Harnessing the Power of AI and Automation to Protect Optical Networks from Fiber Cuts

Telecom Tech Outlook | Tuesday, November 23, 2021

AI opens the door to approaches that increase network availability by detecting and optimizing increasingly complex systems. It can also accommodate new service definitions made possible by just-in-time restoration.

Fremont, CA: Artificial intelligence (AI) is gaining popularity in the optical community as researchers look for new ways to use it to solve intriguing problems. Proactive detection of fiber cuts is one of these applications.

Using the Power of Data to Predict the Future

AI opens the door to approaches that increase network availability by detecting and optimizing increasingly complex systems. It can also accommodate new service definitions made possible by just-in-time restoration. To realize this potential for innovation, we need to monitor the optical infrastructure on a large scale. This becomes a tangible reality thanks to advanced coherent technology, which provides a significant amount of data without the need for additional hardware.

Companies use fiber as a sensor to detect nearby mechanical events, such as an excavator starting to dig, to help prevent fiber cuts.

Modern optical variable bit rate transponders send data on both orthogonal polarizations of light. Their receivers monitor any changes in these two polarizations in order to correctly decode the transmitted information. Thanks to existing filters utilized in advanced digital signal processing, the receiver can also monitor these fluctuations by reconstructing and tracking the so-called state-of-polarization (SOP).

Identifying Event Signature

AI-enabled machines have the ability to learn patterns in a way that no other machine can. They can be trained, for instance, to classify and distinguish between people's signatures on the basis of handcrafted marks. The number of letters in a signature, the presence of an underline, as well as the space between two letters, are all features that can be classified.

Companies can use these capabilities to help prevent fiber cuts in a similar way. One can begin by collecting thousands of SOP measurements in order to populate the AI database. Each of these SOP measurements lasts about two seconds and acts as a signature, allowing companies to identify a specific type of event.

With SOP measurements, event categorization is determined by a set of easily interpretable features, such as the minimum amplitude value required for a given measurement period.