Nokia has transformed a 2,000 km production fibre route into an active distributed sensor network using optical tomography.
Operators are responsible for vast stretches of fibre optic cable across land and seabed routes and often lack detailed visibility into the physical condition of those links. Traditionally, administrators can only tell if a connection is up or down, leaving the infrastructure vulnerable to undetected damage or tampering.
High-speed fibre networks are the invisible arteries of modern society, and many organisations rely on sections of fibre they do not own. Those third-party segments remain opaque, hiding risks such as fibre tapping that can undermine cybersecurity and data integrity.
Physical threats — from a ship dragging an anchor to nearby construction or excessive cable strain — eventually cause breaks that trigger costly, time-consuming repairs. Reactive maintenance is expensive and inefficient.
Implementing active sensor grids
Nokia Bell Labs addressed this problem with optical network tomography, converting passive fibre strands into an active, distributed sensor network. The team validated the approach on a live 2,000-kilometre production network in collaboration with Nordic research and education network operators CSC, Sikt, and SUNET.
Sylvain Almonacil, Research Engineer at Nokia Bell Labs, said: “We are not adding new sensors. We are turning the transponder itself into a sensor to see the network from the inside.”
Rather than relying on basic continuity checks, this architecture analyses the quality of transmitted light with high precision to infer what happens to the cable along its path. The core technique monitors tiny changes in the light’s state of polarization. Vibrations, temperature fluctuations, or mechanical stress change that polarization by altering the light’s orientation.
The trial uses a Nokia PSC 6S silicon engine to detect these subtle variations. This hardware combines advanced algorithms to optimize network capacity, automate processes with AI, and proactively detect anomalies. Acting as sensitive sensors, the engines continuously monitor polarization states and map physical stresses on the cable in near real time, without impacting data transmission speeds.
Edge computing and multi-domain analytics
Coherent transceivers at each end of the fibre act as primary edge sensors, performing high-frequency measurements that capture raw polarization data directly from the optical signal. Embedding this capability into existing infrastructure enables monitoring across multiple operational domains.
Edge hardware streams raw data to centralized processing algorithms that work like medical tomography. By correlating minute changes observed at both ends of the fibre link, these algorithms can precisely locate and quantify the physical disturbance along the route.
| Operational metric | Legacy network monitoring | Optical tomography approach |
| Fault detection | Reactive troubleshooting after a break. | Predictive alerts before physical damage occurs. |
| Location precision | Estimates spanning hundreds of kilometres. | Pinpoints exact geographic segments. |
| Domain visibility | Limited to owned, internal domains. | End-to-end multi-domain tracking. |
| Security posture | Blind to passive physical tapping. | Detects anomalous physical interference promptly. |
Optical network tomography helps secure infrastructure segments the operator does not physically own. By converting the standard transponder into an internal sensor, operators gain an inside view of the entire route and can monitor the fibre itself.
Previously, administrators could only verify the performance of segments they configured. The new end-to-end protocol ensures a route behaves as intended across ownership boundaries.
Operators can now follow specific wavelengths through different optical regions, including stretches managed by third-party telecom providers. This significantly improves situational awareness, delivering alerts that describe environmental vibrations and exact geographic coordinates of disturbances.
Production trial outcomes
The trial used 2,000 km of SUNET’s fibre infrastructure as a real-world production environment carrying live traffic for universities and research institutions across the Nordic region. This exposed the system to environmental noise and complex operational variables. Notably, the deployment required no additional dedicated sensors and caused no interference with primary customer data.
After processing operational data over three weeks, the Nokia Bell Labs team reported that the digital tomography estimates matched physical measurements across the multi-domain network. The trial accurately mapped fibre types and exact span lengths along the entire route.
This approach changes how telecom infrastructure is managed. Locating a fibre cut traditionally can take days and heavy capital resources; optical tomography reduces the search area from hundreds of kilometres to a specific span. Alerts about nearby digging or other activities allow operators to act before physical damage occurs.
Administrators also gain deeper visibility into network topology and can detect anomalous behavior like fibre tapping early, addressing security threats before they escalate into outages. The transponder-based sensors detect the precise moment light polarization diverges from learned AI baselines.
Nationwide fibre networks can become distributed acoustic and seismic sensor grids, continuously mapping physical stress on cables in near real time. That data is valuable for monitoring other infrastructure such as pipelines and for providing early warnings about geological events.
Environmental information is extracted continuously at network endpoints and fed into a centralized software platform that ingests the constant stream of polarization change data. The platform learns baseline vibrations from roads, railways, and weather patterns so AI can flag anomalies that indicate genuine physical threats.
In short, this technology enables operators to rethink their physical assets, converting standard data channels into self-protecting sensing networks.
See also: Broadcom silicon bridges AI data centres and edge
Want to learn more about AI and big data from industry leaders? Attend the AI & Big Data Expo events in Amsterdam, California, and London. The event series is part of TechEx and is co-located with related technology conferences such as IoT Tech Expo and Cyber Security & Cloud Expo.
Telecoms is published by TechForge Media. Explore other enterprise technology events and webinars organized by the same publisher.