arXiv: CITADEL: CSI-Based Jamming Detection and Open-Set Classification for IIoT Networks

AI Analysis

As a senior EU regulatory compliance analyst, I summarize the following regulatory change for compliance professionals.

This publication introduces CITADEL, a novel framework for detecting jamming attacks in Industrial Internet of Things (IIoT) networks using Channel State Information (CSI) and open-set classification. While not a regulation itself, this research directly impacts compliance with the EU AI Safety framework, particularly for systems classified as high-risk under the AI Act. The framework’s ability to identify unknown attack types raises new requirements for robustness, transparency, and continuous monitoring of AI-driven network security tools.

Organizations in critical infrastructure sectors—such as energy, manufacturing, transportation, and healthcare—that deploy IIoT networks are most affected. This includes operators of smart factories, power grids, and automated logistics systems. Compliance teams must assess whether their existing jamming detection systems rely on closed-set classification, which may now be considered insufficient under evolving AI safety standards.

Compliance teams should immediately review their AI-based network security solutions to determine if they incorporate open-set detection capabilities. They should document how their systems handle unknown or novel attacks, update risk assessments to reflect this new technical capability, and engage with notified bodies to ensure alignment with upcoming AI Act implementing acts. Proactive adaptation to this research will reduce future regulatory friction and liability exposure.