Access to this article is restricted until 12 months after publication by request of the publisher. Restriction lift date: 2026-07-10
Poster: Using machine learning to infer network structure from security metadata
Loading...
Files
Date
2025-07-10
Authors
Khalid, Asfa
Murphy, Seán Óg
Sreenan, Cormac J.
Roedig, Utz
Journal Title
Journal ISSN
Volume Title
Publisher
Springer Nature Switzerland
Published Version
Abstract
In distributed cloud-edge environments, data-driven decision-making is essential for enhancing operational efficiency and maintaining a competitive advantage. Achieving this requires strong guarantees of data integrity and authenticity, as any compromise can lead to inaccurate insights, loss of trust, and financial damage. To address the cybersecurity risks posed by data transmission across complex, heterogeneous networks, Data Confidence Fabrics have been introduced. These enhance data security by generating metadata at each stage of transmission and storing it using distributed ledgers, which ensures the immutability and verifiability of this metadata. However, despite these benefits, the public accessibility of ledgers introduces significant privacy concerns. While previous research has focused on hostname obfuscation to protect network structure, timestamps often remain exposed, creating an exploitable vulnerability. We demonstrate that one can use K-means clustering on exposed timestamp patterns to reconstruct the obfuscated network structure, even when hostnames are fully obfuscated. Our findings reveal a critical gap in existing metadata protection mechanisms and highlight the need for defense against timestamp-based inference attacks.
Description
Keywords
Data Confidence Fabrics , K-means , Machine learning , Network structure , Security metadata
Citation
Khalid, A., Murphy, S. Ó., Sreenan, C. J., Roedig, U. (2025) 'Poster: Using machine learning to infer network structure from security metadata', in Egele, M., Moonsamy, V., Gruss, D. and Carminati, M. (eds.) Detection of Intrusions and Malware, and Vulnerability Assessment. DIMVA 2025. Lecture Notes in Computer Science, vol 15748, pp. 93-99. Springer, Cham. https://doi.org/10.1007/978-3-031-97623-0_6
Link to publisher’s version
Collections
Copyright
© 2025 The Author(s), under exclusive license to Springer Nature Switzerland AG.