Background
Incomplete statistics show about 300,000 optical cable faults occur annually in China, including more than 7,000 interruptions on level-2 trunk and above. In access network scenarios, if primary and backup routes share the same cable or duct, protection can fail and reduce service reliability. Therefore, proactively identifying whether parts of primary and backup protection paths and east-west transmission ring paths share physical cables can eliminate most hidden risks.
Pilot validation in a local network
Jiangxi Mobile and FiberHome addressed the problem of many routing security risks in existing networks that cannot be detected automatically. Recently they validated a feasibility solution for intelligent same-cable and same-duct detection in the Jingdezhen local network. The pilot demonstrated that the solution can promptly detect same-route cable issues and support remediation of identified risks.
Evolution from OTDR big-data detection to intelligent same-duct detection
In 2022, Jiangxi Mobile and FiberHome completed development of same-cable detection software based on OTDR big-data analysis, deployed it in production networks, and validated it with large-scale field data. In Jingdezhen, Jiujiang, Nanchang and the secondary trunk network, the detection rate for same-cable scenarios reached 96%. However, due to the detection principle, the solution could not identify same-route scenarios that involved different cables, which limited practical applicability.
Building on the experience from the same-cable detection 1.0 project, Jiangxi Mobile and FiberHome developed an intelligent same-duct detection 2.0 project. Unlike the earlier approach, this project uses a model training approach: it analyzes vibration signal features of live network cables, designs a classifier network that recognizes and filters all types of vibration events, and implements a high-precision time-frequency signal similarity comparison algorithm. The result is an AI-driven automated same-duct detection solution.
Performance and adaptability
The detection scheme generalizes well across different physical environments. Through algorithmic optimization and classification, the solution can maintain recognition accuracy above 93%. It can identify risks for common deployment environments such as buried and aerial cables, and automatically distinguish among all types of vibration events in live networks, including vehicle traffic, excavation, and wind. On the hardware side, the solution imposes no restrictions on network equipment vendors, network types, network hierarchy, or cable models, maximizing applicability and removing vendor or industry barriers.
On the management platform, the solution can interface with the operator transmission operations and maintenance workstation to forward analyzed data to upper-layer workbenches, support account hierarchies, and allow departmental and city-level operations administrators to perform independent data operations and analysis, enabling autonomous analysis workflows.
Outlook
As network security and automation requirements increase, building secure and efficient optical networks remains an urgent challenge. Future work includes automating network management tools, digitizing dormant resource information, and improving operations efficiency and resource visualization to support more intelligent network operations.
