Topology-Aware Graph Neural Networks for Distributed Node Fault Identification

Abstract

This paper proposes a fault identification model based on graph neural networks for node fault detection in distributed systems. The method models the distributed system as a graph structure and fuses node features with topological relationships to accurately represent node states in complex architectures. The model employs a multi-layer graph neural network to aggregate and propagate information across nodes, significantly enhancing its ability to identify potential faulty nodes. In the experimental section, a standard graph-structured dataset is used for testing. The proposed model is compared with several mainstream approaches. Results show that it outperforms other methods in terms of accuracy, AUC, and F1 score. In addition, the paper conducts extended experiments to evaluate the impact of varying neighbor counts and noise disturbances on model stability. These results further confirm the adaptability and robustness of the proposed approach in complex environments.