EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Machine Learning-Based Risk-Adaptive Attribute- Based Access Control Framework for Secure IoT Networks

Rashmin Prajapati, Sweta S. Panchal, Neha Soni, Sandipkumar R. Panchal

International Research Journal of Multidisciplinary Technovation · 30.05.2026 · 495-519

DOI: 10.54392/irjmt26329 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Research Journal of Multidisciplinary Technovation, online erschienen am 30.05.2026. Verfasst von Rashmin Prajapati und 3 weiteren Autor:innen. Die Arbeit nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Internet of Things (IoT) communication networks have been growing rapidly, which exposes security threats to the heterogeneity of the devices, changing circumstances, and changing attack patterns. Classical Role-Based Access Control (RBAC) and attribute-based access control (ABAC) models are not flexible and do not give real-time risk indicators and responsiveness to decision making. In this paper, a risk-adaptive attribute-based access control (RAd-ABAC) framework based on machine learning is proposed as a part of a secure IoT communication system. The model combines the modeling of subject, object, and environmental attribute with dynamic risk scoring performed by a supervised machine learning model. A ternary authorization scheme, Permit, Review, and Deny, is presented to enhance the decision granularity as well as be able to respond to borderline access requests better than binary authorization schemes. IoT23 and TON IoT benchmark datasets containing 45,000 samples converted into ABAC compatible feature vectors were used in the proposed framework evaluation. The experimental findings indicate that the proposed RAd-ABAC framework with a proposed accuracy of 94.6 percent, macro-average AUC of 0.97 and a false positive rate of 4.2. The false positive rate decreased by 8.3 percentage points, reaching 4.2%, which corresponds to a 66.4% relative reduction than the Static ABAC and is considered an absolute change of 8.3 and a relative change of 66.4. The scalability analysis also demonstrated near-linear scaling in processing performance of average latency of less than 5.4 ms per request. Results show that risk scoring using ML has the potential of enhancing the accuracy, flexibility, and reliability of authorization in distributed IoT communication systems.

Zitieren

APA

Prajapati, R., Panchal, S. S., Soni, N., & Panchal, S. R. (2026). Machine Learning-Based Risk-Adaptive Attribute- Based Access Control Framework for Secure IoT Networks. International Research Journal of Multidisciplinary Technovation, 495-519. https://doi.org/10.54392/irjmt26329

Vancouver

Prajapati R, Panchal SS, Soni N, Panchal SR. Machine Learning-Based Risk-Adaptive Attribute- Based Access Control Framework for Secure IoT Networks. International Research Journal of Multidisciplinary Technovation. 2026:495-519. doi:10.54392/irjmt26329

BibTeX
@article{Prajapati2026,
  author = {Prajapati, Rashmin and Panchal, Sweta S. and Soni, Neha and Panchal, Sandipkumar R.},
  title = {Machine Learning-Based Risk-Adaptive Attribute- Based Access Control Framework for Secure IoT Networks},
  journal = {International Research Journal of Multidisciplinary Technovation},
  year = {2026},
  pages = {495--519},
  doi = {10.54392/irjmt26329}
}
Kostenlos starten →