Subsea Fiber Is Becoming the World's Largest Listening Network: Not Spycraft, Physics
Distributed acoustic sensing (DAS) turns ordinary fiber-optic cable into a continuous vibration monitor. Oil and gas well surveys, earthquake monitoring, whale tracking and anti-submarine ship detection all run on the same technology chain, and the ocean is rapidly losing its privacy.
The argument · tap a timestamp to hear it
The scattering that makes the sky blue is also the source of the eavesdropping
Attenuation in optical fiber comes mainly from scattering. Brillouin scattering is produced by low-frequency phonons, which temperature or strain can induce; Raman scattering comes from molecular vibration and is commonly used for temperature measurement. But the most significant of the three is Rayleigh scattering, where light is reflected back by density fluctuations smaller than its own wavelength. Those fluctuations act like tiny mirrors scattered along the entire length of the fiber, sending a small fraction of the light back the way it came. Rayleigh scattering transfers no energy and does not change the frequency of the light. It is the reason the sky is blue, and it is the source of every signal DAS uses: both OTDR and, later, DAS are picking up this reflected light.
Polarization sensing failed because too many causes produce the same change
In the 1980s, OTDR was commonly used to locate breaks in a cable: send a pulse of light, measure the arrival time of the Rayleigh backscatter, and pinpoint the break or the high-loss section. In 1981, AJ Rogers at the UK's Central Electricity Research Laboratories proposed polarization optical time domain reflectometry, hoping to work backward from changes in polarization to external factors such as temperature, strain and sound. In the 1980s that path failed: too many things change polarization, so a given change could not be attributed to a single cause. The Rayleigh signal is the sum of all the fluctuations along the whole fiber, so the signal is over-averaged, and temperature sensing did not work either. The industry turned instead to Raman and Brillouin scattering: weaker signals, but ones whose Stokes and anti-Stokes frequency shifts map deterministically onto temperature and acoustic strain.
The navy did not commercialize DAS; oil wells did
Three breakthroughs turned Rayleigh backscatter from a diagnostic tool into a sensing method: lasers with narrower linewidths, meaning purer light; computing power able to sample tens of thousands of times per second; and a clear commercial demand. The US Navy did some early work here, and it remains semi-classified to this day. What actually broke DAS through into commercial use was the oil and gas industry: in the mid-1990s, oil companies began using fiber to monitor wellbores. Downhole temperatures run around 200°C and pressures reach up to 2,000 bar, a harsh enough environment to wear out any electronics; the fiber approach puts all the sensitive equipment on the surface, leaving nothing but cable down the hole.
Fiber VSP won on speed, not on accuracy
The oil and gas industry first used Raman scattering and the anti-Stokes signal for distributed temperature sensing (DTS), then in the late 2000s applied Rayleigh-scattering DAS to vertical seismic profiling (VSP). Conventional VSP relies on an array of geophones lowered into the well, roughly seven of them strung together per run, with the whole array repositioned after each measurement, which is slow and expensive. The fiber approach has a slightly lower signal-to-noise ratio than wireline geophones, but it is already close, and the killer application is speed: in some cases it finishes in minutes where the geophone approach takes 10 hours. The early challenge was data processing, since DAS can generate up to 10,000 times as much data per second as an ordinary temperature sensor.
Seismic stations are too expensive to deploy; dark fiber is already in the ground
Teams at Berkeley and Stanford applied DAS to earthquake monitoring in the early 2010s. Conventional seismic stations are costly, which rules out deploying them in bulk; meanwhile telecom and technology companies have laid vast amounts of unused dark fiber, which Berkeley Lab's Ajo-Franklin estimates at roughly 1 million kilometers worldwide, growing by another 100,000 to 200,000 kilometers a year. These cables float freely underground rather than being anchored in place, and earthquakes are extremely low-frequency signals, so picking them out of the noise of cars and trains is hard. Even so, the teams successfully tracked hundreds of small earthquakes both on campus and in the field, some with epicenters as far away as Mexico.
One cable already on the seafloor is worth ten thousand seismometers
Seismic monitoring networks are even scarcer on the ocean floor than on land. Giuseppe Marra's team at the UK's National Physical Laboratory used laser interferometry, which is related to DAS but not quite the same thing, to detect a major earthquake in central Italy from within the UK, and did it with a cable that had not been laid for earthquake monitoring. The Berkeley team went further: Nate Lindsey used a four-day maintenance window on a subsea fiber cable in Monterey Bay to turn 20 kilometers of cable into 10,000 sensors, recording a small earthquake and also revealing multiple fault zones on the seafloor. A four-day window, an existing cable, ten thousand listening points: that is the paradigm that turns dark fiber into an asset.
It can hear elephants on the railway tracks, and it can hear whale song
The applications quickly outgrew earthquakes. Back in 1993, Taylor and Lee at Texas A&M filed a patent for intrusion detection based on Rayleigh backscatter, covering large-perimeter protection scenarios such as airports and high-speed rail. Indian Railways even uses it to detect elephants walking onto the tracks. The most striking use is marine biology: by 2022, DAS could track the songs of baleen whales in the open ocean using existing subsea cables. The undersea acoustic environment is noisier than land, with internal waves, sediment transport and storms, but with grid search and Bayesian filtering, tracking is now close to automated.
A ship can switch off its transponder, but not its acoustic signature
In 2021, Diane Rivet's team in France found that DAS could track an oil tanker at a depth of 2,000 meters, that it worked best in 85 meters of shallow water, and that it could hear a tanker's engine and equipment noise from 2 kilometers away. Conventional passive sonar arrays, civilian or military, are reported to cost anywhere from hundreds of millions to billions of dollars; DAS covers a wider area using cable that is already there, and it will only get stronger as computing power and AI recognition improve. A ship can switch off its transponder to evade satellites, but it cannot switch off its acoustic signature, and threats like an anchor dragging along the seabed can also be flagged in advance. The host's conclusion: perhaps before long, the ocean will be less private than it has ever been.
In their own words · checked verbatim
Now imagine a new technology that can turn all that into a massive network of listening sensors.
Rather than sending probing signals into the sky or water, we send them into a glass fiber. We parse and analyze what comes back.
While some preliminary work was done by the US Navy - and remains somewhat shrouded - DAS did not find its big break commercially until it was adopted by the oil and gas industry.
But the real killer app is that fiber makes it far faster and easier to run a seismic survey. In some cases, just minutes as compared to up to 10 hours for a geophone-based survey.
People didn’t believe this would work ... They always assumed that an uncoupled optical fiber would generate too much signal noise to be useful
Eileen Martin14:46
Perhaps before long, DAS techniques can make the ocean a far less private place than it ever has been before. It is a game changer.
Figures
| Total fiber-optic mileage worldwide | About 5 million route-kilometers | 0:02 |
| Downhole temperature and pressure in an oil well | About 200°C / up to 2,000 bar | 8:30 |
| What one conventional geophone VSP run requires | About 7 geophones strung on a single steel cable | 10:34 |
| Time for DAS to complete a VSP survey | Minutes, vs 10 hours for the geophone approach | 11:38 |
| DAS data volume per second | Up to 10,000 times that of an ordinary temperature sensor | 11:38 |
| Scale of the Monterey Bay subsea experiment | 20 kilometers of cable turned into 10,000 sensors | 15:46 |
| Longest listening range for the French team's tanker tracking | In 85 meters of shallow water, it picked up tanker engine noise from 2 kilometers away | 17:50 |
| Cost of a military passive sonar array | Hundreds of millions to billions of dollars | 18:55 |
Glossary
- DAS (Distributed Acoustic Sensing)
- Uses the way the Rayleigh backscatter signal in an optical fiber changes with external vibration and strain to turn an entire cable into a continuously distributed vibration sensor.
- Dark Fiber
- Fiber that has been laid but carries no traffic, mostly pre-installed by telecom companies as spare capacity, and available for temporary use as DAS.
- OTDR (Optical Time Domain Reflectometry)
- A diagnostic tool that sends pulses into a fiber and analyzes the arrival time of the backscattered return to locate breaks or high-loss points.
- VSP (Vertical Seismic Profiling)
- A technique that fires acoustic waves into the ground and receives the reflected signals inside the well, used to image the rock around a reservoir.
- Geophone
- The electronic vibration sensor lowered into a well in conventional seismic surveying: expensive, fragile, and limited to discrete measurement points.
- Rayleigh Scattering
- The scattering of light by particles or density fluctuations smaller than its wavelength; the reason the sky is blue, and the source of the DAS signal.
How to listen
Investors in subsea cable and ocean-sensing plays, engineers working on earthquake monitoring or underwater target identification, and policy researchers who care about maritime security.
The first 4 minutes on fiber physics and the taxonomy of scattering can be skipped. Everything from the oil and gas VSP section onward (roughly 7:25) is the good part.