idnits 2.17.1 draft-ietf-rmcat-wireless-tests-10.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 294 has weird spacing: '...nternet fix...' -- The document date (March 9, 2020) is 1507 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: '40ms' is mentioned on line 329, but not defined == Missing Reference: '150ms' is mentioned on line 329, but not defined == Missing Reference: 'Uplink' is mentioned on line 348, but not defined == Missing Reference: 'Downlink' is mentioned on line 348, but not defined -- Looks like a reference, but probably isn't: '4' on line 818 -- Looks like a reference, but probably isn't: '8' on line 818 -- Looks like a reference, but probably isn't: '12' on line 818 -- Looks like a reference, but probably isn't: '16' on line 818 -- Looks like a reference, but probably isn't: '20' on line 818 == Unused Reference: 'RFC8174' is defined on line 966, but no explicit reference was found in the text == Outdated reference: A later version (-14) exists of draft-ietf-rmcat-eval-criteria-11 Summary: 0 errors (**), 0 flaws (~~), 8 warnings (==), 6 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group Z. Sarker 3 Internet-Draft Ericsson AB 4 Intended status: Informational X. Zhu 5 Expires: September 10, 2020 J. Fu 6 W. Tan 7 Cisco Systems 8 M. Ramalho 9 AcousticComms 10 March 9, 2020 12 Evaluation Test Cases for Interactive Real-Time Media over Wireless 13 Networks 14 draft-ietf-rmcat-wireless-tests-10 16 Abstract 18 The Real-time Transport Protocol (RTP) is a common transport choice 19 for interactive multimedia communication applications. The 20 performance of these applications typically depends on a well- 21 functioning congestion control algorithm. To ensure a seamless and 22 robust user experience, a well-designed RTP-based congestion control 23 algorithm should work well across all access network types. This 24 document describes test cases for evaluating performances of 25 candidate congestion control algorithms over cellular and Wi-Fi 26 networks. 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 September 10, 2020. 45 Copyright Notice 47 Copyright (c) 2020 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. Cellular Network Specific Test Cases . . . . . . . . . . . . 3 64 2.1. Varying Network Load . . . . . . . . . . . . . . . . . . 6 65 2.1.1. Network Connection . . . . . . . . . . . . . . . . . 6 66 2.1.2. Simulation Setup . . . . . . . . . . . . . . . . . . 7 67 2.2. Bad Radio Coverage . . . . . . . . . . . . . . . . . . . 9 68 2.2.1. Network connection . . . . . . . . . . . . . . . . . 9 69 2.2.2. Simulation Setup . . . . . . . . . . . . . . . . . . 9 70 2.3. Desired Evaluation Metrics for cellular test cases . . . 10 71 3. Wi-Fi Networks Specific Test Cases . . . . . . . . . . . . . 10 72 3.1. Bottleneck in Wired Network . . . . . . . . . . . . . . . 12 73 3.1.1. Network topology . . . . . . . . . . . . . . . . . . 12 74 3.1.2. Test setup . . . . . . . . . . . . . . . . . . . . . 13 75 3.1.3. Typical test scenarios . . . . . . . . . . . . . . . 14 76 3.1.4. Expected behavior . . . . . . . . . . . . . . . . . . 15 77 3.2. Bottleneck in Wi-Fi Network . . . . . . . . . . . . . . . 15 78 3.2.1. Network topology . . . . . . . . . . . . . . . . . . 15 79 3.2.2. Test setup . . . . . . . . . . . . . . . . . . . . . 15 80 3.2.3. Typical test scenarios . . . . . . . . . . . . . . . 17 81 3.2.4. Expected behavior . . . . . . . . . . . . . . . . . . 18 82 3.3. Other Potential Test Cases . . . . . . . . . . . . . . . 19 83 3.3.1. EDCA/WMM usage . . . . . . . . . . . . . . . . . . . 19 84 3.3.2. Effect of heterogeneous link rates . . . . . . . . . 19 85 4. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 20 86 5. Security Considerations . . . . . . . . . . . . . . . . . . . 20 87 6. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 20 88 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 20 89 7.1. Normative References . . . . . . . . . . . . . . . . . . 20 90 7.2. Informative References . . . . . . . . . . . . . . . . . 21 91 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 22 93 1. Introduction 95 Wireless networks (both cellular and Wi-Fi [IEEE802.11]) are an 96 integral and increasingly more significant part of the Internet. 97 Typical application scenarios for interactive multimedia 98 communication over wireless include from video conferencing calls in 99 a bus or train as well as live media streaming at home. It is well 100 known that the characteristics and technical challenges for 101 supporting multimedia services over wireless are very different from 102 those of providing the same service over a wired network. Although 103 the basic test cases as defined in [I-D.ietf-rmcat-eval-test] have 104 covered many common effects of network impairments for evaluating 105 RTP-based congestion control schemes, they remain to be tested over 106 characteristics and dynamics unique to a given wireless environment. 107 For example, in cellular networks, the base station maintains 108 individual queues per radio bearer per user hence it leads to a 109 different nature of interactions between traffic flows of different 110 users. This contrasts with a typical wired network setting where 111 traffic flows from all users share the same queue at the bottleneck. 112 Furthermore, user mobility patterns in a cellular network differ from 113 those in a Wi-Fi network. Therefore, it is important to evaluate the 114 performance of proposed candidate RTP-based congestion control 115 solutions over cellular mobile networks and over Wi-Fi networks 116 respectively. 118 The draft [I-D.ietf-rmcat-eval-criteria] provides the guideline for 119 evaluating candidate algorithms and recognizes the importance of 120 testing over wireless access networks. However, it does not describe 121 any specific test cases for performance evaluation of candidate 122 algorithms. This document describes test cases specifically 123 targeting cellular and Wi-Fi networks. 125 2. Cellular Network Specific Test Cases 127 A cellular environment is more complicated than its wireline 128 counterpart since it seeks to provide services in the context of 129 variable available bandwidth, location dependencies and user 130 mobilities at different speeds. In a cellular network, the user may 131 reach the cell edge which may lead to a significant amount of 132 retransmissions to deliver the data from the base station to the 133 destination and vice versa. These radio links will often act as a 134 bottleneck for the rest of the network and will eventually lead to 135 excessive delays or packet drops. An efficient retransmission or 136 link adaptation mechanism can reduce the packet loss probability but 137 there will remain some packet losses and delay variations. Moreover, 138 with increased cell load or handover to a congested cell, congestion 139 in the transport network will become even worse. Besides, there 140 exist certain characteristics that distinguish the cellular network 141 from other wireless access networks such as Wi-Fi. In a cellular 142 network -- 144 o The bottleneck is often a shared link with relatively few users. 146 * The cost per bit over the shared link varies over time and is 147 different for different users. 149 * Leftover/unused resources can be consumed by other greedy 150 users. 152 o Queues are always per radio bearer hence each user can have many 153 such queues. 155 o Users can experience both Inter and Intra Radio Access Technology 156 (RAT) handovers (see [HO-def-3GPP] for the definition of 157 "handover"). 159 o Handover between cells or change of serving cells (as described in 160 [HO-LTE-3GPP] and [HO-UMTS-3GPP]) might cause user plane 161 interruptions which can lead to bursts of packet losses, delay 162 and/or jitter. The exact behavior depends on the type of radio 163 bearer. Typically, the default best-effort bearers do not 164 generate packet loss, instead, packets are queued up and 165 transmitted once the handover is completed. 167 o The network part decides how much the user can transmit. 169 o The cellular network has variable link capacity per user. 171 * It can vary as fast as a period of milliseconds. 173 * It depends on many factors (such as distance, speed, 174 interference, different flows). 176 * It uses complex and smart link adaptation which makes the link 177 behavior ever more dynamic. 179 * The scheduling priority depends on the estimated throughput. 181 o Both Quality of Service (QoS) and non-QoS radio bearers can be 182 used. 184 Hence, a real-time communication application operating over a 185 cellular network needs to cope with a shared bottleneck link and 186 variable link capacity, events like handover, non-congestion related 187 loss, abrupt changes in bandwidth (both short term and long term) due 188 to handover, network load and bad radio coverage. Even though 3GPP 189 has defined QoS bearers [QoS-3GPP] to ensure high-quality user 190 experience, it is still preferable for real-time applications to 191 behave in an adaptive manner. 193 Different mobile operators deploy their own cellular networks with 194 their own set of network functionalities and policies. Usually, a 195 mobile operator network includes a range of radio access technologies 196 such as 3G and 4G/LTE. Looking at the specifications of such radio 197 technologies it is evident that only the more recent radio 198 technologies can support the high bandwidth requirements from real- 199 time interactive video applications. The future real-time 200 interactive application will impose even greater demand on cellular 201 network performance which makes 4G (and beyond) radio technologies 202 more suitable for such genre of application. 204 The key factors in defining test cases for cellular networks are: 206 o Shared and varying link capacity 208 o Mobility 210 o Handover 212 However, these factors are typically highly correlated in a cellular 213 network. Therefore, instead of devising separate test cases for 214 individual important events, we have divided the test case into two 215 categories. It should be noted that the goal of the following test 216 cases is to evaluate the performance of candidate algorithms over the 217 radio interface of the cellular network. Hence it is assumed that 218 the radio interface is the bottleneck link between the communicating 219 peers and that the core network does not introduce any extra 220 congestion along the path. Consequently, this draft has kept as out 221 of scope the combination of multiple access technologies involving 222 both cellular and Wi-Fi users. In this latter case the shared 223 bottleneck is likely at the wired backhaul link. These test cases 224 further assume a typical real-time telephony scenario where one real- 225 time session consists of one voice stream and one video stream. 227 Even though it is possible to carry out tests over operational 228 cellular networks (e.g., LTE/5G), and actually such tests are already 229 available today, these tests cannot in general be carried out in a 230 deterministic fashion to ensure repeatability. The main reason is 231 that these networks are controlled by cellular operators and there 232 exist various amounts of competing traffic in the same cell(s). In 233 practice, it is only in underground mines that one can carry out near 234 deterministic testing. Even there, it is not guaranteed either as 235 workers in the mines may carry with them their personal mobile 236 phones. Furthermore, the underground mining setting may not reflect 237 typical usage patterns in an urban setting. We, therefore, recommend 238 that a cellular network simulator is used for the test cases defined 239 in this document, for example -- the LTE simulator in [NS-3]. 241 2.1. Varying Network Load 243 The goal of this test is to evaluate the performance of the candidate 244 congestion control algorithm under varying network load. The network 245 load variation is created by adding and removing network users a.k.a. 246 User Equipments (UEs) during the simulation. In this test case, each 247 user/UE in the media session is an endpoint following RTP-based 248 congestion control. User arrivals follow a Poisson distribution 249 proportional to the length of the call, to keep the number of users 250 per cell fairly constant during the evaluation period. At the 251 beginning of the simulation, there should be enough time to warm-up 252 the network. This is to avoid running the evaluation in an empty 253 network where network nodes are having empty buffers, low 254 interference at the beginning of the simulation. This network 255 initialization period should be excluded from the evaluation period. 256 Typically, the evaluation period starts 30 seconds after test 257 initialization. 259 This test case also includes user mobility and some competing 260 traffic. The latter includes both the same types of flows (with same 261 adaptation algorithms) and different types of flows (with different 262 services and congestion control schemes). The investigated 263 congestion control algorithms should show maximum possible network 264 utilization and stability in terms of rate variations, lowest 265 possible end to end frame latency, network latency and Packet Loss 266 Rate (PLR) at different cell load level. 268 2.1.1. Network Connection 270 Each mobile user is connected to a fixed user. The connection 271 between the mobile user and fixed user consists of a cellular radio 272 access, an Evolved Packet Core (EPC) and an Internet connection. The 273 mobile user is connected to the EPC using cellular radio access 274 technology which is further connected to the Internet. At the other 275 end, the fixed user is connected to the Internet via wired connection 276 with sufficiently high bandwidth, for instance, 10 Gbps, so that the 277 system bottleneck is on the cellular radio access interface. The 278 wired connection to in this setup does not introduce any network 279 impairments to the test; it only adds 10 ms of one-way propagation 280 delay. 282 The path from the fixed user to the mobile users is defined as 283 "Downlink" and the path from the mobile users to the fixed user is 284 defined as "Uplink". We assume that only uplink or downlink is 285 congested for mobile users. Hence, we recommend that the uplink and 286 downlink simulations are run separately. 288 uplink 289 ++))) +--------------------------> 290 ++-+ ((o)) 291 | | / \ +-------+ +------+ +---+ 292 +--+ / \----+ +-----+ +----+ | 293 / \ +-------+ +------+ +---+ 294 UE BS EPC Internet fixed 295 <--------------------------+ 296 downlink 298 Figure 1: Simulation Topology 300 2.1.2. Simulation Setup 302 The values enclosed within "[ ]" for the following simulation 303 attributes follow the same notion as in [I-D.ietf-rmcat-eval-test]. 304 The desired simulation setup is as follows -- 306 1. Radio environment: 308 A. Deployment and propagation model: 3GPP case 1 (see 309 [HO-deploy-3GPP]) 311 B. Antenna: Multiple-Input and Multiple-Output (MIMO), 2D or 3D 312 antenna pattern. 314 C. Mobility: [3km/h, 30km/h] 316 D. Transmission bandwidth: 10MHz 318 E. Number of cells: multi-cell deployment (3 Cells per Base 319 Station (BS) * 7 BS) = 21 cells 321 F. Cell radius: 166.666 Meters 323 G. Scheduler: Proportional fair with no priority 325 H. Bearer: Default bearer for all traffic. 327 I. Active Queue Management (AQM) settings: AQM [on,off] 329 2. End-to-end Round Trip Time (RTT): [40ms, 150ms] 331 3. User arrival model: Poisson arrival model 332 4. User intensity: 334 * Downlink user intensity: {0.7, 1.4, 2.1, 2.8, 3.5, 4.2, 4.9, 335 5.6, 6.3, 7.0, 7.7, 8.4, 9,1, 9.8, 10.5} 337 * Uplink user intensity : {0.7, 1.4, 2.1, 2.8, 3.5, 4.2, 4.9, 338 5.6, 6.3, 7.0} 340 5. Simulation duration: 91s 342 6. Evaluation period: 30s-60s 344 7. Media traffic: 346 1. Media type: Video 348 a. Media direction: [Uplink, Downlink] 350 b. Number of Media source per user: One (1) 352 c. Media duration per user: 30s 354 d. Media source: same as defined in Section 4.3 of 355 [I-D.ietf-rmcat-eval-test] 357 2. Media Type: Audio 359 a. Media direction: Uplink and Downlink 361 b. Number of Media source per user: One (1) 363 c. Media duration per user: 30s 365 d. Media codec: Constant Bit Rate (CBR) 367 e. Media bitrate: 20 Kbps 369 f. Adaptation: off 371 8. Other traffic models: 373 * Downlink simulation: Maximum of 4Mbps/cell (web browsing or 374 FTP traffic following default TCP congestion control 375 [RFC5681]) 377 * Unlink simulation: Maximum of 2Mbps/cell (web browsing or FTP 378 traffic following default TCP congestion control [RFC5681]) 380 2.2. Bad Radio Coverage 382 The goal of this test is to evaluate the performance of candidate 383 congestion control algorithm when users visit part of the network 384 with bad radio coverage. The scenario is created by using a larger 385 cell radius than that in the previous test case. In this test case, 386 each user/UE in the media session is an endpoint following RTP-based 387 congestion control. User arrivals follow a Poisson distribution 388 proportional to the length of the call, to keep the number of users 389 per cell fairly constant during the evaluation period. At the 390 beginning of the simulation, there should be enough amount of time to 391 warm-up the network. This is to avoid running the evaluation in an 392 empty network where network nodes are having empty buffers, low 393 interference at the beginning of the simulation. This network 394 initialization period should be excluded from the evaluation period. 395 Typically, the evaluation period starts 30 seconds after test 396 initialization. 398 This test case also includes user mobility and some competing 399 traffic. The latter includes the same kind of flows (with same 400 adaptation algorithms). The investigated congestion control 401 algorithms should result in maximum possible network utilization and 402 stability in terms of rate variations, lowest possible end to end 403 frame latency, network latency and Packet Loss Rate (PLR) at 404 different cell load levels. 406 2.2.1. Network connection 408 Same as defined in Section 2.1.1 410 2.2.2. Simulation Setup 412 The desired simulation setup is the same as the Varying Network Load 413 test case defined in Section 2.1 except the following changes: 415 1. Radio environment: Same as defined in Section 2.1.2 except the 416 following: 418 A. Deployment and propagation model: 3GPP case 3 (see 419 [HO-deploy-3GPP]) 421 B. Cell radius: 577.3333 Meters 423 C. Mobility: 3km/h 425 2. User intensity = {0.7, 1.4, 2.1, 2.8, 3.5, 4.2, 4.9, 5.6, 6.3, 426 7.0} 428 3. Media traffic model: Same as defined in Section 2.1.2 430 4. Other traffic models: 432 * Downlink simulation: Maximum of 2Mbps/cell (web browsing or 433 FTP traffic following default TCP congestion control 434 [RFC5681]) 436 * Unlink simulation: Maximum of 1Mbps/cell (web browsing or FTP 437 traffic following default TCP congestion control [RFC5681]) 439 2.3. Desired Evaluation Metrics for cellular test cases 441 The evaluation criteria document [I-D.ietf-rmcat-eval-criteria] 442 defines the metrics to be used to evaluate candidate algorithms. 443 Considering the nature and distinction of cellular networks we 444 recommend that at least the following metrics be used to evaluate the 445 performance of the candidate algorithms: 447 o Average cell throughput (for all cells), shows cell utilizations. 449 o Application sending and receiving bitrate, goodput. 451 o Packet Loss Rate (PLR). 453 o End-to-end Media frame delay. For video, this means the delay 454 from capture to display. 456 o Transport delay. 458 o Algorithm stability in terms of rate variation. 460 3. Wi-Fi Networks Specific Test Cases 462 Given the prevalence of Internet access links over Wi-Fi, it is 463 important to evaluate candidate RTP-based congestion control 464 solutions over test cases that include Wi-Fi access links. Such 465 evaluations should highlight the inherently different characteristics 466 of Wi-Fi networks in contrast to their wired counterparts: 468 o The wireless radio channel is subject to interference from nearby 469 transmitters, multipath fading, and shadowing. These effects lead 470 to fluctuations in the link throughput and sometimes an error- 471 prone communication environment. 473 o Available network bandwidth is not only shared over the air 474 between concurrent users but also between uplink and downlink 475 traffic due to the half-duplex nature of the wireless transmission 476 medium. 478 o Packet transmissions over Wi-Fi are susceptible to contentions and 479 collisions over the air. Consequently, traffic load beyond a 480 certain utilization level over a Wi-Fi network can introduce 481 frequent collisions over the air and significant network overhead, 482 as well as packet drops due to buffer overflow at the 483 transmitters. This, in turn, leads to excessive delay, 484 retransmissions, packet losses and lower effective bandwidth for 485 applications. Note further that the collision-induced delay and 486 loss patterns are qualitatively different from those caused by 487 congestion over a wired connection. 489 o The IEEE 802.11 standard (i.e., Wi-Fi) supports multi-rate 490 transmission capabilities by dynamically choosing the most 491 appropriate modulation and coding scheme (MCS) for the given 492 received signal strength. A different choice in the MCS Index 493 leads to different physical-layer (PHY-layer) link rates and 494 consequently different application-layer throughput. 496 o The presence of legacy devices (e.g., ones operating only in IEEE 497 802.11b) at a much lower PHY-layer link rate can significantly 498 slow down the rest of a modern Wi-Fi network. As discussed in 499 [Heusse2003], the main reason for such anomaly is that it takes 500 much longer to transmit the same packet over a slower link than 501 over a faster link, thereby consuming a substantial portion of air 502 time. 504 o Handover from one Wi-Fi Access Point (AP) to another may lead to 505 excessive packet delays and losses during the process. 507 o IEEE 802.11e has introduced the Enhanced Distributed Channel 508 Access (EDCA) mechanism to allow different traffic categories to 509 contend for channel access using different random back-off 510 parameters. This mechanism is a mandatory requirement for the Wi- 511 Fi Multimedia (WMM) certification in Wi-Fi Alliance. It allows 512 for prioritization of real-time application traffic such as voice 513 and video over non-urgent data transmissions (e.g., file 514 transfer). 516 In summary, the presence of Wi-Fi access links in different network 517 topologies can exert different impact on the network performance in 518 terms of application-layer effective throughput, packet loss rate, 519 and packet delivery delay. These, in turn, will influence the 520 behavior of end-to-end real-time multimedia congestion control. 522 Unless otherwise mentioned, the test cases in this section choose the 523 PHY- and MAC-layer parameters based on the IEEE 802.11n Standard. 524 Statistics collected from enterprise Wi-Fi networks show that the two 525 dominant physical modes are 802.11n and 802.11ac, accounting for 41% 526 and 58% of connected devices. As Wi-Fi standards evolve over time -- 527 for instance, with the introduction of the emerging Wi-Fi 6 (based on 528 IEEE 802.11ax) products -- the PHY- and MAC-layer test case 529 specifications need to be updated accordingly to reflect such 530 changes. 532 Typically, a Wi-Fi access network connects to a wired infrastructure. 533 Either the wired or the Wi-Fi segment of the network can be the 534 bottleneck. The following sections describe basic test cases for 535 both scenarios separately. The same set of performance metrics as in 536 [I-D.ietf-rmcat-eval-test]) should be collected for each test case. 538 We recommend to carry out the test cases as defined in this document 539 using a simulator, such as [NS-2] or [NS-3]. When feasible, it is 540 encouraged to perform testbed-based evaluations using Wi-Fi access 541 points and endpoints running up-to-date IEEE 802.11 protocols, such 542 as 802.11ac and the emerging Wi-Fi 6, so as to verify the viability 543 of the candidate schemes. 545 3.1. Bottleneck in Wired Network 547 The test scenarios below are intended to mimic the setup of video 548 conferencing over Wi-Fi connections from the home. Typically, the 549 Wi-Fi home network is not congested and the bottleneck is present 550 over the wired home access link. Although it is expected that test 551 evaluation results from this section are similar to those as in 552 [I-D.ietf-rmcat-eval-test], it is still worthwhile to run through 553 these tests as sanity checks. 555 3.1.1. Network topology 557 Figure 2 shows the network topology of Wi-Fi test cases. The test 558 contains multiple mobile nodes (MNs) connected to a common Wi-Fi 559 access point (AP) and their corresponding wired clients on fixed 560 nodes (FNs). Each connection carries either a RTP-based media flow 561 or a TCP traffic flow. Directions of the flows can be uplink (i.e., 562 from mobile nodes to fixed nodes), downlink (i.e., from fixed nodes 563 to mobile nodes), or bi-directional. The total number of 564 uplink/downlink/bi-directional flows for RTP-based media traffic and 565 TCP traffic are denoted as N and M, respectively. 567 Uplink 568 +----------------->+ 569 +------+ +------+ 570 | MN_1 |)))) /=====| FN_1 | 571 +------+ )) // +------+ 572 . )) // . 573 . )) // . 574 . )) // . 575 +------+ +----+ +-----+ +------+ 576 | MN_N | ))))))) | | | |========| FN_N | 577 +------+ | | | | +------+ 578 | AP |=========| FN0 | 579 +----------+ | | | | +----------+ 580 | MN_tcp_1 | )))) | | | |======| FN_tcp_1 | 581 +----------+ +----+ +-----+ +----------+ 582 . )) \\ . 583 . )) \\ . 584 . )) \\ . 585 +----------+ )) \\ +----------+ 586 | MN_tcp_M |))) \=====| FN_tcp_M | 587 +----------+ +----------+ 588 +<-----------------+ 589 Downlink 591 Figure 2: Network topology for Wi-Fi test cases 593 3.1.2. Test setup 595 o Test duration: 120s 597 o Wi-Fi network characteristics: 599 * Radio propagation model: Log-distance path loss propagation 600 model (see [NS3WiFi]) 602 * PHY- and MAC-layer configuration: IEEE 802.11n 604 * MCS Index at 11: 16-QAM 1/2, Raw Data Rate at 52Mbps 606 o Wired path characteristics: 608 * Path capacity: 1Mbps 610 * One-Way propagation delay: 50ms. 612 * Maximum end-to-end jitter: 30ms 614 * Bottleneck queue type: Drop tail. 616 * Bottleneck queue size: 300ms. 618 * Path loss ratio: 0%. 620 o Application characteristics: 622 * Media Traffic: 624 + Media type: Video 626 + Media direction: See Section 3.1.3 628 + Number of media sources (N): See Section 3.1.3 630 + Media timeline: 632 - Start time: 0s. 634 - End time: 119s. 636 * Competing traffic: 638 + Type of sources: long-lived TCP or CBR over UDP 640 + Traffic direction: See Section 3.1.3 642 + Number of sources (M): See Section 3.1.3 644 + Congestion control: Default TCP congestion control [RFC5681] 645 or constant-bit-rate (CBR) traffic over UDP. 647 + Traffic timeline: See Section 3.1.3 649 3.1.3. Typical test scenarios 651 o Single uplink RTP-based media flow: N=1 with uplink direction and 652 M=0. 654 o One pair of bi-directional RTP-based media flows: N=2 (i.e., one 655 uplink flow and one downlink flow); M=0. 657 o One pair of bi-directional RTP-based media flows: N=2; one uplink 658 on-off CBR flow over UDP: M=1 (uplink). The CBR flow has ON time 659 at t=0s-60s and OFF time at t=60s-119s. 661 o One pair of bi-directional RTP-based media flows: N=2; one uplink 662 off-on CBR flow over UDP: M=1 (uplink). The CBR flow has OFF time 663 at t=0s-60s and ON time at t=60s-119s. 665 o One RTP-based media flow competing against one long-live TCP flow 666 in the uplink direction: N=1 (uplink) and M = 1(uplink). The TCP 667 flow has start time at t=0s and end time at t=119s. 669 3.1.4. Expected behavior 671 o Single uplink RTP-based media flow: the candidate algorithm is 672 expected to detect the path capacity constraint, to converge to 673 the bottleneck link capacity, and to adapt the flow to avoid 674 unwanted oscillations when the sending bit rate is approaching the 675 bottleneck link capacity. No excessive oscillations in the media 676 rate should be present. 678 o Bi-directional RTP-based media flows: the candidate algorithm is 679 expected to converge to the bottleneck capacity of the wired path 680 in both directions despite the presence of measurement noise over 681 the Wi-Fi connection. In the presence of background TCP or CBR 682 over UDP traffic, the rate of RTP-based media flows should adapt 683 promptly to the arrival and departure of background traffic flows. 685 o One RTP-based media flow competing with long-live TCP flow in the 686 uplink direction: the candidate algorithm is expected to avoid 687 congestion collapse and to stabilize at a fair share of the 688 bottleneck link capacity. 690 3.2. Bottleneck in Wi-Fi Network 692 The test cases in this section assume that the wired segment along 693 the media path is well-provisioned whereas the bottleneck exists over 694 the Wi-Fi access network. This is to mimic the application scenarios 695 typically encountered by users in an enterprise environment or at a 696 coffee house. 698 3.2.1. Network topology 700 Same as defined in Section 3.1.1 702 3.2.2. Test setup 704 o Test duration: 120s 706 o Wi-Fi network characteristics: 708 * Radio propagation model: Log-distance path loss propagation 709 model (see [NS3WiFi]) 711 * PHY- and MAC-layer configuration: IEEE 802.11n 712 * MCS Index at 11: 16-QAM 1/2, Raw Data Rate at 52Mbps 714 o Wired path characteristics: 716 * Path capacity: 100Mbps. 718 * One-Way propagation delay: 50ms. 720 * Maximum end-to-end jitter: 30ms. 722 * Bottleneck queue type: Drop tail. 724 * Bottleneck queue size: 300ms. 726 * Path loss ratio: 0%. 728 o Application characteristics: 730 * Media Traffic: 732 + Media type: Video 734 + Media direction: See Section 3.2.3. 736 + Number of media sources (N): See Section 3.2.3. 738 + Media timeline: 740 - Start time: 0s. 742 - End time: 119s. 744 * Competing traffic: 746 + Type of sources: long-lived TCP or CBR over UDP. 748 + Number of sources (M): See Section 3.2.3. 750 + Traffic direction: See Section 3.2.3. 752 + Congestion control: Default TCP congestion control [RFC5681] 753 or constant-bit-rate (CBR) traffic over UDP. 755 + Traffic timeline: See Section 3.2.3. 757 3.2.3. Typical test scenarios 759 This section describes a few test scenarios that are deemed as 760 important for understanding the behavior of a candidate RTP-based 761 congestion control scheme over a Wi-Fi network. 763 a. Multiple RTP-based media flows sharing the wireless downlink: 764 N=16 (all downlink); M = 0. This test case is for studying the 765 impact of contention on the multiple concurrent media flows. For 766 an 802.11n network, given the MCS Index of 11 and the 767 corresponding link rate of 52Mbps, the total application-layer 768 throughput (assuming reasonable distance, low interference and 769 infrequent contentions caused by competing streams) is around 770 20Mbps. A total of N=16 RTP-based media flows (with a maximum 771 rate of 1.5Mbps each) are expected to saturate the wireless 772 interface in this experiment. Evaluation of a given candidate 773 scheme should focus on whether the downlink media flows can 774 stabilize at a fair share of the total application-layer 775 throughput. 777 b. Multiple RTP-based media flows sharing the wireless uplink: N = 778 16 (all uplink); M = 0. When multiple clients attempt to 779 transmit media packets uplink over the Wi-Fi network, they 780 introduce more frequent contentions and potential collisions. 781 Per-flow throughput is expected to be lower than that in the 782 previous downlink-only scenario. Evaluation of a given candidate 783 scheme should focus on whether the uplink flows can stabilize at 784 a fair share of the total application-layer throughput. 786 c. Multiple bi-directional RTP-based media flows: N = 16 (8 uplink 787 and 8 downlink); M = 0. The goal of this test is to evaluate the 788 performance of the candidate scheme in terms of bandwidth 789 fairness between uplink and downlink flows. 791 d. Multiple bi-directional RTP-based media flows with on-off CBR 792 traffic over UDP: N = 16 (8 uplink and 8 downlink); M = 5 793 (uplink). The goal of this test is to evaluate the adaptation 794 behavior of the candidate scheme when its available bandwidth 795 changes due to the departure of background traffic. The 796 background traffic consists of several (e.g., M=5) CBR flows 797 transported over UDP. These background flows are ON at time 798 t=0-60s and OFF at time t=61-120s. 800 e. Multiple bi-directional RTP-based media flows with off-on CBR 801 traffic over UDP: N = 16 (8 uplink and 8 downlink); M = 5 802 (uplink). The goal of this test is to evaluate the adaptation 803 behavior of the candidate scheme when its available bandwidth 804 changes due to the arrival of background traffic. The background 805 traffic consists of several (e.g., M=5) parallel CBR flows 806 transported over UDP. These background flows are OFF at time 807 t=0-60s and ON at times t=61-120s. 809 f. Multiple bi-directional RTP-based media flows in the presence of 810 background TCP traffic: N=16 (8 uplink and 8 downlink); M = 5 811 (uplink). The goal of this test is to evaluate how RTP-based 812 media flows compete against TCP over a congested Wi-Fi network 813 for a given candidate scheme. TCP flows have start time at t=40s 814 and end time at t=80s. 816 g. Varying number of RTP-based media flows: A series of tests can be 817 carried out for the above test cases with different values of N, 818 e.g., N = [4, 8, 12, 16, 20]. The goal of this test is to 819 evaluate how a candidate scheme responds to varying traffic load/ 820 demand over a congested Wi-Fi network. The start times of the 821 media flows are randomly distributes within a window of t=0-10s; 822 their end times are randomly distributed within a window of 823 t=110-120s. 825 3.2.4. Expected behavior 827 o Multiple downlink RTP-based media flows: each media flow is 828 expected to get its fair share of the total bottleneck link 829 bandwidth. Overall bandwidth usage should not be significantly 830 lower than that experienced by the same number of concurrent 831 downlink TCP flows. In other words, the behavior of multiple 832 concurrent TCP flows will be used as a performance benchmark for 833 this test scenario. The end-to-end delay and packet loss ratio 834 experienced by each flow should be within an acceptable range for 835 real-time multimedia applications. 837 o Multiple uplink RTP-based media flows: overall bandwidth usage by 838 all media flows should not be significantly lower than that 839 experienced by the same number of concurrent uplink TCP flows. In 840 other words, the behavior of multiple concurrent TCP flows will be 841 used as a performance benchmark for this test scenario. 843 o Multiple bi-directional RTP-based media flows with dynamic 844 background traffic carrying CBR flows over UDP: the media flows 845 are expected to adapt in a timely fashion to the changes in 846 available bandwidth introduced by the arrival/departure of 847 background traffic. 849 o Multiple bi-directional RTP-based media flows with dynamic 850 background traffic over TCP: during the presence of TCP background 851 flows, the overall bandwidth usage by all media flows should not 852 be significantly lower than those achieved by the same number of 853 bi-directional TCP flows. In other words, the behavior of 854 multiple concurrent TCP flows will be used as a performance 855 benchmark for this test scenario. All downlink media flows are 856 expected to obtain similar bandwidth as each other. The 857 throughput of each media flow is expected to decrease upon the 858 arrival of TCP background traffic and, conversely, increase upon 859 their departure. Both reactions should occur in a timely fashion, 860 for example, within 10s of seconds. 862 o Varying number of bi-directional RTP-based media flows: the test 863 results for varying values of N -- while keeping all other 864 parameters constant -- is expected to show steady and stable per- 865 flow throughput for each value of N. The average throughput of 866 all media flows is expected to stay constant around the maximum 867 rate when N is small, then gradually decrease with increasing 868 value of N till it reaches the minimum allowed rate, beyond which 869 the offered load to the Wi-Fi network exceeds its capacity (i.e., 870 with a very large value of N). 872 3.3. Other Potential Test Cases 874 3.3.1. EDCA/WMM usage 876 The EDCA/WMM mechanism defines prioritized QoS for four traffic 877 classes (or Access Categories). RTP-based real-time media flows 878 should achieve better performance in terms of lower delay and fewer 879 packet losses with EDCA/WMM enabled when competing against non- 880 interactive background traffic such as file transfers. When most of 881 the traffic over Wi-Fi is dominated by media, however, turning on WMM 882 may degrade performance since all media flows now attempt to access 883 the wireless transmission medium more aggressively, thereby causing 884 more frequent collisions and collision-induced losses. This is a 885 topic worthy of further investigation. 887 3.3.2. Effect of heterogeneous link rates 889 As discussed in [Heusse2003], the presence of clients operating over 890 slow PHY-layer link rates (e.g., a legacy 802.11b device) connected 891 to a modern network may adversely impact the overall performance of 892 the network. Additional test cases can be devised to evaluate the 893 effect of clients with heterogeneous link rates on the performance of 894 the candidate congestion control algorithm. Such test cases, for 895 instance, can specify that the PHY-layer link rates for all clients 896 span over a wide range (e.g., 2Mbps to 54Mbps) for investigating its 897 effect on the congestion control behavior of the real-time 898 interactive applications. 900 4. IANA Considerations 902 This memo includes no request to IANA. 904 5. Security Considerations 906 The security considerations in [I-D.ietf-rmcat-eval-criteria] and the 907 relevant congestion control algorithms apply. The principles for 908 congestion control are described in [RFC2914], and in particular, any 909 new method must implement safeguards to avoid congestion collapse of 910 the Internet. 912 Given the difficulty of deterministic wireless testing, it is 913 recommended and expected that the tests described in this document 914 would be done via simulations. However, in the case where these test 915 cases are carried out in a testbed setting, the evaluation should 916 take place in a controlled lab environment. In the testbed, the 917 applications, simulators and network nodes ought to be well-behaved 918 and should not impact the desired results. It is important to take 919 appropriate caution to avoid leaking non-responsive traffic with 920 unproven congestion avoidance behavior onto the open Internet. 922 6. Acknowledgments 924 The authors would like to thank Ingemar Johansson for contributing to 925 the cellular test cases during the earlier stage of this draft. 927 The authors would like to thank Tomas Frankkila, Magnus Westerlund, 928 Kristofer Sandlund, Sergio Mena de la Cruz, and Mirja Kuehlewind for 929 their valuable inputs and review comments regarding this draft. 931 7. References 933 7.1. Normative References 935 [HO-deploy-3GPP] 936 TS 25.814, 3GPP., "Physical layer aspects for evolved 937 Universal Terrestrial Radio Access (UTRA)", October 2006, 938 . 941 [I-D.ietf-rmcat-eval-criteria] 942 Singh, V., Ott, J., and S. Holmer, "Evaluating Congestion 943 Control for Interactive Real-time Media", draft-ietf- 944 rmcat-eval-criteria-11 (work in progress), February 2020. 946 [I-D.ietf-rmcat-eval-test] 947 Sarker, Z., Singh, V., Zhu, X., and M. Ramalho, "Test 948 Cases for Evaluating RMCAT Proposals", draft-ietf-rmcat- 949 eval-test-10 (work in progress), May 2019. 951 [IEEE802.11] 952 IEEE, "Standard for Information technology-- 953 Telecommunications and information exchange between 954 systems Local and metropolitan area networks--Specific 955 requirements Part 11: Wireless LAN Medium Access Control 956 (MAC) and Physical Layer (PHY) Specifications", 2012. 958 [NS3WiFi] "Wi-Fi Channel Model in ns-3 Simulator", 959 . 962 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 963 Control", RFC 5681, DOI 10.17487/RFC5681, September 2009, 964 . 966 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 967 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 968 May 2017, . 970 7.2. Informative References 972 [Heusse2003] 973 Heusse, M., Rousseau, F., Berger-Sabbatel, G., and A. 974 Duda, "Performance anomaly of 802.11b", in Proc. 23th 975 Annual Joint Conference of the IEEE Computer and 976 Communications Societies, (INFOCOM'03), March 2003. 978 [HO-def-3GPP] 979 TR 21.905, 3GPP., "Vocabulary for 3GPP Specifications", 980 December 2009, . 983 [HO-LTE-3GPP] 984 TS 36.331, 3GPP., "E-UTRA- Radio Resource Control (RRC); 985 Protocol specification", December 2011, 986 . 989 [HO-UMTS-3GPP] 990 TS 25.331, 3GPP., "Radio Resource Control (RRC); Protocol 991 specification", December 2011, 992 . 995 [NS-2] "ns-2", December 2014, 996 . 998 [NS-3] "ns-3 Network Simulator", . 1000 [QoS-3GPP] 1001 TS 23.203, 3GPP., "Policy and charging control 1002 architecture", June 2011, . 1005 [RFC2914] Floyd, S., "Congestion Control Principles", BCP 41, 1006 RFC 2914, DOI 10.17487/RFC2914, September 2000, 1007 . 1009 Authors' Addresses 1011 Zaheduzzaman Sarker 1012 Ericsson AB 1013 Laboratoriegraend 11 1014 Luleae 97753 1015 Sweden 1017 Phone: +46 107173743 1018 Email: zaheduzzaman.sarker@ericsson.com 1020 Xiaoqing Zhu 1021 Cisco Systems 1022 12515 Research Blvd., Building 4 1023 Austin, TX 78759 1024 USA 1026 Email: xiaoqzhu@cisco.com 1028 Jiantao Fu 1029 Cisco Systems 1030 771 Alder Drive 1031 Milpitas, CA 95035 1032 USA 1034 Email: jianfu@cisco.com 1035 Wei-Tian Tan 1036 Cisco Systems 1037 510 McCarthy Blvd 1038 Milpitas, CA 95035 1039 USA 1041 Email: dtan2@cisco.com 1043 Michael A. Ramalho 1044 AcousticComms Consulting 1045 6310 Watercrest Way Unit 203 1046 Lakewood Ranch, FL 34202-5211 1047 USA 1049 Phone: +1 732 832 9723 1050 Email: mar42@cornell.edu