Hydrophone Drift and the 12 ms Budget
- Published
- Reading time
- 2 min
- Author
- Mateo Ferreira
Every Bathys mooring reports within 12 ms. Most of that budget is physics; the rest is fighting hydrophone drift. A look inside the acoustic latency chain.
The number on the Bathys landing page is 12 ms average acoustic latency. People assume it is a marketing figure. It is a budget, and this is how it is spent.
Where the milliseconds go
| Stage | Budget | Notes |
|---|---|---|
| Sensor sample to packet | 0.8 ms | 4 Hz pressure, 48 kHz hydrophone, on-hull ADC |
| Acoustic relay hop | 6.5 ms | Mooring to nearest relay buoy, mean 9.6 km path |
| Relay to surface uplink | 2.1 ms | Buoy radio to vessel or shore station |
| Engine ingest and evaluation | 1.4 ms | Normalization, topology lookup, threshold evaluation |
| Console render | 1.2 ms | Operator-visible |
That is 12.0 ms on a good day. The acoustic hop dominates, and the acoustic hop is where drift lives.
What drift does to the budget
A hydrophone element that has drifted in sensitivity does not change the speed of sound. It changes how many retransmissions the relay needs to decode a packet. A 3 dB sensitivity loss on a relay path at the edge of its range roughly doubles the retransmission rate, and every retransmission costs a full hop.
In the Java trench deployment last year, two moorings crept from 11 ms to 19 ms over four months. The cause was biofouling on the hydrophone face, which is drift by another name.
Compensation
Three mechanisms keep the budget honest:
- Reference pings. Every relay buoy emits a calibrated ping hourly. Each mooring compares received level against the known source level and logs the delta as a drift estimate.
- Adaptive gain. When the drift estimate exceeds 1.5 dB the hull firmware raises front-end gain, up to a 6 dB ceiling, and raises a maintenance flag in the system log.
- Path re-routing. If a mooring’s effective latency exceeds 15 ms for more than an hour, the engine re-routes it to the next-nearest relay, even at a longer geometric path, because a clean longer hop beats a noisy short one.
The honest number
Across all live Bathys moorings the trailing 30-day mean is 11.7 ms. The 95th percentile is 16.2 ms. We publish the mean because it is what the array feels like to operate; we monitor the tail because it is where the hulls that need a visit are hiding.
