.png){kind=icon}
.png){kind=icon}
Return
Optimizing Network Security in IoT Networks: Leveraging Graph Learning for Zero-Trust Authentication
.png){kind=icon}
PRE
AI
DOI：10.1109/JIOT.2026.3682602.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The rapid growth of Internet of Things (IoT) networks in 5G, 6G, and edge computing has drastically increased the number of interconnected, resource-constrained devices, making them prime targets for security breaches. Despite advances in on-chip and software protection, IoT systems remain vulnerable to dynamic, stealthy attacks. A single compromised node can disrupt local operations and propagate malicious behavior across the network, leading to data leakage, denial-of-service, or even system-wide shutdowns. Traditional static security and periodic monitoring often fail to mitigate such threats in real time, particularly given the scale, heterogeneity, and latency constraints of modern IoT ecosystems. To address these challenges, we propose a lightweight zero-trust-based dynamic trust evaluation framework with adaptive access revocation for distributed IoT systems. Unlike prior works, our framework modifies a standard graph neural network (GNN) by integrating node-aware temporal attention, improving sensitivity to time-dependent anomalies. The system ingests hardware performance counters (HPCs), network-level features, and internode interaction data to dynamically compute probabilistic trust scores for each node. These scores directly enforce adaptive access control policies, selectively revoking read/write privileges or isolating compromised nodes in real time. We evaluated the framework in a custom simulation with 20–100 heterogeneous IoT nodes under scenarios including malware injection, data exfiltration, and covert channel attacks. Results show 99% threat detection accuracy and 95% policy enforcement accuracy, demonstrating a scalable, adaptive defense mechanism for next-generation IoT networks.
Keywords:
Edge devices
graph recurrent neural networks (GRNNs)
Internet of Things (IoT) networks
zero-trust authentication
Journal
IF:
8.9
Papers:
1.4W
Citations:
7.8W