Machine learning-based framework for dynamic risk assessment and early warning in wireless network security environments
Data and Metadata · 13.05.2026 · 5 · 803
DOI: 10.56294/dm2026803 ↗Kurz gefasst
Fachartikel in der Zeitschrift Data and Metadata, online erschienen am 13.05.2026. Verfasst von Hua Tian. Die Arbeit nennt 19 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Introduction
Wireless Sensor Networks (WSNs) are integral to modern digital infrastructures, enabling real-time communication among devices, sensors, and control systems, but remain vulnerable to complex cyberattacks. Existing research on wireless network security is frequently restricted in its capacity to react to new attack patterns and has a high false alarm rate in dynamic networks.
Objective
This research aims to develop a dynamic evaluation and early warning system for wireless network security risks using a hybrid machine learning framework.
Method
The proposed KMA-ERSM model combines the Komodo Mlipir Algorithm (KMA) and Efficient Random Support Machine (ERSM), which integrates Enhanced Random Forest (ERF) and Support Vector Machine (SVM) for precise classification. Network traffic data were collected from the CSE-CIC-IDS-2018 and CIC-IDS-2017 datasets, covering both benign and malicious flows. Data preprocessing included K-Nearest Neighbour (KNN) imputation and Min-Max normalization to ensure complete, normalized data. Principal Component Analysis (PCA) was applied to extract key patterns representing anomalous network behavior. The KMA dynamically selects influential features and tunes parameters. Classification of Network Traffic Using SVM, WSN Security Classification using ERF. Finally, ERSM identifies complex attack patterns and establishes accurate classification boundaries.
Results
Experimental results demonstrate that the proposed KMA-ERSM outperforms existing models. KMA-ERSM achieved 98,4 % accuracy, 98,8 % F1-score, 98,3 % recall, 99,2 % precision[user2.1], 97,5 % AUC-ROC, and 2 % FPR.
Conclusion
The hybrid framework effectively captures complex traffic patterns, ensuring reliable detection of malicious and benign activities while supporting adaptive, real-time risk alerts. This approach enhances proactive threat detection, reduces false alarms, and strengthens resilience in dynamic wireless network environments.
Zitieren
APA
Tian, H. (2026). Machine learning-based framework for dynamic risk assessment and early warning in wireless network security environments. Data and Metadata, 5, 803. https://doi.org/10.56294/dm2026803
Vancouver
Tian H. Machine learning-based framework for dynamic risk assessment and early warning in wireless network security environments. Data and Metadata. 2026;5:803. doi:10.56294/dm2026803
BibTeX
@article{Tian2026,
author = {Tian, Hua},
title = {Machine learning-based framework for dynamic risk assessment and early warning in wireless network security environments},
journal = {Data and Metadata},
year = {2026},
volume = {5},
pages = {803},
doi = {10.56294/dm2026803}
}
EUVIMED