pith. sign in

arxiv: 2511.19213 · v2 · submitted 2025-11-24 · 💻 cs.NI

Characterizing the Impact of Active Queue Management on Speed Test Measurements

Siddhant Ray , Taveesh Sharma , Jonatas Marques , Paul Schmitt , Francesco Bronzino , Nick Feamster This is my paper

Pith reviewed 2026-05-17 04:54 UTC · model grok-4.3

classification 💻 cs.NI
keywords Active Queue ManagementSpeed Test MeasurementsLatency Under LoadNetwork PerformanceCoDelDrop-tail QueuingThroughput Variance
0
0 comments X p. Extension

The pith

Active queue management schemes produce high variance in speed test measurements of throughput and latency under load.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper performs an empirical study in a controlled laboratory environment to examine how different Active Queue Management (AQM) schemes affect speed test results. It compares metrics like throughput and latency under load using CoDel, FQ-CoDel, Stochastic Fair Queuing, and a standard drop-tail baseline. The key finding is that these measurements show high variance depending on the AQM scheme and load conditions. This is important because new speed test features aim to measure user-perceived network responsiveness, but their reliability depends on understanding AQM effects. If the variance is as described, then speed test results used for policy decisions require careful interpretation and calibration of the testing platforms.

Core claim

In controlled experiments, speed test measurements of throughput and latency under load exhibit high variance when comparing different AQM schemes such as CoDel, FQ-CoDel, and Stochastic Fair Queuing against a drop-tail baseline, across varying load conditions. This demonstrates that AQM configurations play a critical role in determining the values of emerging latency metrics like latency under load or working latency.

What carries the argument

Laboratory-controlled comparisons of throughput and latency under load distributions across AQM schemes including CoDel, FQ-CoDel, SFQ, and drop-tail queuing.

If this is right

  • Speed test platforms must account for AQM when interpreting and reporting latency under load metrics.
  • High variance means measurements may not be consistent or comparable across networks with different queue management.
  • Results from speed tests could mislead policy or regulatory decisions if AQM effects are ignored.
  • The design of future speed test tools should consider sensitivity to basic network configurations like AQM.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Real networks might show even more complex interactions if AQM implementations differ from the lab versions tested.
  • Developers of speed test apps could add options to simulate or report under different AQM to improve accuracy.
  • This work points to the need for standardized testing environments that include common AQM to make results more robust.

Load-bearing premise

That the chosen laboratory setup with specific AQM implementations and load patterns is representative of real production networks and that AQM is the main driver of the observed measurement variance.

What would settle it

Running equivalent speed tests in actual production networks with known AQM deployments and checking if the high variance in latency and throughput metrics across schemes still holds.

Figures

Figures reproduced from arXiv: 2511.19213 by Francesco Bronzino, Jonatas Marques, Nick Feamster, Paul Schmitt, Siddhant Ray, Taveesh Sharma.

Figure 1
Figure 1. Figure 1: Throughput measurements without burst shaping and without cross traffic has lower variability across different AQM algorithms but often underuti￾lizes the link capacity. that adding such latency merely offsets the latency measure￾ments by a constant factor, without affecting throughput or relative latency under load. Therefore, in this paper we only present results without added latency. Finally, we evalua… view at source ↗
Figure 3
Figure 3. Figure 3: Throughput measurement with burst shaping [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Latency measurement with burst shaping without cross traffic is still stable due to lower compet￾ing load on the link. necessary to allow speed test tools to maximize link utiliza￾tion and accurately measure throughput, especially at higher bandwidths. NDT speed tests on the lab testbed with burst shaping and TCP Cubic cross traffic: For these measurements we no_aqm codelfq_codel sfq 10.00 20.00 30.00 40.0… view at source ↗
Figure 8
Figure 8. Figure 8: Instantaneous upload throughput measure [PITH_FULL_IMAGE:figures/full_fig_p006_8.png] view at source ↗
Figure 6
Figure 6. Figure 6: Latency measurement with burst shaping and [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 9
Figure 9. Figure 9: Instantaneous upload throughput measure [PITH_FULL_IMAGE:figures/full_fig_p006_9.png] view at source ↗
read the original abstract

Present day speed test tools measure peak throughput, but often fail to capture the user-perceived responsiveness of a network connection under load. Recently, platforms such as NDT, Ookla Speedtest and Cloudflare Speed Test have introduced metrics such as ``latency under load'' or ``working latency'' to fill this gap. Yet, the sensitivity of these metrics to basic network configurations such as Active Queue Management (AQM) remains poorly understood. In this work, we conduct an empirical study of the impact of AQM on speed test measurements in a laboratory setting. Using controlled experiments, we compare the distribution of throughput and latency under different load measurements across different AQM schemes, including CoDel, FQ-CoDel and Stochastic Fair Queuing (SFQ). On comparing with a standard drop-tail baseline, we find that measurements have high variance across AQM schemes and load conditions. These results highlight the critical role of AQM in shaping how emerging latency metrics should be interpreted, and underscore the need for careful calibration of speed test platforms before their results are used to guide policy or regulatory outcomes.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 3 minor

Summary. The paper conducts a controlled laboratory empirical study comparing the effects of Active Queue Management schemes (CoDel, FQ-CoDel, SFQ) versus a drop-tail baseline on speed-test measurements of throughput and latency-under-load metrics. Using synthetic loads, it reports that these metrics exhibit high variance across AQM schemes and load conditions, arguing that AQM must be accounted for when interpreting results from platforms such as NDT, Ookla, and Cloudflare Speed Test.

Significance. If the attribution of variance to AQM holds after addressing experimental controls, the work is significant for network measurement research. It provides concrete evidence that emerging latency metrics are sensitive to queuing behavior, which has direct implications for how speed-test data should be used in policy and regulatory contexts. The controlled lab design is a clear strength, enabling direct comparison of AQM effects without confounding variables from live networks.

major comments (3)
  1. [§3 (Experimental Setup)] §3 (Experimental Setup): The central claim that variance is driven primarily by AQM requires explicit demonstration that non-AQM factors (traffic generator parameters, buffer sizes, hardware timing) are held constant across runs. Without reported repetition counts, exclusion criteria, or validation that synthetic loads reproduce production traffic statistics, the observed differences could arise from uncontrolled variables rather than queuing discipline.
  2. [§4 (Results)] §4 (Results): The manuscript asserts 'high variance' across schemes but does not report statistical tests, number of runs, or measures of dispersion (e.g., inter-quartile ranges or confidence intervals). This omission makes it impossible to assess whether the reported differences are statistically reliable or merely consistent with experimental noise.
  3. [§5 (Discussion)] §5 (Discussion): The generalizability argument is load-bearing for the policy implications yet rests on the untested assumption that lab conditions with the chosen AQM implementations and load patterns are representative. A concrete test—such as replaying captured production traces or varying buffer sizes—would be needed to rule out implementation artifacts.
minor comments (3)
  1. [Abstract] Abstract: The phrase 'high variance' would be more informative if accompanied by at least one quantitative example (e.g., latency range or coefficient of variation) to convey magnitude to readers.
  2. [Figures] Figures: Ensure that all plots include error bars or box-plot whiskers and that legends explicitly label each AQM scheme and load condition to avoid ambiguity in visual comparison.
  3. Reproducibility: Consider releasing the traffic-generation scripts and AQM configuration files as supplementary material to allow independent verification of the reported variance.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback, which has helped us improve the clarity and rigor of the manuscript. We have revised the paper to address the concerns about experimental controls, statistical reporting, and generalizability. Point-by-point responses to the major comments follow.

read point-by-point responses

  1. Referee: [§3 (Experimental Setup)] §3 (Experimental Setup): The central claim that variance is driven primarily by AQM requires explicit demonstration that non-AQM factors (traffic generator parameters, buffer sizes, hardware timing) are held constant across runs. Without reported repetition counts, exclusion criteria, or validation that synthetic loads reproduce production traffic statistics, the observed differences could arise from uncontrolled variables rather than queuing discipline.

    Authors: We agree that explicit controls are essential to attribute variance to AQM. In the revised manuscript, Section 3 now includes a dedicated subsection on experimental controls, with a table listing fixed parameters (traffic generator configured for constant bit-rate flows with 1500-byte packets, buffer size fixed at 1000 packets for all schemes, identical hardware and timing sources across runs). Each configuration was repeated 30 times; we report means with standard deviations and note that no runs were excluded. We also added a validation paragraph comparing synthetic load statistics (e.g., flow size distribution and inter-arrival times) to samples from public speed-test traces, confirming close alignment. revision: yes

  2. Referee: [§4 (Results)] §4 (Results): The manuscript asserts 'high variance' across schemes but does not report statistical tests, number of runs, or measures of dispersion (e.g., inter-quartile ranges or confidence intervals). This omission makes it impossible to assess whether the reported differences are statistically reliable or merely consistent with experimental noise.

    Authors: We accept this criticism of the original presentation. The revised Section 4 now states that 30 independent runs were performed per AQM-load pair. Figures have been updated to include box plots with inter-quartile ranges and whiskers, and we report 95% confidence intervals on the mean throughput and latency values. We added a statistical analysis subsection performing one-way ANOVA across schemes followed by post-hoc Tukey tests, with p-values confirming that differences between AQM variants and the drop-tail baseline are statistically significant (p < 0.01) rather than noise. revision: yes

  3. Referee: [§5 (Discussion)] §5 (Discussion): The generalizability argument is load-bearing for the policy implications yet rests on the untested assumption that lab conditions with the chosen AQM implementations and load patterns are representative. A concrete test—such as replaying captured production traces or varying buffer sizes—would be needed to rule out implementation artifacts.

    Authors: We acknowledge the importance of this point for the strength of our policy-related claims. In the revision we have expanded the limitations paragraph in §5 to explicitly list the assumptions of our synthetic loads and single testbed. We added a small sensitivity experiment varying buffer size (500 vs. 2000 packets) under one AQM scheme and show that the relative ordering of variance remains consistent. However, a full replay of production traces would require new data collection and analysis that exceeds the scope and resources of the current study; we have therefore framed the policy implications more cautiously and listed trace-replay validation as future work. revision: partial

Circularity Check

0 steps flagged

No circularity: purely empirical comparison with no derivations or self-referential predictions

full rationale

The paper describes a laboratory-based empirical study that runs controlled experiments comparing throughput and latency distributions across AQM schemes (CoDel, FQ-CoDel, SFQ) versus a drop-tail baseline under varying loads. No equations, models, fitted parameters, or predictions appear in the provided text. Claims rest directly on experimental observations rather than any reduction to inputs by construction, self-citation chains, or renamed known results. The central finding of high variance is presented as an outcome of the measurements themselves, making the work self-contained against external benchmarks with no load-bearing circular steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Empirical measurement study; no mathematical derivations, free parameters, or new postulated entities are introduced.

pith-pipeline@v0.9.0 · 5502 in / 1074 out tokens · 40095 ms · 2026-05-17T04:54:09.913916+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

49 extracted references · 49 canonical work pages · 2 internal anchors

  1. [1]

    Understanding Speedtest Methodology

    2023. Understanding Speedtest Methodology. https://www.ookla. com/articles/ookla-speedtest-methodology. Whitepaper, Accessed: 2025-10-02

    work page 2023

  2. [2]

    Dale Alexander and Lika Døhl Diouf. 2022. Illustrating Internet Speed Divides in the Caribbean During COVID-19.FOCUS Magazine of the Caribbean Development and Cooperation Committee (CDCC)(12 2022). https://hdl.handle.net/11362/48956

    work page 2022

  3. [3]

    Apple Inc. 2025. Apple Speedtest. https://support.apple.com/ HT212313. Accessed: 2025-10-02

    work page 2025

  4. [4]

    Babak Arzani, Theophilus Benson, David Maltz, and Lucian Popa

    work page

  5. [5]

    InProceedings of the ACM SIGMETRICS Conference

    Speedtest at Scale: Measurement and Biases in Crowdsourced Performance Data. InProceedings of the ACM SIGMETRICS Conference. ACM, 307–318. doi:10.1145/2896377.2901480

    work page doi:10.1145/2896377.2901480

  6. [6]

    Steven Bauer, David D Clark, and William Lehr. 2010. Understanding broadband speed measurements. TPRC

    work page 2010

  7. [7]

    Bischof, André Callado, Srikanth Sundaresan, and Nick Feamster

    Zachary S. Bischof, André Callado, Srikanth Sundaresan, and Nick Feamster. 2019. Characterizing latency in broadband access networks. InProceedings of the Passive and Active Measurement Conference (PAM). Springer, 65–77. doi:10.1007/978-3-030-15986-3_5

    work page doi:10.1007/978-3-030-15986-3_5 2019

  8. [8]

    Zdravko Bozakov, Aaron Schulman, and Srikanth Sundaresan. 2018. M-Lab: An Open Platform for Large-Scale Network Measurement. In Proceedings of the ACM SIGCOMM Conference (Demo Session). ACM, 80–81. doi:10.1145/3234200.3234239

    work page doi:10.1145/3234200.3234239 2018

  9. [9]

    Center for Rural Pennsylvania. 2018. Broadband Availability and Ac- cess. https://www.rural.pa.gov/publications/broadband.cfm Accessed November 13, 2024

    work page 2018

  10. [10]

    David D Clark and Sara Wedeman. 2021. Measurement, meaning and purpose: Exploring the M-Lab NDT dataset. InTPRC49: The 49th Research Conference on Communication, Information and Internet Policy. https://dx.doi.org/10.2139/ssrn.3898339

    work page doi:10.2139/ssrn.3898339 2021

  11. [11]

    David D Clark and Sara Wedeman. 2024. Measurement of Inter- net access latency: A cross-dataset comparison. InProceedings of the TPRC2024 The Research Conference on Communications, Information and Internet Policy. http://dx.doi.org/10.2139/ssrn.4909679

    work page doi:10.2139/ssrn.4909679 2024

  12. [12]

    Dave Täht. [n. d.]. The new features and flaws of speedtest.net, ookla and cloudflare. https://blog.cerowrt.org/post/speedtests/. Accessed: 2025-10-01

    work page 2025

  13. [13]

    Nick Feamster and Jason Livingood. 2019. Internet Speed Measurement: Current Challenges and Future Recommendations. arXiv:1905.02334 [cs.NI] https://arxiv.org/abs/1905.02334

    work page arXiv 2019

  14. [14]

    S. Floyd. 2001. A report on recent developments in TCP congestion control.Comm. Mag.39, 4 (April 2001), 84–90. doi:10.1109/35.917508

    work page doi:10.1109/35.917508 2001

  15. [15]

    Floyd and V

    S. Floyd and V. Jacobson. 1993. Random early detection gateways for congestion avoidance.IEEE/ACM Transactions on Networking1, 4 (1993), 397–413. doi:10.1109/90.251892

    work page doi:10.1109/90.251892 1993

  16. [16]

    Jim Gettys and Kathleen Nichols. 2011. Bufferbloat: Dark Buffers in the Internet: Networks without effective AQM may again be vulnerable to congestion collapse.Queue9, 11 (nov 2011), 40–54. doi:10.1145/ 2063166.2071893

    work page arXiv 2011

  17. [17]

    Toke Høiland-Jørgensen. [n. d.]. Bufferbloat and Beyond. ([n. d.])

    work page

  18. [18]

    Toke Høiland-Jørgensen, Bengt Ahlgren, Per Hurtig, and Anna Brun- strom. 2016. Measuring Latency Variation in the Internet. InProceed- ings of the 12th International on Conference on Emerging Networking EXperiments and Technologies(Irvine, California, USA)(CoNEXT ’16). Association for Computing Machinery, New York, NY, USA, 473–480. doi:10.1145/2999572.2999603

    work page doi:10.1145/2999572.2999603 2016

  19. [19]

    Toke Høiland-Jørgensen, Paul McKenney, dave.taht@gmail.com, Jim Gettys, and Eric Dumazet. 2018. The Flow Queue CoDel Packet Scheduler and Active Queue Management Algorithm. RFC 8290. doi:10.17487/RFC8290

    work page doi:10.17487/rfc8290 2018

  20. [20]

    Toke Høiland-Jørgensen, Dave Täht, and Jonathan Morton. 2018. Piece of CAKE: A Comprehensive Queue Management Solution for Home Gateways. arXiv:1804.07617 [cs.NI] https://arxiv.org/abs/1804.07617

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  21. [21]

    Hanyang Jiang, Henry Shaowu Yuchi, Elizabeth Belding, Ellen Zegura, and Yao Xie. 2023. Mobile Internet Quality Estimation using Self- Tuning Kernel Regression. arXiv:2311.05641 [stat.AP] https://arxiv. org/abs/2311.05641

    work page arXiv 2023

  22. [22]

    Hyeongseong Lee, Udit Paul, Arpit Gupta, Elizabeth Belding, and Mengyang Gu. 2023. Analyzing Disparity and Temporal Progression of Internet Quality through Crowdsourced Measurements with Bias- Correction. arXiv:2310.16136 [stat.AP] https://arxiv.org/abs/2310. 16136

    work page arXiv 2023

  23. [23]

    Linux man-pages project. 2025. tc - traffic control. https://man7. org/linux/man-pages/man8/tc.8.html Accessed through ‘man tc‘ on a Linux system

    work page 2025

  24. [24]

    Marwell, and Nick Feamster

    Kyle MacMillan, Tarun Mangla, James Saxon, Nicole P. Marwell, and Nick Feamster. 2023. A Comparative Analysis of Ookla Speedtest and Measurement Labs Network Diagnostic Test (NDT7).Proc. ACM Meas. Anal. Comput. Syst.7, 1, Article 19 (March 2023), 26 pages. doi:10.1145/3579448

    work page doi:10.1145/3579448 2023

  25. [25]

    McKenney

    Paul E. McKenney. 1990. Stochastic Fairness Queuing. InIEEE IN- FOCOM’90 Proceedings. The Institute of Electrical and Electronics Engineers, Inc., San Francisco, 733–740

    work page 1990

  26. [26]

    Measurement Lab. 2025. NDT7 Protocol Specification. https://github. com/m-lab/ndt-server/blob/main/spec/ndt7-protocol.md. Accessed: 2025-10-02

    work page 2025

  27. [27]

    Measurement Lab. 2025. Network Diagnostic Tool (NDT). https: //www.measurementlab.net/tests/ndt/. Accessed: 2025-10-02

    work page 2025

  28. [28]

    Cise Midoglu, Leonhard Wimmer, Andra Lutu, Ozgu Alay, and Carsten Griwodz. 2018. MONROE-Nettest: A Configurable Tool for Dissecting Speed Measurements in Mobile Broadband Networks. arXiv:1710.07805 [cs.NI] https://arxiv.org/abs/1710.07805

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  29. [29]

    Syed Tauhidun Nabi, Zhuowei Wen, Brooke Ritter, and Shaddi Hasan

    work page

  30. [30]

    InProceedings of the 2024 ACM on Internet Measurement Conference(Madrid, Spain) (IMC ’24)

    Red is Sus: Automated Identification of Low-Quality Service Availability Claims in the US National Broadband Map. InProceedings of the 2024 ACM on Internet Measurement Conference(Madrid, Spain) (IMC ’24). Association for Computing Machinery, New York, NY, USA, 2–18. doi:10.1145/3646547.3688441

    work page doi:10.1145/3646547.3688441 2024

  31. [31]

    Netflix. 2025. Fast.com Internet Speed Test. https://fast.com/. Accessed: 2025-10-02

    work page 2025

  32. [32]

    Kathleen Nichols and Van Jacobson. 2012. Controlling queue delay. Commun. ACM55, 7 (jul 2012), 42–50. doi:10.1145/2209249.2209264

    work page doi:10.1145/2209249.2209264 2012

  33. [33]

    Ookla. 2025. Speedtest by Ookla. https://www.speedtest.net/. Ac- cessed: 2025-10-02

    work page 2025

  34. [34]

    2025.Responsiveness under Working Conditions

    Christoph Paasch, Randall Meyer, Stuart Cheshire, and Will Hawkins. 2025.Responsiveness under Working Conditions. Internet-Draft draft- ietf-ippm-responsiveness-07. Internet Engineering Task Force. https: 7 Ray et al. //datatracker.ietf.org/doc/draft-ietf-ippm-responsiveness/07/ Work in Progress

    work page 2025

  35. [35]

    Rong Pan, Preethi Natarajan, Chiara Piglione, Mythili Suryanarayana Prabhu, Vijay Subramanian, Fred Baker, and Bill VerSteeg. 2013. PIE: A lightweight control scheme to address the bufferbloat problem. In 2013 IEEE 14th International Conference on High Performance Switching and Routing (HPSR). 148–155. doi:10.1109/HPSR.2013.6602305

    work page doi:10.1109/hpsr.2013.6602305 2013

  36. [36]

    Udit Paul, Jiamo Liu, David Farias-llerenas, Vivek Adarsh, Arpit Gupta, and Elizabeth Belding. 2022. Characterizing Internet Ac- cess and Quality Inequities in California M-Lab Measurements. In Proceedings of the 5th ACM SIGCAS/SIGCHI Conference on Comput- ing and Sustainable Societies(Seattle, WA, USA)(COMPASS ’22). As- sociation for Computing Machinery,...

    work page doi:10.1145/3530190.3534813 2022

  37. [38]

    Aditya Rao, Arpit Gupta, Nick Feamster, and Balachander Krishna- murthy. 2020. Understanding Bottlenecks in the Internet Last Mile. In Proceedings of the ACM Internet Measurement Conference (IMC). ACM, 321–335. doi:10.1145/3419394.3423642

    work page doi:10.1145/3419394.3423642 2020

  38. [39]

    Juan Restrepo, John Rula, and Alberto Dainotti. 2021. Crowdsourced Internet Measurement: Bias and Representativeness. InProceedings of the Passive and Active Measurement Conference (PAM). Springer, 123–136. doi:10.1007/978-3-030-72582-2_8

    work page doi:10.1007/978-3-030-72582-2_8 2021

  39. [40]

    Jaimes, Shahram Taj, and Md

    Reinaldo Sanchez-Arias, Luis G. Jaimes, Shahram Taj, and Md. Selim Habib. 2023. Understanding the State of Broadband Connectivity: An Analysis of Speedtests and Emerging Technologies.IEEE Access11 (2023), 101580–101603. doi:10.1109/ACCESS.2023.3313231

    work page doi:10.1109/access.2023.3313231 2023

  40. [41]

    Fatih Berkay Sarpkaya, Fraida Fund, and Shivendra Panwar. 2025. To Adopt or Not to Adopt L4S-Compatible Congestion Control? Under- standing Performance in a Partial L4S Deployment. InPassive and Active Measurement, Cecilia Testart, Roland van Rijswijk-Deij, and Burkhard Stiller (Eds.). Springer Nature Switzerland, Cham, 217–246

    work page 2025

  41. [42]

    James Saxon and Dan A. Black. 2022. What we can learn from selected, unmatched data: Measuring internet inequality in Chicago.Comput- ers, Environment and Urban Systems98 (2022), 101874. doi:10.1016/j. compenvurbsys.2022.101874

    work page doi:10.1016/j 2022

  42. [43]

    Koen De Schepper, Olga Albisser, Olivier Tilmans, and Bob Briscoe

    work page

  43. [44]

    arXiv:2209.01078 [cs.NI] https://arxiv.org/abs/2209.01078

    Dual Queue Coupled AQM: Deployable Very Low Queuing Delay for All. arXiv:2209.01078 [cs.NI] https://arxiv.org/abs/2209.01078

    work page arXiv

  44. [45]

    Taveesh Sharma, Paul Schmitt, Francesco Bronzino, Nick Feamster, and Nicole P. Marwell. 2024. Beyond Data Points: Regionalizing Crowd- sourced Latency Measurements.Proc. ACM Meas. Anal. Comput. Syst. 8, 3, Article 34 (Dec. 2024), 24 pages. doi:10.1145/3700416

    work page doi:10.1145/3700416 2024

  45. [46]

    Shreedhar and George Varghese

    M. Shreedhar and George Varghese. 1995. Efficient fair queueing using deficit round robin. InProceedings of the Conference on Applications, Technologies, Architectures, and Protocols for Computer Communication (Cambridge, Massachusetts, USA)(SIGCOMM ’95). Association for Computing Machinery, New York, NY, USA, 231–242. doi:10.1145/ 217382.217453

    work page arXiv 1995

  46. [47]

    Jacob Smith, Srikanth Sundaresan, Will Scott, Siddharth Eravuchira, and Nick Feamster. 2022. Understanding Netflix’s Fast.com: A Study of Web-Based Speed Measurement. InProceedings of the ACM Internet Measurement Conference (IMC). ACM, 458–470. doi:10.1145/3517745. 3561442

    work page doi:10.1145/3517745 2022

  47. [48]

    Srikanth Sundaresan, Walter de Donato, Nick Feamster, Renata Teix- eira, Sam Crawford, and Antonio Pescapè. 2011. Broadband In- ternet performance: A view from the edge. InProceedings of the Passive and Active Measurement Conference (PAM). Springer, 9–22. doi:10.1007/978-3-642-19260-9_2

    work page doi:10.1007/978-3-642-19260-9_2 2011

  48. [49]

    Srikanth Sundaresan, Xiaohong Deng, Yun Feng, Danny Lee, and Amogh Dhamdhere. 2017. Challenges in inferring internet conges- tion using throughput measurements. InProceedings of the 2017 In- ternet Measurement Conference(London, United Kingdom)(IMC ’17). Association for Computing Machinery, New York, NY, USA, 43–56. doi:10.1145/3131365.3131382

    work page doi:10.1145/3131365.3131382 2017

  49. [50]

    " " Runs ndt7 - client locally while capturing PCAP remotely via SSH

    Greg White. 2025.Operational Guidance on Coexistence with Classic ECN during L4S Deployment. Internet-Draft draft-ietf-tsvwg-l4sops-08. Internet Engineering Task Force. https://datatracker.ietf.org/doc/ draft-ietf-tsvwg-l4sops/08/ Work in Progress. A Ethics Statement This research was conducted in a lab environment using our own testbed infrastructure. Al...

    work page 2025