Duplicate-Aware Normal-Only Anomaly Detection for Wi-Fi/MQTT IoMT Traffic
International Journal of Computer Science and Modern Engineering · 09.11.2026 · 1 (1) · 1-20
DOI: 10.4018/ijcsme.426864 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computer Science and Modern Engineering, erschienen am 09.11.2026. Verfasst von Rakan AlTome. Die Arbeit nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 3.0).
Abstract (Originalsprache)
This paper presents a duplicate-aware, normal-only anomaly-detection protocol for Wi-Fi/MQTT IoMT traffic using CICIoMT2024. Its primary contribution is an auditable evaluation procedure rather than a new detector: features are derived only from profiling-normal traffic, duplicate vectors and cross-pool overlaps are audited, normal splits are hash-disjoint, preprocessing and training use normal data only, and the threshold is calibrated on held-out normal validation data. Under the deduplicated protocol, OCSVM achieves F1 = 0.9989, MCC = 0.9737, balanced accuracy = 0.9751, FPR = 0.04976, and zero false negatives across seven attack families. However, logistic-regression and IAT-only diagnostics reveal strong profiling-versus-attack separability; IAT alone reaches F1 = 0.9886. The results are therefore interpreted as a leakage-controlled benchmark for this CICIoMT2024 branch, not evidence of universal attack generalization. False positives also vary substantially across normal operating modes.
Zitieren
APA
AlTome, R. (2026). Duplicate-Aware Normal-Only Anomaly Detection for Wi-Fi/MQTT IoMT Traffic. International Journal of Computer Science and Modern Engineering, 1(1), 1-20. https://doi.org/10.4018/ijcsme.426864
Vancouver
AlTome R. Duplicate-Aware Normal-Only Anomaly Detection for Wi-Fi/MQTT IoMT Traffic. International Journal of Computer Science and Modern Engineering. 2026;1(1):1-20. doi:10.4018/ijcsme.426864
BibTeX
@article{AlTome2026,
author = {AlTome, Rakan},
title = {Duplicate-Aware Normal-Only Anomaly Detection for Wi-Fi/MQTT IoMT Traffic},
journal = {International Journal of Computer Science and Modern Engineering},
year = {2026},
volume = {1},
number = {1},
pages = {1--20},
doi = {10.4018/ijcsme.426864}
}
EUVIMED