ZORAN-Sec: A Zero Trust–Compliant Intrusion Detection Framework for Open RAN With a Lightweight Transformer
IEEE Open Journal of the Communications Society · 2026 · 7 · 6308-6333
DOI: 10.1109/ojcoms.2026.3699599 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Open Journal of the Communications Society, 2026. Verfasst von Loreen Mahmoud und 2 weiteren Autor:innen. Die Arbeit nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Information Security Education and Awareness (ISEA) Phase-3 Project, Ministry of Electronics and Information Technology (MeitY), Government of India, in July 2024.
Abstract (Originalsprache)
Open Radio Access Networks (O-RAN) address vendor lock-in and limited flexibility in traditional Radio Access Networks by introducing open interfaces, disaggregation, and programmability. While these properties improve scalability and interoperability, they also significantly expand the security attack surface of the RAN. Zero Trust Architecture (ZTA) has emerged as a suitable security paradigm for O-RAN, as it enforces continuous verification, least privilege, and adaptive policy enforcement. However, existing ZTA-enabled O-RAN security solutions remain limited by three key challenges: intrusion detection methods that operate on individual samples and ignore temporal attack context, inference latency that violates near–real-time control requirements, and high computational overhead that prevents deployment on resource-constrained edge components. This paper presents ZORAN-Sec, a Zero Trust–compliant intrusion detection framework that addresses these gaps through system-level co-design rather than isolated algorithmic improvements. ZORAN-Sec operationalises ZTA’s Policy Enforcement, Decision, and Information Points by explicitly mapping them to O-RAN entities and coupling detection outputs to adaptive policy enforcement. At its core, we introduce LightSecFormer, a lightweight encoder-only transformer trained from scratch on tabular sequential O-RAN traffic, avoiding the domain mismatch of NLP-derived distilled or pruned variants, with linear embedding and CLS-token pooling calibrated for edge constraints. Empirical evaluation across four heterogeneous datasets (5G-NIDD, NetsLab-5G-ORAN-IDD v1/v2, and N-BaIoT) demonstrates strong detection fidelity in binary and multiclass settings, with inference latency well below O-RAN near-real-time control constraints and competitive or superior performance relative to established baselines. Statistical validation across ten independent runs confirms that performance gains are reproducible and not attributable to favourable initialisation. Cross-dataset generalisation experiments reveal strong forward transferability across evolving traffic distributions, and adversarial robustness analysis under FGSM and PGD perturbations demonstrates graceful degradation under increasing attack strength. Separately, a prototype Near-RT RIC testbed deployment using a real srsRAN gNB and FlexRIC Near-RT RIC confirms operational compliance with near-real-time constraints under live E2SM-KPM telemetry ingestion, with measured P99 xApp processing latency of 0.75 ms ( $13\times $ below the 10 ms Near-RT RIC budget) and stable resource consumption of 501 MB RAM, and mean CPU usage under half a CPU core. Full live O-RAN deployment with over-the-air RF validation remains a direction for future work. Together, these results substantiate architectural viability, policy-driven enforcement integration, and operational robustness within O-RAN security workflows.
Förderung
- Information Security Education and Awareness (ISEA) Phase-3 Project
- Ministry of Electronics and Information Technology (MeitY), Government of India, in July 2024
Zitieren
APA
Mahmoud, L., Liyanage, M., & Gangopadhyay, S. (2026). ZORAN-Sec: A Zero Trust–Compliant Intrusion Detection Framework for Open RAN With a Lightweight Transformer. IEEE Open Journal of the Communications Society, 7, 6308-6333. https://doi.org/10.1109/ojcoms.2026.3699599
Vancouver
Mahmoud L, Liyanage M, Gangopadhyay S. ZORAN-Sec: A Zero Trust–Compliant Intrusion Detection Framework for Open RAN With a Lightweight Transformer. IEEE Open Journal of the Communications Society. 2026;7:6308-6333. doi:10.1109/ojcoms.2026.3699599
BibTeX
@article{Mahmoud2026,
author = {Mahmoud, Loreen and Liyanage, Madhusanka and Gangopadhyay, Sugata},
title = {ZORAN-Sec: A Zero Trust–Compliant Intrusion Detection Framework for Open RAN With a Lightweight Transformer},
journal = {IEEE Open Journal of the Communications Society},
year = {2026},
volume = {7},
pages = {6308--6333},
doi = {10.1109/ojcoms.2026.3699599}
}
EUVIMED