Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Unsupervised Anomaly Detection Framework for Multimodal Data in Industrial Control Systems.

Yunsung Kim1, Gyeongdeok An1, Kihyun Kim2

  • 1Department of Information Security, Hoseo University, Asan-Si 31499, Republic of Korea.

Sensors (Basel, Switzerland)
|June 26, 2026
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

Classification of Systems-I01:26

Classification of Systems-I

Linearity is a system property characterized by a direct input-output relationship, combining homogeneity and additivity.
Homogeneity dictates that if an input x(t) is multiplied by a constant c, the output y(t) is multiplied by the same constant. Mathematically, this is expressed as:

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Clock Glitch Fault Attacks on Deep Neural Networks and Their Countermeasures.

Sensors (Basel, Switzerland)·2025
Same author

SSIM-Based Autoencoder Modeling to Defeat Adversarial Patch Attacks.

Sensors (Basel, Switzerland)·2024
See all related articles

This study introduces an unsupervised multimodal anomaly detection framework for industrial control systems (ICSs). Latent feature fusion effectively integrates sensor and network data, improving detection of cyber-physical threats.

Area of Science:

  • Cyber-Physical Systems Security
  • Machine Learning for Anomaly Detection
  • Industrial Control Systems (ICSs)

Background:

  • Industrial control systems (ICSs) generate both physical process and network data.
  • Existing anomaly detection methods often focus on single data sources (sensor or network), limiting threat detection.
  • Multimodal approaches are needed to leverage complementary information from diverse data streams.

Purpose of the Study:

  • To propose an unsupervised multimodal anomaly detection framework for ICSs.
  • To investigate and compare anomaly score fusion and latent feature fusion strategies.
  • To enhance the detection of diverse cyber-physical attack indicators.

Main Methods:

  • Developed autoencoder-based single-modality models for sensor and network data.
Keywords:
anomaly detectionautoencodercross-modal complementarityindustrial control system (ICS)multimodal datasensor–network fusiontime synchronizationunsupervised learning

Related Experiment Videos

  • Extracted anomaly scores and latent features, then temporally aligned them.
  • Implemented and compared anomaly score fusion and latent feature fusion, including CCA-derived features for the latter.

    • Main Results:

    • Latent feature fusion demonstrated stable and superior performance across various sensor-network encoder combinations.
    • The GRU-CNN combination achieved a high F1-score of 0.9166 and ROC-AUC of 0.9795.
    • Complementarity analysis confirmed that latent feature fusion successfully integrated evidence from both modalities, improving missed detection rates.

    Conclusions:

    • Latent feature fusion is an effective multimodal strategy for unsupervised anomaly detection in ICSs.
    • The proposed framework enhances the ability to capture diverse attack indicators by integrating sensor and network data.
    • This approach offers a robust solution for securing critical industrial infrastructure against sophisticated cyber threats.