RFC 8869

Evaluation Test Cases for Interactive Real-Time Media over Wireless Networks

📅 published January 2021, from ietf

Okay bestie, let's get into RFC 8869 💖. Its title is Evaluation Test Cases for Interactive Real-Time Media over Wireless Networks 🎀. It's Informational, which means it's guidance and not a Standards Track spec 🌸. It was published in January 2021 with ISSN 2070-1721. The authors are Z. Sarker from Ericsson AB, X. Zhu, and J. Fu from Cisco Systems ✨. There's no "Obsoletes" or "Updates" line in the header, so this doc doesn't replace or change any other RFC 🦋.

Abstract ✨

Okay so RTP (the Real-time Transport Protocol) is the go-to transport for interactive multimedia apps 💅. Those apps depend a lot on congestion control working well. For a seamless, robust experience 🫶, an RTP-based congestion control algorithm needs to work well on every kind of access network. This document describes test cases for checking how candidate congestion control algorithms perform over cellular and Wi-Fi networks 🔥.

Status of This Memo 📜

This document is not an Internet Standards Track specification, so it's published for informational purposes 🌷.

It's a product of the Internet Engineering Task Force (IETF) 💜. It represents the consensus of the IETF community, got public review, and was approved for publication by the Internet Engineering Steering Group (IESG). Not every IESG-approved doc is a candidate for any level of Internet Standard, so see Section 2 of RFC 7841 for the details 🙈.

Info about this doc's current status, any errata, and how to give feedback is at https://www.rfc-editor.org/info/rfc8869 🦋.

Copyright (c) 2021 IETF Trust and the persons identified as the document authors. All rights reserved ✨.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) as in effect on the publication date. Please read these carefully, because they explain your rights and restrictions for this document 📖. Code Components extracted from this document must include Simplified BSD License text, as described in Section 4.e of the Trust Legal Provisions, and they come without warranty, as described in the Simplified BSD License 💅.

Table of Contents 🗂️

   1.  Introduction
   2.  Cellular Network Specific Test Cases
     2.1.  Varying Network Load
       2.1.1.  Network Connection
       2.1.2.  Simulation Setup
       2.1.3.  Expected Behavior
     2.2.  Bad Radio Coverage
       2.2.1.  Network Connection
       2.2.2.  Simulation Setup
       2.2.3.  Expected Behavior
     2.3.  Desired Evaluation Metrics for Cellular Test Cases
   3.  Wi-Fi Networks Specific Test Cases
     3.1.  Bottleneck in Wired Network
       3.1.1.  Network Topology
       3.1.2.  Test/Simulation Setup
       3.1.3.  Typical Test Scenarios
       3.1.4.  Expected Behavior
     3.2.  Bottleneck in Wi-Fi Network
       3.2.1.  Network Topology
       3.2.2.  Test/Simulation Setup
       3.2.3.  Typical Test Scenarios
       3.2.4.  Expected Behavior
     3.3.  Other Potential Test Cases
       3.3.1.  EDCA/WMM usage
       3.3.2.  Effect of Heterogeneous Link Rates
   4.  IANA Considerations
   5.  Security Considerations
   6.  References
     6.1.  Normative References
     6.2.  Informative References
   Contributors
   Acknowledgments
   Authors' Addresses

1. Introduction ✨

Wireless networks, both cellular and Wi-Fi [IEEE802.11], are a huge and growing part of the Internet 🌍. Typical scenarios for interactive multimedia over wireless include video conferencing calls on a bus or train, and live media streaming at home 🚌. It's well known that supporting multimedia over wireless is super different from doing the same thing over a wired network 💀. The basic test cases in [RFC8867] already cover lots of common network impairments for evaluating RTP-based congestion control schemes. Still, they need testing against the characteristics and dynamics unique to each wireless environment 🦋. For example, in cellular networks the base station keeps individual queues per radio bearer per user. That means traffic flows from different users interact in a different way. On a typical wired network, though, traffic from every user shares the same queue at the bottleneck 🧚. User mobility patterns also differ between cellular and Wi-Fi networks 🌸. So it's important to evaluate proposed candidate RTP-based congestion control solutions over cellular mobile networks and over Wi-Fi networks, separately 👑.

[RFC8868] gives guidelines for evaluating candidate algorithms and recognizes how important it is to test over wireless access networks 💜. However, it doesn't describe any specific test cases for evaluating candidate algorithms' performance. This document does, and the test cases target cellular and Wi-Fi networks specifically 🎀.

2. Cellular Network Specific Test Cases 📡

A cellular environment is way more complicated than its wireline counterpart 😭. It's trying to provide services with variable available bandwidth, location dependencies, and user mobility at different speeds. When a user reaches the cell edge, it can lead to a lot of retransmissions to deliver data from the base station to the destination and back 🔥. These radio links often act as the bottleneck for the rest of the network, and they can eventually cause excessive delays or packet drops 💅. An efficient retransmission or link adaptation mechanism can reduce the packet loss probability, but some packet losses and delay variations will still stick around 🫠. Plus, when cell load goes up or a user hands over to a congested cell, congestion in the transport network gets even worse 🍓. There are also certain characteristics that set cellular apart from other wireless access networks like Wi-Fi. In a cellular network:

  • The bottleneck is often a shared link with relatively few users.
    • The cost per bit over the shared link varies over time and is different for different users.
    • Leftover or unused resources can be consumed by other greedy users.
  • Queues are always per radio bearer, hence each user can have many such queues.
  • Users can experience both inter- and intra-Radio Access Technology (RAT) handovers (see [HO-def-3GPP] for the definition of "handover").
  • Handover between cells or change of serving cells (as described in [HO-LTE-3GPP] and [HO-UMTS-3GPP]) might cause user plane interruptions, which can lead to bursts of packet losses, delay, and/or jitter. The exact behavior depends on the type of radio bearer. Typically, the default best-effort bearers do not generate packet loss; instead, packets are queued up and transmitted once the handover is completed 🥲.
  • The network part decides how much the user can transmit 🙅.
  • The cellular network has variable link capacity per user 📶.
    • It can vary as fast as a period of milliseconds ⚡.
    • It depends on many factors, such as distance, speed, interference, and different flows 🌪️.
    • It uses complex and smart link adaptation, which makes the link behavior ever more dynamic 🤯.
    • The scheduling priority depends on the estimated throughput.
  • Both Quality of Service (QoS) and non-QoS radio bearers can be used 🎀.

So a real-time communication app running over a cellular network has to cope with a shared bottleneck link and variable link capacity 🫶. It also has to deal with events like handover, loss that isn't caused by congestion, and sudden bandwidth changes, both short term and long term. Those changes come from handover, network load, and bad radio coverage 😵‍💫. 3GPP has defined QoS bearers [QoS-3GPP] to ensure a high-quality user experience, but real-time apps still benefit a lot from adapting on their own 🌷.

Different mobile operators deploy their own cellular networks, each with their own set of network functionalities and policies 🏗️. A mobile operator network usually includes a range of radio access technologies, such as 3G and 4G/LTE. Looking at the specs for these radio technologies, it's obvious that only the more recent ones can support the high bandwidth needs of real-time interactive video apps 💖. Future real-time interactive apps will put even more demand on cellular network performance, which makes 4G (and beyond) radio technologies a better fit for this kind of app 🚀.

The key factors for defining test cases for cellular networks are:

  • Shared and varying link capacity
  • Mobility
  • Handover

But these factors are usually highly correlated in a cellular network 🧩. So instead of making separate test cases for each important event, we split the test cases into two categories 🍰. It's worth noting that the goal of the following test cases is to evaluate how candidate algorithms perform over the radio interface of the cellular network 📱. So it's assumed that the radio interface is the bottleneck link between the communicating peers, and that the core network doesn't add any extra congestion along the path 🛣️. That's also why this document leaves out the combination of multiple access technologies involving both cellular and Wi-Fi users. In that case, the shared bottleneck is likely to be the wired backhaul link 🔌. These test cases also assume a typical real-time telephony scenario, where one real-time session consists of one voice stream and one video stream 📞.

Even though it's possible to run tests over operational cellular networks (for example, LTE/5G), and such tests are already available today, they can't generally be run in a deterministic way to ensure repeatability 😤. The main reason is that these networks are controlled by cellular operators, and there's various amounts of competing traffic in the same cell or cells 🌐. In practice, the only place you can get close to deterministic testing is underground mines ⛏️. Even there it isn't guaranteed, because workers in the mines may carry their personal mobile phones 📱. Also, the underground mining setting may not reflect typical usage patterns in an urban setting 🏙️. So, we recommend using a cellular network simulator for the test cases defined in this document, for example -- the LTE simulator in [NS-3] 🧚‍♀️💕

2.1. Varying Network Load ✨

okay so bestie, this first test is all about seeing how a candidate congestion control algorithm holds up when the network load keeps changing 🦋 the way it works is that users get added and removed during the simulation. Each one is a User Equipment (UE) 📱, and each user in the media session is an endpoint running RTP-based congestion control. The arrivals follow a Poisson distribution that's proportional to the length of the call, so the number of users per cell stays fairly steady during the evaluation period 💅

Before anything is measured, the network needs a warm-up 🔥 if you start with an empty network, the nodes have empty buffers and really low interference, which is not a realistic picture at all. So this initialization period gets excluded from the evaluation, and the evaluation typically starts 30 seconds after test initialization 🧚

This test also throws in user mobility and some competing traffic 🌈 the competition includes flows of the same types (with the same adaptation algorithms) and flows of different types (with different services and congestion control schemes). Very messy, very realistic, very cute 🎀

2.1.1. Network Connection 🌸

so each mobile user is connected to a fixed user 🫶 the path between them goes through a cellular radio access network, an Evolved Packet Core (EPC), and then the Internet. The mobile user reaches the EPC over cellular radio access technology, and that connects onward to the Internet. On the other side, the fixed user hangs off the Internet via a wired connection with plenty of bandwidth, for example 10 Gbps, so the bottleneck is the cellular radio access interface and not the wire 💖 That wired link adds no network impairments to the test. It only adds 10 ms of one-way propagation delay ⭐

The direction from the fixed user toward the mobile users is called "downlink", and the direction from the mobile users back to the fixed user is called "uplink" 💜 We assume only uplink or downlink is congested for mobile users, so we recommend running the uplink and downlink simulations separately 🙈

                         uplink
        ++)))        +-------------------------->
        ++-+      ((o))
        |  |       / \     +-------+     +------+    +---+
        +--+      /   \----+       +-----+      +----+   |
                 /     \   +-------+     +------+    +---+
         UE         BS        EPC        Internet    fixed
                     <--------------------------+
                              downlink

                   Figure 1: Simulation Topology

2.1.2. Simulation Setup 🎀

the bracketed values "[ ]" in the simulation attributes follow the same notion as in [RFC8867] ✨ here's the setup we want:

Radio environment 📡

  • Deployment and propagation model: 3GPP case 1 (see [HO-deploy-3GPP])
  • Antenna: Multiple-Input and Multiple-Output (MIMO), 2D or 3D antenna pattern
  • Mobility: [3 km/h, 30 km/h]
  • Transmission bandwidth: 10 MHz
  • Number of cells: multi-cell deployment (3 cells per Base Station (BS) * 7 BS) = 21 cells
  • Cell radius: 166.666 meters
  • Scheduler: Proportional fair with no priority
  • Bearer: Default bearer for all traffic
  • Active Queue Management (AQM) settings: AQM [on, off]

End-to-end Round Trip Time (RTT): [40 ms, 150 ms] 💅

User arrival model: Poisson arrival model 🦋

User intensity 🌈

  • Downlink user intensity: {0.7, 1.4, 2.1, 2.8, 3.5, 4.2, 4.9, 5.6, 6.3, 7.0, 7.7, 8.4, 9,1, 9.8, 10.5}
  • Uplink user intensity: {0.7, 1.4, 2.1, 2.8, 3.5, 4.2, 4.9, 5.6, 6.3, 7.0}

Simulation duration: 91 s ⏱️

Evaluation period: 30 s - 60 s 🧚

Media traffic 🎬

  • Media type: Video
    • Media direction: [uplink, downlink]
    • Number of media sources per user: One (1)
    • Media duration per user: 30 s
    • Media source: same as defined in Section 4.3 of [RFC8867]
  • Media type: Audio
    • Media direction: [uplink, downlink]
    • Number of media sources per user: One (1)
    • Media duration per user: 30 s
    • Media codec: Constant Bit Rate (CBR)
    • Media bitrate: 20 Kbps
    • Adaptation: off

Other traffic models 🍓

  • Downlink simulation: Maximum of 4 Mbps/cell (web browsing or FTP traffic following default TCP congestion control [RFC5681])
  • Uplink simulation: Maximum of 2 Mbps/cell (web browsing or FTP traffic following default TCP congestion control [RFC5681])

2.1.3. Expected Behavior 👑

the algorithms being investigated should get maximum possible network utilization and stability in terms of rate variations 🔥 they should also get the lowest possible end-to-end frame latency, network latency, and Packet Loss Rate (PLR) across different cell load levels. Goals, ngl 😭

2.2. Bad Radio Coverage 📶

this one checks how the candidate algorithm does when users wander into a part of the network with bad radio coverage 🙈 the scenario gets built with a bigger cell radius than the last test. Each user/UE is again an endpoint following RTP-based congestion control 💜 arrivals are Poisson and proportional to call length, so the users per cell stay fairly constant during evaluation 🦋 the same warm-up rule applies: enough time to fill the network, with that initialization period excluded from evaluation, which typically starts 30 seconds after test initialization 🌸

This test also includes user mobility and some competing traffic, and the competition is made of the same kind of flows (with same adaptation algorithms) ✨

2.2.1. Network Connection 🎀

Same as defined in Section 2.1.1. 🫶

2.2.2. Simulation Setup 💖

the desired setup is the same as the Varying Network Load test case in Section 2.1, with the following changes 🌈

Radio environment: Same as defined in Section 2.1.2 except for the following 💅

  • Deployment and propagation model: 3GPP case 3 (see [HO-deploy-3GPP])
  • Cell radius: 577.3333 meters
  • Mobility: 3 km/h

User intensity: {0.7, 1.4, 2.1, 2.8, 3.5, 4.2, 4.9, 5.6, 6.3, 7.0} 🦋

Media traffic model: Same as defined in Section 2.1.2 🎬

Other traffic models 🍓

  • Downlink simulation: Maximum of 2 Mbps/cell (web browsing or FTP traffic following default TCP congestion control [RFC5681])
  • Uplink simulation: Maximum of 1 Mbps/cell (web browsing or FTP traffic following default TCP congestion control [RFC5681])

2.2.3. Expected Behavior ⭐

same vibe as before: the investigated algorithms should get maximum possible network utilization and stability in terms of rate variations 🔥 plus the lowest possible end-to-end frame latency, network latency, and Packet Loss Rate (PLR) at different cell load levels 😭

2.3. Desired Evaluation Metrics for Cellular Test Cases 📊

the evaluation criteria document [RFC8868] is where the metrics for judging candidate algorithms are defined 📖 because cellular networks are their own special kind of chaotic, we recommend that at least the following metrics get used to evaluate the candidates 🧚

  • Average cell throughput (for all cells), which shows cell utilization. 🍓
  • Application sending and receiving bitrate, goodput. 💜
  • Packet Loss Rate (PLR). 🙈
  • End-to-end media frame delay. For video, this means the delay from capture to display. 🎀
  • Transport delay. 🌸
  • Algorithm stability in terms of rate variation. 👑

3. Wi-Fi Networks Specific Test Cases ✨

okay bestie, let's talk about Wi-Fi 📶🦋 Given how many Internet access links are over Wi-Fi these days, it is absolutely essential that we test RTP-based congestion control candidates over test cases that include Wi-Fi access links 💅 These evaluations should highlight how Wi-Fi is inherently different from its wired counterparts, and it's giving main character energy:

  • The wireless radio channel is subject to interference from nearby transmitters, multipath fading, and shadowing 🌧️ These effects cause the link throughput to fluctuate and sometimes make the communication environment error-prone 😭
  • Available bandwidth is not only shared over the air between concurrent users but also between uplink and downlink traffic, because the wireless medium is half-duplex 🔄
  • Packet transmissions over Wi-Fi are susceptible to contention and collisions over the air 💥 So traffic load beyond a certain utilization level can cause frequent collisions, significant network overhead, and packet drops from buffer overflow at the transmitters. That leads to excessive delay, retransmissions, packet losses, and lower effective bandwidth for apps. Also, the collision-induced delay and loss patterns are qualitatively different from congestion over a wired connection 🙈
  • The IEEE 802.11 standard (aka Wi-Fi) supports multi-rate transmission by dynamically choosing the most appropriate modulation and coding scheme (MCS) for the received signal strength 📡 A different MCS Index leads to different physical-layer (PHY-layer) link rates and therefore different application-layer throughput 🌈
  • Legacy devices operating only in IEEE 802.11b, at a much lower PHY-layer link rate, can significantly slow down the rest of a modern Wi-Fi network 🐢 As discussed in [Heusse2003], the main reason for this anomaly is that it takes much longer to transmit the same packet over a slower link than over a faster one, so it eats up a substantial portion of air time 🍓
  • Handover from one Wi-Fi Access Point (AP) to another may lead to excessive packet delays and losses during the process 🧚
  • IEEE 802.11e introduced the Enhanced Distributed Channel Access (EDCA) mechanism, which lets different traffic categories contend for channel access using different random back-off parameters 🎀 This mechanism is a mandatory requirement for Wi-Fi Multimedia (WMM) certification in the Wi-Fi Alliance 👑 It lets real-time traffic like voice and video be prioritized over non-urgent data transmissions (e.g., file transfer) 💜

In summary, Wi-Fi access links in different network topologies can have different impacts on network performance in terms of application-layer effective throughput, packet loss rate, and packet delivery delay 📊 And those, in turn, will influence how end-to-end real-time multimedia congestion control behaves 🫶

Unless otherwise mentioned, the test cases in this section choose the PHY- and MAC-layer parameters based on the IEEE 802.11n standard 📘 Stats from enterprise Wi-Fi networks show the two dominant physical modes are 802.11n and 802.11ac, accounting for 41% and 58% of connected devices, respectively 💖 As Wi-Fi standards evolve (for instance, with the emerging Wi-Fi 6 (based on IEEE 802.11ax) products), the PHY- and MAC-layer test case specifications need to be updated accordingly to reflect those changes ⭐

Typically, a Wi-Fi access network connects to a wired infrastructure 🏠 Either the wired or the Wi-Fi segment can be the bottleneck. The following sections describe basic test cases for both scenarios separately. The same set of performance metrics as in [RFC8867]) should be collected for each test case 🌸

We recommend carrying out the test cases defined in this document using a simulator, such as [NS-2] or [NS-3] 💻 When feasible, it is encouraged to perform testbed-based evaluations using Wi-Fi access points and endpoints running up-to-date IEEE 802.11 protocols, such as 802.11ac and the emerging Wi-Fi 6, so the viability of the candidate schemes can be verified 🔥

3.1. Bottleneck in Wired Network 🏡

The test scenarios below are meant to mimic video conferencing over Wi-Fi from the home 🎥 Typically, the Wi-Fi home network is not congested, and the bottleneck is over the wired home access link. Although test evaluation results from this section are expected to be similar to those in [RFC8867], it's still worth running through these tests as sanity checks 🫧

3.1.1. Network Topology 🗺️

Figure 2 shows the network topology of Wi-Fi test cases 🦋 The test contains multiple mobile nodes (MNs) connected to a common Wi-Fi AP, plus their corresponding wired clients on fixed nodes (FNs). Each connection carries either an RTP-based media flow or a TCP traffic flow. Directions of the flows can be uplink (i.e., from mobile nodes to fixed nodes), downlink (i.e., from fixed nodes to mobile nodes), or bidirectional. The total number of uplink/downlink/bidirectional flows for RTP-based media traffic and TCP traffic are denoted as N and M, respectively ✨

                           Uplink
                     +----------------->+
    +------+                                       +------+
    | MN_1 |))))                             /=====| FN_1 |
    +------+    ))                          //     +------+
        .        ))                        //         .
        .         ))                      //          .
        .          ))                    //           .
    +------+         +----+         +-----+        +------+
    | MN_N | ))))))) |    |         |     |========| FN_N |
    +------+         |    |         |     |        +------+
                     | AP |=========| FN0 |
   +----------+      |    |         |     |      +----------+
   | MN_tcp_1 | )))) |    |         |     |======| FN_tcp_1 |
   +----------+      +----+         +-----+      +----------+
         .          ))                 \\             .
         .         ))                   \\            .
         .        ))                     \\           .
   +----------+  ))                       \\     +----------+
   | MN_tcp_M |)))                         \=====| FN_tcp_M |
   +----------+                                  +----------+
                    +<-----------------+
                            Downlink

      Figure 2: Network Topology for Wi-Fi Test Cases

3.1.2. Test/Simulation Setup 🧪

Test duration: 120 s ⏱️

Wi-Fi network characteristics:

  • Radio propagation model: Log-distance path loss propagation model (see [NS3WiFi])
  • PHY- and MAC-layer configuration: IEEE 802.11n
  • MCS Index at 11: Raw data rate at 52 Mbps, 16-QAM (Quadrature amplitude modulation) and 1/2 coding rate

Wired path characteristics:

  • Path capacity: 1 Mbps
  • One-way propagation delay: 50 ms
  • Maximum end-to-end jitter: 30 ms
  • Bottleneck queue type: Drop tail
  • Bottleneck queue size: 300 ms
  • Path loss ratio: 0%

Application characteristics:

  • Media traffic:
    • Media type: Video
    • Media direction: See Section 3.1.3
    • Number of media sources (N): See Section 3.1.3
    • Media timeline:
      • Start time: 0 s
      • End time: 119 s
  • Competing traffic:
    • Type of sources: Long-lived TCP or CBR over UDP
    • Traffic direction: See Section 3.1.3
    • Number of sources (M): See Section 3.1.3
    • Congestion control: Default TCP congestion control [RFC5681] or CBR traffic over UDP
    • Traffic timeline: See Section 3.1.3

3.1.3. Typical Test Scenarios 🎬

Single uplink RTP-based media flow: N=1 with uplink direction and M=0 💜

One pair of bidirectional RTP-based media flows: N=2 (i.e., one uplink flow and one downlink flow); M=0 🌈

One pair of bidirectional RTP-based media flows: N=2; one uplink on-off CBR flow over UDP: M=1 (uplink). The CBR flow has ON time at t=0s-60s and OFF time at t=60s-119s 🔁

One pair of bidirectional RTP-based media flows: N=2; one uplink off-on CBR flow over UDP: M=1 (uplink). The CBR flow has OFF time at t=0s-60s and ON time at t=60s-119s 🔂

One RTP-based media flow competing against one long-lived TCP flow in the uplink direction: N=1 (uplink) and M=1 (uplink). The TCP flow has start time at t=0s and end time at t=119s 🥊

3.1.4. Expected Behavior 🎯

Single uplink RTP-based media flow: The candidate algorithm is expected to detect the path capacity constraint, converge to the bottleneck link capacity, and adapt the flow to avoid unwanted oscillations as the sending bit rate approaches the bottleneck link capacity. No excessive oscillations in the media rate should be present 🌊

Bidirectional RTP-based media flows: The candidate algorithm is expected to converge to the bottleneck capacity of the wired path in both directions despite measurement noise over the Wi-Fi connection. In the presence of background TCP or CBR over UDP traffic, the rate of RTP-based media flows should adapt promptly to the arrival and departure of background traffic flows 🦩

One RTP-based media flow competing with a long-lived TCP flow in the uplink direction: The candidate algorithm is expected to avoid congestion collapse and stabilize at a fair share of the bottleneck link capacity 🤝

3.2. Bottleneck in Wi-Fi Network 📶

The test cases in this section assume the wired segment along the media path is well-provisioned, while the bottleneck exists over the Wi-Fi access network 🏢 This mimics the app scenarios people usually run into in an enterprise environment or at a coffee house ☕

3.2.1. Network Topology 🗺️

Same as defined in Section 3.1.1. 🫶

3.2.2. Test/Simulation Setup 🧪

Test duration: 120 s ⏱️

Wi-Fi network characteristics:

  • Radio propagation model: Log-distance path loss propagation model (see [NS3WiFi])
  • PHY- and MAC-layer configuration: IEEE 802.11n
  • MCS Index at 11: Raw data rate at 52 Mbps, 16-QAM (Quadrature amplitude modulation) and 1/2 coding rate

Wired path characteristics:

  • Path capacity: 100 Mbps
  • One-Way propagation delay: 50 ms
  • Maximum end-to-end jitter: 30 ms
  • Bottleneck queue type: Drop tail
  • Bottleneck queue size: 300 ms
  • Path loss ratio: 0%

Application characteristics:

  • Media traffic:
    • Media type: Video
    • Media direction: See Section 3.2.3
    • Number of media sources (N): See Section 3.2.3
    • Media timeline:
      • Start time: 0 s
      • End time: 119 s
  • Competing traffic:
    • Type of sources: long-lived TCP or CBR over UDP
    • Number of sources (M): See Section 3.2.3
    • Traffic direction: See Section 3.2.3
    • Congestion control: Default TCP congestion control [RFC5681] or CBR traffic over UDP
    • Traffic timeline: See Section 3.2.3

3.2.3. Typical Test Scenarios 📶✨

Okay bestie, this section lists a few test scenarios that are considered super important for figuring out how a candidate RTP-based congestion control scheme behaves over a Wi-Fi network 💅

  • Multiple RTP-based media flows sharing the wireless downlink: N=16 (all downlink); M=0. This one is all about studying how contention affects multiple concurrent media flows 🐝 For an 802.11n network with an MCS Index of 11 and a link rate of 52 Mbps, the total application-layer throughput (assuming reasonable distance, low interference, and infrequent contentions caused by competing streams) is around 20 Mbps. A total of N=16 RTP-based media flows (each with a maximum rate of 1.5 Mbps) is expected to saturate the wireless interface in this experiment. When evaluating a candidate scheme, focus on whether the downlink media flows can settle into a fair share of the total application-layer throughput 🎀
  • Multiple RTP-based media flows sharing the wireless uplink: N=16 (all uplink); M=0. When multiple clients try to send media packets uplink over the Wi-Fi network, things get messy fast 😭 They create more frequent contentions and potential collisions. Per-flow throughput is expected to be lower than in the downlink-only scenario above. Evaluation of a given candidate scheme should focus on whether the uplink flows can stabilize at a fair share of the total application-layer throughput 🦋
  • Multiple bidirectional RTP-based media flows: N=16 (8 uplink and 8 downlink); M=0. The goal here is to see how the candidate scheme handles bandwidth fairness between uplink and downlink flows 🫶
  • Multiple bidirectional RTP-based media flows with on-off CBR traffic over UDP: N=16 (8 uplink and 8 downlink); M=5 (uplink). The goal is to check how the candidate scheme adapts when its available bandwidth changes because background traffic leaves 🌸 The background traffic consists of several (e.g., M=5) CBR flows transported over UDP. These background flows are ON at time t=0-60s and OFF at time t=61-120s.
  • Multiple bidirectional RTP-based media flows with off-on CBR traffic over UDP: N=16 (8 uplink and 8 downlink); M=5 (uplink). This is the mirror image of the test above 🔥 The goal is to check how the candidate scheme adapts when its available bandwidth changes because background traffic arrives. The background traffic consists of several (e.g., M=5) parallel CBR flows transported over UDP. These background flows are OFF at time t=0-60s and ON at times t=61-120s.
  • Multiple bidirectional RTP-based media flows in the presence of background TCP traffic: N=16 (8 uplink and 8 downlink); M=5 (uplink). The goal is to see how RTP-based media flows compete against TCP over a congested Wi-Fi network for a given candidate scheme 👑 TCP flows have a start time at t=40s and an end time at t=80s.
  • Varying number of RTP-based media flows: A series of tests can be run for the test cases above with different values of N, e.g., N=[4, 8, 12, 16, 20] 🧚 The goal is to see how a candidate scheme responds to varying traffic load and demand over a congested Wi-Fi network. The start times of the media flows are randomly distributed within a window of t=0-10s, and their end times are randomly distributed within a window of t=110-120s.

3.2.4. Expected Behavior 🌈

Here's what we're hoping to see in each of those scenarios, no cap 💖

  • Multiple downlink RTP-based media flows: Each media flow is expected to get its fair share of the total bottleneck link bandwidth. Overall bandwidth usage should not be significantly lower than what the same number of concurrent downlink TCP flows would get. In other words, the behavior of multiple concurrent TCP flows is used as the performance benchmark for this scenario 📊 The end-to-end delay and packet loss ratio experienced by each flow should be within an acceptable range for real-time multimedia applications ✨
  • Multiple uplink RTP-based media flows: Overall bandwidth usage by all media flows should not be significantly lower than what the same number of concurrent uplink TCP flows would get. Again, the behavior of multiple concurrent TCP flows is the performance benchmark for this scenario 🎀
  • Multiple bidirectional RTP-based media flows with dynamic background traffic carrying CBR flows over UDP: The media flows are expected to adapt in a timely fashion to the changes in available bandwidth introduced by the arrival or departure of background traffic 🦋
  • Multiple bidirectional RTP-based media flows with dynamic background traffic over TCP: While the TCP background flows are present, the overall bandwidth usage by all media flows should not be significantly lower than what the same number of bidirectional TCP flows would achieve. Once more, the behavior of multiple concurrent TCP flows is the benchmark 🍓 All downlink media flows are expected to obtain similar bandwidth as each other. The throughput of each media flow is expected to decrease when TCP background traffic arrives and, conversely, increase when it departs. Both reactions should occur in a timely fashion, for example, within 10s of seconds ⏱️
  • Varying number of bidirectional RTP-based media flows: The test results for varying values of N, while keeping all other parameters constant, are expected to show steady and stable per-flow throughput for each value of N 💜 The average throughput of all media flows is expected to stay constant around the maximum rate when N is small, then gradually decrease as N increases until it reaches the minimum allowed rate. Beyond that point, the offered load to the Wi-Fi network exceeds its capacity (i.e., with a very large value of N) 🥺

3.3. Other Potential Test Cases 🧩

3.3.1. EDCA/WMM usage 🚦

The EDCA/WMM mechanism defines prioritized QoS for four traffic classes (or Access Categories). RTP-based real-time media flows should achieve better performance in terms of lower delay and fewer packet losses with EDCA/WMM enabled when they compete against non-interactive background traffic such as file transfers 📁 When most of the traffic over Wi-Fi is dominated by media, however, turning on WMM may degrade performance. All the media flows now attempt to access the wireless transmission medium more aggressively, which causes more frequent collisions and collision-induced losses 🙈 This is a topic worthy of further investigation. 🔍

As discussed in [Heusse2003], the presence of clients operating over slow PHY-layer link rates (e.g., a legacy 802.11b device) connected to a modern network may adversely impact the overall performance of the network 😬 Additional test cases can be devised to evaluate the effect of clients with heterogeneous link rates on the performance of the candidate congestion control algorithm. Such test cases, for instance, can specify that the PHY-layer link rates for all clients span over a wide range (e.g., 2 Mbps to 54 Mbps). This helps investigate its effect on the congestion control behavior of real-time interactive applications 💅

4. IANA Considerations 📋

This document has no IANA actions. 🌸

5. Security Considerations 🚨

The security considerations in [RFC8868] and the relevant congestion control algorithms apply. The principles for congestion control are described in [RFC2914], and in particular, any new method must implement safeguards to avoid congestion collapse of the Internet. This is serious, bestie 💀

Given the difficulty of deterministic wireless testing, it is recommended and expected that the tests described in this document would be done via simulations. However, in the case where these test cases are carried out in a testbed setting, the evaluation should take place in a controlled lab environment. In the testbed, the applications, simulators, and network nodes ought to be well-behaved and should not impact the desired results. It is important to take appropriate caution to avoid leaking nonresponsive traffic with unproven congestion avoidance behavior onto the open Internet 🚫🌐 Please, no chaos on the open Internet 😭

6. References 📚

6.1. Normative References 🔗

[HO-deploy-3GPP]
           3GPP, "Physical layer aspects for evolved Universal
           Terrestrial Radio Access (UTRA)", TS 25.814, October 2006,
           <http://www.3gpp.org/ftp/specs/
           archive/25_series/25.814/25814-710.zip>.

[IEEE802.11]
           IEEE, "Standard for Information technology--
           Telecommunications and information exchange between
           systems Local and metropolitan area networks--Specific
           requirements Part 11: Wireless LAN Medium Access Control
           (MAC) and Physical Layer (PHY) Specifications",
           IEEE 802.11-2012,
           <https://ieeexplore.ieee.org/document/7786995>.

[NS3WiFi]  "ns3::YansWifiChannel Class Reference",
           <https://www.nsnam.org/doxygen/
           classns3_1_1_yans_wifi_channel.html>.

[RFC5681]  Allman, M., Paxson, V., and E. Blanton, "TCP Congestion
           Control", RFC 5681, DOI 10.17487/RFC5681, September 2009,
           <https://www.rfc-editor.org/info/rfc5681>.

[RFC8867]  Sarker, Z., Singh, V., Zhu, X., and M. Ramalho, "Test
           Cases for Evaluating Congestion Control for Interactive
           Real-Time Media", RFC 8867, DOI 10.17487/RFC8867, January
           2021, <https://www.rfc-editor.org/info/rfc8867>.

[RFC8868]  Singh, V., Ott, J., and S. Holmer, "Evaluating Congestion
           Control for Interactive Real-Time Media", RFC 8868,
           DOI 10.17487/RFC8868, January 2021,
           <https://www.rfc-editor.org/info/rfc8868>.

6.2. Informative References

Okay so these are the informative refs bestie, the extra reading that backs up the doc without being required to implement it 📚✨

Contributors 🫶

These lovelies helped design, build, and verify the proposed test cases in the early days of this work, and they really leveled up this spec 💖

Ingemar Johansson (ingemar.s.johansson@ericsson.com) of Ericsson AB helped describe and validate the cellular test cases during the earlier stage of this document. 📶

Wei-Tian Tan (dtan2@cisco.com) of Cisco Systems designed and set up a Wi-Fi testbed for evaluating parallel video conferencing streams, and the proposed Wi-Fi test cases are based on that work. He also suggested extra test cases to consider, like the impact of EDCA/WMM usage. Main character energy 👑

Michael A. Ramalho (mar42@cornell.edu) of AcousticComms Consulting (previously at Cisco Systems) applied lessons from Cisco's internal experimentation to the draft versions of the document. He also worked on validating the proposed test cases in a virtual-machine-based lab setting. 🧚

Acknowledgments 🌈

The authors want to thank Tomas Frankkila, Magnus Westerlund, Kristofer Sandlund, Sergio Mena de la Cruz, and Mirja Kühlewind for their valuable inputs and review comments on this document. Iconic reviewers, all of them ✨💅

Authors' Addresses 🎀

Zaheduzzaman Sarker Ericsson AB Torshamnsgatan 23 SE-164 83 Stockholm Sweden

Phone: +46 10 717 37 43 Email: zaheduzzaman.sarker@ericsson.com

Xiaoqing Zhu Cisco Systems Building 4 12515 Research Blvd Austin, TX 78759 United States of America

Email: xiaoqzhu@cisco.com

Jiantao Fu Cisco Systems 771 Alder Drive Milpitas, CA 95035 United States of America

Email: jianfu@cisco.com

📦 where this page comes from