How Ordinary Fiber Optic Cables Can Be Repurposed as Hidden Microphones

Background and Core Findings

The study, titled Hiding an Ear in Plain Sight: On the Practicality and Implications of Acoustic Eavesdropping with Telecom Fiber Optic Cables and presented at NDSS 2026, shows that fiber optic cables are naturally sensitive to acoustic vibrations. Sound induces minute structural deformation in the fiber, causing phase shifts in the laser beam that can be monitored to reconstruct the original audio, even across distances greater than 50 m.

To amplify the weak acoustic signal, the authors built a Sensory Receptor : a 65 mm hollow PET cylinder wound with 15 m of fiber. The cylinder converts pressure waves into longitudinal strain on the fiber, dramatically improving sound capture while remaining indistinguishable from a typical FTTH splitter box.

Attack Conditions and Feasibility

An attacker must gain physical access to both the user‑side ONU and the optical distribution network (ODN). This is realistic in FTTH deployments because ISP technicians, third‑party contractors, or malicious insiders routinely handle these endpoints. The vector is therefore viable for organized APT operations.

Experimental Results

Word‑error rate (WER) at 2 m distance: below 20 % (over 80 % of content correctly transcribed) using OpenAI Whisper and NVIDIA Parakeet.

Speaker localisation accuracy indoors: average error of 77 cm.

Acoustic event detection (typing, coughing, alarms) after deep‑learning model tuning: 83 % accuracy.

Real‑office scenario with a 50 m fiber run between two rooms: optimal placement (fiber box under a desk) achieved a WER of only 9 %.

Why Existing Countermeasures Fail

RF scanners are ineffective because the fiber sensor emits no radio frequency signals.

Ultrasonic jammers provide no benefit; placing a commercial jammer 10 cm from the device did not degrade speech recognition, whereas traditional microphones were fully suppressed.

Passive nature : the device requires no power and leaves no electronic signature, allowing long‑term silent deployment inside the splitter box.

Mitigation Recommendations

Physical‑layer protection : install polished fiber connectors, introduce Fresnel reflections to create blind spots for DAS, and deploy optical isolators to block Rayleigh back‑scatter.

Wiring optimisation : minimise redundant indoor fiber loops and avoid routing fibers near resonant surfaces such as desks or walls.

Acoustic insulation : add sound‑absorbing material to walls and ceilings surrounding fiber runs.

Security Impact Assessment

Corporate boardrooms : interception of confidential negotiations and strategic decisions.

Government agencies : exposure of sensitive policy discussions and diplomatic talks.

Diplomatic venues : embassies and consulates, where “no RF emission” is a core security assumption.

These environments rely on the belief that fiber provides physical isolation; the demonstrated attack overturns that assumption. An adversary needs only to compromise a single fiber endpoint, potentially by planting a disguised splitter box during installation, to create an almost undetectable, long‑lasting listening node.

Conclusion

The research uncovers a long‑overlooked security blind spot: fiber optic networks themselves act as hidden acoustic sensors. High‑security settings must rethink physical‑layer defenses, recognising that the absence of RF emissions does not guarantee immunity from eavesdropping.