Tidewall

And know where in the first pulse

A strain event or an acoustic excursion above the adaptive threshold produces: something is happening at cable-kilometre 412.6, ±10 m.

This is distributed acoustic sensing's structural advantage over every hydrophone field ever built. The along-cable position comes from optical time-of-flight, not from triangulating arrival times. There is no geometry to solve, no sound-speed profile to assume, no left-right ambiguity. Every subsequent stage starts from a position better than what an array converges to after minutes of tracking.

Fish, current, earthquake, trawl, anchor, or vehicle

These are mechanically different events and the strain waveform separates them cleanly. An anchor drag is a 0.1–3 Hz sawtooth with stick-slip release transients, advancing along the cable at 0.5–3 kt. A trawl door is a continuous 4–40 Hz rumble tracking at 2–4 kt, in two parallel tracks at door spacing. A grapnel snag is an impulsive step in DC strain, then a hold, then either release or a rising ramp to break. A burial plough has a characteristic 12–20 Hz tooth-passing line. An ROV or UUV thruster is a narrowband blade-rate tone with harmonics, near-stationary.

Fit the hyperbola, get the track

This is the most useful single fact about distributed acoustic sensing, and the whole attribution chain is built on it. A vessel passing over a cable draws an unmistakable hyperbola on the distance-versus-time waterfall. Its apex is the closest point of approach in both position and time; its asymptotic slope gives speed over ground; its curvature at the apex gives the CPA range.

Fit it and you have a track — course, speed and range — from a single sensing line, with no array, no beamforming and no second sensor. CPA position to ±15 m, CPA time to ±4 s, speed to ±0.4 kt, range to ±12%, crossing angle to ±8°.

Three independent lines. Any one alone is an accusation.

Acoustic signature. The tetrahedral nodes extract shaft rate, blade rate, gear mesh, diesel firing rate and turbo-alternator lines, plus DEMON cadence, which recovers shaft rate and blade count directly. Shaft rate times blade count is close to hull-specific. Matched against a library of 26,000 signatures, confidence against a hull seen before is typically 0.88.

And the library builds itself, legally and for free. Every AIS-squawking vessel that crosses an installed cable delivers a labelled training example — the acoustic signature on one side, the MMSI, name, dimensions and loading condition on the other. A single cable across a busy lane collects tens of thousands of labelled examples in eighteen months. Nobody has to run a collection ship.

AIS correlation — and non-correlation. If the hyperbola's fitted CPA gates against an AIS track, the vessel is named and the case closes in seconds. If nothing gates — if there is a hull over your cable and nothing is squawking — that is the alarm. "Dark hull, kilometre 412.6, 3.1 kt, loitering, 40 minutes" is the product.

Tool marks. If the intrusion touches the asset, the strain record fingerprints the implement — grapnel versus plough versus stockless anchor — with the drag azimuth, the applied tension and the moment of parting. With a timestamped position and an acoustic signature, that is physical evidence, not suspicion.

Raw sample to conclusion, hash-chained

Every stage is written into REVENANT LEDGER with the raw strain and acoustic windows retained alongside the derived products. This is what the commercial market is actually buying. An operator arguing with an insurer, or a foreign ministry demarching a flag state, needs a record an opposing expert cannot pull apart. A screenshot of a waterfall is not that. A tamper-evident chain from raw sample to conclusion is.

Thirty days of listening before a single threshold is set

Establish the segment's ambient strain and acoustic environment — traffic patterns, biologic seasonality, current noise, seismic background. Every detection threshold in the system is adaptive to this baseline, and a system installed without it will drown in false alarms. This phase is the most commonly skipped and the most commonly regretted.

Landing stations, or nothing at all on the surface

Interrogators at the landing stations or in-line every 80 km; nodes every 4 to 12 km, laid from a cable ship, a STYX, or a LEVIATHAN payload module. The expeditionary fit leaves no landing station and no shore footprint.

The library starts at 26,000 and grows on its own

AIS is correlated to acoustic signature continuously from the day the system is energised. There is no separate collection campaign and no collection ship.

Continuous strain, acoustics and seismics

Adaptive thresholds, automated characterisation, and a false-alarm rate of 0.4 per 100 km per day after filtering — low enough that an operator's watch floor will still be reading alerts in month six.

In time for a VHF call, or in time for a démarche

An anchor loitering over the asset triggers a warning while the problem is still free to fix. A dark hull triggers an intelligence alert in 90 seconds.

And the grid navigates whatever arrives

Cue a STYX, a LEVIATHAN, or a national asset onto a named track. Every surveyed node is also a one-way-travel-time reference, so a vehicle operating inside a TIDEWALL field navigates to 45 m where an unaided one would be nautical miles out. The sensor field and the navigation infrastructure are the same objects.

Hand it to the lawyers

Sign, retain and export the evidence chain — to the operator's counsel, the flag state, or the alliance. This is the deliverable. Everything before it is instrumentation.