Protecting Subsea Cables with Photonics and AI

Subsea telecommunications cable operators protect global internet infrastructure by turning unlit "dark fibers" into continuous underwater motion sensors. By combining Coherent Optical Time-Domain Reflectometry (C-OTDR) and Distributed Acoustic Sensing (DAS) with artificial intelligence, operators can detect and locate physical threats in real time. This article explains how light pulses measure microscopic vibrations along the ocean floor and how machine learning models distinguish between destructive ship anchor drags and seabed earthquakes to prevent global connectivity outages.

The Role of Dark Fibers

Subsea cables contain multiple optical fibers, not all of which are actively used for transmitting internet traffic. Operators reserve unlit fibers, known as dark fibers, for network expansion, redundancy, and cable monitoring. Because these fibers run the entire length of the cable, they provide an ideal pathway for continuous environmental sensing without interfering with active data transmission on neighboring fibers.

How C-OTDR and DAS Work

Distributed Acoustic Sensing (DAS) is the overarching technology that turns an optical fiber into a series of thousands of virtual microphones, while Coherent Optical Time-Domain Reflectometry (C-OTDR) is the specific laser interrogation technique used to achieve this over long distances.

The process works through the following steps:

  1. Laser Pulse Transmission: A DAS interrogator unit on shore sends coherent laser pulses down a dark fiber.
  2. Rayleigh Backscattering: Every glass fiber contains microscopic imperfections. As the laser pulse travels through the fiber, a tiny fraction of the light bounces off these imperfections and travels back toward the shore. This phenomenon is called Rayleigh backscattering.
  3. Strain Detection: When an external acoustic wave or physical vibration hits the subsea cable, it microscopically stretches or compresses the glass fiber at that exact location.
  4. Phase Measurement: The C-OTDR system measures the phase and frequency changes of the returning backscattered light. By calculating the time it took for the light to return, the system pinpoints the exact location of the vibration along the cable within a few meters.

AI Classification of Threats

The ocean floor is noisy, filled with the sounds of marine life, ocean currents, and normal shipping traffic. A single DAS unit generates terabytes of acoustic data daily. Operators use artificial intelligence and machine learning algorithms to filter this background noise and classify specific threats in real time.

Detecting Anchor Drags

Ship anchors are the leading cause of subsea cable damage. AI models are trained to recognize the specific acoustic signatures of maritime threats:

  • Vessel Tracking: The AI first identifies the low-frequency propeller and engine noise of a ship passing directly over the cable route.
  • Impact and Scrape: If the ship drops an anchor, the DAS records a sharp acoustic spike followed by a continuous, rhythmic, high-amplitude scraping vibration as the anchor drags across the seabed toward the cable.
  • Preemptive Action: When the AI confirms an anchor drag pattern, it immediately alerts the cable operator and pulls Automatic Identification System (AIS) maritime data to identify the specific vessel. Operators can then contact the ship's captain or coast guard to halt the vessel before the anchor severs the cable.

Detecting Seabed Earthquakes

Subsea earthquakes and underwater landslides can snap cables across vast distances. DAS and AI turns the cable into an array of seismometers:

  • Wave Identification: Earthquakes generate primary (P) waves, which travel fast and compress the ground, followed by slower, more destructive secondary (S) waves that shear the ground side-to-side.
  • Pattern Recognition: AI algorithms instantly differentiate the sudden, broad-spectrum, multi-kilometer acoustic footprint of a seismic P-wave from the localized, slow-moving signature of an anchor drag.
  • Automated Rerouting: Upon detecting the initial P-wave of a seabed earthquake, the AI triggers automated network management systems. Operators can proactively reroute global internet traffic to alternative cable paths milliseconds before the destructive S-waves or resulting underwater landslides sever the physical cable.