Case Studies Industry Technology

Fiber Cut Detection: How Existing Fiber Can Hear the Excavator Before the Cut

By BiiSensing Team September 23, 2026 7 min read

On the morning of September 21, a construction crew cut a fiber optic cable between New Brunswick and North Brunswick, New Jersey. Within hours, the FAA ordered ground stops at Newark and Philadelphia, and approach control for five airports in the Northeast went dark. One buried cable, one machine, one day of air traffic. The incident is the clearest recent argument for treating fiber cut detection as an operational discipline rather than an insurance clause.

For anyone who operates buried infrastructure, the detail that matters is how the cut happened. According to NJ Transit, quoted by NBC New York, “the utility markings at the site may not have been accurate.” The crew was working about 10 feet, roughly 3 meters, from those markings. In other words, the map may have been wrong. Locating the cable on paper did not protect it. Sensing the cable itself with distributed acoustic sensing might have.

One cut, five airports, seven hours

NBC News reported that the crew cut the cable at around 9:45 a.m. The failure took down the Philadelphia TRACON, which controls approaches to Newark, Teterboro, Philadelphia, LaGuardia and JFK. Newark cancelled roughly 619 flights and Philadelphia another 197. Delays ran up to 150 minutes, and the disruption lasted about seven hours.

The cable the crew hit was a backup line that runs alongside the Amtrak track between Newark and New Brunswick. The TRACON still went down. Logical redundancy protects against equipment failure, but it does not protect against a physical threat on the route itself. FAA administrator Bryan Bedford put the repair in perspective: “That fiber break is probably 13 hours to repair.”

fiber cut detection: excavator digging trench
Foto de Built Robotics en Unsplash

Why fiber cut detection has to start before the cut

Cuts like this one are routine. The Common Ground Alliance’s DIRT report, covered by Cabling Installation & Maintenance, counted 221,717 damage incidents to buried utilities in the United States in 2025. Telecom, MSO and CATV networks took 51% of them. The leading causes were:

  • No notification to the One Call Center or 811: 22.3%
  • Excavator failed to maintain clearance after verifying the marks: 14.8%
  • Facility not marked because of locator error: 14.2%
  • Facility marked inaccurately because of locator error: 9.2%

The last two are locator errors, and together they add up to about 23% of all incidents. In those cases the excavator followed the rules and the map was wrong, exactly as NJ Transit described. Fiber cut detection that depends on the accuracy of the plan inherits the plan’s errors. Detection that depends on the cable itself does not.

The economics point the same way. Paul Dickinson, former president of the Fiber Optic Sensing Association, told Fierce Network that underground excavation causes more than 80 billion dollars in damage per year in the United States. Fierce’s own headline was careful: fiber sensing “might have helped.” That nuance is worth keeping.

What distributed acoustic sensing hears along a telecom route

Distributed acoustic sensing, or DAS, turns a standard telecom fiber into a continuous vibration sensor. An interrogator at one end sends laser pulses down the fiber and reads the light that scatters back. When the ground near the cable vibrates, the backscatter changes at that point. The timing of the return then tells the interrogator where along the route it happened.

On a telecom route, the signatures that matter are mechanical excavation, hand digging and direct handling of the cable. Telling an excavator apart from a truck on a nearby road is the core classification problem. It depends on soil, burial depth and, above all, on training data recorded on that specific route. The goal of fiber cut detection here is not to log the cut. Instead, it is to catch the digging that precedes it, while there is still time to make a phone call or send a patrol.

Two limits apply. First, the fiber has to be close enough to the threat. A cable in a deep duct or an aerial section hears differently than one in a shallow trench. Second, classification is only as good as its training. A route next to a railway, like the one in New Jersey, has a loud and constant background. The system therefore needs time on site to learn what normal sounds like. Anyone who promises zero false alarms on a route like that has not operated one.

You do not need a dedicated fiber for this. Where a cable has unused strands, dark fiber sensing uses one of them while the rest keep carrying traffic. Where no spare strand exists, the ITU has now defined how to sense on the live one.

fiber cut detection: excavator digging trench
Foto de Built Robotics en Unsplash

Sensing on a live network: what ITU-T G.681 changes

The most common objection from a network operator is simple: nobody wants a sensing laser on a fiber that carries customer traffic. Recommendation ITU-T G.681, “Distributed fibre optic sensing system for terrestrial optical transmission systems,” addresses that case directly. The ITU approved it on November 29, 2025, and it is in force. It applies to systems that are in service.

In practice, there is now an international reference for running sensing and transmission on the same network. Fiber cut detection therefore does not require laying new cable, leasing a separate route, or waiting for the next civil works window. It can start on the fiber that already sits in the ground.

Latin America: theft and sabotage, not just excavators

In the United States the dominant threat is an excavator that did not call 811. In Latin America the same physical exposure takes a different form. On September 8, 2026, someone cut the fiber link between Puan and San Miguel Arcángel in Buenos Aires province, Argentina. Four towns lost internet for several hours, until repairs finished near 10 p.m. The cooperative COSSMA found “multiple marks and cuts made by hand” and ruled out an accident, as El Ojo Ameghinense reported.

The regional numbers back the anecdote. A survey by cet.la, the Centro de Estudios de Telecomunicaciones de América Latina, reached operators in 14 countries. Of those, 63% perceived a significant or moderate increase in theft or damage to their infrastructure over the last five years. Fiber optic cable was the target in 30% of cases. Users stay offline for between 3 and 72 hours, and documented losses reach 46 million dollars. Moreover, 90% of the companies had to hire private security, according to Consumotic.

Private guards cover points. A DAS system covers the whole route. Hand digging and cable handling are quieter than an excavator, but they happen right on top of the cable. For fiber cut detection in the region, the threat model has to include hands as well as machines. The response, likewise, has to include a patrol dispatch.

The same logic is moving offshore. Brazil’s Navy is evaluating unmanned vessels, drones and acoustic sensors to protect the submarine cables that carry more than 95% of the country’s internet traffic to the world. According to Revista Sociedade Militar, DAS is one of the fronts under study, with an application that Petrobras has already demonstrated.

Where to start

The New Jersey cut and the Argentine cut share a root cause: a cable whose only protection was a line on a plan. If you operate a fiber route, a pipeline with fiber alongside it, or a rail corridor, the sequence for a fiber cut detection project is short.

  1. Map where the fiber is actually vulnerable, not where the plan says it runs. Shallow sections, shared corridors and stretches with a history of theft go first.
  2. Check for unused strands. Dark fiber sensing on a spare strand is the simplest start. Where none exist, G.681 gives a framework for the live fiber.
  3. Run a proof of concept on one real segment. Let the system learn that segment’s background, then test it with controlled excavation and manual handling.
  4. Define who receives the alarm and what they do with it. A detection that reaches nobody is a log entry, not protection.

BiiSensing approaches fiber cut detection in that order with infrastructure operators in Latin America. The team assesses the feasibility of a specific route, runs proofs of concept on installed fiber, and supports the project from survey to commissioning. If the only thing between a cable that matters to you and the next crew is paint on the pavement, that is the conversation to have now.

Frequently asked questions

What is fiber cut detection?

It means sensing the cable itself with distributed acoustic sensing, which turns a standard telecom fiber into a continuous vibration sensor. The goal is not to log the cut but to catch what precedes it, such as mechanical excavation, hand digging or handling of the cable, while there is still time to make a phone call or send a patrol.

Why are utility markings not enough to protect buried fiber?

Locator errors, with facilities left unmarked or marked inaccurately, are among the leading causes of damage to buried utilities in the Common Ground Alliance DIRT data. In those cases the excavator followed the rules and the map was wrong. Detection that depends on the plan inherits its errors, and detection that depends on the cable does not.

What limits fiber cut detection?

The fiber has to be close to the threat: a cable in a deep duct or an aerial section hears differently than one in a shallow trench. Classification is only as good as its training, so a route with a loud, constant background next to a railway needs time on site to learn what normal sounds like.

Does fiber cut detection need a dedicated fiber?

No. Where a cable has unused strands, dark fiber sensing uses one of them while the rest keep carrying traffic. Where no spare strand exists, Recommendation ITU-T G.681 now gives an international reference for running sensing and transmission on the same network while it is in service.

Share