GNN-based trust propagation and intelligent certificate revocation decision mechanism for large-scale IoT networks

Abstract

Abstract The rapid expansion of Internet of Things deployments has introduced significant challenges in trust management and certificate lifecycle administration. Traditional Public Key Infrastructure mechanisms struggle with the scalability and responsiveness demands of large-scale IoT environments. This paper proposes a graph neural network-based framework that integrates trust propagation with intelligent certificate revocation decision-making. We develop a graph attention-based trust propagation model that captures relational dynamics among IoT devices through multi-head attention mechanisms with explicit temporal decay factors. Additionally, we design an adaptive revocation decision algorithm that synthesizes trust embeddings, behavioral anomaly indicators, and topological features to generate risk scores for certificate management. Experimental evaluation across networks comprising up to 102,400 devices demonstrates that our approach achieves trust propagation accuracy exceeding 89% and revocation decision F1 scores of 0.904, with median response latency under five seconds. The proposed framework outperforms the evaluated baseline methods, including traditional reputation-based approaches and standard graph convolutional networks, in both accuracy and computational efficiency within the considered experimental settings, providing a practical solution for securing large-scale IoT infrastructures.