Knowledge-Graph-Assisted Fault Localization for Cloud-Native Open RAN and Edge Infrastructure

Abstract

Cloud-native Open Radio Access Network (Open RAN) and edge computing infrastructures have become essential for supporting scalable and intelligent 5G services. However, the distributed nature of these environments makes fault localization and root-cause identification increasingly complex due to the large volume of telemetry data, dynamic resource allocation, and interdependent network components. This paper proposes a Knowledge-Graph-Assisted Fault Localization framework that integrates cloudnative observability with knowledge graph modeling to improve fault diagnosis in Open RAN and edge infrastructures. The framework constructs a knowledge graph representing relationships among network functions, services, containers, and edge nodes, while continuously collecting logs, metrics, and events from the infrastructure. Graph-based reasoning is applied to correlate anomalies and identify the underlying root cause of failures with high accuracy. Experimental evaluation demonstrates that the proposed approach reduces fault localization time, improves diagnosis accuracy, minimizes service disruptions, and enhances overall network reliability compared with conventional monitoring techniques. The framework provides an efficient and scalable solution for intelligent fault management in next-generation cloud-native Open RAN environments.