A Machine Learning Approach to TCP Throughput Prediction
Abstract
TCP throughput predictionis an important capability for networks where multiple paths exist between data senders and receivers. In this paper, we describe a new lightweight method for TCP throughput prediction. Our predictor uses Support Vector Regression (SVR); prediction is based on both prior file transfer history and measurements of simple path properties. We evaluate our predictor in a laboratory setting where ground truth can be measured with perfect accuracy. We report the performance of our predictor fororacularandpracticalmeasurements of path properties over a wide range of traffic conditions and transfer sizes. For bulk transfers in heavy traffic usingoracularmeasurements, TCP throughput is predicted within 10% of the actual value 87% of the time, representing nearly a threefold improvement in accuracy over prior history-based methods. Forpracticalmeasurements of path properties, predictions can be made within 10% of the actual value nearly 50% of the time, approximately a 60% improvement over history-based methods, and with much lower measurement traffic overhead. We implement our predictor in a tool calledPathPerf, test it in the wide area, and show thatPathPerfpredicts TCP throughput accurately over diverse wide area paths.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Several ML-based approaches have addressed limitations of inferring congestion by estimating throughput, RTT, and mobility in TCP-based networks, table entries rate in NDNs, and congestion level in DTNs.
“For the aforementioned reasons, several ML-based approaches have addressed the limitations of inferring the congestion in various network architectures by estimating different network parameters: throughput [238, 316, 371], RTT [22,128], and mobility [309] in TCP-based networks, table entries rate in NDNs [230], and congestion level in DTNs [412].”
From A comprehensive survey on machine learning for networking: Evolution, applications and research opportunities (2018) · cites this paper as [316]
TCP, as well as some multicast and multipath protocols, infer the congestion state from estimated network parameters such as RTT and throughput to adapt their behavior.
“TCP, as well as some multicast and multipath protocols, infer the congestion state from estimated network parameters (e.g.RTT and throughput) to adapt their behavior.”
From A comprehensive survey on machine learning for networking: Evolution, applications and research opportunities (2018) · cites this paper as [316]
Cites
The Nature of Statistical Learning Theory
Vapnik, V. N. (1995) · doi:10.1007/978-1-4757-2440-0 · Cited by 39,138 · see citing works
A tutorial on support vector regression
Smola, A. J., & Schölkopf, B. (2004) · Statistics and Computing · doi:10.1023/b:stco.0000035301.49549.88 · Cited by 13,482 · see citing works
Advances in kernel methods: support vector learning
Schölkopf, B., Burges, C. J. C., & Smola, A. J. (1999) · doi:10.5555/299094 · Cited by 5,809 · see citing works
Congestion avoidance and control
Jacobson, V. L. (1995) · ACM SIGCOMM Computer Communication Review · doi:10.1145/205447.205462 · Cited by 5,306 · see citing works
Congestion avoidance and control
Jacobson, V. (1988) · Conference on Applications, Technologies, Architectures, and Protocols for Computer Communication · doi:10.1145/52324.52356 · Cited by 4,650 · see citing works
Making Large-Scale SVM Learning PracticalPreprint
Joachims, T. (1998) · RePEc: Research Papers in Economics · doi:10.17877/de290r-14262 · Cited by 4,514 · see citing works
Padhye, J. D., Firoiu, V., Towsley, D., & Kurose, J. (1998) · ACM SIGCOMM Conference (SIGCOMM) · doi:10.1145/285237.285291 · Cited by 2,107 · see citing works
Andersen, D. G., Balakrishnan, H., Kaashoek, F., et al. (2001) · doi:10.1145/502034.502048 · Cited by 1,874 · see citing works
The macroscopic behavior of the TCP congestion avoidance algorithm
Mathis, M., Semke, J., Mahdavi, J., & Ott, T. J. (1997) · ACM SIGCOMM Computer Communication Review · doi:10.1145/263932.264023 · Cited by 1,496 · see citing works
End-to-end Internet packet dynamics
Paxson, V. (1999) · IEEE/ACM Transactions on Networking · doi:10.1109/90.779192 · Cited by 912 · see citing works
End-to-end available bandwidth: Measurement methodology, dynamics, and relation with TCP throughput
Jain, M., & Dovrolis, C. (2003) · IEEE/ACM Transactions on Networking · doi:10.1109/tnet.2003.815304 · Cited by 774 · see citing works
State transition analysis: a rule-based intrusion detection approach
Ilgun, K., Kemmerer, R. A., & Porras, P. A. (1995) · IEEE Transactions on Software Engineering · doi:10.1109/32.372146 · Cited by 753 · see citing works
Padhye, J. D., Firoiu, V., Towsley, D., & Kurose, J. (1998) · ACM SIGCOMM Computer Communication Review · doi:10.1145/285243.285291 · Cited by 716 · see citing works
A measurement study of available bandwidth estimation tools
Strauss, J., Katabi, D., & Kaashoek, F. (2003) · doi:10.1145/948205.948211 · Cited by 628 · see citing works
Measurements and analysis of end-to-end Internet dynamics
Lawrence Berkeley National Lab. (LBNL), B., & Paxson, V. (1997) · doi:10.2172/551971 · Cited by 572 · see citing works
Cardwell, N., Savage, S., & Anderson, T. E. (2002) · IEEE Conference on Computer Communications (INFOCOM) · doi:10.1109/infcom.2000.832574 · Cited by 473 · see citing works
On the constancy of internet path properties
Zhang⋆, Y., & Duffield, N. G. (2001) · doi:10.1145/505202.505228 · Cited by 468 · see citing works
On the characteristics and origins of internet flow rates
Zhang⋆, Y., Breslau, L., Paxson, V., et al. (2002) · ACM SIGCOMM Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications (SIGCOMM) · doi:10.1145/633025.633055 · Cited by 329 · see citing works
Self-configuring network traffic generation
Sommers, J., & Barford, P. (2004) · ACM SIGCOMM Internet Measurement Conference (IMC) · doi:10.1145/1028788.1028798 · Cited by 196 · see citing works
Self-citationOn the predictability of large transfer TCP throughput
He, Q., Dovrolis, C., & Ammar, M. (2005) · ACM SIGCOMM Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications (SIGCOMM) · doi:10.1145/1080091.1080110 · Cited by 163 · see citing works
Improving accuracy in end-to-end packet loss measurement
Sommers, J., Barford, P., Duffield, N. G., et al. (2005) · ACM SIGCOMM Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications (SIGCOMM) · doi:10.1145/1080091.1080111 · Cited by 157 · see citing works
Self-citationAnalyzing stability in wide-area network performance
Balakrishnan, H., Stemm, M., Seshan, S., et al. (1997) · ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS) · doi:10.1145/258612.258631 · Cited by 143 · see citing works
Predicting the performance of wide area data transfers
Vazhkudai, S., Schopf, J. M., & Foster, I. T. (2002) · Proceedings - IEEE International Parallel and Distributed Processing Symposium · doi:10.1109/ipdps.2002.1015510 · Cited by 142 · see citing works
Comparison of Public End-to-End Bandwidth Estimation Tools on High-Speed Links
Shriram, A., Murray, M., Hyun, Y., et al. (2005) · Lecture notes in computer science · doi:10.1007/978-3-540-31966-5_24 · Cited by 140 · see citing works
A stochastic model of TCP/IP with stationary random losses
Altman, E., Avrachenkov, K. E., & Barakat, C. (2005) · IEEE/ACM Transactions on Networking · doi:10.1109/tnet.2005.845536 · Cited by 137 · see citing works
Stochastic modeling of TCP over lossy links
Abouzeid, A. A., Roy, S., & Azizoglu, M. (2002) · IEEE Conference on Computer Communications (INFOCOM) · doi:10.1109/infcom.2000.832572 · Cited by 93 · see citing works
Multivariate Resource Performance Forecasting in the Network Weather Service
Swany, M., & Wolski, R. (2002) · doi:10.5555/762761.762812 · Cited by 85 · see citing works
Characterizing and Predicting TCP Throughput on the Wide Area Network
Dong, L., Qiao, Y., Dinda, P. A., & Bustamante, F. E. (2005) · IEEE International Conference on Distributed Computing Systems (ICDCS) · doi:10.1109/icdcs.2005.17 · Cited by 75 · see citing works
On the predictability of large transfer TCP throughput
He, Q., Dovrolis, C., & Ammar, M. (2005) · ACM SIGCOMM Computer Communication Review · doi:10.1145/1090191.1080110 · Cited by 73 · see citing works
On the characteristics and origins of internet flow rates
Zhang⋆, Y., Breslau, L., Paxson, V., et al. (2002) · ACM SIGCOMM Computer Communication Review · doi:10.1145/964725.633055 · Cited by 71 · see citing works
A Geometric Approach to Improving Active Packet Loss Measurement
Sommers, J., Barford, P., Duffield, N. G., et al. (2008) · IEEE/ACM Transactions on Networking · doi:10.1109/tnet.2007.900412 · Cited by 59 · see citing works
Self-citationMultivariate Resource Performance Forecasting in the Network Weather Service
Swany, M., & Wolski, R. (2002) · Proceedings - Supercomputing · doi:10.1109/sc.2002.10039 · Cited by 53 · see citing works
Critical path analysis of TCP transactions
Barford, P., & Crovella, M. E. (2001) · IEEE/ACM Transactions on Networking · doi:10.1109/90.929848 · Cited by 51 · see citing works
Self-citationPredicting TCP throughput from non-invasive network sampling
Goyal, M., Guérin, R. A., & Rajan, R. (2003) · Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM) · doi:10.1109/infcom.2002.1019259 · Cited by 51 · see citing works
A Proposed Framework for Calibration of Available Bandwidth Estimation Tools
Sommers, J., Barford, P., & Willinger, W. (2006) · Proceedings - IEEE Symposium on Computers and Communications/IEEE Symposium on Computers and Communications · doi:10.1109/iscc.2006.15 · Cited by 48 · see citing works
Self-citationSVM learning of IP address structure for latency prediction
Beverly, R., Sollins, K. R., & Berger, A. (2006) · doi:10.1145/1162678.1162682 · Cited by 33 · see citing works
Predicting short-transfer latency from TCP arcana: a trace-based validation
Arlitt, M. F., Krishnamurthy, B., & Mogul, J. C. (2005) · doi:10.5555/1251086.1251105 · Cited by 24 · see citing works
Predicting short-transfer latency from TCP arcana
Arlitt, M. F., Krishnamurthy, B., & Mogul, J. C. (2005) · doi:10.1145/1330107.1330132 · Cited by 14 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Boutaba, R., Salahuddin, M. A., Limam, N., Ayoubi, S., Shahriar, N., Estrada-Solano, F., & Caicedo, O. M. (2018) · Journal of Internet Services and Applications · doi:10.1186/s13174-018-0087-2 · Cited by 1,036 · see citing works
Learning-Based and Data-Driven TCP Design for Memory-Constrained IoT
Li, W., Zhou, F., Meleis, W., & Chowdhury, K. (2016) · 2016 International Conference on Distributed Computing in Sensor Systems (DCOSS) · doi:10.1109/dcoss.2016.8 · Cited by 29 · see citing works
- Authors
- Mirza, M., Sommers, J., Barford, P., & Zhu, X.
- Affiliations
- University of Wisconsin–Madison (US)
- Colgate University (US)
- Venue
- IEEE/ACM Transactions on Networking
- Published
- 2010-08
- Volume
- 18
- Issue
- 4
- Pages
- 1026-1039
- ISSN
- 1063-6692 (ISSN-L)
- 1558-2566
- OpenAlex
- W2136348220
- Cited by
- 132
- References
- 38
- Citations by year
- 2026: 5 · 2025: 3 · 2024: 6 · 2023: 6 · 2022: 11 · 2021: 8 · 2020: 13 · 2019: 10 · 2018: 12 · 2017: 5 · 2016: 15 · 2015: 13 · 2014: 4 · 2013: 4 · 2012: 6
- FWCI
- 7.53
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- IEEE/ACM Transactions on Networking (publishedVersion)
- Funders
- Princeton University
- University of Utah
- Topics
- Physical Sciences › Computer Science › Computer Networks and Communications › Network Traffic and Congestion Control
- Physical Sciences › Computer Science › Computer Networks and Communications › Software-Defined Networks and 5G
- Physical Sciences › Computer Science › Artificial Intelligence › Internet Traffic Analysis and Secure E-voting
- Keywords
- support vector regression, multiple paths, path properties, wide area networks, bulk transfer, heavy traffic
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
