A False Sense of Security?: Revisiting the State of Machine Learning-Based Industrial Intrusion Detection
Open accessThe DOI matches its Crossref record; 40 of the 41 references it cites were found in a registry or an index.
Abstract
Anomaly-based intrusion detection promises to detect novel or unknown attacks on industrial control systems by modeling expected system behavior and raising corresponding alarms for any deviations. As manually creating these behavioral models is tedious and error-prone, research focuses on machine learning to train them automatically, achieving detection rates upwards of 99%. However, these approaches are typically trained not only on benign traffic but also on attacks and then evaluated against the same type of attack used for training. Hence, their actual, real-world performance on unknown (not trained on) attacks remains unclear. In turn, the reported near-perfect detection rates of machine learning-based intrusion detection might create a false sense of security. To assess this situation and clarify the real potential of machine learning-based industrial intrusion detection, we develop an evaluation methodology and examine multiple approaches from literature for their performance on unknown attacks (excluded from training). Our results highlight an ineffectiveness in detecting unknown attacks, with detection rates dropping to between 3.2% and 14.7% for some types of attacks. Moving forward, we derive recommendations for further research on machine learning-based approaches to ensure clarity on their ability to detect unknown attacks.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“This empirically demonstrates that many ML-based IDS solutions behave more like implicit signature-based detectors that memorize trained attack patterns, rather than robust models of normal behavior, and thus exhibit limited generalization capability to unknown attacks.”
From AI-Based Anomaly Detection in Industrial Control and Cyber–Physical Systems: A Data-Type-Oriented Systematic Review (2025) · cites this paper as [66]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Applying RF, SVM, and BLSTM-based IDS models to the Gas Pipeline dataset, they showed that although high F1 scores are achieved for attacks seen during training, recall for unseen attacks in some attack families deteriorates to as low as 3.2-14.7%.”
From AI-Based Anomaly Detection in Industrial Control and Cyber–Physical Systems: A Data-Type-Oriented Systematic Review (2025) · cites this paper as [66]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Their experiments, in which certain attack types were deliberately excluded from training, showed that recall for unseen attack families could drop to as low as 3.2-14.7%.”
From AI-Based Anomaly Detection in Industrial Control and Cyber–Physical Systems: A Data-Type-Oriented Systematic Review (2025) · cites this paper as [66]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“As pointed out by Kus et al. [66] in Section 4.3.6, many network-based ML models implicitly learn the 'signatures' of attacks present in the training set rather than truly modeling the distribution of normal behavior.”
From AI-Based Anomaly Detection in Industrial Control and Cyber–Physical Systems: A Data-Type-Oriented Systematic Review (2025) · cites this paper as [66]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“This addresses precisely the type of stealth, and zero-day attacks that traditional, purely statistical IDS tend to miss, and directly mitigates the limitations highlighted by Kus et al. [66].”
From AI-Based Anomaly Detection in Industrial Control and Cyber–Physical Systems: A Data-Type-Oriented Systematic Review (2025) · cites this paper as [66]
Experimental — accuracy not yet measured
4 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Outside the Closed World: On Using Machine Learning for Network Intrusion Detection
Sommer, R., & Paxson, V. (2010) · 2010 IEEE Symposium on Security and Privacy · doi:10.1109/sp.2010.25 · Cited by 2,085
Survey of intrusion detection systems: Techniques, datasets and challenges
Khraisat, A., Gondal, I., Vamplew, P., & Kamruzzaman, J. (2019) · Cybersecurity · doi:10.1186/s42400-019-0038-7 · Cited by 2,024
Cyber‐physical attacks and defences in the smart grid: a survey
He, H., & Yan, J. (2016) · IET Cyber-Physical Systems: Theory & Applications · doi:10.1049/iet-cps.2016.0019 · Cited by 454
Cybersecurity for industrial control systems: A survey
Bhamare, D., Zolanvari, M., Erbad, A., Jain, R., Khan, K., & Meskin, N. (2020) · Computers & Security · doi:10.1016/j.cose.2019.101677 · Cited by 394
The Cybersecurity Landscape in Industrial Control Systems
McLaughlin, S., Konstantinou, C., Wang, X., Davi, L., Sadeghi, A.-R., Maniatakos, M., & Karri, R. (2016) · Proceedings of the IEEE · doi:10.1109/jproc.2015.2512235 · Cited by 391
Challenges and Opportunities in Securing the Industrial Internet of Things
Serror, M., Hack, S., Henze, M., Schuba, M., & Wehrle, K. (2021) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2020.3023507 · Cited by 362
Self-citationFuture research challenges in wireless sensor and actuator networks targeting industrial automation
Akerberg, J., Gidlund, M., & Bjorkman, M. (2011) · 2011 9th IEEE International Conference on Industrial Informatics · doi:10.1109/indin.2011.6034912 · Cited by 343
Generation of a new IDS test dataset: Time to retire the KDD collection
Creech, G., & Hu, J. (2013) · 2013 IEEE Wireless Communications and Networking Conference (WCNC) · doi:10.1109/wcnc.2013.6555301 · Cited by 320
An Ensemble Deep Learning-Based Cyber-Attack Detection in Industrial Control System
Al-Abassi, A., Karimipour, H., Dehghantanha, A., & Parizi, R. M. (2020) · IEEE Access · doi:10.1109/access.2020.2992249 · Cited by 320
Toward Credible Evaluation of Anomaly-Based Intrusion-Detection Methods
Tavallaee, M., Stakhanova, N., & Ghorbani, A. A. (2010) · IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews) · doi:10.1109/tsmcc.2010.2048428 · Cited by 280
Deep Learning-based Anomaly Detection in Cyber-physical Systems: Progress and Opportunities
Luo, Y., Xiao, Y., Cheng, L., Peng, G., & Yao, D. (. (2022) · ACM Computing Surveys · doi:10.1145/3453155 · Cited by 268
Multi-level Anomaly Detection in Industrial Control Systems via Package Signatures and LSTM Networks
Feng, C., Li, T., & Chana, D. (2017) · 2017 47th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN) · doi:10.1109/dsn.2017.34 · Cited by 231
Mokhtari, S., Abbaspour, A., Yen, K. K., & Sargolzaei, A. (2021) · Electronics · doi:10.3390/electronics10040407 · Cited by 197
A survey of intrusion detection on industrial control systems
Hu, Y., Yang, A., Li, H., Sun, Y., & Sun, L. (2018) · International Journal of Distributed Sensor Networks · doi:10.1177/1550147718794615 · Cited by 185
Zhou, C., Huang, S., Xiong, N., Yang, S.-H., Li, H., Qin, Y., & Li, X. (2015) · IEEE Transactions on Systems, Man, and Cybernetics: Systems · doi:10.1109/tsmc.2015.2415763 · Cited by 168
Refaeilzadeh, P., Tang, L., & Liu, H. (2016) · Springer eBooks · doi:10.1007/978-1-4899-7993-3_565-2 · Cited by 168
Umer, M. A., Junejo, K. N., Jilani, M. T., & Mathur, A. P. (2022) · International Journal of Critical Infrastructure Protection · doi:10.1016/j.ijcip.2022.100516 · Cited by 164
Towards an Infrastructure Enabling the Internet of Production
Pennekamp, J., Glebke, R., Henze, M., Meisen, T., Quix, C., Hai, R., Gleim, L., Niemietz, P., Rudack, M., Knape, S., et al. (2019) · 2019 IEEE International Conference on Industrial Cyber Physical Systems (ICPS) · doi:10.1109/icphys.2019.8780276 · Cited by 135
Self-citationRaman, M. R. G., Ahmed, C. M., & Mathur, A. P. (2021) · Cybersecurity · doi:10.1186/s42400-021-00095-5 · Cited by 125
A Systematic Framework to Generate Invariants for Anomaly Detection in Industrial Control Systems
Feng, C., Palleti, V. R., Mathur, A., & Chana, D. (2019) · Proceedings 2019 Network and Distributed System Security Symposium · doi:10.14722/ndss.2019.23265 · Cited by 120
Behaviour-Based Attack Detection and Classification in Cyber Physical Systems Using Machine Learning
Junejo, K. N., & Goh, J. (2016) · Proceedings of the 2nd ACM International Workshop on Cyber-Physical System Security · doi:10.1145/2899015.2899016 · Cited by 109
Truth Will Out: Departure-Based Process-Level Detection of Stealthy Attacks on Control Systems
Aoudi, W., Iturbe, M., & Almgren, M. (2018) · Proceedings of the 2018 ACM SIGSAC Conference on Computer and Communications Security · doi:10.1145/3243734.3243781 · Cited by 106
An Autocorrelation-based LSTM-Autoencoder for Anomaly Detection on Time-Series Data
Homayouni, H., Ghosh, S., Ray, I., Gondalia, S., Duggan, J., & Kahn, M. G. (2020) · 2020 IEEE International Conference on Big Data (Big Data) · doi:10.1109/bigdata50022.2020.9378192 · Cited by 66
Machine Learning for Reliable Network Attack Detection in SCADA Systems
Lopez Perez, R., Adamsky, F., Soua, R., & Engel, T. (2018) · 2018 17th IEEE International Conference On Trust, Security And Privacy In Computing And Communications/ 12th IEEE International Conference On Big Data Science And Engineering (TrustCom/BigDataSE) · doi:10.1109/trustcom/bigdatase.2018.00094 · Cited by 62
Secure Low Latency Communication for Constrained Industrial IoT Scenarios
Hiller, J., Henze, M., Serror, M., Wagner, E., Richter, J. N., & Wehrle, K. (2018) · Conference on Local Computer Networks · doi:10.1109/lcn.2018.8638027 · Cited by 59
Self-citationA Deep Learning-based Framework for Conducting Stealthy Attacks in Industrial Control SystemsPreprint
Feng, C., Li, T., Zhu, Z., & Chana, D. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1709.06397 · Cited by 58
A three-tiered intrusion detection system for industrial control systems
Anthi, E., Williams, L., Burnap, P., & Jones, K. (2021) · Journal of Cybersecurity · doi:10.1093/cybsec/tyab006 · Cited by 46
Reproducible and Adaptable Log Data Generation for Sound Cybersecurity Experiments
Uetz, R., Hemminghaus, C., Hackländer, L., Schlipper, P., & Henze, M. (2021) · Annual Computer Security Applications Conference · doi:10.1145/3485832.3488020 · Cited by 43
Self-citationEvaluation of Machine Learning Algorithms Used on Attacks Detection in Industrial Control Systems
Arora, P., Kaur, B., & Teixeira, M. A. (2021) · Journal of The Institution of Engineers (India) Series B · doi:10.1007/s40031-021-00563-z · Cited by 41
Machine Learning Algorithm for Detection of False Data Injection Attack in Power System
Kumar, A., Saxena, N., & Choi, B. J. (2021) · International Conference on Information Networking · doi:10.1109/icoin50884.2021.9333913 · Cited by 33
Industrial Control Intrusion Detection Approach Based on Multiclassification GoogLeNet-LSTM Model
Chu, A., Lai, Y., & Liu, J. (2019) · Security and Communication Networks · doi:10.1155/2019/6757685 · Cited by 32
Anomaly Detection in Critical Infrastructure Using Probabilistic Neural Network
Raman, M. R. G., Somu, N., & Mathur, A. P. (2019) · Communications in computer and information science · doi:10.1007/978-981-15-0871-4_10 · Cited by 29
Anomaly-Based Intrusion Detection System for Cyber-Physical System Security
Colelli, R., Magri, F., Panzieri, S., & Pascucci, F. (2021) · Mediterranean Conference on Control and Automation (MED) · doi:10.1109/med51440.2021.9480182 · Cited by 28
Security threats of Internet-reachable ICS
Abe, S., Fujimoto, M., Horata, S., Uchida, Y., & Mitsunaga, T. (2016) · Annual Conference of the Society of Instrument and Control Engineers of Japan (SICE) · doi:10.1109/sice.2016.7749239 · Cited by 27
Umer, M., Ahmed, C. M., Jilani, M. T., & Mathur, A. P. (2021) · Strathprints: The University of Strathclyde institutional repository (University of Strathclyde) · doi:10.48550/arxiv.2107.05127 · Cited by 22
Domain-Based Fuzzing for Supervised Learning of Anomaly Detection in Cyber-Physical Systems
Wijaya, H., Aniche, M. F., & Mathur, A. P. (2020) · doi:10.1145/3387940.3391486 · Cited by 20
The Nuts and Bolts of Deploying Process-Level IDS in Industrial Control Systems
Almgren, M., Aoudi, W., Gustafsson, R., Krahl, R., & Lindhé, A. (2018) · Annual Industrial Control System Security Workshop (ICSS) · doi:10.1145/3295453.3295456 · Cited by 17
Erba, A., & Tippenhauer, N. O. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.03586 · Cited by 15
The Art of Intrusion Detection
(2009) · Springer eBooks · doi:10.1007/978-3-540-79698-5_9 · Cited by 3
Kharitonov, A. N., & Zimmermann, A. P. (2019) · doi:10.1145/3327961.3329531 · Cited by 2
Umer, M., Ahmed, C. M., Jilani, M. T., & Mathur, A. P. (2021) · doi:10.1145/3462633.3483976
Research works printed but not linked: 18
- [2] Chuadhry Mujeeb Ahmed et al. 2020. Challenges in Machine Learning Based Approaches for Real-Time Anomaly Detection in Industrial Control Systems. In ACM CPSS . -
- [8] Simon D. Duque Anton et al. 2019. Anomaly-based Intrusion Detection in Industrial Data with SVM and Random Forests. In SoftCOM . -
- [11] Daniel Arp et al. 2022. Dos and Don'ts of Machine Learning in Computer Security. In USENIX SEC . -
- [13] Yuqi Chen et al. 2018. Learning from Mutants: Using Code Mutation to Learn and Monitor Invariants of a Cyber-Physical System. In IEEE SP . -
- [17] Markus Dahlmanns et al. 2020. Easing the Conscience with OPC UA: An InternetWide Study on Insecure Deployments. In ACM IMC . -
- [18] Markus Dahlmanns et al. 2022. Missed Opportunities: Measuring the Untapped TLS Support in the Industrial Internet of Things. In ACM ASIACCS . -
- [23] Jonathan Goh et al. 2017. Anomaly Detection in Cyber Physical Systems Using Recurrent Neural Networks. In IEEE HASE . -
- [29] Veeramreddy Jyothsna et al. 2011. A Review of Anomaly based Intrusion Detection Systems. Int. J. Comput. Appl. 28, 7. -
- [32] Tim Krause et al. 2021. Cybersecurity in Power Grids: Challenges and Opportunities. Sensors 21, 18. -
- [34] Qin Lin et al. 2018. TABOR: A Graphical Model-based Approach for Anomaly Detection in Industrial Control Systems. In ACM ASIACCS . -
- [39] Thomas H. Morris et al. 2015. Industrial control system simulation and data logging for intrusion detection system research. In SCSS . -
- [41] Feargus Pendlebury et al. 2019. TESSERACT: Eliminating Experimental Bias in Malware Classification across Space and Time. In USENIX SEC . -
- [42] Jan Pennekamp et al. 2021. Collaboration is not Evil: A Systematic Look at Security Research for Industrial Use. In LASER . -
- [44] Panagiotis Radoglou-Grammatikis et al. 2020. DIDEROT: An Intrusion Detection and Prevention System for DNP3-Based SCADA Systems. In ARES . -
- [49] Robin Sommer. 2003. Bro: An open source network intrusion detection system. In Security, E-learning, E-Services, 17. DFN-Arbeitstagung über Kommunikationsnetze . -
- [56] David I. Urbina et al. 2016. Limiting the Impact of Stealthy Attacks on Industrial Control Systems. In ACM CCS . -
- [59] Konrad Wolsing et al. 2020. Poster: Facilitating Protocol-independent Industrial Intrusion Detection Systems. In ACM CCS . -
- [60] Konrad Wolsing et al. 2021. IPAL: Breaking up Silos of Protocol-dependent and Domain-specific Industrial Intrusion Detection Systems. arXiv:2111.03438. - · arXiv
Cited by
Deep Learning vs. Machine Learning for Intrusion Detection in Computer Networks: A Comparative Study
Ali, M. L., Thakur, K., Schmeelk, S., DeBello, J. E., & Dragos, D. M. (2025) · Applied Sciences · doi:10.3390/app15041903 · Cited by 165
“Unlike traditional rule-based approaches, which rely on static signatures and predefined rules, machine learning-based IDS analyze network traffic data dynamically [17].”
Wolsing, K., Wagner, E., Saillard, A., & Henze, M. (2022) · International Symposium on Research in Attacks, Intrusions and Defenses (RAID) · doi:10.1145/3545948.3545968 · Cited by 55
“Second, IIDS research across different industrial domains suffers from an evaluation bias, where vast inconsistencies in used evaluation methodologies make comparisons across publications hardly possible [49, 78, 114].”
BackgroundSelf-citationApplications of Machine Learning in Cyber Security: A Review
Vourganas, I., & Michala, A. L. (2024) · Journal of Cybersecurity and Privacy · doi:10.3390/jcp4040045 · Cited by 35
Simulation of multi-stage attack and defense mechanisms in smart grids
Sen, Ö., Ivanov, B., Kloos, C., Zöll, C., Lutat, P., Henze, M., Ulbig, A., & Andres, M. (2024) · International Journal of Critical Infrastructure Protection · doi:10.1016/j.ijcip.2024.100727 · Cited by 31
“ML-based IDSs employ sophisticated methods for detecting anomalies [9].”
Self-citationTechniques for Enhancing Security in Industrial Control Systems
Varadharajan, V., Tupakula, U. K., & Karmakar, K. K. (2024) · ACM Transactions on Cyber-Physical Systems · doi:10.1145/3630103 · Cited by 30
Metadata
- Authors
- Kus, D., Wagner, E., Pennekamp, J., Wolsing, K., Fink, I. B., Dahlmanns, M., Wehrle, K., & Henze, M.
- Affiliations
- RWTH Aachen University (DE)
- Fraunhofer Institute for Communication, Information Processing and Ergonomics (DE)
- ORCID
- Venue
- Proceedings of the 8th ACM on Cyber-Physical System Security Workshop
- Publisher
- ACM
- Published
- 2022-05-30
- Pages
- 73-84
- OpenAlex
- W4280520332
- Cited by
- 28
- References
- 38
- Citations by year
- 2026: 1 · 2025: 5 · 2024: 9 · 2023: 10 · 2022: 3
- FWCI
- 6.54
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- ACM Workshop on Cyber-Physical System Security (CPSS) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Fraunhofer-Publica (Fraunhofer-Gesellschaft) (submittedVersion)
- RWTH Publications (RWTH Aachen) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Deutsche Forschungsgemeinschaft (EXC-2023 - 390621612)
- Topics
- Physical Sciences › Computer Science › Computer Networks and Communications › Network Security and Intrusion Detection
- Physical Sciences › Engineering › Control and Systems Engineering › Smart Grid Security and Resilience
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Immune Systems Applications
- Keywords
- anomaly-based intrusion detection, industrial control systems, unknown attack detection, machine learning, behavioral modeling, evaluation methodology, detection rate, false sense of security, industrial cybersecurity
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
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.