   is the problem of adapting the transmission of an audio/visual data
   flow, encapsulated within an RTP transport session, from one sender
   to one receiver, so that it matches the available network bandwidth.
   Such adaptation needs to be done in a way that limits the disruption
   to the user experience caused by both packet loss and excessive rate
   changes.  Congestion control for multicast flows is outside the scope
   of this memo.  Multicast traffic needs different solutions, since the
   available bandwidth estimator for a group of receivers will differ
   from that for a single receiver, and because multicast congestion
   control has to consider issues of fairness across groups of receivers
   that do not apply to unicast flows.

   Congestion control for unicast RTP traffic can be implemented in one
   of two places in the protocol stack.  One approach is to run the RTP
   traffic over a congestion controlled transport protocol, for example
   over TCP, and to adapt the media encoding to match the dictates of
   the transport-layer congestion control algorithm.  This is safe for
   the network, but can be suboptimal for the media quality unless the
   transport protocol is designed to support real-time media flows.  We
   do not consider this class of applications further in this memo, as
   their network safety is guaranteed by the underlying transport.

   Alternatively, RTP flows can be run over a non-congestion controlled
   transport protocol, for example UDP, performing rate adaptation at
   the application layer based on RTP Control Protocol (RTCP) feedback.
   With a well-designed, network-aware, application, this allows highly
   effective media quality adaptation, but there is potential to disrupt
   the network's operation if the application does not adapt its sending
   rate in a timely and effective manner.  We consider this class of
   applications in this memo.

   Congestion control relies on monitoring the delivery of a media flow,
   and responding to adapt the transmission of that flow when there are
   signs that the network path is congested.  Network congestion can be
   detected in one of three ways: 1) a receiver can infer the onset of
   congestion by observing an increase in one-way delay caused by queue
   build-up within the network; 2) if Explicit Congestion Notification
   (ECN) [RFC3168] is supported, the network can signal the presence of
   congestion by marking packets using ECN Congestion Experienced (CE)
   marks; or 3) in the extreme case, congestion will cause packet loss
   that can be detected by observing a gap in the received RTP sequence
   numbers.

   Once the onset of congestion is observed, the receiver has to send
   feedback to the sender to indicate that the transmission rate needs
   to be reduced.  How the sender reduces the transmission rate is
   highly dependent on the media codec being used, and is outside the
   scope of this memo.

   There are several ways in which a receiver can send feedback to a
   media sender within the RTP framework:

   o  The base RTP specification [RFC3550] defines RTCP Reception Report
      (RR) packets to convey reception quality feedback information, and
      Sender Report (SR) packets to convey information about the media
      transmission.  RTCP SR packets contain data that can be used to
      reconstruct media timing at a receiver, along with a count of the
      total number of octets and packets sent.  RTCP RR packets report
      on the fraction of packets lost in the last reporting interval,
      the cumulative number of packets lost, the highest sequence number
      received, and the inter-arrival jitter.  The RTCP RR packets also
      contain timing information that allows the sender to estimate the
      network round trip time (RTT) to the receivers.  RTCP reports are
      sent periodically, with the reporting interval being determined by
      the number of SSRCs used in the session and a configured session
      bandwidth estimate (the number of SSRCs used is usually two in a
      unicast session, one for each participant, but can be greater if
      the participants send multiple media streams).  The interval
      between reports sent from each receiver tends to be on the order
      of a few seconds on average, although it varies with the session
      bandwidth, and sub-second reporting intervals are possible in high
      bandwidth sessions, and it is randomised to avoid synchronisation
      of reports from multiple receivers.  RTCP RR packets allow a
      receiver to report ongoing network congestion to the sender.
      However, if a receiver detects the onset of congestion part way
      through a reporting interval, the base RTP specification contains
      no provision for sending the RTCP RR packet early, and the
      receiver has to wait until the next scheduled reporting interval.

   o  The RTCP Extended Reports (XR) [RFC3611] allow reporting of more
      complex and sophisticated reception quality metrics, but do not
      change the RTCP timing rules.  RTCP extended reports of potential
      interest for congestion control purposes are the extended packet
      loss, discard, and burst metrics [RFC3611], [RFC7002], [RFC7097],
      [RFC7003], [RFC6958]; and the extended delay metrics [RFC6843],
      [RFC6798].  Other RTCP Extended Reports that could be helpful for
      congestion control purposes might be developed in future.

   o  Rapid feedback about the occurrence of congestion events can be
      achieved using the Extended RTP Profile for RTCP-Based Feedback
      (RTP/AVPF) [RFC4585] (or its secure variant, RTP/SAVPF [RFC5124])
      in place of the RTP/AVP profile [RFC3551].  This modifies the RTCP
      timing rules to allow RTCP reports to be sent early, in some cases
      immediately, provided the RTCP transmission rate keeps within its
      bandwidth allocation.  It also defines transport-layer feedback
      messages, including negative acknowledgements (NACKs), that can be
      used to report on specific congestion events.  RTP Codec Control
      Messages [RFC5104] extend the RTP/AVPF profile with additional
      feedback messages that can be used to influence that way in which
      rate adaptation occurs, but do not further change the dynamics of
      how rapidly feedback can be sent.  Use of the RTP/AVPF profile is
      dependent on signalling.

   o  Finally, Explicit Congestion Notification (ECN) for RTP over UDP
      [RFC6679] can be used to provide feedback on the number of packets
      that received an ECN Congestion Experienced (CE) mark.  This RTCP
      extension builds on the RTP/AVPF profile to allow rapid congestion
      feedback when ECN is supported.

   In addition to these mechanisms for providing feedback, the sender
   can include an RTP header extension in each packet to record packet
   transmission times.  There are two methods: [RFC5450] represents the
   transmission time in terms of a time-offset from the RTP timestamp of
   the packet, while [RFC6051] includes an explicit NTP-format sending
   timestamp (potentially more accurate, but a higher header overhead). times [RFC5450].  Accurate sending transmission timestamps can
   be helpful for estimating queuing delays, to get an early indication
   of the onset of congestion.

   Taken together, these various mechanisms allow receivers to provide
   feedback on the senders when congestion events occur, with varying
   degrees of timeliness and accuracy.  The key distinction is between
   systems that use only the basic RTCP mechanisms, without RTP/AVPF
   rapid feedback, and those that use the RTP/AVPF extensions to respond
   to congestion more rapidly.

4.  RTP Circuit Breakers for Systems Using the RTP/AVP Profile

3.  Terminology

   The feedback mechanisms defined in [RFC3550] key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and available under the
   RTP/AVP profile [RFC3551] "OPTIONAL" in this
   document are the minimum that can be assumed for a
   baseline circuit breaker mechanism that is suitable for all unicast
   applications of RTP.  Accordingly, for an RTP circuit breaker to be
   useful, it needs to be able to detect that an RTP flow is causing
   excessive congestion using interpreted as described in RFC 2119 [RFC2119].
   This interpretation of these key words applies only basic RTCP features, without needing
   RTCP XR feedback when written in
   ALL CAPS.  Mixed- or the RTP/AVPF profile for rapid RTCP reports.

   RTCP is a fundamental part of the RTP protocol, and the mechanisms
   described here rely on the implementation lower-case uses of RTCP.  Implementations
   that claim to support RTP, but that do these key words are not implement RTCP, cannot use
   the circuit breaker mechanisms described to be
   interpreted as carrying special significance in this memo.  Such
   implementations SHOULD NOT be used on networks that might be subject
   to congestion unless equivalent mechanisms are defined using some
   non-RTCP feedback channel to report congestion and signal

   The definition of the RTP circuit breaker conditions.

   Three potential congestion signals are available from the basic RTCP
   SR/RR packets and are reported for each synchronisation source (SSRC) is specified in terms of
   the RTP session:

   1.  The sender can estimate following variables:

   o  Td is the network round-trip time once per deterministic RTCP reporting interval, based on as defined in
      Section 6.3.1 of [RFC3550].

   o  Tdr is the contents and timing sender's estimate of the deterministic RTCP SR
       and RR packets.

   2.  Receivers report reporting
      interval, Td, calculated by a jitter estimate (the statistical variance receiver of the RTP data packet inter-arrival time) calculated over it is sending.
      Tdr is not known at the RTCP
       reporting interval.  Due to sender, but can be estimated by executing
      the nature algorithm in Section 6.2 of [RFC3550] using the jitter calculation
       ([RFC3550], section 6.4.4), average RTCP
      packet size seen at the jitter is only meaningful for RTP
       flows that send a single data packet for each RTP timestamp value
       (i.e., audio flows, or video flows where each packet comprises
       one video frame).

   3.  Receivers report sender, the fraction number of RTP data packets lost during members reported in
      the
       RTCP reporting interval, receiver's SR/RR report blocks, and whether the cumulative number of RTP packets
       lost over the entire RTP session.

   These congestion signals limit receiver is
      sending SR or RR packets.  Tdr is recalculated when each new RTCP
      SR/RR report is received, but the possible media timeout circuit breakers, since
   they give breaker
      (see Section 4.2) is only limited visibility into reconsidered when Tdr increases.

   o  Tr is the behaviour of network round-trip time, calculated by the network.

   RTT estimates are widely used sender using
      the algorithm in congestion control algorithms, Section 6.4.1 of [RFC3550] and smoothed using an
      exponentially weighted moving average as a
   proxy for queuing delay measures in delay-based congestion control or
   to determine connection timeouts. Tr = (0.8 * Tr) + (0.2 *
      Tr_new) where Tr_new is the latest RTT estimates derived estimate obtained from an
      RTCP SR
   and RR packets sent according to report.  The weight is chosen so old estimates decay over k
      intervals.

   o  k is the RTP/AVP timing rules are too
   infrequent to be useful though, and don't give enough information to
   distinguish non-reporting threshold (see Section 4.2).

   o  Tf is the media framing interval at the sender.  For applications
      sending at a delay change due to routing updates from queuing delay
   caused by congestion.  Accordingly, we cannot use constant frame rate, Tf is the RTT estimate
   alone as an RTP circuit breaker.

   Increased jitter can be inter-frame interval.
      For applications that switch between a signal small set of transient network congestion, but
   in possible frame
      rates, for example when sending speech with comfort noise, where
      comfort noise frames are sent less often than speech frames, Tf is
      set to the highly aggregated form reported in RTCP RR packets, it offers
   insufficient information to estimate the extent or persistence longest of
   congestion.  Jitter reports are a useful early warning the inter-frame intervals of potential
   network congestion, the different
      frame rates.  For applications that send periodic frames but provide an insufficiently strong signal
      dynamically vary their frame rate, Tf is set to be the largest inter-
      frame interval used as a circuit breaker.

   The remaining congestion signals are in the packet loss fraction and last 10 seconds.  For applications that
      send less than one frame every 10 seconds, or that have no concept
      of periodic frames (e.g., text conversation [RFC4103], or pointer
      events [RFC2862]), Tf is set to the time interval since the
      previous frame when each frame is sent.

   o  G is the frame group size.  That is, the
   cumulative number of packets lost.  If considered carefully, these
   can be effective indicators frames that congestion is occurring in networks
   where packet loss is primarily due to queue overflows, although loss
   caused are
      coded together based on a particular sending rate setting.  If the
      codec used by non-congestive packet corruption can distort the result in
   some networks.  TCP congestion control [RFC5681] intentionally tries sender can change its rate on each frame, G = 1;
      otherwise G is set to fill the router queues, and uses number of frames before the resulting packet loss as
   congestion feedback.  An RTP flow competing with TCP traffic will
   therefore expect codec can
      adjust to see a non-zero packet loss fraction the new rate.  For codecs that has to
   be related to TCP dynamics to estimate available capacity.  This
   behaviour have the concept of TCP a
      group-of-pictures (GoP), G is reflected in likely the congestion circuit breaker
   below, and will affect GoP length.

   o  T_rr_interval is the design of any RTP congestion control
   protocol.

   Two packet loss regimes can be observed: 1) minimal interval between RTCP RR packets show a
   non-zero packet loss fraction, while the extended highest sequence
   number received continues to increment; and 2) RR packets show a loss
   fraction reports, as
      defined in Section 3.4 of zero, but the extended highest sequence number received
   does not increment even though the sender has been transmitting RTP
   data packets.  The former corresponds to the TCP congestion avoidance
   state, and indicates a congested path that [RFC4585]; it is still delivering data; only meaningful for
      implementations of RTP/AVPF profile [RFC4585] or the latter corresponds to a TCP timeout, and RTP/SAVPF
      profile [RFC5124].

   o  X is most likely due to estimated throughput a
   path failure.  A third condition is that data is being sent but no
   RTCP feedback is received at all, corresponding to TCP connection would achieve over a failure of the
   reverse path.  We derive circuit breaker conditions for these loss
   regimes
      path, in bytes per second.

   o  s is the following.

4.1.  RTP/AVP Circuit Breaker #1: Media Timeout

   If size of RTP data packets are being sent, but in bytes.  If the RTCP SR or RR packets
   reporting on that SSRC indicate a non-increasing extended highest
   sequence number received, this is an indication that those RTP data
      packets are not reaching being sent vary in size, then the receiver.  This could average size over the
      packet comprising the last 4 * G frames MUST be a short-term
   issue affecting only a few packets, perhaps caused used (this is
      intended to be comparable to the four loss intervals used in
      [RFC5348]).

   o  p is the loss event rate, between 0.0 and 1.0, that would be seen
      by a slow-to-open
   firewall or TCP connection over a transient connectivity problem, but if particular path.  When used in the issue
   persists, it RTP
      congestion circuit breaker, this is approximated as described in
      Section 4.3.

   o  t_RTO is the retransmission timeout value that would be used by a sign of a more ongoing and significant problem.
   Accordingly, if
      TCP connection over a sender particular path, in seconds.  This MUST be
      approximated using t_RTO = 4 * Tr when used as part of the RTP data packets receives CB_INTERVAL or
   more consecutive RTCP SR or RR packets from
      congestion circuit breaker.

   o  b is the same receiver (see
   Section 4.5), and those packets correspond to its transmission and
   have a non-increasing extended highest sequence number received
   field, then of packets that sender SHOULD cease transmission (see Section 4.6).
   The extended highest sequence number received field are acknowledged by a single TCP
      acknowledgement.  Following [RFC3448], it is non-increasing
   if the sender receives at least CB_INTERVAL consecutive RTCP SR or RR
   packets RECOMMENDED that report the same
      value for this field, but it has sent
   RTP data packets, at a rate b = 1 is used as part of at least one per RTT, that would have
   caused an increase in the reported value if they had reached RTP congestion circuit breaker.

4.  RTP Circuit Breakers for Systems Using the
   receiver. RTP/AVP Profile

   The rationale for waiting feedback mechanisms defined in [RFC3550] and available under the
   RTP/AVP profile [RFC3551] are the minimum that can be assumed for CB_INTERVAL or more consecutive RTCP
   packets with a non-increasing extended highest sequence number
   baseline circuit breaker mechanism that is to
   give enough time suitable for transient reception problems all unicast
   applications of RTP.  Accordingly, for an RTP circuit breaker to resolve
   themselves, but be
   useful, it needs to stop problem flows quickly enough be able to avoid detect that an RTP flow is causing
   serious ongoing network congestion.  A single RTCP report showing no
   reception could be caused by a transient fault, and so will not cease
   transmission.  Waiting for more than CB_INTERVAL consecutive
   excessive congestion using only basic RTCP
   reports before stopping a flow might avoid some false positives, but
   could lead to problematic flows running features, without needing
   RTCP XR feedback or the RTP/AVPF profile for rapid RTCP reports.

   RTCP is a long time period
   (potentially tens fundamental part of seconds, depending on the RTCP reporting
   interval) before being cut off.  Equally, an application that sends
   few packets when RTP protocol, and the packet loss rate is high runs mechanisms
   described here rely on the risk implementation of RTCP.  Implementations
   that claim to support RTP, but that do not implement RTCP, cannot use
   the
   media timeout circuit breaker triggers inadvertently.  The chosen
   timeout interval is a trade-off between these extremes. mechanisms described in this memo.  Such
   implementations SHOULD NOT be used on networks that might be subject
   to congestion unless equivalent mechanisms are defined using some
   non-RTCP feedback channel to report congestion and signal circuit
   breaker conditions.  The rationale for enforcing a minimum sending rate below which the
   media RTCP timeout circuit breaker (Section 4.1)
   will not trigger is if an implementation of this memo attempts to avoid spurious interwork
   with an endpoint that does not support RTCP.  Implementations that
   sometimes need to interwork with endpoints that do not support RTCP
   need to disable the RTP circuit breaker triggers when breakers if they don't receive some
   confirmation via signalling that the number remote endpoint implements RTCP
   (the presence of packets sent per an SDP "a=rtcp:" attribute in an answer might be
   such an indication).

   Three potential congestion signals are available from the basic RTCP
   reporting interval is small (e.g., a telephony application sends only
   two RTP comfort noise
   SR/RR packets during a five second RTCP reporting
   interval, and both are lost; this is 100% packet loss, but it seems
   extreme to terminate reported for each synchronisation source (SSRC)
   in the RTP session). session:

   1.  The one packet sender can estimate the network round-trip time once per RTT bound
   derives from [RFC5405].

4.2.  RTP/AVP Circuit Breaker #2: RTCP Timeout

   In addition to media timeouts, as were discussed in Section 4.1, an
   RTP session has
       reporting interval, based on the possibility contents and timing of an RTCP timeout.  This can occur
   when RTP data packets are being sent, but there are no RTCP reports
   returned from the receiver.  This is either due to SR
       and RR packets.

   2.  Receivers report a failure jitter estimate (the statistical variance of
       the
   receiver to send RTCP reports, or a failure of RTP data packet inter-arrival time) calculated over the return path that
   is preventing those RTCP
       reporting from being delivered.  In either
   case, it is not safe interval.  Due to continue transmission, since the sender has
   no way nature of knowing if it the jitter calculation
       ([RFC3550], section 6.4.4), the jitter is causing congestion.  Accordingly, an only meaningful for RTP
   sender
       flows that has not received any RTCP SR or RTCP RR packets reporting
   on the SSRC it is using send a single data packet for three each RTP timestamp value
       (i.e., audio flows, or more of its RTCP reporting
   intervals SHOULD cease transmission (see Section 4.6).  When
   calculating video flows where each packet comprises
       one video frame).

   3.  Receivers report the timeout, fraction of RTP data packets lost during the deterministic
       RTCP reporting interval,
   Td, without the randomization factor, and using the fixed minimum
   interval of Tmin=5 seconds, MUST be used.  The rationale for this
   choice of timeout is as described in Section 6.2 cumulative number of [RFC3550] ("so
   that implementations which do not use RTP packets
       lost over the entire RTP session.

   These congestion signals limit the possible circuit breakers, since
   they give only limited visibility into the behaviour of the network.

   RTT estimates are widely used in congestion control algorithms, as a
   proxy for queuing delay measures in delay-based congestion control or
   to determine connection timeouts.  RTT estimates derived from RTCP SR
   and RR packets sent according to the RTP/AVP timing rules are too
   infrequent to be useful for congestion control, and don't give enough
   information to distinguish a delay change due to routing updates from
   queuing delay caused by congestion.  Accordingly, we cannot use the
   RTT estimate alone as an RTP circuit breaker.

   Increased jitter can be a signal of transient network congestion, but
   in the highly aggregated form reported in RTCP RR packets, it offers
   insufficient information to estimate the extent or persistence of
   congestion.  Jitter reports are a useful early warning of potential
   network congestion, but provide an insufficiently strong signal to be
   used as a circuit breaker.

   The remaining congestion signals are the packet loss fraction and the
   cumulative number of packets lost.  If considered carefully, these
   can be effective indicators that congestion is occurring in networks
   where packet loss is primarily due to queue overflows, although loss
   caused by non-congestive packet corruption can distort the result in
   some networks.  TCP congestion control [RFC5681] intentionally tries
   to fill the router queues, and uses the resulting packet loss as
   congestion feedback.  An RTP flow competing with TCP traffic will
   therefore expect to see a non-zero packet loss fraction that has to
   be related to TCP dynamics to estimate available capacity.  This
   behaviour of TCP is reflected in the congestion circuit breaker
   below, and will affect the design of any RTP congestion control
   protocol.

   Two packet loss regimes can be observed: 1) RTCP RR packets show a
   non-zero packet loss fraction, while the extended highest sequence
   number received continues to increment; and 2) RR packets show a loss
   fraction of zero, but the extended highest sequence number received
   does not increment even though the sender has been transmitting RTP
   data packets.  The former corresponds to the TCP congestion avoidance
   state, and indicates a congested path that is still delivering data;
   the latter corresponds to a TCP timeout, and is most likely due to a
   path failure.  A third condition is that data is being sent but no
   RTCP feedback is received at all, corresponding to a failure of the
   reverse path.  We derive circuit breaker conditions for these loss
   regimes in the following.

4.1.  RTP/AVP Circuit Breaker #1: RTCP Timeout

   An RTCP timeout can occur when RTP data packets are being sent, but
   there are no RTCP reports returned from the receiver.  This is either
   due to a failure of the receiver to send RTCP reports, or a failure
   of the return path that is preventing those RTCP reporting from being
   delivered.  In either case, it is not safe to continue transmission,
   since the sender has no way of knowing if it is causing congestion.

   An RTP sender that has not received any RTCP SR or RTCP RR packets
   reporting on the SSRC it is using, for a time period of at least
   three times its deterministic RTCP reporting interval, Td, without
   the randomization factor, and using the fixed minimum interval of
   Tmin=5 seconds, SHOULD cease transmission (see Section 4.5).  The
   rationale for this choice of timeout is as described in Section 6.2
   of [RFC3550] ("so that implementations which do not use the reduced
   value for transmitting RTCP packets are not timed out by other
   participants prematurely"), as updated by Section 6.1.4 of
   [I-D.ietf-avtcore-rtp-multi-stream] to account for the use of the
   RTP/AVPF profile [RFC4585] or the RTP/SAVPF profile [RFC5124].

   To reduce the risk of premature timeout, implementations SHOULD NOT
   configure the RTCP bandwidth such that Td is larger than 5 seconds.
   Similarly, implementations that use the RTP/AVPF profile [RFC4585] or
   the RTP/SAVPF profile [RFC5124] SHOULD NOT configure T_rr_interval to
   values larger than 4 seconds (the reduced limit for T_rr_interval
   follows Section 6.1.3 of [I-D.ietf-avtcore-rtp-multi-stream]).

   The choice of three RTCP reporting intervals as the timeout is made
   following Section 6.3.5 of RFC 3550 [RFC3550].  This specifies that
   participants in an RTP session will timeout and remove an RTP sender
   from the list of active RTP senders if no RTP data packets have been
   received from that RTP sender within the last two RTCP reporting
   intervals.  Using a timeout of three RTCP reporting intervals is
   therefore large enough that the other participants will have timed
   out the sender if a network problem stops the reduced value data packets it is
   sending from reaching the receivers, even allowing for loss of some
   RTCP packets.

   If a sender is transmitting a large number of RTP media streams, such
   that the corresponding RTCP SR or RR packets are not timed out by other participants
   prematurely"), as updated by Section 6.1.4 of
   [I-D.ietf-avtcore-rtp-multi-stream] too large to account for fit
   into the use of network MTU, the RTP
   /AVPF profile [RFC4585] receiver will generate RTCP SR or RR
   packets in a round-robin manner.  In this case, the RTP/SAVPF profile [RFC5124].

   To reduce sender SHOULD
   treat receipt of an RTCP SR or RR packet corresponding to any SSRC it
   sent on the risk same 5-tuple of premature timeout, implementations SHOULD NOT
   configure source and destination IP address, port,
   and protocol, as an indication that the receiver and return path are
   working, preventing the RTCP bandwidth such timeout circuit breaker from triggering.

4.2.  RTP/AVP Circuit Breaker #2: Media Timeout

   If RTP data packets are being sent, but the RTCP SR or RR packets
   reporting on that Td SSRC indicate a non-increasing extended highest
   sequence number received, this is larger than 5 seconds.
   Similarly, implementations an indication that use those RTP data
   packets are not reaching the RTP/AVPF profile [RFC4585] receiver.  This could be a short-term
   issue affecting only a few RTP packets, perhaps caused by a slow to
   open firewall or a transient connectivity problem, but if the RTP/SAVPF profile [RFC5124] SHOULD NOT configure T_rr_interval to
   values larger than 4 seconds (the reduced limit for T_rr_interval
   follows Section 6.1.3 issue
   persists, it is a sign of [I-D.ietf-avtcore-rtp-multi-stream]). a more ongoing and significant problem (a
   "media timeout").

   The time needed to declare a media timeout depends on the parameters
   Tdr, Tr, Tf, and on the non-reporting threshold k.  The choice value of three RTCP reporting intervals as the timeout k is made
   following Section 6.3.5
   chosen so that when Tdr is large compared to Tr and Tf, receipt of RFC 3550 [RFC3550].  This specifies at
   least k RTCP reports with non-increasing extended highest sequence
   number received gives reasonable assurance that
   participants in an RTP session will timeout the forward path has
   failed, and remove an RTP sender
   from that the list of active RTP senders if no RTP data packets have not been
   received from that lost by chance.
   The RECOMMENDED value for k is 5 reports.

   When Tdr < Tf, then RTP sender within data packets are being sent at a rate less
   than one per RTCP reporting interval of the last two receiver, so the extended
   highest sequence number received can be expected to be non-increasing
   for some receiver RTCP reporting intervals.  Using a timeout of three  Similarly, when Tdr <
   Tr, some receiver RTCP reporting intervals might pass before the RTP
   data packets arrive at the receiver, also leading to reports where
   the extended highest sequence number received is non-increasing.
   Both issues require the media timeout interval to be scaled relative
   to the threshold, k.

   The media timeout RTP circuit breaker is therefore large enough as follows.  When
   starting sending, calculate MEDIA_TIMEOUT using:

      MEDIA_TIMEOUT = ceil(k * max(Tf, Tr, Tdr) / Tdr)

   When a sender receives an RTCP packet indicating that the other participants will have timed
   out media it's
   sending is being received, then it cancels the media timeout circuit
   breaker.  If it is still sending, then it MUST calculate a new value
   for MEDIA_TIMEOUT, and set a new media timeout circuit breaker.

   If a sender if receives an RTCP packet indicating that its media was not
   received, it MUST calculate a network problem stops new value for MEDIA_TIMEOUT.  If the data packets
   new value is larger than the previous, is replaces MEDIA_TIMEOUT with
   the new value, extending the media timeout circuit breaker; otherwise
   it is
   sending from reaching keeps the receivers, even allowing for loss original value of some
   RTCP packets.

   If a sender MEDIA_TIMEOUT.  This process is transmitting a large number of RTP media streams, such
   that known
   as reconsidering the corresponding media timeout circuit breaker.

   If MEDIA_TIMEOUT consecutive RTCP SR or RR packets are too large to fit
   into received indicating
   that the network MTU, media being sent was not received, and the receiver will generate RTCP SR or RR
   packets in a round-robin manner.  In this case, media timeout
   circuit breaker has not been cancelled, then the media timeout
   circuit breaker triggers.  When the media timeout circuit breaker
   triggers, the sender SHOULD
   treat receipt of an RTCP SR or RR packet corresponding to any SSRC it
   sent on the same 5-tuple of source and destination IP address, port,
   and protocol, as cease transmission (see Section 4.5).

   When stopping sending an indication that the receiver and return path are
   working, preventing RTP stream, a sender MUST cancel the RTCP
   corresponding media timeout circuit breaker from triggering. breaker.

4.3.  RTP/AVP Circuit Breaker #3: Congestion

   If RTP data packets are being sent, and the corresponding RTCP SR or
   RR packets show non-zero packet loss fraction and increasing extended
   highest sequence number received, then those RTP data packets are
   arriving at the receiver, but some degree of congestion is occurring.
   The RTP/AVP profile [RFC3551] states that:

      If best-effort service is being used, RTP receivers SHOULD monitor
      packet loss to ensure that the packet loss rate is within
      acceptable parameters.  Packet loss is considered acceptable if a
      TCP flow across the same network path and experiencing the same
      network conditions would achieve an average throughput, measured
      on a reasonable time scale, that is not less than the RTP flow is
      achieving.  This condition can be satisfied by implementing
      congestion control mechanisms to adapt the transmission rate (or
      the number of layers subscribed for a layered multicast session),
      or by arranging for a receiver to leave the session if the loss
      rate is unacceptably high.

      The comparison to TCP cannot be specified exactly, but is intended
      as an "order-of-magnitude" comparison in time scale and
      throughput.  The time scale on which TCP throughput is measured is
      the round-trip time of the connection.  In essence, this
      requirement states that it is not acceptable to deploy an
      application (using RTP or any other transport protocol) on the
      best-effort Internet which consumes bandwidth arbitrarily and does
      not compete fairly with TCP within an order of magnitude.

   The phase "order of magnitude" in the above means within a factor of
   ten, approximately.  In order to implement this, it is necessary to
   estimate the throughput a TCP connection would achieve over the path.
   For a long-lived TCP Reno connection, it has been shown that the TCP
   throughput can be estimated using the following equation [Padhye]:

                                     s
     X = --------------------------------------------------------------
         R*sqrt(2*b*p/3) + (t_RTO * (3*sqrt(3*b*p/8) * p * (1+32*p^2)))

   where:

   X  is the transmit rate in bytes/second.

   s  is the packet size in bytes.  If data packets vary in size, then
      the average size is to be used.

   R  is the round trip time in seconds.

   p  is the loss event rate, between 0 and 1.0, of the number of loss
      events as a fraction of the number of packets transmitted.

   t_RTO  is the TCP retransmission timeout value in seconds, generally
      approximated by setting t_RTO = 4*R.

   b  is the number of packets that are acknowledged by a single TCP
      acknowledgement; [RFC3448] recommends that the use of b=1 since many TCP implementations do not use delayed acknowledgements.
   throughput, X, in bytes per second, can be estimated using [Padhye]:

                                  s
      X = -------------------------------------------------------------
          Tr*sqrt(2*b*p/3)+(t_RTO * (3*sqrt(3*b*p/8) * p * (1+32*p*p)))

   This is the same approach to estimated TCP throughput that is used in
   [RFC3448].  Under conditions of low packet loss the second term on
   the denominator is small, so this formula can be approximated with
   reasonable accuracy as follows [Mathis]:

                s
      X = -----------------
                R * sqrt(2*b*p/3) ----------------
          Tr*sqrt(2*b*p/3)

   It is RECOMMENDED that this simplified throughout equation be used,
   since the reduction in accuracy is small, and it is much simpler to
   calculate than the full equation.  Measurements have shown that the
   simplified TCP throughput equation is effective as an RTP circuit
   breaker for multimedia flows sent to hosts on residential networks
   using ADSL and cable modem links [Singh].  The data shows that the
   full TCP throughput equation tends to be more sensitive to packet
   loss and triggers the RTP circuit breaker earlier than the simplified
   equation.  Implementations that desire this extra sensitivity MAY use
   the full TCP throughput equation in the RTP circuit breaker.  Initial
   measurements in LTE networks have shown that the extra sensitivity is
   helpful in that environment, with the full TCP throughput equation
   giving a more balanced circuit breaker response than the simplified
   TCP equation [Sarker]; other networks might see similar behaviour.

   No matter what TCP throughput equation is chosen, two parameters need
   to be estimated and reported to the sender in order to calculate the
   throughput: the round trip time, R, Tr, and the loss event rate, p (the
   packet size, s, is known to the sender).  The round trip time can be
   estimated from RTCP SR and RR packets.  This is done too infrequently
   for accurate statistics, but is the best that can be done with the
   standard RTCP mechanisms.

   Report blocks in RTCP SR or RR packets contain the packet loss
   fraction, rather than the loss event rate, so p cannot be reported
   (TCP typically treats the loss of multiple packets within a single
   RTT as one loss event, but RTCP RR packets report the overall
   fraction of packets lost, and does not report when the packet losses
   occurred).  Using the loss fraction in place of the loss event rate
   can overestimate the loss.  We believe that this overestimate will
   not be significant, given that we are only interested in order of
   magnitude comparison ([Floyd] section 3.2.1 shows that the difference
   is small for steady-state conditions and random loss, but using the
   loss fraction is more conservative in the case of bursty loss).

   The congestion circuit breaker is therefore: when a sender that is
   transmitting more than at least one RTP packet per RTT every max(Tdr, Tr) seconds
   receives an RTCP SR or RR packet that contains a report block for an
   SSRC it is using, the sender MUST record the value of the fraction
   lost field in the report block and the time since the last report
   block was received for that SSRC.  If more than CB_INTERVAL (see Section 4.5)
   below) report blocks have been received for that SSRC, the sender
   MUST calculate the average fraction lost over the last CB_INTERVAL
   reporting intervals, and then estimate the TCP throughput that would
   be achieved over the path using the chosen TCP throughput equation
   and the measured values of the round-trip time, R, Tr, the loss event
   rate, p (as approximated (approximated by the average fraction lost), lost, as is described
   below), and the packet size, s.  This estimate of
   the TCP throughput is then compared with the actual sending rate.  If
   the actual sending rate is more than ten times the TCP throughput
   estimate, then the congestion circuit breaker is triggered.

   The average fraction lost is calculated based on the sum, over the
   last CB_INTERVAL reporting intervals, of the fraction lost in each
   reporting interval multiplied by the duration of the corresponding
   reporting interval, divided by the total duration of the last
   CB_INTERVAL reporting intervals.

   The rationale for enforcing a minimum sending rate below which the
   congestion circuit breaker will not trigger is to avoid spurious
   circuit breaker triggers when the number of packets sent per RTCP
   reporting interval is small, and hence the fraction lost samples are
   subject to measurement artefacts.  The one packet per RTT bound
   derives from [RFC5405].

   When the congestion circuit breaker is triggered, the sender SHOULD
   cease transmission (see Section 4.6).  However, if the sender is able
   to reduce its sending rate by a factor of (approximately) ten, then
   it MAY first reduce its sending rate by this factor (or some larger
   amount) to see if that resolves the congestion.  If the sending rate
   is reduced in this way and the congestion circuit breaker triggers
   again after the next CB_INTERVAL RTCP reporting intervals, the sender
   MUST then cease transmission.  An example estimate of such a rate reduction
   might be a video conferencing system that backs off to sending audio
   only, before completely dropping the call.  If such a reduction in TCP throughput,
   X, is then compared with the actual sending rate resolves the congestion problem, of the sender MAY
   gradually increase RTP stream.
   If the actual sending rate at which it sends data after a reasonable
   amount of time has passed, provided it takes care not to cause the
   problem to recur ("reasonable" RTP stream is intentionally not defined here).

   The RTCP reporting interval of more than 10 * X,
   then the media sender does not affect how
   quickly congestion circuit breaker can trigger. is triggered.

   The timing average fraction lost is calculated based on the RTCP reporting interval of the receiver that generates the SR/
   RR packets from which sum, over the loss rate and RTT estimate are derived
   (note that RTCP requires all participants in a session to have
   similar
   last CB_INTERVAL reporting intervals, else of the participant timeout rules fraction lost in
   [RFC3550] will not work, so this each
   reporting interval is likely similar to that
   of multiplied by the sender).  If duration of the incoming RTCP SR or RR packets are using a
   reduced minimum RTCP corresponding
   reporting interval (as specified in Section 6.2 interval, divided by the total duration of RFC 3550 [RFC3550] or the RTP/AVPF profile [RFC4585]), then that
   reduced RTCP last
   CB_INTERVAL reporting interval intervals.  The CB_INTERVAL parameter is used when determining if set
   to:

      CB_INTERVAL =
         ceil(3*min(max(10*G*Tf, 10*Tr, 3*Tdr), max(15, 3*Td))/(3*Tdr))

   The parameters that feed into CB_INTERVAL are chosen to give the
   circuit breaker is triggered.

   As in Section 4.1
   congestion control algorithm time to react to congestion.  They give
   at least three RTCP reports, ten round trip times, and Section 4.2, we use CB_INTERVAL reporting
   intervals ten groups of
   frames to avoid triggering adjust the circuit breaker on transient
   failures.  This circuit breaker is a worst-case condition, and
   congestion control needs rate to be performed reduce the congestion to keep well within this
   bound. a reasonable
   level.  It is expected that the circuit breaker a responsive congestion control algorithm
   will only begin to respond with the next group of frames after it receives
   indication of congestion, so CB_INTERVAL ought to be
   triggered if a much longer
   interval than the usual congestion control fails for some reason. response.

   If there are more media streams that can be reported in a single RTCP
   SR or RR packet, the RTP/AVPF profile [RFC4585] or if the size RTP/SAVPF [RFC5124] is used,
   and the T_rr_interval parameter is used to reduce the frequency of a complete
   regular RTCP SR or RR packet
   exceeds the network MTU, reports, then the receiver will report on a subset of
   sources value Tdr in each reporting interval, with the subsets selected round-
   robin across multiple intervals so that all sources are eventually
   reported [RFC3550].  When generating such round-robin RTCP reports,
   priority SHOULD above expression for
   the CB_INTERVAL parameter MUST be given to reports on sources that have high packet
   loss rates, to ensure that senders are aware of network congestion
   they are causing (this replaced by max(T_rr_interval,
   Tdr).

   The CB_INTERVAL parameter is an update to [RFC3550]).

4.4.  RTP/AVP Circuit Breaker #4: Media Usability

   Applications that use RTP are generally tolerant to some amount of
   packet loss.  How much packet loss can be tolerated will depend calculated on joining the application, media codec, session, and
   recalculated on receipt of each RTCP packet, after checking whether
   the media timeout circuit breaker or the congestion circuit breaker
   has been triggered.

   To ensure a timely response to persistent congestion, implementations
   SHOULD NOT configure the amount of error correction and
   packet loss concealment RTCP bandwidth such that Tdr is applied.  There is an upper bound on larger than
   5 seconds.  Similarly, implementations that use the amount RTP/AVPF profile
   [RFC4585] or the RTP/SAVPF profile [RFC5124] SHOULD NOT configure
   T_rr_interval to values larger than 4 seconds (the reduced limit for
   T_rr_interval follows Section 6.1.3 of loss can be corrected, however, beyond
   [I-D.ietf-avtcore-rtp-multi-stream]).

   The rationale for enforcing a minimum sending rate below which the media
   becomes unusable.  Similarly, many applications have some upper bound
   on the media capture
   congestion circuit breaker will not trigger is to play-out latency that can be tolerated before avoid spurious
   circuit breaker triggers when the application becomes unusable. number of packets sent per RTCP
   reporting interval is small, and hence the fraction lost samples are
   subject to measurement artefacts.  The latency bound will depend on
   the application, but typical values can range of at least one packet
   every max(Tdr, Tr) seconds is derived from the order one packet per RTT
   minimum sending rate of a few
   hundred milliseconds for voice telephony and interactive conferencing
   applications, up to several seconds TCP [RFC5405], adapted for some video-on-demand systems.

   As a final circuit breaker, use with RTP senders SHOULD monitor where
   the reported
   packet loss and delay to estimate whether RTCP reporting interval is decoupled from the media network RTT.

   When the congestion circuit breaker is likely to be
   suitable for triggered, the intended purpose. sender SHOULD
   cease transmission (see Section 4.5).  However, if the sender is able
   to reduce its sending rate by a factor of (approximately) ten, then
   it MAY first reduce its sending rate by this factor (or some larger
   amount) to see if that resolves the congestion.  If the packet loss sending rate and/or
   latency
   is such that the media has become unusable, reduced in this way and has remained
   unusable for a significant time period, then the application SHOULD congestion circuit breaker triggers
   again after the next CB_INTERVAL RTCP reporting intervals, the sender
   MUST then cease transmission.  Similarly, receivers SHOULD monitor the quality  An example of such a rate reduction
   might be a video conferencing system that backs off to sending audio
   only, before completely dropping the media they receive, and if call.  If such a reduction in
   sending rate resolves the quality is unusable for congestion problem, the sender MAY
   gradually increase the rate at which it sends data after a
   significant reasonable
   amount of time period, they SHOULD terminate has passed, provided it takes care not to cause the session.  This
   memo
   problem to recur ("reasonable" is intentionally not defined here).

   The RTCP reporting interval of the media sender does not define a bound affect how
   quickly congestion circuit breaker can trigger.  The timing is based
   on the packet RTCP reporting interval of the receiver that generates the SR/
   RR packets from which the loss rate or
   latency and RTT estimate are derived
   (note that will result RTCP requires all participants in unusable media, nor does it specify what
   time period a session to have
   similar reporting intervals, else the participant timeout rules in
   [RFC3550] will not work, so this interval is deemed significant, as these are highly application
   dependent.

   Sending media likely similar to that suffers from such high packet loss
   of the sender).  If the incoming RTCP SR or RR packets are using a
   reduced minimum RTCP reporting interval (as specified in Section 6.2
   of RFC 3550 [RFC3550] or latency the RTP/AVPF profile [RFC4585]), then that
   it
   reduced RTCP reporting interval is unusable at used when determining if the receiver
   circuit breaker is both wasteful of resources, and of
   no benefit to triggered.

   If there are more media streams that can be reported in a single RTCP
   SR or RR packet, or if the user size of a complete RTCP SR or RR packet
   exceeds the application.  It also is highly likely
   to be congesting the network, and disrupting other applications.  As
   such, network MTU, then the congestion circuit breaker receiver will almost certainly trigger to
   stop flows where report on a subset of
   sources in each reporting interval, with the media would subsets selected round-
   robin across multiple intervals so that all sources are eventually
   reported [RFC3550].  When generating such round-robin RTCP reports,
   priority SHOULD be unusable due given to reports on sources that have high packet
   loss
   or latency.  However, in pathological scenarios where the congestion
   circuit breaker does not stop the flow, it is desirable rates, to ensure that the RTP
   application cease sending useless traffic.  The role senders are aware of the media
   usability circuit breaker network congestion
   they are causing (this is an update to protect the network in such cases.

4.5.  Choice of [RFC3550]).

4.4.  RTP/AVP Circuit Breaker Interval

   The CB_INTERVAL parameter determines the number of consecutive RTCP
   reporting intervals #4: Media Usability

   Applications that need use RTP are generally tolerant to suffer congestion before the media
   timeout circuit breaker (see Section 4.1) or the congestion circuit
   breaker (see Section 4.3) triggers.  It determines some amount of
   packet loss.  How much packet loss can be tolerated will depend on
   the sensitivity application, media codec, and responsiveness the amount of these circuit breakers.

   The CB_INTERVAL parameter error correction and
   packet loss concealment that is set to min(floor(3+(2.5/Td)), 30) RTCP
   reporting intervals, where Td applied.  There is an upper bound on
   the deterministic calculated RTCP
   interval described in section 6.3.1 amount of [RFC3550].  This expression
   gives an CB_INTERVAL loss can be corrected, however, beyond which the media
   becomes unusable.  Similarly, many applications have some upper bound
   on the media capture to play-out latency that varies as follows:

            Td       |       CB_INTERVAL            | Time can be tolerated before
   the application becomes unusable.  The latency bound will depend on
   the application, but typical values can range from the order of a few
   hundred milliseconds for voice telephony and interactive conferencing
   applications, up to trigger
       --------------+------------------------------+-----------------
       0.016 seconds |  30 RTCP reporting intervals |  0.48 seconds
       0.033 seconds |  30 RTCP reporting intervals |  0.99 seconds
       0.100 seconds |  28 RTCP reporting intervals |  2.80 seconds
       0.500 seconds |   8 RTCP reporting intervals |  4.00 seconds
       1.000 seconds |   5 RTCP reporting intervals |  5.00 seconds
       2.000 seconds |   4 RTCP reporting intervals |  8.00 seconds
       5.000 seconds |   3 RTCP reporting intervals | 15.00 seconds
      10.000 seconds |   3 RTCP reporting intervals | 30.00 several seconds

   If the RTP/AVPF profile [RFC4585] or for some video-on-demand systems.

   As a final circuit breaker, RTP senders SHOULD monitor the RTP/SAVPF [RFC5124] is used, reported
   packet loss and delay to estimate whether the T_rr_interval parameter media is used likely to reduce be
   suitable for the frequency of
   regular RTCP reports, then intended purpose.  If the value Td in packet loss rate and/or
   latency is such that the above expression media has become unusable, and has remained
   unusable for a significant time period, then the CB_INTERVAL parameter MUST be replaced by max(T_rr_interval, Td).

   The CB_INTERVAL parameter is calculated on joining application SHOULD
   cease transmission.  Similarly, receivers SHOULD monitor the session, and
   recalculated on receipt quality
   of each RTCP packet, after checking whether the media timeout circuit breaker or they receive, and if the congestion circuit breaker
   has been triggered.

   To ensure quality is unusable for a timely response to persistent congestion, implementations
   significant time period, they SHOULD NOT configure terminate the RTCP bandwidth such session.  This
   memo intentionally does not define a bound on the packet loss rate or
   latency that Td will result in unusable media, nor does it specify what
   time period is larger than 5
   seconds.  Similarly, implementations deemed significant, as these are highly application
   dependent.

   Sending media that use the RTP/AVPF profile
   [RFC4585] suffers from such high packet loss or latency that
   it is unusable at the RTP/SAVPF profile [RFC5124] SHOULD NOT configure
   T_rr_interval to values larger than 4 seconds (the reduced limit for
   T_rr_interval follows Section 6.1.3 receiver is both wasteful of
   [I-D.ietf-avtcore-rtp-multi-stream]).

   Rationale: If the CB_INTERVAL was always set to the same number resources, and of
   RTCP reporting intervals, this would cause higher rate RTP sessions
   no benefit to trigger the RTP circuit breaker after a shorter time interval than
   lower rate sessions, because the RTCP reporting interval scales based
   on user of the RTP session bandwidth.  This application.  It also is felt to penalise high rate RTP
   sessions too aggressively.  Conversely, scaling CB_INTERVAL according highly likely
   to be congesting the inverse of the RTCP reporting interval, so network, and disrupting other applications.  As
   such, the RTP congestion circuit breaker triggers after a constant time interval, doesn't sufficiently
   protect the network from congestion caused by high-rate flows.  The
   chosen expression for CB_INTERVAL seeks a balance between these two
   extremes.  It causes higher rate RTP sessions subject will almost certainly trigger to persistent
   congestion
   stop flows where the media would be unusable due to trigger high packet loss
   or latency.  However, in pathological scenarios where the RTP congestion
   circuit breaker after a shorter time
   interval than do lower rate RTP sessions, while also making does not stop the flow, it is desirable that the RTP
   application cease sending useless traffic.  The role of the media
   usability circuit breaker for such sessions less sensitive by requiring the
   congestion is to persist for longer numbers of RTCP reporting intervals.

4.6. protect the network in such cases.

4.5.  Ceasing Transmission

   What it means to cease transmission depends on the application, but
   the intention is that the application will stop sending RTP data
   packets to a particular destination 3-tuple (transport protocol,
   destination port, IP address), until the user makes an explicit
   attempt to restart the call.  It is important that a human user is
   involved in the decision to try to restart the call, since that user
   will eventually give up if the calls repeatedly trigger the circuit
   breaker.  This will help avoid problems with automatic redial systems
   from congesting the network.  Accordingly, RTP flows halted by the
   circuit breaker SHOULD NOT be restarted automatically unless the
   sender has received information that the congestion has dissipated.

   It is recognised that the RTP implementation in some systems might
   not be able to determine if a call set-up request was initiated by a
   human user, or automatically by some scripted higher-level component
   of the system.  These implementations SHOULD rate limit attempts to
   restart a call to the same destination 3-tuple as used by a previous
   call that was recently halted by the circuit breaker.  The chosen
   rate limit ought to not exceed the rate at which an annoyed human
   caller might redial a misbehaving phone.

5.  RTP Circuit Breakers and the RTP/AVPF and RTP/SAVPF Profiles

   Use of the Extended RTP Profile for RTCP-based Feedback (RTP/AVPF)
   [RFC4585] allows receivers to send early RTCP reports in some cases,
   to inform the sender about particular events in the media stream.
   There are several use cases for such early RTCP reports, including
   providing rapid feedback to a sender about the onset of congestion.
   The RTP/SAVPF Profile [RFC5124] is a secure variant of the RTP/AVPF
   profile, that is treated the same in the context of the RTP circuit
   breaker.  These feedback profiles are often used with non-compound
   RTCP reports [RFC5506] to reduce the reporting overhead.

   Receiving rapid feedback about congestion events potentially allows
   congestion control algorithms to be more responsive, and to better
   adapt the media transmission to the limitations of the network.  It
   is expected that many RTP congestion control algorithms will adopt
   the RTP/AVPF profile or the RTP/SAVPF profile for this reason,
   defining new transport layer feedback reports that suit their
   requirements.  Since these reports are not yet defined, and likely
   very specific to the details of the congestion control algorithm
   chosen, they cannot be used as part of the generic RTP circuit
   breaker.

   Reduced-size RTCP reports sent under the RTP/AVPF early feedback
   rules that do not contain an RTCP SR or RR packet MUST be ignored by
   the congestion circuit breaker (they do not contain the information
   needed by the congestion circuit breaker algorithm), but MUST be
   counted as received packets for the RTCP timeout circuit breaker.
   Reduced-size RTCP reports sent under the RTP/AVPF early feedback
   rules that contain RTCP SR or RR packets MUST be processed by the
   congestion circuit breaker as if they were sent as regular RTCP
   reports, and counted towards the circuit breaker conditions specified
   in Section 4 of this memo.  This will potentially make the RTP
   circuit breaker trigger earlier than it would if the RTP/AVPF profile
   was not used.

   When using ECN with RTP (see Section 8), 7), early RTCP feedback packets
   can contain ECN feedback reports.  The count of ECN-CE marked packets
   contained in those ECN feedback reports is counted towards the number
   of lost packets reported if the ECN Feedback Report report is sent in
   an compound RTCP packet along with an RTCP SR/RR report packet.
   Reports of ECN-CE packets sent as reduced-size RTCP ECN feedback
   packets without an RTCP SR/RR packet MUST be ignored.

   These rules are intended to allow the use of low-overhead RTP/AVPF
   feedback for generic NACK messages without triggering the RTP circuit
   breaker.  This is expected to make such feedback suitable for RTP
   congestion control algorithms that need to quickly report loss events
   in between regular RTCP reports.  The reaction to reduced-size RTCP
   SR/RR packets is to allow such algorithms to send feedback that can
   trigger the circuit breaker, when desired.

   The RTP/AVPF and RTP/SAVPF profiles include the T_rr_interval
   parameter that can be used to adjust the regular RTCP reporting
   interval.  The use of the T_rr_interval parameter changes the
   behaviour of the RTP circuit breaker, as described in Section 4.

6.  Impact of RTCP Extended Reports (XR)

   RTCP Extended Report (XR) blocks provide additional reception quality
   metrics, but do not change the RTCP timing rules.  Some of the RTCP
   XR blocks provide information that might be useful for congestion
   control purposes, others provided non-congestion-related metrics.
   With the exception of RTCP XR ECN Summary Reports (see Section 8), 7),
   the presence of RTCP XR blocks in a compound RTCP packet does not
   affect the RTP circuit breaker algorithm.  For consistency and ease
   of implementation, only the reception report blocks contained in RTCP
   SR packets, RTCP RR packets, or RTCP XR ECN Summary Report packets,
   are used by the RTP circuit breaker algorithm.

7.  Impact of RTCP Reporting Groups

   An optimisation for grouping RTCP reception statistics and other
   feedback in RTP sessions with large numbers of participants is given
   in [I-D.ietf-avtcore-rtp-multi-stream-optimisation].  This allows one
   SSRC to act as a representative that sends reports on behalf of other
   SSRCs that are co-located in the same endpoint and see identical
   reception quality.  When running the circuit breaker algorithms, an
   endpoint MUST treat a reception report from the representative of the
   reporting group as if a reception report was received from all
   members of that group.

8.  Impact of Explicit Congestion Notification (ECN)

   The use of ECN for RTP flows does not affect the media timeout RTP
   circuit breaker (Section 4.1) 4.2) or the RTCP timeout circuit breaker
   (Section 4.2), 4.1), since these are both connectivity checks that simply
   determinate if any packets are being received.

   ECN-CE marked packets SHOULD be treated as if it were lost for the
   purposes of congestion control, when determining the optimal media
   sending rate for an RTP flow.  If an RTP sender has negotiated ECN
   support for an RTP session, and has successfully initiated ECN use on
   the path to the receiver [RFC6679], then ECN-CE marked packets SHOULD
   be treated as if they were lost when calculating if the congestion-
   based RTP circuit breaker (Section 4.3) has been met.  The count of
   ECN-CE marked RTP packets is returned in RTCP XR ECN summary report
   packets if support for ECN has been initiated for an RTP session.

9.

8.  Impact of Bundled Media and Layered Coding

   The RTP circuit breaker operates on a per-RTP session basis.  An RTP
   sender that participates in several RTP sessions MUST treat each RTP
   session independently with regards to the RTP circuit breaker.

   An RTP sender can generate several media streams within a single RTP
   session, with each stream using a different SSRC.  This can happen if
   bundled media are in use, when using simulcast, or when using layered
   media coding.  By default, each SSRC will be treated independently by
   the RTP circuit breaker.  However, the sender MAY choose to treat the
   flows (or a subset thereof) as a group, such that a circuit breaker
   trigger for one flow applies to the group of flows as a whole, and
   either causes the entire group to cease transmission, or the sending
   rate of the group to reduce by a factor of ten, depending on the RTP
   circuit breaker triggered.  Grouping flows in this way is expected to
   be especially useful for layered flows sent using multiple SSRCs, as
   it allows the layered flow to react as a whole, ceasing transmission
   on the enhancement layers first to reduce sending rate if necessary,
   rather than treating each layer independently.

10.

9.  Security Considerations

   The security considerations of [RFC3550] apply.

   If the RTP/AVPF profile is used to provide rapid RTCP feedback, the
   security considerations of [RFC4585] apply.  If ECN feedback for RTP
   over UDP/IP is used, the security considerations of [RFC6679] apply.

   If non-authenticated RTCP reports are used, an on-path attacker can
   trivially generate fake RTCP packets that indicate high packet loss
   rates, causing the circuit breaker to trigger and disrupting an RTP
   session.  This is somewhat more difficult for an off-path attacker,
   due to the need to guess the randomly chosen RTP SSRC value and the
   RTP sequence number.  This attack can be avoided if RTCP packets are
   authenticated; authentication options are discussed in [RFC7201].

   Timely operation of the RTP circuit breaker depends on the choice of
   RTCP reporting interval.  If the receiver has a reporting interval
   that is overly long, then the responsiveness of the circuit breaker
   decreases.  In the limit, the RTP circuit breaker can be disabled for
   all practical purposes by configuring an RTCP reporting interval that
   is many minutes duration.  This issue is not specific to the circuit
   breaker: long RTCP reporting intervals also prevent reception quality
   reports, feedback messages, codec control messages, etc., from being
   used.  Implementations are expected to impose an upper limit on the
   RTCP reporting interval they are willing to negotiate (based on the
   session bandwidth and RTCP bandwidth fraction) when using the RTP
   circuit breaker, as discussed in Section 4.5.

11. 4.3.

10.  IANA Considerations

   There are no actions for IANA.

12.

11.  Acknowledgements

   The authors would like to thank Bernard Aboba, Harald Alvestrand,
   Gorry Fairhurst, Nazila Fough, Kevin Gross, Cullen Jennings, Randell
   Jesup, Jonathan Lennox, Matt Mathis, Stephen McQuistin, Simon
   Perreault, Eric Rescorla, Abheek Saha, Fabio Verdicchio, and Magnus
   Westerlund for their valuable feedback.

13.

12.  References

13.1.

12.1.  Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997. 1997,
              <http://www.rfc-editor.org/info/rfc2119>.

   [RFC3448]  Handley, M., Floyd, S., Padhye, J., and J. Widmer, "TCP
              Friendly Rate Control (TFRC): Protocol Specification",
              RFC 3448, DOI 10.17487/RFC3448, January 2003. 2003,
              <http://www.rfc-editor.org/info/rfc3448>.

   [RFC3550]  Schulzrinne, H., Casner, S., Frederick, R., and V.
              Jacobson, "RTP: A Transport Protocol for Real-Time
              Applications", STD 64, RFC 3550, DOI 10.17487/RFC3550,
              July 2003. 2003, <http://www.rfc-editor.org/info/rfc3550>.

   [RFC3551]  Schulzrinne, H. and S. Casner, "RTP Profile for Audio and
              Video Conferences with Minimal Control", STD 65, RFC 3551,
              DOI 10.17487/RFC3551, July 2003. 2003,
              <http://www.rfc-editor.org/info/rfc3551>.

   [RFC3611]  Friedman, T., Ed., Caceres, R., Ed., and A. Clark, Ed.,
              "RTP Control Protocol Extended Reports (RTCP XR)",
              RFC 3611, DOI 10.17487/RFC3611, November
              2003. 2003,
              <http://www.rfc-editor.org/info/rfc3611>.

   [RFC4585]  Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey,
              "Extended RTP Profile for Real-time Transport Control
              Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585,
              DOI 10.17487/RFC4585, July
              2006.

13.2. 2006,
              <http://www.rfc-editor.org/info/rfc4585>.

12.2.  Informative References

   [Floyd]    Floyd, S., Handley, M., Padhye, J., and J. Widmer,
              "Equation-Based Congestion Control for Unicast
              Applications", Proceedings of the ACM SIGCOMM
              conference, 2000, DOI 10.1145/347059.347397, August 2000.

   [I-D.ietf-avtcore-rtp-multi-stream-optimisation]
              Lennox, J., Westerlund, M., Wu, W., and C. Perkins,
              "Sending Multiple Media Streams in a Single RTP Session:
              Grouping RTCP Reception Statistics and Other Feedback",
              draft-ietf-avtcore-rtp-multi-stream-optimisation-05 (work
              in progress), February 2015.

   [I-D.ietf-avtcore-rtp-multi-stream]
              Lennox, J., Westerlund, M., Wu, W., and C. Perkins,
              "Sending Multiple Media Streams in a Single RTP Session",
              draft-ietf-avtcore-rtp-multi-stream-07
              draft-ietf-avtcore-rtp-multi-stream-09 (work in progress),
              March
              September 2015.

   [I-D.ietf-tsvwg-circuit-breaker]
              Fairhurst, G., "Network Transport Circuit Breakers",
              draft-ietf-tsvwg-circuit-breaker-00
              draft-ietf-tsvwg-circuit-breaker-05 (work in progress),
              September 2014.
              October 2015.

   [Mathis]   Mathis, M., Semke, J., Mahdavi, J., and T. Ott, "The
              macroscopic behavior of the TCP congestion avoidance
              algorithm", ACM SIGCOMM Computer Communication
              Review 27(3), DOI 10.1145/263932.264023, July 1997.

   [Padhye]   Padhye, J., Firoiu, V., Towsley, D., and J. Kurose,
              "Modeling TCP Throughput: A Simple Model and its Empirical
              Validation", Proceedings of the ACM SIGCOMM
              conference, 1998, DOI 10.1145/285237.285291, August 1998.

   [RFC2862]  Civanlar, M. and G. Cash, "RTP Payload Format for Real-
              Time Pointers", RFC 2862, DOI 10.17487/RFC2862, June 2000,
              <http://www.rfc-editor.org/info/rfc2862>.

   [RFC3168]  Ramakrishnan, K., Floyd, S., and D. Black, "The Addition
              of Explicit Congestion Notification (ECN) to IP",
              RFC 3168, DOI 10.17487/RFC3168, September 2001. 2001,
              <http://www.rfc-editor.org/info/rfc3168>.

   [RFC4103]  Hellstrom, G. and P. Jones, "RTP Payload for Text
              Conversation", RFC 4103, DOI 10.17487/RFC4103, June 2005,
              <http://www.rfc-editor.org/info/rfc4103>.

   [RFC5104]  Wenger, S., Chandra, U., Westerlund, M., and B. Burman,
              "Codec Control Messages in the RTP Audio-Visual Profile
              with Feedback (AVPF)", RFC 5104, DOI 10.17487/RFC5104,
              February 2008. 2008, <http://www.rfc-editor.org/info/rfc5104>.

   [RFC5124]  Ott, J. and E. Carrara, "Extended Secure RTP Profile for
              Real-time Transport Control Protocol (RTCP)-Based Feedback
              (RTP/SAVPF)", RFC 5124, DOI 10.17487/RFC5124, February 2008.
              2008, <http://www.rfc-editor.org/info/rfc5124>.

   [RFC5348]  Floyd, S., Handley, M., Padhye, J., and J. Widmer, "TCP
              Friendly Rate Control (TFRC): Protocol Specification",
              RFC 5348, DOI 10.17487/RFC5348, September 2008,
              <http://www.rfc-editor.org/info/rfc5348>.

   [RFC5405]  Eggert, L. and G. Fairhurst, "Unicast UDP Usage Guidelines
              for Application Designers", BCP 145, RFC 5405,
              DOI 10.17487/RFC5405, November
              2008. 2008,
              <http://www.rfc-editor.org/info/rfc5405>.

   [RFC5450]  Singer, D. and H. Desineni, "Transmission Time Offsets in
              RTP Streams", RFC 5450, DOI 10.17487/RFC5450, March 2009. 2009,
              <http://www.rfc-editor.org/info/rfc5450>.

   [RFC5506]  Johansson, I. and M. Westerlund, "Support for Reduced-Size
              Real-Time Transport Control Protocol (RTCP): Opportunities
              and Consequences", RFC 5506, DOI 10.17487/RFC5506, April 2009.
              2009, <http://www.rfc-editor.org/info/rfc5506>.

   [RFC5681]  Allman, M., Paxson, V., and E. Blanton, "TCP Congestion
              Control", RFC 5681, DOI 10.17487/RFC5681, September 2009.

   [RFC6051]  Perkins, C. and T. Schierl, "Rapid Synchronisation of RTP
              Flows", RFC 6051, November 2010. 2009,
              <http://www.rfc-editor.org/info/rfc5681>.

   [RFC6679]  Westerlund, M., Johansson, I., Perkins, C., O'Hanlon, P.,
              and K. Carlberg, "Explicit Congestion Notification (ECN)
              for RTP over UDP", RFC 6679, DOI 10.17487/RFC6679, August 2012.
              2012, <http://www.rfc-editor.org/info/rfc6679>.

   [RFC6798]  Clark, A. and Q. Wu, "RTP Control Protocol (RTCP) Extended
              Report (XR) Block for Packet Delay Variation Metric
              Reporting", RFC 6798, DOI 10.17487/RFC6798, November 2012. 2012,
              <http://www.rfc-editor.org/info/rfc6798>.

   [RFC6843]  Clark, A., Gross, K., and Q. Wu, "RTP Control Protocol
              (RTCP) Extended Report (XR) Block for Delay Metric
              Reporting", RFC 6843, DOI 10.17487/RFC6843, January 2013. 2013,
              <http://www.rfc-editor.org/info/rfc6843>.

   [RFC6958]  Clark, A., Zhang, S., Zhao, J., and Q. Wu, Ed., "RTP
              Control Protocol (RTCP) Extended Report (XR) Block for
              Burst/Gap Loss Metric Reporting", RFC 6958,
              DOI 10.17487/RFC6958, May 2013. 2013,
              <http://www.rfc-editor.org/info/rfc6958>.

   [RFC7002]  Clark, A., Zorn, G., and Q. Wu, "RTP Control Protocol
              (RTCP) Extended Report (XR) Block for Discard Count Metric
              Reporting", RFC 7002, DOI 10.17487/RFC7002, September 2013.
              2013, <http://www.rfc-editor.org/info/rfc7002>.

   [RFC7003]  Clark, A., Huang, R., and Q. Wu, Ed., "RTP Control
              Protocol (RTCP) Extended Report (XR) Block for Burst/Gap
              Discard Metric Reporting", RFC 7003, DOI 10.17487/RFC7003,
              September 2013. 2013, <http://www.rfc-editor.org/info/rfc7003>.

   [RFC7097]  Ott, J., Singh, V., Ed., and I. Curcio, "RTP Control
              Protocol (RTCP) Extended Report (XR) for RLE of Discarded
              Packets", RFC 7097, DOI 10.17487/RFC7097, January 2014. 2014,
              <http://www.rfc-editor.org/info/rfc7097>.

   [RFC7201]  Westerlund, M. and C. Perkins, "Options for Securing RTP
              Sessions", RFC 7201, DOI 10.17487/RFC7201, April 2014. 2014,
              <http://www.rfc-editor.org/info/rfc7201>.

   [Sarker]   Sarker, Z., Singh, V., and C.S. C. Perkins, "An Evaluation of
              RTP Circuit Breaker Performance on LTE Networks",
              Proceedings of the IEEE Infocom workshop on Communication
              and Networking Techniques for Contemporary Video, 2014,
              April 2014.

   [Singh]    Singh, V., McQuistin, S., Ellis, M., and C.S. C. Perkins,
              "Circuit Breakers for Multimedia Congestion Control",
              Proceedings of the International Packet Video
              Workshop, 2013, DOI 10.1109/PV.2013.6691439, December
              2013.

Authors' Addresses

   Colin Perkins
   University of Glasgow
   School of Computing Science
   Glasgow  G12 8QQ
   United Kingdom

   Email: csp@csperkins.org

   Varun Singh
   Aalto University
   School of Electrical Engineering
   Otakaari 5 A
   Espoo, FIN  02150
   Finland

   Email: varun@comnet.tkk.fi
   URI:   http://www.netlab.tkk.fi/~varun/