<?xmlversion="1.0" encoding="US-ASCII"?>version='1.0' encoding='utf-8'?> <!DOCTYPE rfc SYSTEM"rfc2629.dtd" [ <!ENTITY rfc2119 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.2119.xml"> <!ENTITY rfc3168 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.3168.xml"> <!ENTITY rfc3550 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.3550.xml"> <!ENTITY rfc5348 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.5348.xml"> <!ENTITY rfc5450 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.5450.xml"> <!ENTITY rfc6660 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.6660.xml"> <!ENTITY rfc6679 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.6679.xml"> <!ENTITY rfc6817 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.6817.xml"> <!ENTITY rfc7567 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.7567.xml"> <!ENTITY rfc8033 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8033.xml"> <!ENTITY rfc8290 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8290.xml"> <!ENTITY rfc8174 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8174.xml"> <!ENTITY rfc8593 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8593.xml"> <!ENTITY I-D.ietf-avtcore-cc-feedback-message SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-avtcore-cc-feedback-message.xml"> <!ENTITY I-D.ietf-rmcat-cc-requirements SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-cc-requirements.xml"> <!ENTITY I-D.ietf-rmcat-eval-test SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-eval-test.xml"> <!ENTITY I-D.ietf-rmcat-wireless-tests SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-wireless-tests.xml"> <!ENTITY I-D.ietf-rmcat-cc-codec-interactions SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-cc-codec-interactions.xml"> ]> <?xml-stylesheet type='text/xsl' href='rfc2629.xslt' ?> <?rfc toc="yes" ?> <?rfc compact="yes" ?> <?rfc symrefs="yes" ?>"rfc2629-xhtml.ent"> <rfc number="8698" xmlns:xi="http://www.w3.org/2001/XInclude" category="exp" docName="draft-ietf-rmcat-nada-13"ipr="trust200902"> <!-- What is the category field value-->ipr="trust200902" obsoletes="" updates="" submissionType="IETF" consensus="true" xml:lang="en" tocInclude="true" sortRefs="true" symRefs="true" version="3"> <front> <title abbrev="NADA">NADA:Network-Assisted Dynamic Adaptation (NADA): A Unified Congestion Control Scheme for Real-Time Media </title> <seriesInfo name="RFC" value="8698"/> <author fullname="Xiaoqing Zhu" initials="X" surname="Zhu"> <organization>Cisco Systems</organization> <address> <postal> <street>12515 Research Blvd., Building 4</street> <city>Austin</city> <region>TX</region> <code>78759</code><country>USA</country><country>United States of America</country> </postal> <email>xiaoqzhu@cisco.com</email> </address> </author> <author fullname="RongPan *"Pan" initials="R" surname="Pan"><organization abbrev="Cisco Systems">* Pending affiliation change.</organization><organization>Intel Corporation</organization> <address><email>rong.pan@gmail.com</email><postal> <street>2200 Mission College Blvd </street> <city>Santa Clara</city> <region>CA</region> <code>95054</code> <country>United States of America</country> </postal> <email>rong.pan@intel.com</email> </address> </author> <author fullname="Michael A. Ramalho"initials="M. A."initials="M." surname="Ramalho"> <organizationabbrev="Cisco Systems">Cisco Systems, Inc.</organization>abbrev="AcousticComms">AcousticComms Consulting</organization> <address> <postal><street>8000 Hawkins Road</street> <city>Sarasota</city><street>6310 Watercrest Way Unit 203</street> <city>Lakewood Ranch</city> <region>FL</region><code>34241</code> <country>USA</country><code>34202-5211</code> <country>United States of America</country> </postal> <phone>+1919 476 2038</phone>732 832 9723</phone> <email>mar42@cornell.edu</email> <uri>http://ramalho.webhop.info/</uri> </address> </author> <author fullname="SergioMena de la Cruz"Mena" initials="S. " surname="Mena"> <organization>Cisco Systems</organization> <address> <postal> <street>EPFL, Quartier de l'Innovation, Batiment E</street> <city>Ecublens</city> <region>Vaud</region> <code>1015</code> <country>Switzerland</country> </postal> <email>semena@cisco.com</email> </address> </author><!-- <author fullname="Paul E. Jones" initials="P. E." surname="Jones"> <organization>Cisco Systems</organization> <address> <postal> <street>7025 Kit Creek Rd.</street> <city>Research Triangle Park</city> <region>NC</region> <code>27709</code> <country>USA</country> </postal> <email>paulej@packetizer.com</email> </address> </author> <author fullname="Jiantao Fu" initials="J." surname="Fu"> <organization>Cisco Systems</organization> <address> <postal> <street>707 Tasman Drive</street> <city>Milpitas</city> <region>CA</region> <code>95035</code> <country>USA</country> </postal> <email>jianfu@cisco.com</email> </address> </author> <author fullname="Stefano D'Aronco" initials="S." surname="D'Aronco"> <organization abbrev="ETH">ETH Zurich</organization> <address> <postal> <street>Stefano-Franscini-Platz 5</street> <city>Zurich</city> <region></region> <code>8093</code> <country>Switzerland</country> </postal> <email>stefano.daronco@geod.baug.ethz.ch</email> </address> </author> --><dateyear="2019" />month="February" year="2020"/> <area>TSV</area> <keyword>Multimedia</keyword> <keyword>Congestion Control</keyword> <abstract> <t>This document describesNADA (network-assisted dynamic adaptation),Network-Assisted Dynamic Adaptation (NADA), a novel congestion control scheme for interactive real-time mediaapplications,applications such as video conferencing. In the proposed scheme, the sender regulates its sendingraterate, based on either implicit or explicit congestion signaling, in a unified approach. The scheme can benefit fromexplicit congestion notificationExplicit Congestion Notification (ECN) markings from network nodes. It also maintains consistent sender behavior in the absence of suchmarkings,markings by reacting to queuing delays and packet losses instead. </t> </abstract> </front> <middle> <section anchor="sec-intro"title="Introduction">numbered="true" toc="default"> <name>Introduction</name> <t>Interactive real-time media applications introduce a unique set of challenges for congestion control. Unlike TCP, the mechanism used for real-time media needs to adapt quickly to instantaneous bandwidth changes, accommodate fluctuations in the output of video encoder rate control, and cause low queuing delay over the network. An ideal scheme should also make effective use of all types of congestion signals, including packet loss, queuing delay, and explicit congestion notification (ECN) <xreftarget="RFC3168"></xref>target="RFC3168" format="default"/> markings. The requirements for the congestion control algorithm are outlined in <xreftarget="I-D.ietf-rmcat-cc-requirements"></xref>. It highlightstarget="I-D.ietf-rmcat-cc-requirements" format="default"/>. The requirements highlight that the desired congestion control scheme should 1) avoid flow starvation and attain a reasonable fair share of bandwidth when competing against other flows, 2) adapt quickly, and 3) operate in a stable manner. </t> <t>This document describes an experimental congestion control scheme callednetwork-assisted dynamic adaptationNetwork-Assisted Dynamic Adaptation (NADA). The design of NADA benefits from explicit congestion control signals (e.g., ECN markings) from the network, yet also operates when only implicit congestion indicators (delay and/or loss) are available. Such a unified sender behavior distinguishes NADA from other congestion control schemes for real-time media. In addition, its core congestion control algorithm is designed to guarantee stability for pathround-trip-timesround-trip times (RTTs) below a prescribed bound (e.g.,250ms250 ms with default parameter choices). It further supports weighted bandwidth sharing among competing video flows with different priorities. The signaling mechanism consists of standardRTPReal-time Transport Protocol (RTP) timestamp <xreftarget="RFC3550"></xref>target="RFC3550" format="default"/> andRTCPReal-time Transport Control Protocol (RTCP) feedback reports. The definition of the desired RTCP feedback message is described in detail in <xreftarget="I-D.ietf-avtcore-cc-feedback-message"></xref>target="I-D.ietf-avtcore-cc-feedback-message" format="default"/> so as to support the successful operation of several congestion control schemes for real-time interactive media. </t> </section> <section anchor="sec-term"title="Terminology"> <t>Thenumbered="true" toc="default"> <name>Terminology</name> <t> The key words"MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY","<bcp14>MUST</bcp14>", "<bcp14>MUST NOT</bcp14>", "<bcp14>REQUIRED</bcp14>", "<bcp14>SHALL</bcp14>", "<bcp14>SHALL NOT</bcp14>", "<bcp14>SHOULD</bcp14>", "<bcp14>SHOULD NOT</bcp14>", "<bcp14>RECOMMENDED</bcp14>", "<bcp14>NOT RECOMMENDED</bcp14>", "<bcp14>MAY</bcp14>", and"OPTIONAL""<bcp14>OPTIONAL</bcp14>" in this document are to be interpreted as described inBCP 14BCP 14 <xreftarget="RFC2119"></xref>target="RFC2119"/> <xreftarget="RFC8174"></xref>target="RFC8174"/> when, and only when, they appear in all capitals, as shownhere.</t>here. </t> </section> <section anchor="sec-system-overview"title="System Overview">numbered="true" toc="default"> <name>System Overview</name> <t><xreftarget="fig-system-overview"></xref>target="fig-system-overview" format="default"/> shows the end-to-end system for real-time media transport that NADA operates in. Note that there also exist network nodes along the reverse (potentially uncongested) path that the RTCP feedback reports traverse. Those network nodes are not shown in the figure for the sake of brevity.</t><t><figure anchor="fig-system-overview" title="System Overview"> <artwork><![CDATA[<figure anchor="fig-system-overview"> <name>System Overview</name> <artwork name="" type="" align="left" alt=""><![CDATA[ +---------+ r_vin +--------+ +--------+ +----------+ | Media |<--------| RTP | |Network | | RTP | | Encoder |========>| Sender |=======>| Node |====>| Receiver | +---------+ r_vout +--------+ r_send +--------+ +----------+ /|\ | | | +---------------------------------+ RTCP Feedback Report ]]></artwork></figure></t> <t><list style="symbols"> <t>Media</figure> <dl> <dt>Media encoder with rate controlcapabilities. It encodescapabilities: </dt> <dd>Encodes raw media (audio and video) frames into a compressed bitstreamwhichthat is later packetized into RTP packets. As discussed in <xreftarget="RFC8593"></xref>,target="RFC8593"/>, the actual output rate from the encoder r_vout may fluctuate around the target r_vin. Furthermore, it is possible that the encoder can only react to bit rate changes at rather coarse time intervals, e.g., once every 0.5 seconds.</t> <t> RTP</dd> <dt>RTP sender:responsible</dt> <dd>Responsible for calculating the NADA reference rate based on network congestion indicators (delay, loss, or ECN marking reports from the receiver), for updating the video encoder with a new target rater_vin,r_vin and for regulating the actual sending rate r_send accordingly. The RTP sender also generates a sending timestamp for each outgoing packet.</t> <t>RTP</dd> <dt>RTP receiver:responsible</dt> <dd>Responsible for measuring and estimating end-to-end delay (based on sender timestamp), packet loss (based on RTP sequence number), ECN marking ratios (based on <xreftarget="RFC6679"></xref>),target="RFC6679"/>), and receiving rate (r_recv) of the flow. It calculates the aggregated congestion signal (x_curr) that accounts for queuing delay, ECN markings, and packet losses. The receiver also determines the mode for sender rate adaptation (rmode) based on whether the flow has encountered any standing non-zero congestion. The receiver sends periodic RTCP reports back to the sender, containing values of x_curr, rmode, and r_recv.</t> <t> Network</dd> <dt>Network node with several modes ofoperation. Theoperation: </dt> <dd>The system can work with the default behavior of a simpledrop taildrop-tail queue. It can also benefit from advancedAQMActive Queue Management (AQM) features such as Proportional Integral Controller Enhanced <xref target="RFC8033">(PIE)</xref>, Flow Queue Controlling Queue Delay <xreftarget="RFC8033">PIE</xref>, <xref target="RFC8290">FQ-CoDel</xref>,target="RFC8290">(FQ-CoDel)</xref>, ECN marking based on <xreftarget="RFC7567">RED</xref>,target="RFC7567"> Random Early Detection (RED)</xref>, andPCNPre-Congestion Notification (PCN) marking using a token bucket algorithm(<xref target="RFC6660"></xref>).<xref target="RFC6660"/>. Note that network node operation is out ofcontrolscope for the design of NADA.</t> </list></t></dd> </dl> </section> <section anchor="sec-algorithm"title="Corenumbered="true" toc="default"> <name>Core Congestion ControlAlgorithm">Algorithm</name> <t>Like TCP-Friendly Rate Control(TFRC)<xref target="Floyd-CCR00"></xref>(TFRC) <xreftarget="RFC5348"></xref>,target="FLOYD-CCR00" format="default"/> <xref target="RFC5348" format="default"/>, NADA is a rate-based congestion control algorithm. In its simplest form, the sender reacts to the collection of network congestion indicators in the form of an aggregated congestionsignal,signal and operates in one of two modes:<list style="symbols"> <t>Accelerated ramp-up: when</t> <dl> <dt>Accelerated ramp up: </dt> <dd>When the bottleneck is deemed to be underutilized, the rate increases multiplicatively with respect to the rate of previously successful transmissions. The rate increase multiplier (gamma) is calculated based on the observedround-trip-timeround-trip time and target feedback interval, so as to limit self-inflicted queuing delay.</t> <t>Gradual</dd> <dt>Gradual rate update:in</dt> <dd>In the presence of a non-zero aggregate congestion signal, the sending rate is adjusted in reaction to both its value (x_curr) and its change in value(x_diff).</t> </list> </t>(x_diff). </dd> </dl> <t>This section introduces the list of mathematical notations and describes the core congestion control algorithm at the sender and receiver, respectively. Additional details on recommended practical implementations are described in Sections <xreftarget="sec-receiver"></xref>target="sec-receiver" format="counter"/> and <xreftarget="sec-sender"></xref>.target="sec-sender" format="counter"/>. </t> <section anchor="sec-notation"title = "Mathematical Notations">numbered="true" toc="default"> <name>Mathematical Notations</name> <t>This section summarizes the list of variables and parameters used in the NADA algorithm. <xreftarget="tab-parameters"></xref>target="tab-parameters" format="default"/> also includes the default values for choosing the algorithm parameterseitherto represent either a typical setting in practical applications or a setting based on theoretical and simulation studies. See <xreftarget="sec-discussion-c"></xref>target="sec-discussion-c" format="default"/> for some of the discussions on the impact of parameter values. Additional studies in real-world settings suggested in <xreftarget="sec-experiments"></xref>target="sec-experiments" format="default"/> could gather further insight on how to choose and adapt these parameter values in practical deployment.</t><t><figure anchor="tab-variables" title ="List of variables."> <artwork><![CDATA[ +--------------+-------------------------------------------------+ | Notation | Variable Name | +--------------+-------------------------------------------------+ | t_curr | Current timestamp | | t_last | Last<table align="left" anchor="tab-variables"> <name>List of Variables</name> <thead> <tr> <th align='left'>Notation</th> <th align='left'>Variable Name</th> </tr> </thead> <tbody> <tr> <td align="left">t_curr</td> <td align="left">Current timestamp</td> </tr> <tr> <td align="left">t_last</td> <td align="left">Last time sending/receiving a feedback| | delta | Observedmessage</td> </tr> <tr> <td align="left">delta</td> <td align="left">Observed interval between current and previous| | |feedback reports: delta =t_curr-t_last | | r_ref | Referencet_curr-t_last</td> </tr> <tr> <td align="left">r_ref</td> <td align="left">Reference rate based on networkcongestion | | r_send | Sending rate | | r_recv | Receiving rate | | r_vin | Targetcongestion</td> </tr> <tr> <td align="left">r_send</td> <td align="left">Sending rate</td> </tr> <tr> <td align="left">r_recv</td> <td align="left">Receiving rate</td> </tr> <tr> <td align="left">r_vin</td> <td align="left">Target rate for videoencoder | | r_vout | Outputencoder</td> </tr> <tr> <td align="left">r_vout</td> <td align="left">Output rate from videoencoder | | d_base | Estimatedencoder</td> </tr> <tr> <td align="left">d_base</td> <td align="left">Estimated baselinedelay | | d_fwd | Measureddelay</td> </tr> <tr> <td align="left">d_fwd</td> <td align="left">Measured and filtered one-waydelay | | d_queue | Estimateddelay</td> </tr> <tr> <td align="left">d_queue</td> <td align="left">Estimated queuingdelay | | d_tilde | Equivalentdelay</td> </tr> <tr> <td align="left">d_tilde</td> <td align="left">Equivalent delay after non-linearwarping | | p_mark | Estimatedwarping</td> </tr> <tr> <td align="left">p_mark</td> <td align="left">Estimated packet ECN markingratio | | p_loss | Estimatedratio</td> </tr> <tr> <td align="left">p_loss</td> <td align="left">Estimated packet lossratio | | x_curr | Aggregate congestion signal | | x_prev | Previousratio</td> </tr> <tr> <td align="left">x_curr</td> <td align="left">Aggregate congestion signal</td> </tr> <tr> <td align="left">x_prev</td> <td align="left">Previous value of aggregate congestionsignal | | x_diff | Changesignal</td> </tr> <tr> <td align="left">x_diff</td> <td align="left">Change in aggregate congestion signal w.r.t.| | |its previous value: x_diff = x_curr -x_prev | | rmode | Ratex_prev</td> </tr> <tr> <td align="left">rmode</td> <td align="left">Rate update mode: (0 = acceleratedramp-up; | | |ramp up; 1 = gradualupdate) | | gamma | Rateupdate)</td> </tr> <tr> <td align="left">gamma</td> <td align="left">Rate increase multiplier in accelerated ramp-up| | | mode | | loss_int | Measuredmode</td> </tr> <tr> <td align="left">loss_int</td> <td align="left">Measured average loss interval in packetcount | | loss_exp | Thresholdcount</td> </tr> <tr> <td align="left">loss_exp</td> <td align="left">Threshold value for setting the last observed| | |packet loss toexpiration | | rtt | Estimated round-trip-timeexpiration</td> </tr> <tr> <td align="left">rtt</td> <td align="left">Estimated round-trip time atsender | | buffer_len | Rate shapingsender</td> </tr> <tr> <td align="left">buffer_len</td> <td align="left">Rate-shaping buffer occupancy measured inbytes | +--------------+-------------------------------------------------+ ]]></artwork> </figure></t> <t><figure anchor="tab-parameters" title ="Listbytes</td> </tr> </tbody> </table> <table align="left" anchor="tab-parameters"> <name>List ofalgorithm parametersAlgorithm Parameters andtheir default values."> <artwork><![CDATA[ +--------------+----------------------------------+----------------+ | Notation | Parameter Name |Their DefaultValue | +--------------+----------------------------------+----------------+ | PRIO | WeightValues</name> <thead> <tr> <th align='left'>Notation</th> <th align='left'>Parameter Name</th> <th align='left'>Default Value</th> </tr> </thead> <tbody> <tr> <td align="left">PRIO</td> <td align="left">Weight of priority of theflow | 1.0 | RMIN | Minimumflow</td> <td align="left">1.0</td> </tr> <tr> <td align="left">RMIN</td> <td align="left">Minimum rate of application| 150Kbps | | |supported by mediaencoder | | | RMAX | Maximumencoder</td> <td align="left">150 Kbps</td> </tr> <tr> <td align="left">RMAX</td> <td align="left">Maximum rate of application| 1.5Mbps | | |supported by mediaencoder | | | XREF | Reference congestion level | 10ms | | KAPPA | Scalingencoder</td> <td align="left">1.5 Mbps</td> </tr> <tr> <td align="left">XREF</td> <td align="left">Reference congestion level</td> <td align="left">10 ms</td> </tr> <tr> <td align="left">KAPPA</td> <td align="left">Scaling parameter for gradual| 0.5 | | |rate updatecalculation | | | ETA | Scalingcalculation</td> <td align="left">0.5</td> </tr> <tr> <td align="left">ETA</td> <td align="left">Scaling parameter for gradual| 2.0 | | |rate updatecalculation | | | TAU | Uppercalculation</td> <td align="left">2.0</td> </tr> <tr> <td align="left">TAU</td> <td align="left">Upper bound of RTT in gradual| 500ms | | |rate updatecalculation | | | DELTA | Target feedback interval | 100ms | +..............+..................................+................+ | LOGWIN | Observationcalculation</td> <td align="left">500 ms</td> </tr> <tr> <td align="left">DELTA</td> <td align="left">Target feedback interval</td> <td align="left">100 ms</td> </tr> <tr> <td align="left">LOGWIN</td> <td align="left">Observation window in time for| 500ms | | |calculating packet summary| | | |statistics atreceiver | | | QEPS | Thresholdreceiver</td> <td align="left">500 ms</td> </tr> <tr> <td align="left">QEPS</td> <td align="left">Threshold for determiningqueuing| 10ms | | |queuing delaybuild upbuildup atreceiver | | | DFILT | Boundreceiver</td> <td align="left">10 ms</td> </tr> <tr> <td align="left">DFILT</td> <td align="left">Bound on filteringdelay | 120ms | | GAMMA_MAX | Upperdelay</td> <td align="left">120 ms</td> </tr> <tr> <td align="left">GAMMA_MAX</td> <td align="left">Upper bound on rate increase| 0.5 | | |ratio for acceleratedramp-up | | | QBOUND | Upperramp up</td> <td align="left">0.5</td> </tr> <tr> <td align="left">QBOUND</td> <td align="left">Upper bound on self-inflicted| 50ms | | |queuing delay during rampup | | +..............+..................................+................+ | MULTILOSS | Multiplierup</td> <td align="left">50 ms</td> </tr> <tr> <td align="left">MULTILOSS</td> <td align="left">Multiplier for self-scaling the| 7.0 | | |expiration threshold of the last| | | |observed loss (loss_exp) basedon| | | |on measured average loss interval| | | | (loss_int) | | | QTH | Delay(loss_int)</td> <td align="left">7.0</td> </tr> <tr> <td align="left">QTH</td> <td align="left">Delay threshold for invoking| 50ms | | |non-linearwarping | | | LAMBDA | Scalingwarping</td> <td align="left">50 ms</td> </tr> <tr> <td align="left">LAMBDA</td> <td align="left">Scaling parameter in the| 0.5 | | |exponent of non-linearwarping | | +..............+..................................+................+ | PLRREF | Referencewarping</td> <td align="left">0.5</td> </tr> <tr> <td align="left">PLRREF</td> <td align="left">Reference packet lossratio | 0.01 | | PMRREF | Referenceratio</td> <td align="left">0.01</td> </tr> <tr> <td align="left">PMRREF</td> <td align="left">Reference packet markingratio | 0.01 | | DLOSS | Referenceratio</td> <td align="left">0.01</td> </tr> <tr> <td align="left">DLOSS</td> <td align="left">Reference delay penalty for loss| 10ms | | |when packet loss ratio is at| | | | PLRREF | | | DMARK | ReferencePLRREF</td> <td align="left">10 ms</td> </tr> <tr> <td align="left">DMARK</td> <td align="left">Reference delay penalty for ECN| 2ms | | |marking when packet marking| | | |is atPMRREF | | +..............+..................................+................+ | FPS | FramePMRREF</td> <td align="left">2 ms</td> </tr> <tr> <td align="left">FPS</td> <td align="left">Frame rate of incomingvideo | 30 | | BETA_S | Scalingvideo</td> <td align="left">30</td> </tr> <tr> <td align="left">BETA_S</td> <td align="left">Scaling parameter for modulating| 0.1 | | |outgoing sendingrate | | | BETA_V | Scalingrate</td> <td align="left">0.1</td> </tr> <tr> <td align="left">BETA_V</td> <td align="left">Scaling parameter for modulating| 0.1 | | |video encoder targetrate | | | ALPHA | Smoothingrate</td> <td align="left">0.1</td> </tr> <tr> <td align="left">ALPHA</td> <td align="left">Smoothing factor in exponential| 0.1 | | |smoothing of packet loss and| | | |markingratios | +--------------+----------------------------------+----------------+ ]]></artwork> </figure></t>ratios</td> <td align="left">0.1</td> </tr> </tbody> </table> </section> <sectionanchor = "subsec-receiver-algorithm" title = "Receiver-Side Algorithm">anchor="subsec-receiver-algorithm" numbered="true" toc="default"> <name>Receiver-Side Algorithm</name> <t>The receiver-side algorithm can be outlined as below:</t><t><figure><artwork><![CDATA[ On initialization: set<ul empty="true"> <li>On initialization:</li> <li> <ul empty="true"> <li>set d_base = +INFINITYset</li> <li>set p_loss = 0set</li> <li>set p_mark = 0set</li> <li>set r_recv = 0set</li> <li>set both t_last and t_curr as current time in millisecondsOn</li> </ul> </li> <li>On receiving a media packet:obtain</li> <li> <ul empty="true"> <li>obtain current timestamp t_curr from system clockobtain</li> <li>obtain from packet header sending time stamp t_sentobtain</li> <li>obtain one-way delay measurement: d_fwd = t_curr - t_sentupdate</li> <li>update baseline delay: d_base = min(d_base, d_fwd)update</li> <li>update queuing delay: d_queue = d_fwd - d_baseupdate</li> <li>update packet loss ratio estimate p_lossupdate</li> <li>update packet marking ratio estimate p_markupdate</li> <li>update measurement of receiving rate r_recvOn</li> </ul> </li> <li>On time to send a new feedback report (t_curr - t_last > DELTA):calculate</li> <li> <ul empty="true"> <li>calculate non-linear warping of delay d_tilde if packet loss existscalculate</li> <li>calculate current aggregate congestion signal x_currdetermine</li> <li>determine mode of rate adaptation for sender: rmodesend</li> <li>send feedback containing values of: rmode, x_curr, and r_recvupdate</li> <li>update t_last = t_curr]]></artwork></figure></t></li> </ul> </li> </ul> <t>In order for a delay-based flow to hold its ground when competing against loss-based flows (e.g., loss-based TCP), it is important to distinguish between different levels of observed queuing delay. For instance, over wired connections, a moderate queuing delay value on the order of tens of milliseconds is likely self-inflicted or induced by other delay-based flows, whereas a high queuing delay value of several hundreds of milliseconds may indicate the presence of a loss-based flow that does not refrain from increased delay.</t> <t> If the last observed packet loss is within the expiration window of loss_exp (measured in terms of packet counts), the estimated queuing delay follows a non-linear warping: </t><t><figure><artwork><![CDATA[<artwork name="" type="" align="left" alt=""><![CDATA[ / d_queue, ifd_queue<QTH;d_queue < QTH | d_tilde = < (1) | (d_queue-QTH) \ QTH exp(-LAMBDA ---------------) ,otherwise.otherwise QTH]]></artwork></figure></t>]]></artwork> <t> In Equation (1), the queuing delay value is unchanged when it is below the first threshold QTH;otherwiseotherwise, it is scaled down following a non-linear curve. This non-linear warping is inspired by the delay-adaptive congestion window backoff policy in <xreftarget="Budzisz-TON11"></xref>,target="BUDZISZ-AIMD-CC" format="default"/> so as to "gradually nudge" the controller to operate based on loss-induced congestion signals when competing against loss-based flows. The exact form of the non-linear function has been simplified with respect to <xreftarget="Budzisz-TON11"></xref>.target="BUDZISZ-AIMD-CC" format="default"/>. The value of the threshold QTH should be carefully tuned for different operationalenvironments,environments so as to avoid potential risks of prematurely discounting the congestion signal level. Typically, a higher value of QTH is required in a noisier environment (e.g., over wirelessconnections,connections or where the video stream encounters many time-varying background competing traffic) so as to stay robust against occasional non-congestion-induced delay spikes. Additional insights on how this value can be tuned or auto-tuned should be gathered from carrying out experimental studies in different real-world deployment scenarios. </t> <t>The value of loss_exp is configured to self-scale with the average packet loss interval loss_int with a multiplier MULTILOSS: </t><t><figure> <artwork><![CDATA[<artwork name="" type="" align="left" alt=""><![CDATA[ loss_exp = MULTILOSS * loss_int. ]]></artwork></figure></t><t>Estimation of the average loss interval loss_int, in turn, followsSection 5.4 of the<xreftarget="RFC5348">TCPtarget="RFC5348" sectionFormat ="of" section="5.4">"TCP Friendly Rate Control(TFRC) protocol</xref>.(TFRC): Protocol Specification"</xref>. </t> <t>In practice, it is recommended to linearly interpolate between the warped (d_tilde) and non-warped (d_queue) values of the queuing delay during the transitional period lasting for the duration of loss_int. </t> <t>The aggregate congestion signal is:</t><t><figure> <artwork><![CDATA[<artwork name="" type="" align="left" alt=""><![CDATA[ / p_mark \^2 / p_loss \^2 x_curr = d_tilde + DMARK*|--------| +DLOSS*|--------|.DLOSS*|--------| (2) \ PMRREF / \ PLRREF / ]]></artwork></figure></t><t>Here, DMARK is prescribed a reference delay penalty associated with ECN markings at the reference marking ratio of PMRREF; DLOSS is prescribed a reference delay penalty associated with packet losses at the reference packet loss ratio of PLRREF. The value of DLOSS and DMARK does not depend on configurations at the network node. Since ECN-enabled active queue management schemes typically mark a packet before dropping it, the value of DLOSSSHOULD<bcp14>SHOULD</bcp14> be higher than that of DMARK. Furthermore, the values of DLOSS and DMARK need to be set consistently across all NADA flows sharing the same bottlenecklink,link so that they can compete fairly.</t> <t>In the absence of packet marking and losses, the value of x_curr reduces to the observed queuing delay d_queue. In thatcasecase, the NADA algorithm operates in the regime of delay-based adaptation. </t> <t>Given observed per-packet delay and loss information, the receiver is also in a good position to determine whether or not the network is underutilized and then recommend the corresponding rate adaptation mode for the sender. The criteria for operating in accelerated ramp-up mode are:</t><t><list style="symbols"> <t><ul spacing="normal"> <li> No recent packet losses within the observation window LOGWIN;and</t> <t>and</li> <li> Nobuild-upbuildup of queuing delay: d_fwd-d_base < QEPS for all previous delay samples within the observation windowLOGWIN.</t> </list></t> <t>OtherwiseLOGWIN.</li> </ul> <t>Otherwise, the algorithm operates in graduate update mode. </t> </section> <sectionanchor = "subsec-sender-algorithm" title = "Sender-Side Algorithm">anchor="subsec-sender-algorithm" numbered="true" toc="default"> <name>Sender-Side Algorithm</name> <t>The sender-side algorithm is outlined as follows:</t><t><figure> <artwork><![CDATA[ on initialization: set<ul empty="true"> <li>On initialization:</li> <li> <ul empty="true"> <li>set r_ref = RMINset</li> <li>set rtt = 0set</li> <li>set x_prev = 0set</li> <li>set t_last and t_curr as current system clock timeon</li> </ul> </li> <li>On receiving feedback report:obtain</li> <li> <ul empty="true"> <li>obtain current timestamp from system clock: t_currobtain</li> <li>obtain values of rmode, x_curr, and r_recv from feedback reportupdate</li> <li>update estimation of rttmeasure</li> <li>measure feedback interval: delta = t_curr - t_lastif</li> <li>if rmode == 0:update</li> <li> <ul empty="true"> <li>update r_ref following accelerated ramp-up ruleselse: update</li> </ul> </li> <li>else: </li> <li> <ul empty="true"> <li>update r_ref following gradual update rulesclip</li> </ul> </li> <li>clip rate r_ref within the range of minimum rate (RMIN) and maximum rate (RMAX). </li> <li>set x_prev = x_curr </li> <li>set t_last = t_curr]]></artwork> </figure></t></li> </ul> </li> </ul> <t>In accelerated ramp-up mode, the rate r_ref is updated as follows:</t><t><figure> <artwork><![CDATA[<artwork name="" type="" align="left" alt=""><![CDATA[ QBOUND gamma = min(GAMMA_MAX, ------------------) (3) rtt+DELTA+DFILT r_ref = max(r_ref, (1+gamma) r_recv) (4)]]></artwork></figure></t>]]></artwork> <t>The rate increase multiplier gamma is calculated as a function of the upper bound of self-inflicted queuing delay (QBOUND),round-trip-timeround-trip time (rtt), and target feedback interval(DELTA) and(DELTA); it is bound on the filtering delay for calculating d_queue (DFILT). It has a maximum value of GAMMA_MAX. The rationale behind Equations (3)-(4) is that the longer it takes for the sender to observe self-inflicted queuing delaybuild-up,buildup, the more conservative the sender should be in increasing itsrate, hencerate and, hence, the smaller the rate increase multiplier. </t> <t>In gradual update mode, the rate r_ref is updated as:</t><t><figure> <artwork><![CDATA[<artwork name="" type="" align="left" alt=""><![CDATA[ x_offset = x_curr - PRIO*XREF*RMAX/r_ref (5) x_diff = x_curr - x_prev (6) delta x_offset r_ref = r_ref - KAPPA*-------*------------*r_ref TAU TAU x_diff - KAPPA*ETA*---------*r_ref (7) TAU]]></artwork></figure></t>]]></artwork> <t>The rate changes in proportion to the previous rate decision. It is affected by two terms: the offset of the aggregate congestion signal from its value at equilibrium (x_offset) and its change (x_diff).CalculationThe calculation of x_offset depends on the maximum rate of the flow (RMAX), its weight of priority (PRIO), as well as a reference congestion signal (XREF). The value of XREF is chosen so that the maximum rate of RMAX can be achieved when the observed congestion signal level is below PRIO*XREF. </t> <t> At equilibrium, the aggregated congestion signal stabilizes at x_curr = PRIO*XREF*RMAX/r_ref. This ensures that when multiple flows share the same bottleneck and observe a common value of x_curr, their rates at equilibrium will be proportional to their respective priority levels (PRIO) and the range between minimum and maximum rate. Values of the minimum rate (RMIN) and maximum rate (RMAX) will be provided by the media codec, for instance, as outlined by <xreftarget="I-D.ietf-rmcat-cc-codec-interactions">target="I-D.ietf-rmcat-cc-codec-interactions" format="default"> </xref>. In the absence of such information, the NADA sender will choose a default value of 0 forRMIN,RMIN and3Mbps3 Mbps for RMAX. </t> <t> As mentioned in the sender-side algorithm, the final rate is always clipped within the dynamic range specified by the application: </t><t><figure><artwork><![CDATA[<artwork name="" type="" align="left" alt=""><![CDATA[ r_ref = min(r_ref, RMAX) (8) r_ref = max(r_ref, RMIN) (9)]]></artwork></figure></t>]]></artwork> <t>The above operations ignore many practical issues such as clock synchronization between sender and receiver, the filtering of noise in delay measurements, and base delay expiration. These will be addressed in <xreftarget="sec-practical-nada"></xref>.</t>target="sec-practical-nada" format="default"/>.</t> </section> </section> <section anchor="sec-practical-nada"title = "Practicalnumbered="true" toc="default"> <name>Practical Implementation ofNADA">NADA</name> <section anchor="sec-receiver"title="Receiver-Side Operation">numbered="true" toc="default"> <name>Receiver-Side Operation</name> <t>The receiver continuously monitors end-to-end per-packet statistics in terms of delay, loss, and/or ECN marking ratios. It then aggregates all forms of congestion indicators into the form of an equivalent delay and periodically reports this back to the sender. In addition, the receiver tracks the receiving rate of the flow and includes that in the feedback message.</t> <section anchor="sec-receiver-a"title="Estimationnumbered="true" toc="default"> <name>Estimation ofone-way delayOne-Way Delay andqueuing delay">Queuing Delay</name> <t> The delay estimation process in NADA followsa similaran approachas insimilar to that of earlier delay-based congestion control schemes, such as Low Extra Delay Background Transport (LEDBAT) <xreftarget="RFC6817">LEDBAT</xref>.target="RFC6817" format="default"></xref>. For experimental implementations, instead of relying on RTP timestamps and the transmission time offset RTP header extension <xreftarget="RFC5450"></xref>,target="RFC5450" format="default"/>, the NADA sender can generate its own timestamp based on the local system clock and embed that information in the transport packet header. The NADA receiver estimates the forward delay as having a constant base delay component plus atime varyingtime-varying queuing delay component. The base delay is estimated as the minimum value of one-way delay observed over a relatively long period (e.g., tens of minutes), whereas the individual queuing delay value is taken to be the difference between one-way delay and base delay. By re-estimating the base delay periodically, one can avoid the potential issue of base delay expiration, whereby an earlier measured base delay value is no longer valid due to underlying route changes or a cumulative timing difference introduced by theclock rateclock-rate skew between sender and receiver. All delay estimations are based on sender timestamps with a recommended granularity of 100 microseconds or finer. </t> <t> The individual sample values of queuing delay should be further filtered against various non-congestion-induced noise, such as spikes due to a processing "hiccup" at the network nodes. Therefore, in addition to calculating the value of queuing delay using d_queue = d_fwd - d_base, as expressed in <xreftarget="sec-receiver"></xref>,target="sec-receiver" format="default"/>, the current implementation further employs a minimum filter with a window size of 15 samples over per-packet queuing delay values. </t> </section> <section anchor="sec-receiver-b"title="Estimationnumbered="true" toc="default"> <name>Estimation ofpacket loss/marking ratio">Packet Loss/Marking Ratio</name> <t>The receiver detects packet losses via gaps in the RTP sequence numbers of received packets. For interactive real-time mediaapplicationapplications with stringent latencyconstraintconstraints (e.g., video conferencing), the receiver avoids the packetre-orderingreordering delay by treating out-of-order packets as losses. The instantaneous packet loss ratio p_inst is estimated as the ratio between the number of missing packets over the number of total transmitted packets within the recent observation window LOGWIN. The packet loss ratio p_loss is obtained after exponential smoothing: </t><t><figure> <artwork><![CDATA[<artwork name="" type="" align="left" alt=""><![CDATA[ p_loss = ALPHA*p_inst +(1-ALPHA)*p_loss.(1-ALPHA)*p_loss (10) ]]></artwork></figure></t><t>The filtered result is reported back to the sender as the observed packet loss ratio p_loss. </t> <t>EstimationThe estimation of the packet marking ratio p_mark follows the same procedure as above. It is assumed that ECN marking information at the IP header can be passed to the receiving endpoint, e.g., by following the mechanism described in <xreftarget="RFC6679"></xref>.</t>target="RFC6679" format="default"/>.</t> </section> <sectionanchor = "sec-receiver-c" title = "Estimationanchor="sec-receiver-c" numbered="true" toc="default"> <name>Estimation ofreceiving rate">Receiving Rate</name> <t> It is fairly straightforward to estimate the receiving rate r_recv. NADA maintains a recent observation window with a time span ofLOGWIN,LOGWIN and simply divides the total size of packets arriving during that window over the time span. The receiving rate (r_recv) can be either calculated ateitherthe sender side based on the per-packet feedback from thereceiver,receiver or included as part of the feedback report.</t> </section> </section> <section anchor="sec-sender"title="Sender-Side Operation">numbered="true" toc="default"> <name>Sender-Side Operation</name> <t> <xreftarget="fig-nada-sender"></xref>target="fig-nada-sender" format="default"/> provides a detailed view of the NADA sender. Upon receipt of an RTCP feedback report from the receiver, the NADA sender calculates the reference rate r_ref as specified in <xreftarget="subsec-sender-algorithm"></xref>.target="subsec-sender-algorithm" format="default"/>. It further adjusts both the target rate for the live video encoder r_vin and the sending rate r_send over the network based on the updated value of r_ref andrate shapingrate-shaping buffer occupancy buffer_len. </t> <t> The NADA sender behavior stays the same in the presence of all types of congestion indicators: delay, loss, and ECN marking. This unified approach allows a graceful transition of the scheme as the network shifts dynamically between light and heavy congestion levels. </t><t><figure anchor="fig-nada-sender" title="NADA<figure anchor="fig-nada-sender"> <name>NADA SenderStructure"> <artwork><![CDATA[Structure</name> <artwork name="" type="" align="left" alt=""><![CDATA[ +----------------+ | Calculate | <---- RTCP report | Reference Rate | -----------------+ | r_ref +------------+-------------+ | | \|/ \|/ +-----------------+ +---------------+ | Calculate Video | | Calculate | | Target Rate | | Sending Rate | +-----------------+ +---------------+ | /|\ /|\ | r_vin | | | | \|/ +-------------------+ | +----------+ | buffer_len | r_send | Video | r_vout -----------+ \|/ | Encoder |--------> |||||||||=================> +----------+ -----------+ RTP packetsRate ShapingRate-Shaping Buffer]]></artwork></figure></t>]]></artwork> </figure> <sectionanchor = "sec-sender-c" title = "Rate shaping buffer">anchor="sec-sender-c" numbered="true" toc="default"> <name>Rate-Shaping Buffer</name> <t> The operation of the live video encoder is out of the scope of the design for the congestion control scheme in NADA. Instead, its behavior is treated as a black box. </t> <t> Arate shapingrate-shaping buffer is employed to absorb any instantaneous mismatch between the encoder rate output r_vout and the regulated sending rate r_send. Its current level of occupancy is measured in bytes and is denoted as buffer_len. </t> <t>A largerate shapingrate-shaping buffer contributes to higher end-to-end delay, which may harm the performance of real-time media communications. Therefore, the sender has a strong incentive to prevent therate shapingrate-shaping buffer from building up. The mechanisms adopted are: </t><t><list style="symbols"> <t>To<ul spacing="normal"> <li>To deplete therate shapingrate-shaping buffer faster by increasing the sending rate r_send; and</t> <t>To</li> <li>To limit incoming packets of therate shapingrate-shaping buffer by reducing the video encoder target rate r_vin.</t> </list></t></li> </ul> </section> <sectionanchor = "sec-sender-d" title = "Adjusting video target rateanchor="sec-sender-d" numbered="true" toc="default"> <name>Adjusting Video Target Rate andsending rate">Sending Rate</name> <t> If the level of occupancy in therate shapingrate-shaping buffer is accessible at the sender, such information can be leveraged to further adjust the target rate of the live video encoder r_vin as well as the actual sending rate r_send. The purpose of such adjustments is to mitigate the additional latencies introduced by therate shapingrate-shaping buffer. The amount of rate adjustment can be calculated as follows: </t><t><figure><artwork><![CDATA[<artwork name="" type="" align="left" alt=""><![CDATA[ r_diff_v = min(0.05*r_ref,BETA_V*8*buffer_len*FPS).BETA_V*8*buffer_len*FPS) (11) r_diff_s = min(0.05*r_ref,BETA_S*8*buffer_len*FPS).BETA_S*8*buffer_len*FPS) (12) r_vin = max(RMIN, r_ref -r_diff_v).r_diff_v) (13) r_send = min(RMAX, r_ref +r_diff_s).r_diff_s) (14)]]></artwork></figure></t>]]></artwork> <t> In Equations (11) and (12), the amount of adjustment is calculated as proportional to the size of therate shapingrate-shaping buffer but is bounded by 5% of the reference rate r_ref calculated from network congestion feedback alone. This ensures that the adjustment introduced by therate shapingrate-shaping buffer will not counteract with the core congestion control process. Equations (13) and (14) indicate the influence of therate shapingrate-shaping buffer. A largerate shapingrate-shaping buffer nudges the encoder target rate slightly below-- and(and the sending rate slightlyabove --above) the reference rate r_ref. The final video target rate (r_vin) and sending rate (r_send) are further bounded within the original range of [RMIN, RMAX]. </t> <t> Intuitively, the amount of extra rate offset needed to completely drain therate shapingrate-shaping buffer within the duration of a single video frame is given by 8*buffer_len*FPS, where FPS stands for the reference frame rate of the video. The scaling parameters BETA_V and BETA_S can be tuned to balance between the competing goals of maintaining a smallrate shapingrate-shaping buffer and deviating from the reference rate point. Empirical observations show that therate shapingrate-shaping buffer for a responsive live video encoder typically stays empty and only occasionally holds a large frame (e.g., when an intra-frame is produced) in transit. Therefore, the rate adjustment introduced by this mechanism is expected to be minor. For instance, arate shapingrate-shaping buffer of 2000Bytesbytes will lead to a rate adjustment of48Kbps48 Kbps given the recommended scaling parameters of BETA_V = 0.1 and BETA_S =0.10.1, and the reference frame rate of FPS = 30. </t> </section> </section> <sectionanchor = "sec-feedback" title = "Feedbackanchor="sec-feedback" numbered="true" toc="default"> <name>Feedback MessageRequirements">Requirements</name> <t>The following list of information is required for NADA congestion control to function properly: </t><t><list style="symbols"> <t>Recommended<dl newline="false" spacing="normal"> <dt>Recommended rate adaptation mode (rmode):a</dt> <dd>A 1-bit flag indicating whether the sender should operate in accelerated ramp-up mode (rmode=0) or gradual update mode (rmode=1).</t> <t>Aggregated</dd> <dt>Aggregated congestion signal (x_curr):the</dt> <dd>The most recently updated value, calculated by the receiver according to <xreftarget="subsec-receiver-algorithm"></xref>.target="subsec-receiver-algorithm" format="default"/>. This information can be expressed with a unit of 100microsecondmicroseconds (i.e., 1/10 of a millisecond) in 15 bits. This allows a maximum value of x_curr at approximately 3.27second.</t> <t> Receivingseconds. </dd> <dt>Receiving rate (r_recv):the</dt> <dd>The most recently measured receiving rate according to <xreftarget="sec-receiver-c">target="sec-receiver-c" format="default"> </xref>. This information is expressed with a unit of bits per second (bps) in 32 bits (unsigned int). This allows a maximum rate of approximately4.3Gbps,4.3 Gbps, approximately 1000 timesofthe streaming rate of a typical high-definition (HD) video conferencing session today. This field can be expanded further by a few morebytes, in casebytes if an even higher rateneedneeds to bespecified.</t> </list></t>specified. </dd> </dl> <t> The above list of information can be accommodated by 48 bits, or 6 bytes, in total. They can be either included in the feedback report from thereceiver,receiver or, in the case where all receiver-side calculations are moved to the sender, derived from per-packet information from the feedback message as defined in <xreftarget="I-D.ietf-avtcore-cc-feedback-message"></xref>. Choice oftarget="I-D.ietf-avtcore-cc-feedback-message" format="default"/>. Choosing the feedback message interval DELTA is discussed in <xreftarget="sec-discussion-c"></xref>.target="sec-discussion-c" format="default"/>. A target feedback interval ofDELTA=100msDELTA = 100 ms is recommended. </t> </section> </section> <sectionanchor = "sec-discussions" title = "Discussionsanchor="sec-discussions" numbered="true" toc="default"> <name>Discussions and FurtherInvestigations">Investigations</name> <t>This sectiondiscusseddiscusses the various design choices made by NADA, potential alternative variants of its implementation, and guidelines on how the key algorithm parameters can be chosen. <xreftarget="sec-experiments"></xref>target="sec-experiments" format="default"/> recommends additional experimental setups to further explore these topics. </t> <sectionanchor = "sec-discussion-a" title = "Choiceanchor="sec-discussion-a" numbered="true" toc="default"> <name>Choice ofdelay metrics">Delay Metrics</name> <t> The current design works with relativeone-way-delayone-way delay (OWD) as the main indication of congestion. The value of the relative OWD is obtained by maintaining the minimum value of observed OWD over a relatively long time horizon andsubtractsubtracting that out from the observed absolute OWD value. Such an approach cancels out the fixed difference between the sender and receiver clocks. It has been widely adopted by other delay-based congestion control approaches such as <xreftarget="RFC6817"></xref>.target="RFC6817" format="default"/>. As discussed in <xreftarget="RFC6817"></xref>,target="RFC6817" format="default"/>, the time horizon for tracking the minimum OWD needs to be chosen withcare:care; it must be long enough for an opportunity to observe the minimum OWD with zero standing queue along the path, and it must be sufficiently shortso asenough to timely reflect "true" changes in minimum OWD introduced by route changes and other rare events and to mitigate the cumulative impact of clock rate skew over time. </t> <t> The potential drawback in relying on relative OWD as the congestion signal is that when multiple flows share the same bottleneck, the flow arriving late at the network experiencing a non-empty queue may mistakenly consider the standing queuing delay as part of the fixed path propagation delay. This will lead to slightly unfair bandwidth sharing among the flows. </t> <t>Alternatively, one could move the per-packet statistical handling to the sender instead and use relativeround-trip-timeround-trip time (RTT) in lieu of relative OWD, assuming that per-packet acknowledgments are available. The main drawback of an RTT-based approach is the noise in the measured delay in the reverse direction. </t> <t> Note that the choice of either delay metric (relative OWD vs. RTT) involves no change in the proposed rate adaptation algorithm. Therefore, comparing the pros and cons regarding which delay metric to adopt can be kept as an orthogonal direction of investigation. </t> </section> <section anchor="sec-discussion-b"title = "Methodnumbered="true" toc="default"> <name>Method fordelay, loss,Delay, Loss, andmarking ratio estimation">Marking Ratio Estimation</name> <t>Like other delay-based congestion control schemes, performance of NADA depends on the accuracy of its delay measurement and estimation module.Appendix A in<xreftarget = "RFC6817"></xref>target= "RFC6817" format="default" section="A"/> provides an extensive discussion on this aspect. </t> <t>The current recommended practice of applying minimum filter with a window size of 15 samples suffices in guarding against processing delay outliers observed in wired connections. For wireless connections with a higher packet delay variation (PDV), more sophisticated techniques onde-noising,denoising, outlier rejection, and trend analysis may be needed. </t> <t> More sophisticated methods in packet loss ratio calculation, such as that adopted by <xreftarget="Floyd-CCR00"></xref>,target="FLOYD-CCR00" format="default"/>, will likely be beneficial. These alternatives are part of the experiments this document proposes.</t> </section> <sectionanchor = "sec-discussion-c" title = "Impactanchor="sec-discussion-c" numbered="true" toc="default"> <name>Impact ofparameter values">Parameter Values</name> <t>In the gradual rate update mode, the parameter TAU indicates the upper bound ofround-trip-timeround-trip time (RTT) in the feedback control loop. Typically, the observed feedback interval delta is close to the target feedback interval DELTA, and the relative ratio of delta/TAU versus ETA dictates the relative strength of influence from the aggregate congestion signal offset term (x_offset) versus its recent change (x_diff), respectively. These two terms are analogous to the integral and proportional terms in a proportional-integral (PI) controller. The recommended choice ofTAU=500ms, DELTA=100msTAU = 500 ms, DELTA = 100 ms, and ETA = 2.0 corresponds to a relative ratio of 1:10 between the gains of the integral and proportional terms. Consequently, the rate adaptation is mostly driven by the change in the congestion signal with a long-term shift towards its equilibrium value driven by the offset term. Finally, the scaling parameter KAPPA determines the overall speed of the adaptation and needs to strike a balance between responsiveness and stability. </t> <t> The choice of the target feedback interval DELTA needs to strike the right balance between timely feedback and low RTCP feedback message counts. A target feedback interval ofDELTA=100msDELTA = 100 ms is recommended, corresponding to a feedback bandwidth of16Kbps16 Kbps with 200 bytes per feedback message----- approximately 1.6% overhead for a1Mbps1 Mbps flow. Furthermore, both simulation studies and frequency-domain analysis in <xreftarget="IETF-95"></xref>target="IETF-95" format="default"/> have established that a feedback interval below250ms250 ms (i.e., more frequently than 4 feedback messages per second) will not break up the feedback control loop of NADA congestion control. </t> <t>In calculating the non-linear warping of delay in Equation (1), the current design uses fixed values of QTH for determining whether to perform the non-linearwarping).warping. Its value should be carefully tuned for different operational environments (e.g., over wired vs. wirelessconnections),connections) so as to avoid the potential risk of prematurely discounting the congestion signal level. It is possible to adapt its value based on past observed patterns of queuing delay in the presence of packet losses. It needs to be noted that the non-linear warping mechanism may lead to multiple NADA streams stuck in loss-based mode when competing against each other. </t> <t>In calculating the aggregate congestion signal x_curr, the choice of DMARK and DLOSS influence the steady-state packet loss/marking ratio experienced by the flow at a given available bandwidth. Higher values of DMARK and DLOSS result in lower steady-state loss/markingratios,ratios but are more susceptible to the impact of individual packet loss/marking events. While the value of DMARK and DLOSS are fixed and predetermined in the current design, this document also encourages further explorations of a scheme for automatically tuning these values based on desired bandwidth sharing behavior in the presence of other competing loss-based flows (e.g., loss-based TCP).</t> </section> <section anchor="sec-discussion-d"title = "Sender-basednumbered="true" toc="default"> <name>Sender-Based vs.receiver-based calculation">Receiver-Based Calculation</name> <t>In the current design, the aggregated congestion signal x_curr is calculated at the receiver, keeping the sender operation completely independent of the form of actual network congestion indications (delay, loss, or marking) in use. </t> <t>Alternatively, one can shift receiver-side calculations to the sender, whereby the receiver simply reports on per-packet information via periodic feedback messages as defined in <xreftarget="I-D.ietf-avtcore-cc-feedback-message"></xref>.target="I-D.ietf-avtcore-cc-feedback-message" format="default"/>. Such an approach enables interoperability amongst senders operating on different congestion controlschemes,schemes but requires slightly higher overhead in the feedback messages. See additional discussions in <xreftarget="I-D.ietf-avtcore-cc-feedback-message"></xref>target="I-D.ietf-avtcore-cc-feedback-message" format="default"/> regarding the desired format of the feedback messages and the recommended feedback intervals. </t> </section> <sectionanchor = "sec-discussion-e" title = "Incremental deployment">anchor="sec-discussion-e" numbered="true" toc="default"> <name>Incremental Deployment</name> <t> One nice property of NADA is the consistent video endpoint behavior irrespective of network node variations. This facilitates gradual, incremental adoption of the scheme. </t> <t> Initially, the proposed congestion control mechanism can be implemented without any explicit support from thenetwork,network and relies solely on observed relative one-way delay measurements and packet loss ratios as implicit congestion signals. </t> <t> When ECN is enabled at the network nodes with RED-based marking, the receiver can fold its observations of ECN markings into the calculation of the equivalent delay. The sender can react to these explicit congestion signals without any modification. </t> <t> Ultimately, networks equipped with proactive marking based on the level of token bucketlevelmetering can reap the additional benefits of zero standing queues and lower end-to-end delay and work seamlessly with existing senders and receivers. </t> </section> </section> <sectionanchor = "sec-implementations" title ="Reference Implementations">anchor="sec-implementations" numbered="true" toc="default"> <name>Reference Implementations</name> <t> The NADA scheme has been implemented in both ns-2 <xreftarget="ns-2"></xref>target="NS-2" format="default"/> and ns-3 <xreftarget="ns-3"></xref>target="NS-3" format="default"/> simulation platforms. The implementation in ns-2 hosts the calculations as described in <xreftarget="subsec-receiver-algorithm"></xref>target="subsec-receiver-algorithm" format="default"/> at the receiver side, whereas the implementation in ns-3 hosts these receiver-side calculations at the sender for the sake of interoperability. Extensive ns-2 simulation evaluations of an earlier draft version ofthe draftthis document aredocumentedrecorded in <xreftarget="Zhu-PV13"></xref>.target="ZHU-PV13" format="default"/>. Anopen sourceopen-source implementation of NADA as part ofaan ns-3 module is available at <xreftarget="ns3-rmcat"></xref>.target="NS3-RMCAT" format="default"/>. Evaluation results ofthe current draftthis document based on ns-3 are presented in <xreftarget="IETF-90"></xref>target="IETF-90" format="default"/> and <xreftarget="IETF-91"></xref>target="IETF-91" format="default"/> for wired test cases as documented in <xreftarget="I-D.ietf-rmcat-eval-test"></xref>.target="I-D.ietf-rmcat-eval-test" format="default"/>. Evaluation results of NADA overWiFi-basedWi-Fi-based test cases as defined in <xreftarget="I-D.ietf-rmcat-wireless-tests"></xref>target="I-D.ietf-rmcat-wireless-tests" format="default"/> are presented in <xreftarget="IETF-93"></xref>.target="IETF-93" format="default"/>. These simulation-based evaluations have shown that NADA flows can obtain their fair share of bandwidth when competing against each other. They typically adapt fast in reaction to the arrival and departure of otherflows,flows and can sustain a reasonable throughput when competing against loss-based TCP flows. </t> <t> <xreftarget="IETF-90"></xref>target="IETF-90" format="default"/> describes the implementation and evaluation of NADA in a lab setting. Preliminary evaluation results of NADA in single-flow and multi-flow test scenarioshave beenare presented in <xreftarget="IETF-91"></xref>.target="IETF-91" format="default"/>. </t> <t> A reference implementation of NADA has been carried out by modifying the WebRTC module embedded in the Mozillaopen sourceopen-source browser. Presentations from <xreftarget="IETF-103"></xref>target="IETF-103" format="default"/> and <xreftarget="IETF-105"></xref>target="IETF-105" format="default"/> document real-world evaluations of the modified browser driven by NADA. The experimental settinginvolveinvolves remote connections with endpoints over either home or enterprise wireless networks. These evaluations validate the effectiveness of NADA flows in recovering quickly from throughput drops caused by intermittent delay spikes over the last-hop wireless connections. </t> </section> <sectionanchor = "sec-experiments" title = "Suggested Experiments">anchor="sec-experiments" numbered="true" toc="default"> <name>Suggested Experiments</name> <t> NADA has been extensively evaluated under various test scenarios, including the collection of test cases specified by <xreftarget="I-D.ietf-rmcat-eval-test"></xref>target="I-D.ietf-rmcat-eval-test" format="default"/> and the subset ofWiFi-basedWi-Fi-based test cases in <xreftarget="I-D.ietf-rmcat-wireless-tests"></xref>.target="I-D.ietf-rmcat-wireless-tests" format="default"/>. Additional evaluations have been carried out to characterize how NADA interacts with variousactive queue management (AQM)AQM schemes such as RED,CoDel,Controlling Queue Delay (CoDel), andPIE.Proportional Integral Controller Enhanced (PIE). Most of these evaluations have been carried out in simulators. A few key test cases have been evaluated in lab environments with implementations embedded in video conferencing clients. It is strongly recommended to carry out implementation and experimentation of NADA in real-world settings. Suchexerciseexercises will provide insights on how to choose or automatically adapt the values of the key algorithm parameters (see list in <xreftarget="tab-parameters"></xref>)target="tab-parameters" format="default"/>) as discussed in <xreftarget="sec-discussions"></xref>.target="sec-discussions" format="default"/>. </t> <t>Additional experiments are suggested for the followingscenarios andscenarios, preferably over real-world networks: </t><t><list style="symbols"> <t>Experiments<ul spacing="normal"> <li>Experiments reflecting the setup of a typical WAN connection.</t> <t>Experiments</li> <li>Experiments with ECN marking capability turned on at the network for existing testcases.</t> <t>Experimentscases.</li> <li>Experiments with multiple NADA streams bearing different user-specifiedpriorities.</t> <t>Experimentspriorities.</li> <li>Experiments with additional access technologies, especially over cellular networks such as3G/LTE.</t> <t>Experiments3G/LTE.</li> <li>Experiments with various media source contents, including audio only, audio and video, and application content sharing (e.g.,slide shows). </t> </list></t>slideshows). </li> </ul> </section> <sectionanchor = "sec-iana" title = "IANA Considerations">anchor="sec-iana" numbered="true" toc="default"> <name>IANA Considerations</name> <t>This documentmakeshas norequest of IANA.</t>IANA actions.</t> </section> <sectionanchor = "sec-security" title = "Security Considerations">anchor="sec-security" numbered="true" toc="default"> <name>Security Considerations</name> <t>The rate adaptation mechanism in NADA relies on feedback from the receiver. As such, it is vulnerable to attacks where feedback messages are hijacked, replaced, or intentionally injected with misleading information resulting in denial of service, similar to those that can affect TCP.ItTherefore, it istherefore RECOMMENDED<bcp14>RECOMMENDED</bcp14> that the RTCP feedback message is at least integrity checked. In addition, <xreftarget="I-D.ietf-avtcore-cc-feedback-message"></xref>target="I-D.ietf-avtcore-cc-feedback-message" format="default"/> discusses the potential risk of a receiver providing misleading congestion feedback information and the mechanisms for mitigating such risks.</t> <t>The modification of the sending rate based onsend-side rate shapingthe sender-side rate-shaping buffer may lead to temporary excessive congestion over the network in the presence ofaan unresponsive video encoder. However, thiseffect can be mitigated by limiting the amount of rate modification introduced by the rate shaping buffer, bounding the size of the rate shaping buffer at the sender, and maintaining a maximum allowed sending rate by NADA. </t> </section> <section anchor = "sec-acknowledgments" title = "Acknowledgments"> <t> The authors would like to thank Randell Jesup, Luca De Cicco, Piers O'Hanlon, Ingemar Johansson, Stefan Holmer, Cesar Ilharco Magalhaes, Safiqul Islam, Michael Welzl, Mirja Kuhlewind, Karen Elisabeth Egede Nielsen, Julius Flohr, Roland Bless, Andreas Smas, and Martin Stiemerling for their valuable review comments and helpful input to this specification. </t> </section> <section title="Contributors" anchor="sec-contributors"> <t>The following individuals have contributed to the implementation and evaluation of the proposed scheme, and therefore have helped to validate and substantially improve this specification.</t> <t><list> <t>Paul E. Jones <paulej@packetizer.com> of Cisco Systems implemented an early version of the NADA congestion control scheme and helped with its lab-based testbed evaluations. </t> <t>Jiantao Fu <jianfu@cisco.com> of Cisco Systems helped with the implementation and extensive evaluation of NADA both in Mozilla web browsers and in earlier simulation-based evaluation efforts. </t> <t>Stefano D'Aronco <stefano.daronco@geod.baug.ethz.ch> of ETH Zurich (previously at Ecole Polytechnique Federale de Lausanne when contributing to this work) helped with implementation and evaluation of an early version of NADA in <xref target="ns-3"></xref>. </t> <t>Charles Ganzhorn <charles.ganzhorn@gmail.com> contributed toeffect can be mitigated by limiting thetestbed-based evaluationamount ofNADA during an early stagerate modification introduced by the rate-shaping buffer, bounding the size ofits development.the rate-shaping buffer at the sender, and maintaining a maximum allowed sending rate by NADA. </t></list></t></section> </middle> <back><references title="Normative References"> &rfc2119; <!-- RFCs --> &rfc3168; <!-- ECN --> &rfc3550; <!-- RTP --> &rfc5348; <!-- TFRC --> &rfc6679; <!--- ECN over RTP --> &rfc8174; <!--- Ambiguity of Uppercase vs Lowercase --><displayreference target="I-D.ietf-rmcat-cc-requirements" to="RMCAT-CC"/> <displayreference target="I-D.ietf-rmcat-cc-codec-interactions" to="RMCAT-CC-RTP"/> <displayreference target="I-D.ietf-rmcat-eval-test" to="RMCAT-EVAL-TEST"/> <displayreference target="I-D.ietf-rmcat-wireless-tests" to="WIRELESS-TESTS"/> <displayreference target="I-D.ietf-avtcore-cc-feedback-message" to="RTCP-FEEDBACK"/> <references> <name>References</name> <references> <name>Normative References</name> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.2119.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.3168.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.3550.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.5348.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.6679.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8174.xml"/> </references><references title="Informative References"> &rfc5450; &rfc6660; <!-- PCN --> &rfc6817; <!-- LEDBAT --> &rfc7567; <!-- RED --> &rfc8033; <!-- PIE --> &rfc8290; <!-- FQ-CoDel --> &rfc8593; <!-- Video Traffic Model --> <!--RMCAT related --> &I-D.ietf-rmcat-cc-requirements; &I-D.ietf-rmcat-cc-codec-interactions; &I-D.ietf-rmcat-eval-test; &I-D.ietf-rmcat-wireless-tests; &I-D.ietf-avtcore-cc-feedback-message;<references> <name>Informative References</name> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.5450.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.6660.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.6817.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.7567.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8033.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8290.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8593.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-cc-requirements.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-cc-codec-interactions.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-eval-test.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-wireless-tests.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-avtcore-cc-feedback-message.xml"/> <referenceanchor="Floyd-CCR00"anchor="FLOYD-CCR00" target=""> <front> <title> Equation-basedCongestion Controlcongestion control forUnicast Applicationsunicast applications </title> <seriesInfo name="DOI" value="10.1145/347057.347397"/> <author fullname="Sally Floyd" initials="S." surname="Floyd"><organization></organization><organization/> </author> <author fullname="Mark Handley" initials="M." surname="Handley"><organization></organization><organization/> </author> <author fullname="Jitendra Padhye" initials="J." surname="Padhye"><organization></organization><organization/> </author> <authorfullname="Jorgfullname="Jörg Widmer" initials="J." surname="Widmer"><organization></organization><organization/> </author> <date month="October"year="2000" />year="2000"/> </front><seriesInfo name="ACM<refcontent>ACM SIGCOMM Computer CommunicationsReview" value="vol.Review, vol. 30, no. 4, pp.43-56"/>43-56 </refcontent> </reference> <referenceanchor="Budzisz-TON11"anchor="BUDZISZ-AIMD-CC" target=""> <front> <title> On the Fair Coexistence of Loss- and Delay-Based TCP </title> <seriesInfo name="DOI" value="10.1109/TNET.2011.2159736"/> <author fullname="Lukasz Budzisz" initials="L." surname="Budzisz"><organization></organization><organization/> </author> <author fullname="Rade Stanojevic" initials="R." surname="Stanojevic"><organization></organization><organization/> </author> <author fullname="Arieh Schlote" initials="A." surname="Schlote"><organization></organization><organization/> </author> <author fullname="Fred Baker" initials="F." surname="Baker"><organization></organization><organization/> </author> <author fullname="Robert Shorten" initials="R." surname="Shorten"><organization></organization><organization/> </author> <date month="December"year="2011" />year="2011"/> </front><seriesInfo name="IEEE/ACM<refcontent>IEEE/ACM Transactions onNetworking" value="vol.Networking, vol. 19, no. 6, pp.1811-1824"/>1811-1824 </refcontent> </reference> <referenceanchor="Zhu-PV13"anchor="ZHU-PV13" target=""> <front> <title> NADA: A Unified Congestion Control Scheme for Low-Latency Interactive Video </title> <seriesInfo name="DOI" value=" 10.1109/PV.2013.6691448"/> <author fullname="Xiaoqing Zhu" initials="X." surname="Zhu"><organization></organization><organization/> </author> <author fullname="Rong Pan" initials="R." surname="Pan"><organization></organization><organization/> </author> <date month="December"year="2013" />year="2013"/> </front><seriesInfo name="in Proc.<refcontent>Proc. IEEE International Packet VideoWorkshop (PV'13)" value="SanWorkshop, San Jose, CA,USA"/>USA </refcontent> </reference> <referenceanchor="ns-2" target="http://www.isi.edu/nsnam/ns/">anchor="NS-2" target="http://nsnam.sourceforge.net/wiki/index.php/Main_Page"> <front><title>The Network Simulator - ns-2</title><title>ns-2</title> <author><organization></organization><organization/> </author> <date/>month="December" year="2014"/> </front> </reference> <referenceanchor="ns-3"anchor="NS-3" target="https://www.nsnam.org/"> <front><title>The<title>ns-3 NetworkSimulator - ns-3</title>Simulator</title> <author><organization></organization><organization/> </author><date /></front> </reference> <reference anchor="IETF-90" target="https://tools.ietf.org/agenda/90/slides/slides-90-rmcat-6.pdf"> <front> <title>NADA Update: Algorithm, Implementation, and Test CaseEvalua6onEvaluation Results</title> <author fullname="Xiaoqing Zhu" initials="X." surname="Zhu"><organization></organization><organization/> </author> <author fullname="Michael Ramalho" initials="M." surname="Ramalho"><organization></organization><organization/> </author> <author fullname="Charles Ganzhorn" initials="C." surname="Ganzhorn"><organization></organization><organization/> </author> <author fullname="Paul E. Jones"initials="P. E."initials="P." surname="Jones"><organization></organization><organization/> </author> <author fullname="Rong Pan" initials="R." surname="Pan"><organization></organization><organization/> </author> <datemonth= "July" year = "2014"/>month="July" year="2014"/> </front> <refcontent>IETF 90 </refcontent> </reference> <reference anchor="IETF-91"target="http://www.ietf.org/proceedings/interim/2014/11/09/rmcat/slides/slides-interim-2014-rmcat-1-2.pdf">target="https://www.ietf.org/proceedings/interim/2014/11/09/rmcat/slides/slides-interim-2014-rmcat-1-2.pdf"> <front> <title>NADA Algorithm Update and Test Case Evaluations</title> <author fullname="Xiaoqing Zhu" initials="X." surname="Zhu"><organization></organization><organization/> </author> <author fullname="Rong Pan" initials="R." surname="Pan"><organization></organization><organization/> </author> <author fullname="Michael Ramalho" initials="M." surname="Ramalho"><organization></organization><organization/> </author> <author fullname="Sergio Mena" initials="S." surname="Mena"><organization></organization><organization/> </author> <author fullname="Charles Ganzhorn" initials="C." surname="Ganzhorn"><organization></organization><organization/> </author> <author fullname="Paul E. Jones"initials="P. E."initials="P." surname="Jones"><organization></organization><organization/> </author> <author fullname="Stefano D'Aronco" initials="S." surname="D'Aronco"><organization></organization><organization/> </author> <datemonth= "November" year = "2014"/>month="November" year="2014"/> </front> <refcontent>IETF 91 </refcontent> </reference> <reference anchor="IETF-93" target="https://www.ietf.org/proceedings/93/slides/slides-93-rmcat-0.pdf"> <front> <title>Updates on NADA</title> <author fullname="Xiaoqing Zhu" initials="X." surname="Zhu"><organization></organization><organization/> </author> <author fullname="Rong Pan" initials="R." surname="Pan"><organization></organization><organization/> </author> <author fullname="Michael Ramalho" initials="M." surname="Ramalho"><organization></organization><organization/> </author> <author fullname="Sergio Mena" initials="S." surname="Mena"><organization></organization><organization/> </author> <author fullname="Charles Ganzhorn" initials="C." surname="Ganzhorn"><organization></organization><organization/> </author> <author fullname="Paul E. Jones"initials="P. E."initials="P." surname="Jones"><organization></organization><organization/> </author> <author fullname="Stefano D'Aronco" initials="S." surname="D'Aronco"><organization></organization><organization/> </author> <author fullname="Jiantao Fu" initials="J." surname="Fu"><organization></organization><organization/> </author> <datemonth= "July" year = "2015"/>month="July" year="2015"/> </front> <refcontent>IETF 93 </refcontent> </reference> <reference anchor="IETF-95" target="https://www.ietf.org/proceedings/95/slides/slides-95-rmcat-5.pdf"> <front> <title>Updates on NADA: Stability Analysis and Impact of Feedback Intervals</title> <author fullname="Xiaoqing Zhu" initials="X." surname="Zhu"><organization></organization><organization/> </author> <author fullname="Rong Pan" initials="R." surname="Pan"><organization></organization><organization/> </author> <author fullname="Michael Ramalho" initials="M." surname="Ramalho"><organization></organization><organization/> </author> <author fullname="Sergio Mena" initials="S." surname="Mena"><organization></organization><organization/> </author> <author fullname="Paul E. Jones"initials="P. E."initials="P." surname="Jones"><organization></organization><organization/> </author> <author fullname="Jiantao Fu" initials="J." surname="Fu"><organization></organization><organization/> </author> <author fullname="Stefano D'Aronco" initials="S." surname="D'Aronco"><organization></organization><organization/> </author> <author fullname="Charles Ganzhorn" initials="C." surname="Ganzhorn"><organization></organization><organization/> </author> <datemonth= "April" year = "2016"/>month="April" year="2016"/> </front> <refcontent>IETF 95 </refcontent> </reference> <reference anchor="IETF-103" target="https://datatracker.ietf.org/meeting/103/materials/slides-103-rmcat-nada-implementation-in-mozilla-browser-00"> <front> <title>NADA Implementation in Mozilla Browser</title> <author fullname="Xiaoqing Zhu" initials="X." surname="Zhu"><organization></organization><organization/> </author> <author fullname="Rong Pan" initials="R." surname="Pan"><organization></organization><organization/> </author> <author fullname="Michael Ramalho" initials="M." surname="Ramalho"><organization></organization><organization/> </author> <author fullname="Sergio Mena" initials="S." surname="Mena"><organization></organization><organization/> </author> <author fullname="Paul E. Jones"initials="P. E."initials="P." surname="Jones"><organization></organization><organization/> </author> <author fullname="Jiantao Fu" initials="J." surname="Fu"><organization></organization><organization/> </author> <author fullname="Stefano D'Aronco" initials="S." surname="D'Aronco"><organization></organization><organization/> </author> <datemonth= "November" year = "2018"/>month="November" year="2018"/> </front> <refcontent>IETF 103 </refcontent> </reference> <reference anchor="IETF-105" target="https://datatracker.ietf.org/meeting/105/materials/slides-105-rmcat-nada-update-02.pdf"> <front> <title>NADA Implementation in Mozilla Browser and Draft Update</title> <author fullname="Xiaoqing Zhu" initials="X." surname="Zhu"><organization></organization><organization/> </author> <author fullname="Rong Pan" initials="R." surname="Pan"><organization></organization><organization/> </author> <author fullname="Michael Ramalho" initials="M." surname="Ramalho"><organization></organization><organization/> </author> <author fullname="Sergio Mena" initials="S." surname="Mena"><organization></organization><organization/> </author> <author fullname="Paul E. Jones"initials="P. E."initials="P." surname="Jones"><organization></organization><organization/> </author> <author fullname="Jiantao Fu" initials="J." surname="Fu"><organization></organization><organization/> </author> <author fullname="Stefano D'Aronco" initials="S." surname="D'Aronco"><organization></organization><organization/> </author> <datemonth= "July" year = "2019"/>month="July" year="2019"/> </front> <refcontent>IETF 105 </refcontent> </reference> <referenceanchor="ns3-rmcat"anchor="NS3-RMCAT" target="https://github.com/cisco/ns3-rmcat"> <front><title>NS3 open source module<title>Simulator of IETF RMCAT congestion control protocols</title> <author fullname="Jiantao Fu" initials="J." surname="Fu"><organization></organization><organization/> </author> <author fullname="Sergio Mena" initials="S." surname="Mena"><organization></organization><organization/> </author> <author fullname="Xiaoqing Zhu" initials="X." surname="Zhu"><organization></organization><organization/> </author> <datemonth= "November" year = "2017"/>month="November" year="2017"/> </front> </reference> </references> </references> <section anchor="sec-network-nodes"title="Networknumbered="true" toc="default"> <name>Network NodeOperations">Operations</name> <t>NADA can work with different network queue management schemes and does not assume any specific network node operation. As an example, this appendix describes three variants of queue management behavior at the network node, leading to either implicit or explicit congestion signals. It needs to be acknowledged that NADA has not yet been tested with non-probabilistic ECN marking behaviors. </t> <t> In all three flavors described below, the network queue operates with the simplefirst-in-first-outFirst In, First Out (FIFO) principle. There is no need to maintain per-flow state. The system can scale easily with a large number of video flows and at high link capacity. </t> <section anchor="sec-network-droptail"title="Default behaviornumbered="true" toc="default"> <name>Default Behavior ofdrop tail queues">Drop-Tail Queues</name> <t> In a conventional network withdrop taildrop-tail or RED queues, congestion is inferred from the estimation of end-to-end delay and/or packet loss. Packet drops at the queue are detected at thereceiver,receiver andcontributescontribute to the calculation of the aggregated congestion signal x_curr. No special action is required at the network node. </t> </section> <section anchor="sec-network-ecn"title="RED-basednumbered="true" toc="default"> <name>RED-Based ECNmarking">Marking</name> <t>In this mode, the network node randomly marks the ECN field in the IP packet header following the <xreftarget="RFC7567">Randomtarget="RFC7567" format="default">Random Early Detection (RED) algorithm</xref>. Calculation of the marking probability involves the followingsteps:</t> <t><figure><artwork><![CDATA[steps on packetarrival: updatearrival:</t> <ol spacing="normal" type="1"> <li><t>update smoothed queue size q_avgas:as:</t> <artwork name="" type="" align="left" alt=""><![CDATA[ q_avg = w*q +(1-w)*q_avg. calculate(1-w)*q_avg ]]></artwork> </li> <li><t>calculate marking probability pas:as:</t> <artwork name="" type="" align="left" alt=""><![CDATA[ / 0, if q <q_lo;q_lo | | q_avg - q_lo p= < p_max*--------------, if q_lo <= q <q_hi;q_hi | q_hi - q_lo | \ p = 1, if q >=q_hi. ]]></artwork></figure></t>q_hi ]]></artwork> </li> </ol> <t> Here, q_lo and q_hicorrespondscorrespond to the low and high thresholds of queue occupancy. The maximum marking probability is p_max. </t> <t> The ECNmarkingsmarking events will contribute to the calculation of an equivalent delay x_curr at the receiver. No changes are required at the sender. </t> </section> <sectionanchor = "sec-network-pcn" title = "Randomanchor="sec-network-pcn" numbered="true" toc="default"> <name>Random Early Marking with VirtualQueues">Queues</name> <t> Advanced network nodes may support random early marking based on a token bucket algorithm originally designed for <xreftarget="RFC6660">Pre-Congestiontarget="RFC6660" format="default">Pre-Congestion Notification (PCN)</xref>. The early congestion notification (ECN) bit in the IP header of packetsareis marked randomly. The marking probability is calculated based on atoken-buckettoken bucket algorithm originally designed forthe<xreftarget="RFC6660">Pre-Congestion Notification (PCN)</xref>.target="RFC6660" format="default">PCN</xref>. The target link utilization is set as 90%; the marking probability is designed to grow linearly with the token bucket size when it varies between 1/3 and 2/3 of the full token bucket limit.</t> <t>Calculation of the marking probability involves the followingsteps:</t> <t><figure><artwork><![CDATA[steps upon packetarrival: meterarrival:</t> <ol spacing="normal" type="1"> <li><t>meter packet against token bucket(r,b); update(r,b)</t></li> <li><t>update token levelb_tk; calculateb_tk</t></li> <li><t>calculate the marking probabilityas:as:</t> <artwork name="" type="" align="left" alt=""><![CDATA[ / 0, if b-b_tk <b_lo;b_lo | | b-b_tk-b_lo p = < p_max* --------------, ifb_lo<=b_lo <= b-b_tk<b_hi;< b_hi | b_hi-b_lo | \ 1, ifb-b_tk>=b_hi.b-b_tk >= b_hi ]]></artwork></figure></t></li> </ol> <t> Here, the token bucket lower and upper limits are denoted by b_lo and b_hi, respectively. The parameter b indicates the size of the token bucket. The parameter r is chosen to be below capacity, resulting in slightunder-utilizationunderutilization of the link. The maximum marking probability is p_max. </t> <t>The ECNmarkingsmarking events will contribute to the calculation of an equivalent delay x_curr at the receiver. No changes are required at the sender. The virtual queuing mechanism from the PCN-based marking algorithm will lead to additional benefits such as zero standing queues. </t> </section> </section> <section anchor="sec-acknowledgments" numbered="false" toc="default"> <name>Acknowledgments</name> <t> The authors would like to thank <contact fullname="Randell Jesup"/>, <contact fullname="Luca De Cicco"/>, <contact fullname="Piers O'Hanlon"/>, <contact fullname="Ingemar Johansson"/>, <contact fullname="Stefan Holmer"/>, <contact fullname="Cesar Ilharco Magalhaes"/>, <contact fullname="Safiqul Islam"/>, <contact fullname="Michael Welzl"/>, <contact fullname="Mirja Kuehlewind"/>, <contact fullname="Karen Elisabeth Egede Nielsen"/>, <contact fullname="Julius Flohr"/>, <contact fullname="Roland Bless"/>, <contact fullname="Andreas Smas"/>, and <contact fullname="Martin Stiemerling"/> for their valuable review comments and helpful input to this specification. </t> </section> <section anchor="sec-contributors" numbered="false" toc="default"> <name>Contributors</name> <t>The following individuals contributed to the implementation and evaluation of the proposed scheme and, therefore, helped to validate and substantially improve this specification.</t> <t><contact fullname="Paul E. Jones"/> <paulej@packetizer.com> of Cisco Systems implemented an early version of the NADA congestion control scheme and helped with its lab-based testbed evaluations. </t> <t><contact fullname="Jiantao Fu"/> <jianfu@cisco.com> of Cisco Systems helped with the implementation and extensive evaluation of NADA both in Mozilla web browsers and in earlier simulation-based evaluation efforts. </t> <t><contact fullname="Stefano D'Aronco"/> <stefano.daronco@geod.baug.ethz.ch> of ETH Zurich (previously at Ecole Polytechnique Federale de Lausanne when contributing to this work) helped with the implementation and evaluation of an early version of NADA in <xref target="NS-3" format="default"/>. </t> <t><contact fullname="Charles Ganzhorn"/> <charles.ganzhorn@gmail.com> contributed to the testbed-based evaluation of NADA during an early stage of its development. </t> </section> </back> </rfc>