idnits 2.17.1 draft-ietf-rmcat-nada-11.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 478 has weird spacing: '... delta x_o...' -- The document date (July 25, 2019) is 1729 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Missing Reference: 'RMIN' is mentioned on line 669, but not defined == Missing Reference: 'RMAX' is mentioned on line 669, but not defined == Outdated reference: A later version (-11) exists of draft-ietf-rmcat-wireless-tests-08 Summary: 0 errors (**), 0 flaws (~~), 5 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group X. Zhu 3 Internet-Draft R. Pan 4 Intended status: Experimental M. Ramalho 5 Expires: January 26, 2020 S. Mena 6 P. Jones 7 J. Fu 8 Cisco Systems 9 S. D'Aronco 10 ETH 11 July 25, 2019 13 NADA: A Unified Congestion Control Scheme for Real-Time Media 14 draft-ietf-rmcat-nada-11 16 Abstract 18 This document describes NADA (network-assisted dynamic adaptation), a 19 novel congestion control scheme for interactive real-time media 20 applications, such as video conferencing. In the proposed scheme, 21 the sender regulates its sending rate based on either implicit or 22 explicit congestion signaling, in a unified approach. The scheme can 23 benefit from explicit congestion notification (ECN) markings from 24 network nodes. It also maintains consistent sender behavior in the 25 absence of such markings, by reacting to queuing delays and packet 26 losses instead. 28 Status of This Memo 30 This Internet-Draft is submitted in full conformance with the 31 provisions of BCP 78 and BCP 79. 33 Internet-Drafts are working documents of the Internet Engineering 34 Task Force (IETF). Note that other groups may also distribute 35 working documents as Internet-Drafts. The list of current Internet- 36 Drafts is at https://datatracker.ietf.org/drafts/current/. 38 Internet-Drafts are draft documents valid for a maximum of six months 39 and may be updated, replaced, or obsoleted by other documents at any 40 time. It is inappropriate to use Internet-Drafts as reference 41 material or to cite them other than as "work in progress." 43 This Internet-Draft will expire on January 26, 2020. 45 Copyright Notice 47 Copyright (c) 2019 IETF Trust and the persons identified as the 48 document authors. All rights reserved. 50 This document is subject to BCP 78 and the IETF Trust's Legal 51 Provisions Relating to IETF Documents 52 (https://trustee.ietf.org/license-info) in effect on the date of 53 publication of this document. Please review these documents 54 carefully, as they describe your rights and restrictions with respect 55 to this document. Code Components extracted from this document must 56 include Simplified BSD License text as described in Section 4.e of 57 the Trust Legal Provisions and are provided without warranty as 58 described in the Simplified BSD License. 60 Table of Contents 62 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 63 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 64 3. System Overview . . . . . . . . . . . . . . . . . . . . . . . 3 65 4. Core Congestion Control Algorithm . . . . . . . . . . . . . . 5 66 4.1. Mathematical Notations . . . . . . . . . . . . . . . . . 5 67 4.2. Receiver-Side Algorithm . . . . . . . . . . . . . . . . . 8 68 4.3. Sender-Side Algorithm . . . . . . . . . . . . . . . . . . 10 69 5. Practical Implementation of NADA . . . . . . . . . . . . . . 13 70 5.1. Receiver-Side Operation . . . . . . . . . . . . . . . . . 13 71 5.1.1. Estimation of one-way delay and queuing delay . . . . 13 72 5.1.2. Estimation of packet loss/marking ratio . . . . . . . 13 73 5.1.3. Estimation of receiving rate . . . . . . . . . . . . 14 74 5.2. Sender-Side Operation . . . . . . . . . . . . . . . . . . 14 75 5.2.1. Rate shaping buffer . . . . . . . . . . . . . . . . . 15 76 5.2.2. Adjusting video target rate and sending rate . . . . 16 77 5.3. Feedback Message Requirements . . . . . . . . . . . . . . 16 78 6. Discussions and Further Investigations . . . . . . . . . . . 17 79 6.1. Choice of delay metrics . . . . . . . . . . . . . . . . . 17 80 6.2. Method for delay, loss, and marking ratio estimation . . 18 81 6.3. Impact of parameter values . . . . . . . . . . . . . . . 18 82 6.4. Sender-based vs. receiver-based calculation . . . . . . . 19 83 6.5. Incremental deployment . . . . . . . . . . . . . . . . . 20 84 7. Reference Implementation . . . . . . . . . . . . . . . . . . 20 85 8. Suggested Experiments . . . . . . . . . . . . . . . . . . . . 20 86 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 21 87 10. Security Considerations . . . . . . . . . . . . . . . . . . . 21 88 11. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 22 89 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 22 90 12.1. Normative References . . . . . . . . . . . . . . . . . . 22 91 12.2. Informative References . . . . . . . . . . . . . . . . . 22 92 Appendix A. Network Node Operations . . . . . . . . . . . . . . 25 93 A.1. Default behavior of drop tail queues . . . . . . . . . . 25 94 A.2. RED-based ECN marking . . . . . . . . . . . . . . . . . . 25 95 A.3. Random Early Marking with Virtual Queues . . . . . . . . 26 96 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 27 98 1. Introduction 100 Interactive real-time media applications introduce a unique set of 101 challenges for congestion control. Unlike TCP, the mechanism used 102 for real-time media needs to adapt quickly to instantaneous bandwidth 103 changes, accommodate fluctuations in the output of video encoder rate 104 control, and cause low queuing delay over the network. An ideal 105 scheme should also make effective use of all types of congestion 106 signals, including packet loss, queuing delay, and explicit 107 congestion notification (ECN) [RFC3168] markings. The requirements 108 for the congestion control algorithm are outlined in 109 [I-D.ietf-rmcat-cc-requirements]. 111 This document describes an experimental congestion control scheme 112 called network-assisted dynamic adaptation (NADA). The NADA design 113 benefits from explicit congestion control signals (e.g., ECN 114 markings) from the network, yet also operates when only implicit 115 congestion indicators (delay and/or loss) are available. Such a 116 unified sender behavior distinguishes NADA from other congestion 117 control schemes for real-time media. In addition, its core 118 congestion control algorithm is designed to guarantee stability for 119 path round-trip-times (RTTs) below a prescribed bound (e.g., 250ms 120 with default parameter choices). It further supports weighted 121 bandwidth sharing among competing video flows with different 122 priorities. The signaling mechanism consists of standard RTP 123 timestamp [RFC3550] and RTCP feedback reports. The definition of 124 standardized RTCP feedback message requires future work so as to 125 support the successful operation of several congestion control 126 schemes for real-time interactive media. 128 2. Terminology 130 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 131 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 132 "OPTIONAL" in this document are to be interpreted as described in BCP 133 14 [RFC2119] [RFC8174] when, and only when, they appear in all 134 capitals, as shown here. 136 3. System Overview 138 Figure 1 shows the end-to-end system for real-time media transport 139 that NADA operates in. Note that there also exist network nodes 140 along the reverse (potentially uncongested) path that the RTCP 141 feedback reports traverse. Those network nodes are not shown in the 142 figure for sake of abrevity. 144 +---------+ r_vin +--------+ +--------+ +----------+ 145 | Media |<--------| RTP | |Network | | RTP | 146 | Encoder |========>| Sender |=======>| Node |====>| Receiver | 147 +---------+ r_vout +--------+ r_send +--------+ +----------+ 148 /|\ | 149 | | 150 +---------------------------------+ 151 RTCP Feedback Report 153 Figure 1: System Overview 155 o Media encoder with rate control capabilities. It encodes raw 156 media (audio and video) frames into compressed bitstream which is 157 later packetized into RTP packets. As discussed in 158 [I-D.ietf-rmcat-video-traffic-model], the actual output rate from 159 the encoder r_vout may fluctuate around the target r_vin. 160 Furthermore, it is possible that the encoder can only react to bit 161 rate changes at rather coarse time intervals, e.g., once every 0.5 162 seconds. 164 o RTP sender: responsible for calculating the NADA reference rate 165 based on network congestion indicators (delay, loss, or ECN 166 marking reports from the receiver), for updating the video encoder 167 with a new target rate r_vin, and for regulating the actual 168 sending rate r_send accordingly. The RTP sender also generates a 169 sending timestamp for each outgoing packet. 171 o RTP receiver: responsible for measuring and estimating end-to-end 172 delay (based on sender timestamp), packet loss (based on RTP 173 sequence number), ECN marking ratios (based on [RFC6679]), and 174 receiving rate (r_recv) of the flow. It calculates the aggregated 175 congestion signal (x_curr) that accounts for queuing delay, ECN 176 markings, and packet losses. The receiver also determines the 177 mode for sender rate adaptation (rmode) based on whether the flow 178 has encountered any standing non-zero congestion. The receiver 179 sends periodic RTCP reports back to the sender, containing values 180 of x_curr, rmode, and r_recv. 182 o Network node with several modes of operation. The system can work 183 with the default behavior of a simple drop tail queue. It can 184 also benefit from advanced AQM features such as PIE [RFC8033], FQ- 185 CoDel [RFC8290], ECN marking based on RED [RFC7567], and PCN 186 marking using a token bucket algorithm ([RFC6660]). Note that 187 network node operation is out of control for the design of NADA. 189 4. Core Congestion Control Algorithm 191 Like TCP-Friendly Rate Control (TFRC) [Floyd-CCR00] [RFC5348], NADA 192 is a rate-based congestion control algorithm. In its simplest form, 193 the sender reacts to the collection of network congestion indicators 194 in the form of an aggregated congestion signal, and operates in one 195 of two modes: 197 o Accelerated ramp-up: when the bottleneck is deemed to be 198 underutilized, the rate increases multiplicatively with respect to 199 the rate of previously successful transmissions. The rate 200 increase mutliplier (gamma) is calculated based on observed round- 201 trip-time and target feedback interval, so as to limit self- 202 inflicted queuing delay. 204 o Gradual rate update: in the presence of non-zero aggregate 205 congestion signal, the sending rate is adjusted in reaction to 206 both its value (x_curr) and its change in value (x_diff). 208 This section introduces the list of mathematical notations and 209 describes the core congestion control algorithm at the sender and 210 receiver, respectively. Additional details on recommended practical 211 implementations are described in Section 5.1 and Section 5.2. 213 4.1. Mathematical Notations 215 This section summarizes the list of variables and parameters used in 216 the NADA algorithm. Figure 3 also includes the default values for 217 choosing the algorithm parameters either to represent a typical 218 setting in practical applications or based on theoretical and 219 simulation studies. See Section 6.3 for some of the discussions on 220 the impact of parameter values. In the experimental phase of this 221 draft, additional studies in real-world settings will gather further 222 learnings on how to choose and adapt these parameter values in 223 practical deployment. 225 +--------------+-------------------------------------------------+ 226 | Notation | Variable Name | 227 +--------------+-------------------------------------------------+ 228 | t_curr | Current timestamp | 229 | t_last | Last time sending/receiving a feedback | 230 | delta | Observed interval between current and previous | 231 | | feedback reports: delta = t_curr-t_last | 232 | r_ref | Reference rate based on network congestion | 233 | r_send | Sending rate | 234 | r_recv | Receiving rate | 235 | r_vin | Target rate for video encoder | 236 | r_vout | Output rate from video encoder | 237 | d_base | Estimated baseline delay | 238 | d_fwd | Measured and filtered one-way delay | 239 | d_queue | Estimated queuing delay | 240 | d_tilde | Equivalent delay after non-linear warping | 241 | p_mark | Estimated packet ECN marking ratio | 242 | p_loss | Estimated packet loss ratio | 243 | x_curr | Aggregate congestion signal | 244 | x_prev | Previous value of aggregate congestion signal | 245 | x_diff | Change in aggregate congestion signal w.r.t. | 246 | | its previous value: x_diff = x_curr - x_prev | 247 | rmode | Rate update mode: (0 = accelerated ramp-up; | 248 | | 1 = gradual update) | 249 | gamma | Rate increase multiplier in accelerated ramp-up | 250 | | mode | 251 | loss_int | Measured average loss interval in packet count | 252 | loss_exp | Threshold value for setting the last observed | 253 | | packet loss to expiration | 254 | rtt | Estimated round-trip-time at sender | 255 | buffer_len | Rate shaping buffer occupancy measured in bytes | 256 +--------------+-------------------------------------------------+ 258 Figure 2: List of variables. 260 +--------------+----------------------------------+----------------+ 261 | Notation | Parameter Name | Default Value | 262 +--------------+----------------------------------+----------------+ 263 | PRIO | Weight of priority of the flow | 1.0 264 | RMIN | Minimum rate of application | 150 Kbps | 265 | | supported by media encoder | | 266 | RMAX | Maximum rate of application | 1.5 Mbps | 267 | | supported by media encoder | | 268 | XREF | Reference congestion level | 10ms | 269 | KAPPA | Scaling parameter for gradual | 0.5 | 270 | | rate update calculation | | 271 | ETA | Scaling parameter for gradual | 2.0 | 272 | | rate update calculation | | 273 | TAU | Upper bound of RTT in gradual | 500ms | 274 | | rate update calculation | | 275 | DELTA | Target feedback interval | 100ms | 276 +..............+..................................+................+ 277 | LOGWIN | Observation window in time for | 500ms | 278 | | calculating packet summary | | 279 | | statistics at receiver | | 280 | QEPS | Threshold for determining queuing| 10ms | 281 | | delay build up at receiver | | 282 | DFILT | Bound on filtering delay | 120ms | 283 | GAMMA_MAX | Upper bound on rate increase | 0.5 | 284 | | ratio for accelerated ramp-up | | 285 | QBOUND | Upper bound on self-inflicted | 50ms | 286 | | queuing delay during ramp up | | 287 +..............+..................................+................+ 288 | MULTILOSS | Multiplier for self-scaling the | 7. | 289 | | expiration threshold of the last | | 290 | | observed loss (loss_exp) based on| | 291 | | measured average loss interval | | 292 | | (loss_int) | | 293 | QTH | Delay threshold for invoking | 50ms | 294 | | non-linear warping | | 295 | LAMBDA | Scaling parameter in the | 0.5 | 296 | | exponent of non-linear warping | | 297 +..............+..................................+................+ 298 | PLRREF | Reference packet loss ratio | 0.01 | 299 | PMRREF | Reference packet marking ratio | 0.01 | 300 | DLOSS | Reference delay penalty for loss | 10ms | 301 | | when packet loss ratio is at | | 302 | | PLRREF | | 303 | DMARK | Reference delay penalty for ECN | 2ms | 304 | | marking when packet marking | | 305 | | is at PMRREF | | 306 +..............+..................................+................+ 307 | FPS | Frame rate of incoming video | 30 | 308 | BETA_S | Scaling parameter for modulating | 0.1 | 309 | | outgoing sending rate | | 310 | BETA_V | Scaling parameter for modulating | 0.1 | 311 | | video encoder target rate | | 312 | ALPHA | Smoothing factor in exponential | 0.1 | 313 | | smoothing of packet loss and | | 314 | | marking ratios | 315 +--------------+----------------------------------+----------------+ 317 Figure 3: List of algorithm parameters and their default values. 319 4.2. Receiver-Side Algorithm 321 The receiver-side algorithm can be outlined as below: 323 On initialization: 324 set d_base = +INFINITY 325 set p_loss = 0 326 set p_mark = 0 327 set r_recv = 0 328 set both t_last and t_curr as current time in milliseconds 330 On receiving a media packet: 331 obtain current timestamp t_curr from system clock 332 obtain from packet header sending time stamp t_sent 333 obtain one-way delay measurement: d_fwd = t_curr - t_sent 334 update baseline delay: d_base = min(d_base, d_fwd) 335 update queuing delay: d_queue = d_fwd - d_base 336 update packet loss ratio estimate p_loss 337 update packet marking ratio estimate p_mark 338 update measurement of receiving rate r_recv 340 On time to send a new feedback report (t_curr - t_last > DELTA): 341 calculate non-linear warping of delay d_tilde if packet loss exists 342 calculate current aggregate congestion signal x_curr 343 determine mode of rate adaptation for sender: rmode 344 send feedback containing values of: rmode, x_curr, and r_recv 345 update t_last = t_curr 347 In order for a delay-based flow to hold its ground when competing 348 against loss-based flows (e.g., loss-based TCP), it is important to 349 distinguish between different levels of observed queuing delay. For 350 instance, over wired connections, a moderate queuing delay value on 351 the order of tens of milliseonds is likely self-inflicted or induced 352 by other delay-based flows, whereas a high queuing delay value of 353 several hundreds of milliseconds may indicate the presence of a loss- 354 based flow that does not refrain from increased delay. 356 If the last observed packet loss is within the expiration window of 357 loss_exp (measured in terms of packet counts), the estimated queuing 358 delay follows a non-linear warping: 360 / d_queue, if d_queue |||||||||=================> 618 +----------+ -----------+ RTP packets 619 Rate Shaping Buffer 621 Figure 4: NADA Sender Structure 623 5.2.1. Rate shaping buffer 625 The operation of the live video encoder is out of the scope of the 626 design for the congestion control scheme in NADA. Instead, its 627 behavior is treated as a black box. 629 A rate shaping buffer is employed to absorb any instantaneous 630 mismatch between encoder rate output r_vout and regulated sending 631 rate r_send. Its current level of occupancy is measured in bytes and 632 is denoted as buffer_len. 634 A large rate shaping buffer contributes to higher end-to-end delay, 635 which may harm the performance of real-time media communications. 636 Therefore, the sender has a strong incentive to prevent the rate 637 shaping buffer from building up. The mechanisms adopted are: 639 o To deplete the rate shaping buffer faster by increasing the 640 sending rate r_send; and 642 o To limit incoming packets of the rate shaping buffer by reducing 643 the video encoder target rate r_vin. 645 5.2.2. Adjusting video target rate and sending rate 647 If the level of occupancy in the rate shaping buffer is accessible at 648 the sender, such information can be leveraged to further adjust the 649 target rate of the live video encoder r_vin as well as the actual 650 sending rate r_send. The purpose of such adjustments is to mitigate 651 the additional latencies introduced by the rate shaping buffer. The 652 amount of rate adjustment can be calculated as follows: 654 r_diff_v = min(0.05*r_ref, BETA_V*8*buffer_len*FPS). (11) 655 r_diff_s = min(0.05*r_ref, BETA_S*8*buffer_len*FPS). (12) 656 r_vin = max(RMIN, r_ref - r_diff_v). (13) 657 r_send = min(RMAX, r_ref + r_diff_s). (14) 659 In (11) and (12), the amount of adjustment is calculated as 660 proportional to the size of the rate shaping buffer but is bounded by 661 5% of the reference rate r_ref calculated from network congestion 662 feedback alone. This ensures that the adjustment introduced by the 663 rate shaping buffer will not counteract with the core congestion 664 control process. Equations (13) and (14) indicate the influence of 665 the rate shaping buffer. A large rate shaping buffer nudges the 666 encoder target rate slightly below -- and the sending rate slightly 667 above -- the reference rate r_ref. The final video target rate 668 (r_vin) and sending rate (r_send) are further bounded within the 669 original range of [RMIN, RMAX]. 671 Intuitively, the amount of extra rate offset needed to completely 672 drain the rate shaping buffer within the duration of a single video 673 frame is given by 8*buffer_len*FPS, where FPS stands for the 674 reference frame rate of the video. The scaling parameters BETA_V and 675 BETA_S can be tuned to balance between the competing goals of 676 maintaining a small rate shaping buffer and deviating from the 677 reference rate point. Empirical observations show that the rate 678 shaping buffer for a responsive live video encoder typically stays 679 empty and only occasionally holds a large frame (e.g., when an intra- 680 frame is produced) in transit. Therefore, the rate adjustment 681 introduced by this mechanism is expected to be minor. For instance, 682 a rate shaping buffer of 2000 Bytes will lead to a rate adjustment of 683 48 Kbps given the recommended scaling parameters of BETA_V = 0.1 and 684 BETA_S = 0.1 and reference frame rate of FPS = 30. 686 5.3. Feedback Message Requirements 688 The following list of information is required for NADA congestion 689 control to function properly: 691 o Recommended rate adaptation mode (rmode): a 1-bit flag indicating 692 whether the sender should operate in accelerated ramp-up mode 693 (rmode=0) or gradual update mode (rmode=1). 695 o Aggregated congestion signal (x_curr): the most recently updated 696 value, calculated by the receiver according to Section 4.2. This 697 information is expressed with a unit of 100 microsecond (i.e., 698 1/10 of a millisecond) in 15 bits. This allows a maximum value of 699 x_curr at approximately 3.27 second. 701 o Receiving rate (r_recv): the most recently measured receiving rate 702 according to Section 5.1.3. This information is expressed with a 703 unit of bits per second (bps) in 32 bits (unsigned int). This 704 allows a maximum rate of approximately 4.3Gbps, approximately 1000 705 times of the streaming rate of a typical high-definition (HD) 706 video conferencing session today. This field can be expanded 707 further by a few more bytes, in case an even higher rate need to 708 be specified. 710 The above list of information can be accommodated by 48 bits, or 6 711 bytes, in total. Choice of the feedback message interval DELTA is 712 discussed in Section 6.3 A target feedback interval of DELTA=100ms is 713 recommended. 715 6. Discussions and Further Investigations 717 This section discussed the various design choices made by NADA, 718 potential alternative variants of its implementation, and guidelines 719 on how the key algorithm parameters can be chosen. Section 8 720 recommends additional experimental setups to further explore these 721 topics. 723 6.1. Choice of delay metrics 725 The current design works with relative one-way-delay (OWD) as the 726 main indication of congestion. The value of the relative OWD is 727 obtained by maintaining the minimum value of observed OWD over a 728 relatively long time horizon and subtract that out from the observed 729 absolute OWD value. Such an approach cancels out the fixed 730 difference between the sender and receiver clocks. It has been 731 widely adopted by other delay-based congestion control approaches 732 such as [RFC6817]. As discussed in [RFC6817], the time horizon for 733 tracking the minimum OWD needs to be chosen with care: it must be 734 long enough for an opportunity to observe the minimum OWD with zero 735 standing queue along the path, and sufficiently short so as to timely 736 reflect "true" changes in minimum OWD introduced by route changes and 737 other rare events. 739 The potential drawback in relying on relative OWD as the congestion 740 signal is that when multiple flows share the same bottleneck, the 741 flow arriving late at the network experiencing a non-empty queue may 742 mistakenly consider the standing queuing delay as part of the fixed 743 path propagation delay. This will lead to slightly unfair bandwidth 744 sharing among the flows. 746 Alternatively, one could move the per-packet statistical handling to 747 the sender instead and use relative round-trip-time (RTT) in lieu of 748 relative OWD, assuming that per-packet acknowledgements are 749 available. The main drawback of RTT-based approach is the noise in 750 the measured delay in the reverse direction. 752 Note that the choice of either delay metric (relative OWD vs. RTT) 753 involves no change in the proposed rate adaptation algorithm. 754 Therefore, comparing the pros and cons regarding which delay metric 755 to adopt can be kept as an orthogonal direction of investigation. 757 6.2. Method for delay, loss, and marking ratio estimation 759 Like other delay-based congestion control schemes, performance of 760 NADA depends on the accuracy of its delay measurement and estimation 761 module. Appendix A in [RFC6817] provides an extensive discussion on 762 this aspect. 764 The current recommended practice of applying minimum filter with a 765 window size of 15 samples suffices in guarding against processing 766 delay outliers observed in wired connections. For wireless 767 connections with a higher packet delay variation (PDV), more 768 sophisticated techniques on de-noising, outlier rejection, and trend 769 analysis may be needed. 771 More sophisticated methods in packet loss ratio calculation, such as 772 that adopted by [Floyd-CCR00], will likely be beneficial. These 773 alternatives are part of the experiments this document proposes. 775 6.3. Impact of parameter values 777 In the gradual rate update mode, the parameter TAU indicates the 778 upper bound of round-trip-time (RTT) in feedback control loop. 779 Typically, the observed feedback interval delta is close to the 780 target feedback interval DELTA, and the relative ratio of delta/TAU 781 versus ETA dictates the relative strength of influence from the 782 aggregate congestion signal offset term (x_offset) versus its recent 783 change (x_diff), respectively. These two terms are analogous to the 784 integral and proportional terms in a proportional-integral (PI) 785 controller. The recommended choice of TAU=500ms, DELTA=100ms and ETA 786 = 2.0 corresponds to a relative ratio of 1:10 between the gains of 787 the integral and proportional terms. Consequently, the rate 788 adaptation is mostly driven by the change in the congestion signal 789 with a long-term shift towards its equilibrium value driven by the 790 offset term. Finally, the scaling parameter KAPPA determines the 791 overall speed of the adaptation and needs to strike a balance between 792 responsiveness and stability. 794 The choice of the target feedback interval DELTA needs to strike the 795 right balance between timely feedback and low RTCP feedback message 796 counts. A target feedback interval of DELTA=100ms is recommended, 797 corresponding to a feedback bandwidth of 16Kbps with 200 bytes per 798 feedback message --- approximately 1.6% overhead for a 1 Mbps flow. 799 Furthermore, both simulation studies and frequency-domain analysis in 800 [IETF-95] have established that a feedback interval below 250ms 801 (i.e., more frequently than 4 feedback messages per second) will not 802 break up the feedback control loop of NADA congestion control. 804 In calculating the non-linear warping of delay in (1), the current 805 design uses fixed values of QTH for determining whether to perform 806 the non-linear warping). Its value may need to be tuned for 807 different operational enviornments (e.g., over wired vs. wireless 808 connections). It is possible to adapt its value based on past 809 observed patterns of queuing delay in the presence of packet losses. 810 It needs to be noted that the non-linear warping mechanism may lead 811 to multiple NADA streams stuck in loss-based mode when competing 812 against each other. 814 In calculating the aggregate congestion signal x_curr, the choice of 815 DMARK and DLOSS influence the steady-state packet loss/marking ratio 816 experienced by the flow at a given available bandwidth. Higher 817 values of DMARK and DLOSS result in lower steady-state loss/marking 818 ratios, but are more susceptible to the impact of individual packet 819 loss/marking events. While the value of DMARK and DLOSS are fixed 820 and predetermined in the current design, this document also 821 encourages futher explorations of a scheme for automatically tuning 822 these values based on desired bandwidth sharing behavior in the 823 presence of other competing loss-based flows (e.g., loss-based TCP). 825 6.4. Sender-based vs. receiver-based calculation 827 In the current design, the aggregated congestion signal x_curr is 828 calculated at the receiver, keeping the sender operation completely 829 independent of the form of actual network congestion indications 830 (delay, loss, or marking). Alternatively, one can move the logics of 831 (1) and (2) to the sender. Such an approach requires slightly higher 832 overhead in the feedback messages, which should contain individual 833 fields on queuing delay (d_queue), packet loss ratio (p_loss), packet 834 marking ratio (p_mark), receiving rate (r_recv), and recommended rate 835 adaptation mode (rmode). 837 6.5. Incremental deployment 839 One nice property of NADA is the consistent video endpoint behavior 840 irrespective of network node variations. This facilitates gradual, 841 incremental adoption of the scheme. 843 To start off with, the proposed congestion control mechanism can be 844 implemented without any explicit support from the network, and relies 845 solely on observed one-way delay measurements and packet loss ratios 846 as implicit congestion signals. 848 When ECN is enabled at the network nodes with RED-based marking, the 849 receiver can fold its observations of ECN markings into the 850 calculation of the equivalent delay. The sender can react to these 851 explicit congestion signals without any modification. 853 Ultimately, networks equipped with proactive marking based on token 854 bucket level metering can reap the additional benefits of zero 855 standing queues and lower end-to-end delay and work seamlessly with 856 existing senders and receivers. 858 7. Reference Implementation 860 The NADA scheme has been implemented in [ns-2] and [ns-3] simulation 861 platforms. Extensive ns-2 simulation evaluations of an earlier 862 version of the draft are documented in [Zhu-PV13]. Evaluation 863 results of the current draft over several test cases in 864 [I-D.ietf-rmcat-eval-test] have been presented at recent IETF 865 meetings [IETF-90][IETF-91]. Evaluation results of the current draft 866 over several test cases in [I-D.ietf-rmcat-wireless-tests] have been 867 presented at [IETF-93]. An open source implementation of NADA as 868 part of a ns-3 module is available at [ns3-rmcat] 870 The scheme has also been implemented and evaluated in a lab setting 871 as described in [IETF-90]. Preliminary evaluation results of NADA in 872 single-flow and multi-flow scenarios have been presented in 873 [IETF-91]. 875 8. Suggested Experiments 877 NADA has been extensively evaluated under various test scenarios, 878 including the collection of test cases specified by 879 [I-D.ietf-rmcat-eval-test] and the subset of WiFi-based test cases in 880 [I-D.ietf-rmcat-wireless-tests]. Additional evaluations have been 881 carried out to characterize how NADA interacts with various active 882 queue management (AQM) schemes such as RED, CoDel, and PIE. Most of 883 these evaluations have been carried out in simulators. A few key 884 test cases have been evaluated in lab environments with 885 implementations embedded in video conferencing clients. It is 886 strongly recommended to carry out implementation and experimentation 887 of NADA in real-world settings. Such exercise will provide insights 888 on how to choose to automatically adapte the values of the key 889 algorithm parameters (see list in Figure 3) as discussed in 890 Section 6. 892 Additional experiments are suggested for the following scenarios and 893 preferrably over real-world networks: 895 o Experiments reflecting the set up of a typical WAN connection. 897 o Experiments with ECN marking capability turned on at the network 898 for existing test cases. 900 o Experiments with multiple RMCAT streams bearing different user- 901 specified priorities. 903 o Experiments with additional access technologies, especially over 904 cellular networks such as 3G/LTE. 906 o Experiments with various media source contents, including audio 907 only, audio and video, and application content sharing (e.g., 908 slide shows). 910 9. IANA Considerations 912 This document makes no request of IANA. 914 10. Security Considerations 916 The rate adaptation mechanism in NADA relies on feedback from the 917 receiver. As such, it is vulnerable to attacks where feedback 918 messages are hijacked, replaces, or intentionally injected with 919 misleading information, similar to those that can affect TCP. It is 920 therefore RECOMMENDED that the RTCP feedback message is at least 921 integrity checked. The modification of sending rate based on send- 922 side rate shaping buffer may lead to temporary excessive congestion 923 over the network in the presence of a unresponsive video encoder. 924 However, this effect can be mitigated by limiting the amount of rate 925 modification introduced by the rate shaping buffer, bounding the size 926 of the rate shaping buffer at the sender, and maintaining a maximum 927 allowed sending rate by NADA. 929 11. Acknowledgments 931 The authors would like to thank Randell Jesup, Luca De Cicco, Piers 932 O'Hanlon, Ingemar Johansson, Stefan Holmer, Cesar Ilharco Magalhaes, 933 Safiqul Islam, Michael Welzl, Mirja Kuhlewind, Karen Elisabeth Egede 934 Nielsen, Julius Flohr, Roland Bless, Andreas Smas, and Martin 935 Stiemerling for their various valuable review comments and feedback. 936 Thanks to Charles Ganzhorn for contributing to the testbed-based 937 evaluation of NADA during an early stage of its development. 939 12. References 941 12.1. Normative References 943 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 944 Requirement Levels", BCP 14, RFC 2119, 945 DOI 10.17487/RFC2119, March 1997, 946 . 948 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 949 of Explicit Congestion Notification (ECN) to IP", 950 RFC 3168, DOI 10.17487/RFC3168, September 2001, 951 . 953 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 954 Jacobson, "RTP: A Transport Protocol for Real-Time 955 Applications", STD 64, RFC 3550, DOI 10.17487/RFC3550, 956 July 2003, . 958 [RFC5348] Floyd, S., Handley, M., Padhye, J., and J. Widmer, "TCP 959 Friendly Rate Control (TFRC): Protocol Specification", 960 RFC 5348, DOI 10.17487/RFC5348, September 2008, 961 . 963 [RFC6679] Westerlund, M., Johansson, I., Perkins, C., O'Hanlon, P., 964 and K. Carlberg, "Explicit Congestion Notification (ECN) 965 for RTP over UDP", RFC 6679, DOI 10.17487/RFC6679, August 966 2012, . 968 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 969 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 970 May 2017, . 972 12.2. Informative References 974 [Budzisz-TON11] 975 Budzisz, L., Stanojevic, R., Schlote, A., Baker, F., and 976 R. Shorten, "On the Fair Coexistence of Loss- and Delay- 977 Based TCP", IEEE/ACM Transactions on Networking vol. 19, 978 no. 6, pp. 1811-1824, December 2011. 980 [Floyd-CCR00] 981 Floyd, S., Handley, M., Padhye, J., and J. Widmer, 982 "Equation-based Congestion Control for Unicast 983 Applications", ACM SIGCOMM Computer Communications 984 Review vol. 30, no. 4, pp. 43-56, October 2000. 986 [I-D.ietf-rmcat-cc-codec-interactions] 987 Zanaty, M., Singh, V., Nandakumar, S., and Z. Sarker, 988 "Congestion Control and Codec interactions in RTP 989 Applications", draft-ietf-rmcat-cc-codec-interactions-02 990 (work in progress), March 2016. 992 [I-D.ietf-rmcat-cc-requirements] 993 Jesup, R. and Z. Sarker, "Congestion Control Requirements 994 for Interactive Real-Time Media", draft-ietf-rmcat-cc- 995 requirements-09 (work in progress), December 2014. 997 [I-D.ietf-rmcat-eval-test] 998 Sarker, Z., Singh, V., Zhu, X., and M. Ramalho, "Test 999 Cases for Evaluating RMCAT Proposals", draft-ietf-rmcat- 1000 eval-test-10 (work in progress), May 2019. 1002 [I-D.ietf-rmcat-video-traffic-model] 1003 Zhu, X., Cruz, S., and Z. Sarker, "Video Traffic Models 1004 for RTP Congestion Control Evaluations", draft-ietf-rmcat- 1005 video-traffic-model-07 (work in progress), February 2019. 1007 [I-D.ietf-rmcat-wireless-tests] 1008 Sarker, Z., Johansson, I., Zhu, X., Fu, J., Tan, W., and 1009 M. Ramalho, "Evaluation Test Cases for Interactive Real- 1010 Time Media over Wireless Networks", draft-ietf-rmcat- 1011 wireless-tests-08 (work in progress), July 2019. 1013 [IETF-90] Zhu, X., Ramalho, M., Ganzhorn, C., Jones, P., and R. Pan, 1014 "NADA Update: Algorithm, Implementation, and Test Case 1015 Evalua6on Results", July 2014, 1016 . 1019 [IETF-91] Zhu, X., Pan, R., Ramalho, M., Mena, S., Ganzhorn, C., 1020 Jones, P., and S. D'Aronco, "NADA Algorithm Update and 1021 Test Case Evaluations", November 2014, 1022 . 1025 [IETF-93] Zhu, X., Pan, R., Ramalho, M., Mena, S., Ganzhorn, C., 1026 Jones, P., D'Aronco, S., and J. Fu, "Updates on NADA", 1027 July 2015, . 1030 [IETF-95] Zhu, X., Pan, R., Ramalho, M., Mena, S., Jones, P., Fu, 1031 J., D'Aronco, S., and C. Ganzhorn, "Updates on NADA: 1032 Stability Analysis and Impact of Feedback Intervals", 1033 April 2016, . 1036 [ns-2] "The Network Simulator - ns-2", 1037 . 1039 [ns-3] "The Network Simulator - ns-3", . 1041 [ns3-rmcat] 1042 Fu, J., Mena, S., and X. Zhu, "NS3 open source module of 1043 IETF RMCAT congestion control protocols", November 2017, 1044 . 1046 [RFC6660] Briscoe, B., Moncaster, T., and M. Menth, "Encoding Three 1047 Pre-Congestion Notification (PCN) States in the IP Header 1048 Using a Single Diffserv Codepoint (DSCP)", RFC 6660, 1049 DOI 10.17487/RFC6660, July 2012, 1050 . 1052 [RFC6817] Shalunov, S., Hazel, G., Iyengar, J., and M. Kuehlewind, 1053 "Low Extra Delay Background Transport (LEDBAT)", RFC 6817, 1054 DOI 10.17487/RFC6817, December 2012, 1055 . 1057 [RFC7567] Baker, F., Ed. and G. Fairhurst, Ed., "IETF 1058 Recommendations Regarding Active Queue Management", 1059 BCP 197, RFC 7567, DOI 10.17487/RFC7567, July 2015, 1060 . 1062 [RFC8033] Pan, R., Natarajan, P., Baker, F., and G. White, 1063 "Proportional Integral Controller Enhanced (PIE): A 1064 Lightweight Control Scheme to Address the Bufferbloat 1065 Problem", RFC 8033, DOI 10.17487/RFC8033, February 2017, 1066 . 1068 [RFC8290] Hoeiland-Joergensen, T., McKenney, P., Taht, D., Gettys, 1069 J., and E. Dumazet, "The Flow Queue CoDel Packet Scheduler 1070 and Active Queue Management Algorithm", RFC 8290, 1071 DOI 10.17487/RFC8290, January 2018, 1072 . 1074 [Zhu-PV13] 1075 Zhu, X. and R. Pan, "NADA: A Unified Congestion Control 1076 Scheme for Low-Latency Interactive Video", in Proc. IEEE 1077 International Packet Video Workshop (PV'13) San Jose, CA, 1078 USA, December 2013. 1080 Appendix A. Network Node Operations 1082 NADA can work with different network queue management schemes and 1083 does not assume any specific network node operation. As an example, 1084 this appendix describes three variants of queue management behavior 1085 at the network node, leading to either implicit or explicit 1086 congestion signals. It needs to be acknowledged that NADA has not 1087 yet been tested with non-probabilistic ECN marking behaviors. 1089 In all three flavors described below, the network queue operates with 1090 the simple first-in-first-out (FIFO) principle. There is no need to 1091 maintain per-flow state. The system can scale easily with a large 1092 number of video flows and at high link capacity. 1094 A.1. Default behavior of drop tail queues 1096 In a conventional network with drop tail or RED queues, congestion is 1097 inferred from the estimation of end-to-end delay and/or packet loss. 1098 Packet drops at the queue are detected at the receiver, and 1099 contributes to the calculation of the aggregated congestion signal 1100 x_curr. No special action is required at network node. 1102 A.2. RED-based ECN marking 1104 In this mode, the network node randomly marks the ECN field in the IP 1105 packet header following the Random Early Detection (RED) algorithm 1106 [RFC7567]. Calculation of the marking probability involves the 1107 following steps: 1109 on packet arrival: 1110 update smoothed queue size q_avg as: 1111 q_avg = w*q + (1-w)*q_avg. 1113 calculate marking probability p as: 1115 / 0, if q < q_lo; 1116 | 1117 | q_avg - q_lo 1118 p= < p_max*--------------, if q_lo <= q < q_hi; 1119 | q_hi - q_lo 1120 | 1121 \ p = 1, if q >= q_hi. 1123 Here, q_lo and q_hi corresponds to the low and high thresholds of 1124 queue occupancy. The maximum marking probability is p_max. 1126 The ECN markings events will contribute to the calculation of an 1127 equivalent delay x_curr at the receiver. No changes are required at 1128 the sender. 1130 A.3. Random Early Marking with Virtual Queues 1132 Advanced network nodes may support random early marking based on a 1133 token bucket algorithm originally designed for Pre-Congestion 1134 Notification (PCN) [RFC6660]. The early congestion notification 1135 (ECN) bit in the IP header of packets are marked randomly. The 1136 marking probability is calculated based on a token-bucket algorithm 1137 originally designed for the Pre-Congestion Notification (PCN) 1138 [RFC6660]. The target link utilization is set as 90%; the marking 1139 probability is designed to grow linearly with the token bucket size 1140 when it varies between 1/3 and 2/3 of the full token bucket limit. 1142 Calculation of the marking probability involves the following steps: 1144 upon packet arrival: 1145 meter packet against token bucket (r,b); 1147 update token level b_tk; 1149 calculate the marking probability as: 1151 / 0, if b-b_tk < b_lo; 1152 | 1153 | b-b_tk-b_lo 1154 p = < p_max* --------------, if b_lo<= b-b_tk =b_hi. 1159 Here, the token bucket lower and upper limits are denoted by b_lo and 1160 b_hi, respectively. The parameter b indicates the size of the token 1161 bucket. The parameter r is chosen to be below capacity, resulting in 1162 slight under-utilization of the link. The maximum marking 1163 probability is p_max. 1165 The ECN markings events will contribute to the calculation of an 1166 equivalent delay x_curr at the receiver. No changes are required at 1167 the sender. The virtual queuing mechanism from the PCN-based marking 1168 algorithm will lead to additional benefits such as zero standing 1169 queues. 1171 Authors' Addresses 1173 Xiaoqing Zhu 1174 Cisco Systems 1175 12515 Research Blvd., Building 4 1176 Austin, TX 78759 1177 USA 1179 Email: xiaoqzhu@cisco.com 1181 Rong Pan * 1182 * Pending affiliation change. 1184 Email: rong.pan@gmail.com 1185 Michael A. Ramalho 1186 Cisco Systems, Inc. 1187 8000 Hawkins Road 1188 Sarasota, FL 34241 1189 USA 1191 Phone: +1 919 476 2038 1192 Email: mramalho@cisco.com 1194 Sergio Mena de la Cruz 1195 Cisco Systems 1196 EPFL, Quartier de l'Innovation, Batiment E 1197 Ecublens, Vaud 1015 1198 Switzerland 1200 Email: semena@cisco.com 1202 Paul E. Jones 1203 Cisco Systems 1204 7025 Kit Creek Rd. 1205 Research Triangle Park, NC 27709 1206 USA 1208 Email: paulej@packetizer.com 1210 Jiantao Fu 1211 Cisco Systems 1212 707 Tasman Drive 1213 Milpitas, CA 95035 1214 USA 1216 Email: jianfu@cisco.com 1218 Stefano D'Aronco 1219 ETH Zurich 1220 Stefano-Franscini-Platz 5 1221 Zurich 8093 1222 Switzerland 1224 Email: stefano.daronco@geod.baug.ethz.ch