Pith. sign in

REVIEW 54 references

Detecting Flow Gaps in Data Streams

T0 review · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read GapFilter uses sequence-number-only matching and a civilian-suspect bucket split to detect major per-flow sequence gaps, beating a Cuckoo-filter baseline in F1, speed, and memory.

arxiv 2505.13945 v1 pith:2XZHRLWF submitted 2025-05-20 cs.DB

classification cs.DB
keywords accuracymonitoringspeedgapfilterdataflowgapsmemory
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Data streams are long lists of items, each tagged with a flow ID and a sequence number, like frames in a video or packets in a network session. Existing stream summaries track how many items a flow has or how much value it carries. This paper adds a third task: watching how the sequence number jumps between consecutive items of the same flow. A jump larger than a threshold, for example from 100 to 120, is a major gap that indicates a large chunk of data was lost or reordered.

GapFilter keeps a small table of buckets. Instead of storing the 13-byte flow ID, it stores only a few recent sequence numbers per bucket. When a new item arrives, the algorithm finds the stored number closest to its sequence; if the difference is between 5 and 30, it reports a major gap. Because normal flows are common and big gaps are rare, the closest number is usually the same flow's previous number. A second, accuracy-oriented version splits each bucket into civilian and suspect regions, so flows that once had a gap are watched more carefully and are not pushed out by large normal flows. SIMD instructions make the lookup and rearrangement constant time.

Tests on four internet traces with synthetic packet loss show higher F1 scores than a Cuckoo-filter Straw-man at equal memory. The authors claim their accuracy-oriented version needs about 1/32 of the memory to reach the same accuracy, and their speed version is faster. The paper also gives a theoretical recall bound, but that bound assumes a flow in the bucket is always detected, which side-steps the hardest part of the matching problem.

Extended reading notes

Core claim

The paper's central claim is that GapFilter is the first solution for monitoring the variation of value per key, and that it achieves high accuracy with minimal memory: 'GapFilter-AO requires, on average, 1/32 of the memory to match the accuracy of the Straw-man solution. GapFilter-SO operates at a speed 3 times faster than the Straw-man solution' (abstract and Section 1.2). The theoretical counterpart is 'GapFilter secures high accuracy with minimal memory usage' via Theorem 1's recall bound.

Load-bearing premise

The load-bearing premise is the Section 5 assumption that a major gap is reported whenever the flow's record is still present in its bucket: 'the major gap's being reported indicates that the flow is contained in the existing items in the ith bucket.' This lets the proof ignore the algorithm's actual matching step, where the stored sequence number closest to the incoming one may belong to a different flow with a similar randomized range, causing false matches and missed gaps. The entire memory advantage comes from dropping flow IDs and trusting this nearest-sequence match, so without a bound on matching errors the theorem does not establish the headline accuracy claim.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The central analysis rests on six stated or implicit assumptions: uniform hashing, Zipf flow sizes, uniform gap ratio, the strong premise that resident flows are always detected, an untested separation of randomized sequence ranges, and the realism of the synthetic loss model. The first three are conventional modeling choices; the fourth is a load-bearing simplification that removes matching errors from the theorem. The empirical claims are grounded partly in synthetic loss injected by the authors, whose realism is assumed.

free parameters (4)
  • w (cells per bucket) = 8
    Chosen in Section 6.2 as the best setting across CAIDA and MAWI; used for all headline results.
  • lf (fingerprint length) = 8 bits
    Tuned in Section 6.2 on MACCDC and IMC; only GapFilter-AO uses fingerprints.
  • suspect:civilian ratio (s:c) = 3:5
    Selected in Section 6.2 as the most robust setting; only GapFilter-AO uses this split.
  • T1 and T2 thresholds = T1=5, T2=30
    Set by the authors in Section 6.1.1 to define neglect, normal, minor gap, and major gap; they set the detection semantics rather than being fitted to maximize F1, but they affect all reported numbers.
assumptions (6)
  • standard math Flows are mapped uniformly at random to buckets by the hash function.
    Assumed in Lemma 1 and throughout the analysis; real hash functions approximate this, but it is not verified for the datasets.
  • domain assumption Flow sizes follow a Zipf distribution with 1 < alpha <= 3.
    Theorem 1's bound is for Zipf streams; the experimental datasets are not checked for a Zipf fit.
  • domain assumption All flows share the same gap ratio beta.
    Section 5 states this assumption; in real traffic, gap ratios differ per flow.
  • ad hoc to paper If a flow is resident in its bucket, every major gap in it is reported.
    Section 5 bullet: 'the major gap's being reported indicates that the flow is contained in the existing items in the ith bucket.' This assumes away matching errors from sequence-number collisions and nearest-neighbor misassignment, which is the core detection problem.
  • domain assumption Randomized sequence numbers keep concurrently active flows' ranges in a bucket sufficiently separated for nearest-neighbor matching to be correct.
    Section 4.4.2 adds the hash bias b(FID) to separate flows, but no bound on residual overlap is given; fingerprints reduce but do not eliminate collisions.
  • domain assumption The synthetic item-loss model (consecutive with probability b^j, single with probability p) is representative of real network loss patterns.
    Section 6.1.1; ground truth for the IMC dataset comes from this model, so the F1 results inherit its realism.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Detecting Flow Gaps in Data Streams." pith.science (2026). https://pith.science/paper/2XZHRLWF

@misc{pith2026250513945,
  author       = {Pith},
  title        = {Pith review of: Detecting Flow Gaps in Data Streams},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2XZHRLWF}},
  note         = {Machine review of arXiv:2505.13945}
}
read the original abstract

Data stream monitoring is a crucial task which has a wide range of applications. The majority of existing research in this area can be broadly classified into two types, monitoring value sum and monitoring value cardinality. In this paper, we define a third type, monitoring value variation, which can help us detect flow gaps in data streams. To realize this function, we propose GapFilter, leveraging the idea of Sketch for achieving speed and accuracy. To the best of our knowledge, this is the first work to detect flow gaps in data streams. Two key ideas of our work are the similarity absorption technique and the civilian-suspect mechanism. The similarity absorption technique helps in reducing memory usage and enhancing speed, while the civilian-suspect mechanism further boosts accuracy by organically integrating broad monitoring of overall flows with meticulous monitoring of suspicious flows.We have developed two versions of GapFilter. Speed-Oriented GapFilter (GapFilter-SO) emphasizes speed while maintaining satisfactory accuracy. Accuracy-Oriented GapFilter (GapFilter-AO) prioritizes accuracy while ensuring considerable speed. We provide a theoretical proof demonstrating that GapFilter secures high accuracy with minimal memory usage. Further, extensive experiments were conducted to assess the accuracy and speed of our algorithms. The results reveal that GapFilter-AO requires, on average, 1/32 of the memory to match the accuracy of the Straw-man solution. GapFilter-SO operates at a speed 3 times faster than the Straw-man solution. All associated source code has been open-sourced and is available on GitHub.

Figures

Figures reproduced from arXiv: 2505.13945 by the authors.

Figure 1
Figure 1. Illustration of GapFilter-SO. This is a GapFilter [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Monitoring operations in GapFilter-AO. 5 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Concurrency circumstances of all/abnormal flows on different datasets. [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Distributions of flow length and gap size. 6.1.3 Metrics. • Precision Rate (PR): Precision rate is the ratio of the number of correctly reported instances to the number of reported instances. • Recall Rate (RR): Recall rate is the ratio of the number of cor￾rectly repo…
Figure 5
Figure 5. Figure 5: Effect of 𝑤 on GapFilter-SO on different datasets. 10 0 10 1 10 2 Memory Usage (KB) 0.4 0.6 0.8 F1 Score Ours-AO(s=1,c=7) Ours-AO(s=3,c=5) Ours-AO(s=5,c=3) Ours-AO(s=7,c=1) (a) CAIDA 10 0 10 1 10 2 Memory Usage (KB) 0.6 0.7 0.8 0.9 F1 Score Ours-AO(s=1,c=7) Ours-AO(s=3…
Figure 6
Figure 6. Figure 6: Effect of s/c on GapFilter-AO on different datasets. Effect of w on GapFilter-SO ( [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Effect of 𝑙 𝑓 on GapFilter-AO on different datasets. The GapFilter-SO with 𝑤 = 8 performs best. This is attributable to the two counteracting effects on accuracy as 𝑤 gets larger: the risk of different 𝑆𝐸𝑄 colliding in a bucket increases, while the probabil￾ity that to…
Figure 8
Figure 8. Figure 8: 𝐹1-Score of algorithms on different datasets. 10 0 10 1 10 2 Memory Usage (KB) 0.0 0.2 0.4 0.6 0.8 1.0 Recall Ours-AO Ours-SO Straw-man (a) CAIDA 10 0 10 1 10 2 Memory Usage (KB) 0.0 0.2 0.4 0.6 0.8 1.0 Recall Ours-AO Ours-SO Straw-man (b) MAWI 10 0 10 1 10 2 Memory Us…
Figure 9
Figure 9. Figure 9: RR of algorithms in Real-time Report Scenario on different datasets. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: PR of algorithms in Real-time Report Scenario on different datasets. [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: Throughput of algorithms on different datasets. [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: Effect of optimizations on different datasets. [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]
Figure 13
Figure 13. Figure 13: The effect of 𝑟 and 𝑏 on IMC dataset. Effect of 𝑟 on performance ( [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 51 canonical work pages

  1. [1]

    https://github.com/ GapFilter/GapFilter

    The source codes of ours and other related algorithms . https://github.com/ GapFilter/GapFilter

  2. [2]

    Anomaly detection based on data stream monitoring and prediction with improved gauss- ian process regression algorithm

    Jingyue Pang, Datong Liu, Haitao Liao, Yu Peng, and Xiyuan Peng. Anomaly detection based on data stream monitoring and prediction with improved gauss- ian process regression algorithm. In 2014 International Conference on Prognostics and Health Management, pages 1–7. IEEE, 2014

  3. [3]

    Spotlight: Detecting anomalies in streaming graphs

    Dhivya Eswaran, Christos Faloutsos, Sudipto Guha, and Nina Mishra. Spotlight: Detecting anomalies in streaming graphs. InProceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 1378–1386, 2018

  4. [4]

    Bursts- ketch: Finding bursts in data streams

    Ruijie Miao, Zheng Zhong, Jiarui Guo, Zikun Li, Tong Yang, and Bin Cui. Bursts- ketch: Finding bursts in data streams. IEEE Transactions on Knowledge and Data Engineering, 2022

  5. [5]

    Fast memory-efficient anomaly detection in streaming heterogeneous graphs

    Emaad Manzoor, Sadegh M Milajerdi, and Leman Akoglu. Fast memory-efficient anomaly detection in streaming heterogeneous graphs. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1035–1044, 2016

  6. [6]

    {SketchLib}: Enabling efficient sketch-based monitoring on programmable switches

    Hun Namkung, Zaoxing Liu, Daehyeok Kim, Vyas Sekar, and Peter Steenkiste. {SketchLib}: Enabling efficient sketch-based monitoring on programmable switches. In 19th USENIX Symposium on Networked Systems Design and Im- plementation (NSDI 22), pages 743–759, 2022

  7. [7]

    Bitsense: Universal and nearly zero-error optimization for sketch counters with compressive sensing

    Rui Ding, Shibo Yang, Xiang Chen, and Qun Huang. Bitsense: Universal and nearly zero-error optimization for sketch counters with compressive sensing. In Proc. of SIGCOMM, 2023

  8. [8]

    Machine learning based concept drift detection for predictive maintenance

    Jan Zenisek, Florian Holzinger, and Michael Affenzeller. Machine learning based concept drift detection for predictive maintenance. Computers & Industrial Engineering, 137:106031, 2019

Show all 54 references
  1. [9]

    Big data and stream processing platforms for industry 4.0 requirements mapping for a predictive maintenance use case

    Radhya Sahal, John G Breslin, and Muhammad Intizar Ali. Big data and stream processing platforms for industry 4.0 requirements mapping for a predictive maintenance use case. Journal of manufacturing systems , 54:138–151, 2020

  2. [10]

    Data-driven maintenance: combining predictive maintenance and mixed reality-supported remote assis- tance

    Josef Wolfartsberger, Jan Zenisek, and Norbert Wild. Data-driven maintenance: combining predictive maintenance and mixed reality-supported remote assis- tance. Procedia Manufacturing, 45:307–312, 2020

  3. [11]

    An e-customer behavior model with online analytical mining for internet marketing planning

    Irene SY Kwan, Joseph Fong, and Hing K Wong. An e-customer behavior model with online analytical mining for internet marketing planning. Decision Support Systems, 41(1):189–204, 2005

  4. [12]

    Customer behavior analysis using real-time data processing: A case study of digital signage-based online stores

    Ganjar Alfian, Muhammad Fazal Ijaz, Muhammad Syafrudin, M Alex Syaekhoni, Norma Latif Fitriyani, and Jongtae Rhee. Customer behavior analysis using real-time data processing: A case study of digital signage-based online stores. Asia Pacific Journal of Marketing and Logistics ,...

  5. [13]

    An architecture for the real-time data stream monitoring in iot

    Mario José Diván and María Laura Sánchez Reynoso. An architecture for the real-time data stream monitoring in iot. Multimedia Big Data Computing for IoT Applications: Concepts, Paradigms and Solutions , pages 59–100, 2020

  6. [14]

    Reversible sketches: enabling monitoring and analysis over high-speed data streams

    Robert Schweller, Zhichun Li, Yan Chen, and etal. Reversible sketches: enabling monitoring and analysis over high-speed data streams. TON, 2007

  7. [15]

    On- board vehicle data stream monitoring using minefleet and fast resource con- strained monitoring of correlation matrices

    Hillol Kargupta, Vasundhara Puttagunta, Martin Klein, and Kakali Sarkar. On- board vehicle data stream monitoring using minefleet and fast resource con- strained monitoring of correlation matrices. New Generation Computing, 25:5–32, 2006. 13 Siyuan Dong, Yuxuan Tian, Wenhan Ma...

  8. [16]

    An improved data stream summary: the count-min sketch and its applications

    Graham Cormode and Shan Muthukrishnan. An improved data stream summary: the count-min sketch and its applications. Journal of Algorithms, 55(1):58–75, 2005

  9. [17]

    New directions in traffic measurement and accounting: Focusing on the elephants, ignoring the mice

    Cristian Estan and George Varghese. New directions in traffic measurement and accounting: Focusing on the elephants, ignoring the mice. ACM Transactions on Computer Systems (TOCS), 21(3):270–313, 2003

  10. [18]

    Finding frequent items in data streams

    Moses Charikar, Kevin Chen, and Martin Farach-Colton. Finding frequent items in data streams. In International Colloquium on Automata, Languages, and Pro- gramming, pages 693–703. Springer, 2002

  11. [19]

    Per-flow traffic measurement through randomized counter sharing

    Tao Li, Shigang Chen, and Yibei Ling. Per-flow traffic measurement through randomized counter sharing. IEEE/ACM Transactions on Networking, 20(5):1622– 1634, 2012

  12. [20]

    Sequential dependencies

    Lukasz Golab, Howard Karloff, Flip Korn, Avishek Saha, and Divesh Srivastava. Sequential dependencies. Proceedings of the VLDB Endowment , 2(1):574–585, 2009

  13. [21]

    Effective and complete discovery of order dependencies via set-based axiomatization

    Jaroslaw Szlichta, Parke Godfrey, Lukasz Golab, Mehdi Kargar, and Divesh Sri- vastava. Effective and complete discovery of order dependencies via set-based axiomatization. arXiv preprint arXiv:1608.06169, 2016

  14. [22]

    Data auditor: Exploring data quality and semantics using pattern tableaux

    Lukasz Golab, Howard Karloff, Flip Korn, and Divesh Srivastava. Data auditor: Exploring data quality and semantics using pattern tableaux. Proceedings of the VLDB Endowment, 3(1-2):1641–1644, 2010

  15. [23]

    Issues in data stream management

    Lukasz Golab and M Tamer Özsu. Issues in data stream management. ACM Sigmod Record, 32(2):5–14, 2003

  16. [24]

    Update-pattern-aware modeling and process- ing of continuous queries

    Lukasz Golab and M Tamer Özsu. Update-pattern-aware modeling and process- ing of continuous queries. In Proceedings of the 2005 ACM SIGMOD international conference on Management of data , pages 658–669, 2005

  17. [25]

    New techniques for assessing audio and video quality in real-time interactive communications

    Jim Mullin, Lucy Smallwood, Anna Watson, and Gillian Wilson. New techniques for assessing audio and video quality in real-time interactive communications. IHM-HCI Tutorial, pages 1–63, 2001

  18. [26]

    A performance study of ethernet audio video bridging (avb) for industrial real-time communication

    Jahanzaib Imtiaz, Jürgen Jasperneite, and Lixue Han. A performance study of ethernet audio video bridging (avb) for industrial real-time communication. In 2009 IEEE Conference on Emerging Technologies & Factory Automation , pages 1–8. IEEE, 2009

  19. [27]

    A survey on steganography techniques in real time audio signals and evaluation.International Journal of Computer Science Issues (IJCSI) , 9(1):30, 2012

    Abdulaleem Z Al-Othmani, Azizah Abdul Manaf, and Akram M Zeki. A survey on steganography techniques in real time audio signals and evaluation.International Journal of Computer Science Issues (IJCSI) , 9(1):30, 2012

  20. [28]

    New techniques for assess- ing audio and video quality in real-time interactive communication

    Anna Watson, J Mullin, L Smallwood, and G Wilson. New techniques for assess- ing audio and video quality in real-time interactive communication. Tutorial at IHM-HCI, Lille, France. http://citeseerx. ist. psu. edu/viewdoc/download , 2001

  21. [29]

    Evaluating audio and video quality in low-cost multimedia conferencing systems

    Anna Watson and Martina Angela Sasse. Evaluating audio and video quality in low-cost multimedia conferencing systems. Interacting with computers, 8(3):255– 275, 1996

  22. [30]

    Audio and speech quality survey of the opus codec in web real-time commu- nication

    Oliver Jokisch, Michael Maruschke, Martin Meszaros, and Viktor Iaroshenko. Audio and speech quality survey of the opus codec in web real-time commu- nication. In Proc. 27th Conference on Electronic Speech Signal Processing (ESSV) , volume 3, pages 254–62, 2016

  23. [31]

    Gerardo Rubino. Quantifying the quality of audio and video transmissions over the internet: the psqa approach.Communication Networks And Computer Systems: A Tribute to Professor Erol Gelenbe , pages 235–250, 2006

  24. [32]

    A study of real-time packet video quality using random neural networks

    Samir Mohamed and Gerardo Rubino. A study of real-time packet video quality using random neural networks. IEEE transactions on circuits and systems for video technology, 12(12):1071–1083, 2002

  25. [33]

    Detecting malicious packet dropping in the presence of collisions and channel errors in wireless ad hoc networks

    Thaier Hayajneh, Prashant Krishnamurthy, David Tipper, and Taehoon Kim. Detecting malicious packet dropping in the presence of collisions and channel errors in wireless ad hoc networks. In 2009 IEEE International Conference on Communications, pages 1–6. IEEE, 2009

  26. [34]

    A distributed protocol for detection of packet dropping attack in mobile ad hoc networks

    Jaydip Sen, M Girish Chandra, P Balamuralidhar, SG Harihara, and Harish Reddy. A distributed protocol for detection of packet dropping attack in mobile ad hoc networks. In 2007 IEEE International Conference on Telecommunications and Malaysia International Conference on Communi...

  27. [35]

    Resisting malicious packet drop- ping in wireless ad hoc networks

    Mike Just, Evangelos Kranakis, and Tao Wan. Resisting malicious packet drop- ping in wireless ad hoc networks. In Ad-Hoc, Mobile, and Wireless Networks: Second International Conference, ADHOC-NOW2003, Montreal, Canada, October 8-10, 2003. Proceedings 2 , pages 151–163. Springer, 2003

  28. [36]

    Design an anomaly based fuzzy intrusion detection system for packet dropping attack in mobile ad hoc networks

    Alka Chaudhary, VN Tiwari, and Anil Kumar. Design an anomaly based fuzzy intrusion detection system for packet dropping attack in mobile ad hoc networks. In 2014 IEEE International Advance Computing Conference (IACC) , pages 256–261. IEEE, 2014

  29. [37]

    A novel technique to detect malicious packet dropping attacks in wireless sensor networks

    J Sebastian Terence and Geethanjali Purushothaman. A novel technique to detect malicious packet dropping attacks in wireless sensor networks. Journal of Information Processing Systems, 15(1):203–216, 2019

  30. [38]

    Privacy-preserving and truthful detection of packet dropping attacks in wireless ad hoc networks

    Tao Shu and Marwan Krunz. Privacy-preserving and truthful detection of packet dropping attacks in wireless ad hoc networks. IEEE Transactions on mobile computing, 14(4):813–828, 2014

  31. [39]

    A provenance based mechanism to identify malicious packet dropping adversaries in sensor networks

    Salmin Sultana, Elisa Bertino, and Mohamed Shehab. A provenance based mechanism to identify malicious packet dropping adversaries in sensor networks. In 2011 31st International Conference on Distributed Computing Systems Workshops, pages 332–338. IEEE, 2011

  32. [40]

    An efficient cross-layer approach for malicious packet dropping detection in manets

    Leovigildo S’nchez-Casado, Pedro García-Teodoro, et al. An efficient cross-layer approach for malicious packet dropping detection in manets. In 2012 IEEE 11th International Conference on Trust, Security and Privacy in Computing and Communications, pages 231–238. IEEE, 2012

  33. [41]

    A model of data forwarding in manets for light- weight detection of malicious packet dropping

    Leovigildo Sánchez-Casado, Gabriel Maciá-Fernández, Pedro Garcia-Teodoro, and Roberto Magán-Carrión. A model of data forwarding in manets for light- weight detection of malicious packet dropping. Computer Networks, 87:44–58, 2015

  34. [42]

    A stochastic approach for packet dropping attacks detection in mobile ad hoc networks

    Mohammad Rmayti, Rida Khatoun, Youcef Begriche, Lyes Khoukhi, and Do- minique Gaiti. A stochastic approach for packet dropping attacks detection in mobile ad hoc networks. Computer Networks, 121:53–64, 2017

  35. [43]

    Cross-layer detection of sinking behavior in wireless ad hoc networks using svm and fda

    John Felix Charles Joseph, Bu-Sung Lee, Amitabha Das, and Boon-Chong Seet. Cross-layer detection of sinking behavior in wireless ad hoc networks using svm and fda. IEEE Transactions on Dependable and Secure Computing , 8(2):233–245, 2010

  36. [44]

    Detecting blackhole attack on aodv-based mobile ad hoc networks by dynamic learning method

    Satoshi Kurosawa, Hidehisa Nakayama, Nei Kato, Abbas Jamalipour, and Yoshiaki Nemoto. Detecting blackhole attack on aodv-based mobile ad hoc networks by dynamic learning method. Int. J. Netw. Secur., 5(3):338–346, 2007

  37. [45]

    New directions in traffic measurement and accounting: Focusing on the elephants, ignoring the mice

    Cristian Estan and George Varghese. New directions in traffic measurement and accounting: Focusing on the elephants, ignoring the mice. ACM Trans. Comput. Syst., 2003

  38. [46]

    Elastic sketch: adaptive and fast network- wide measurements

    Tong Yang, Jie Jiang, Peng Liu, and etal. Elastic sketch: adaptive and fast network- wide measurements. In SIGCOMM, 2018

  39. [47]

    Heavykeeper: An accurate algorithm for finding top-𝑘 elephant flows

    Tong Yang, Haowei Zhang, Jinyang Li, Junzhi Gong, Steve Uhlig, Shigang Chen, and Xiaoming Li. Heavykeeper: An accurate algorithm for finding top-𝑘 elephant flows. IEEE/ACM Transactions on Networking, 27(5):1845–1858, 2019

  40. [48]

    https://catalog.caida.org/dataset/passive_ 2016_pcap, 2016

    Anonymized Internet Traces 2016. https://catalog.caida.org/dataset/passive_ 2016_pcap, 2016

  41. [49]

    http://mawi.wide.ad.jp/mawi/

    MAWI Working Group Traffic Archive. http://mawi.wide.ad.jp/mawi/

  42. [50]

    https://www.netresec.com/?page= MACCDC

    Capture files from Mid-Atlantic CCDC . https://www.netresec.com/?page= MACCDC

  43. [51]

    Network traffic charac- teristics of data centers in the wild

    Theophilus Benson, Aditya Akella, and David A Maltz. Network traffic charac- teristics of data centers in the wild. In𝑃𝑟𝑜𝑐. IMC, pages 267–280, 2010

  44. [52]

    http://burtleburtle.net/bob/hash/evahash.html

    Hash website. http://burtleburtle.net/bob/hash/evahash.html

  45. [53]

    Cuckoo filter: Practically better than bloom

    Bin Fan, Dave G Andersen, Michael Kaminsky, and Michael D Mitzenmacher. Cuckoo filter: Practically better than bloom. In Proceedings of the 10th ACM International on Conference on emerging Networking Experiments and Technologies, pages 75–88, 2014

  46. [54]

    Cuckoo hashing

    Rasmus Pagh and Flemming Friche Rodler. Cuckoo hashing. Journal of Algo- rithms, 51(2):122–144, 2004. 14

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.