A review of machine learning-based failure management in optical networks

In the present big data era, data is growing exponentially, and optical networks serve as the backbone for high-capacity and long-distance data transmission. Optical networks are always large scale, comprise massive components, and cover a wide area. Considering this, instances of failure will cause extremely serious consequences, such as massive data loss, large-scale computing interruption, and core information transfer blocking. Therefore, failure management in optical networks is crucial to ensure the stable operation, maintain the service status, and, in the event of a failure, recover the failure rapidly. Recently, techniques from machine learning (ML) have been widely studied to address the problems in optical network failure management.

In this article, the authors reviewed the applications of ML to failure management in optical networks from infancy to the near term. First, they introduced the background of failure management and interpreted the typical tasks. The key physical objects that need health monitoring in optical networks and the potential failure categories for each object are listed in detail. Then various ML algorithms applied for failure management are depicted. ML algorithms are strongly dependent on data, and thus, the data sources with data content and extracted information in optical networks are also discussed. Following that, they surveyed the existing schemes of ML-based failure management in terms of alarm analysis, failure prediction, failure detection, failure localization, and failure identification. Finally, the future scope of this topic is envisioned from the perspective of data, models, tasks, and emerging techniques.

See the article:

Wang D S, Zhang C Y, Chen W B, et al. A review of machine learning-based failure management in

optical networks. Sci China Inf Sci, 2022, 65(11): 211302

https://doi.org/10.1007/s11432-022-3557-9

---

---