Dynamic Security Permission Management for Power IoT Edge Devices Based on Kolmogorov-Arnold Network and Zero Trust Architecture

The escalating role of the Power Internet of Things (PIoT) in energy management highlights the urgent need for advanced security solutions to protect against evolving cyber threats. Integrating the Kolmogorov-Arnold Network (KAN) with the Zero Trust security architecture, this research presents a dynamic multi-attribute security permission management approach for P IoT edge devices. Critical to the integrity and stability of power systems, our method leverages KAN to predict trust levels based on device attributes such as ontological information, environmental context, and dynamic interaction behaviors, enabling the dynamic adjustment of access control rules. Our prototype system, demonstrating a prediction accuracy exceeding 96%, underscores the practicality of our approach in real-time security permission management as device attributes evolve. This study contributes significantly to the field by enhancing PIoT security, ensuring resilience against potential attacks.