JAMSTEC streamed sonar data at a maximum of about 58 kbps from an autonomous underwater vehicle working 7,000 to 7,500 meters deep in the Japan Trench in November 2025 (JAMSTEC, Urashima 8000 press release 2026), while the only open international standard for underwater acoustic messaging, NATO’s JANUS, still runs at a baseline of 80 bps (see section 4). Between those two numbers sits the whole physics of the problem: sound in seawater travels at about 1,500 m/s, and acoustic signals lose 0.1 to 4 dB per km where radio loses 3.5 to 5 dB per meter (Schirripa Spagnolo et al., Sensors 2020). This roundup is about communication links between subsea vehicles, sensors and buoys, not ocean noise, which we covered in our underwater acoustics and ocean sonar statistics; for the wired alternative on the seafloor, see our submarine fiber optic cable statistics. We aggregated data from the Woods Hole Oceanographic Institution (WHOI), the NATO Centre for Maritime Research and Experimentation (CMRE), the Japan Agency for Marine-Earth Science and Technology (JAMSTEC), the IEEE Journal of Oceanic Engineering, NOAA, peer-reviewed papers in PubMed Central and arXiv, and manufacturer spec sheets that we label as vendor claims.
TL;DR
- About 58 kbps of sonar data sent acoustically from 7,000 to 7,500 m depth during a November 2025 dive (JAMSTEC, Urashima 8000 press release 2026).
- About 0.5 seconds of two-way travel time over a link of just 400 m (Qarabaqi and Stojanovic, Statistical Modeling of a Shallow Water Acoustic Communication Channel 2009).
- Links of 100 km or more carry 1 kbps or less, while links of 1 km or less reach up to 100 kbps (Xu, Irigoien and Alouini, arXiv survey 2024).
- 600 km of error-free transmission, but at only 37.5 bits per second (Northwestern Polytechnical University study, reported by the South China Morning Post 2025).
- JANUS promulgated as NATO STANAG 4748 on March 24, 2017, and still the only open digital underwater acoustic standard as of 2024 (Petroccia and Alves, NATO CMRE, OCEANS 2024).
- Up to 34.5 kbps for JANUS fast-mode payloads at 96 to 134 kHz over 11 to 415 m (Li and Zheng, IEEE Journal of Oceanic Engineering 2022).
- 5,388 bps top burst rate for the WHOI Micro-Modem, but 3,343 bps measured single-channel throughput (WHOI Acoustic Communications Group 2026).
- 10 uW standby draw for a low-power research modem that transmits on about 120 mW (Sanchez et al., Sensors 2012).
- Acoustic attenuation of 0.1 to 4 dB per km vs 3.5 to 5 dB per meter for radio in water (Schirripa Spagnolo et al., Sensors 2020).
- USD 727.3 million global market in 2024, growing 7.7% a year to 2034 (Global Market Insights 2025).
1. Physics of the Channel: Sound Speed, Latency and Echoes
Every acoustic link pays a time penalty that radio engineers never see. At about 1,500 m/s, a message needs roughly 0.67 seconds to cover one kilometer one way (1,000 m divided by 1,500 m/s), so a question and its acknowledgment across 1 km cost well over a second before any processing. In a measured shallow-water channel, the two-way travel time over only 400 m was about 0.5 seconds (Qarabaqi and Stojanovic, Statistical Modeling of a Shallow Water Acoustic Communication Channel 2009).
The second problem is multipath: copies of the same signal arrive by different routes, bouncing off the surface and the seabed, and pile on top of each other. It is the same echo problem that speech systems fight in rooms (see our acoustic echo cancellation statistics), but stretched over milliseconds and constantly moving. Coherence time, meaning how long the channel stays similar enough to be treated as fixed, sets how often a receiver must re-learn it. Motion makes it worse: JAMSTEC’s acoustic technology group notes that a vehicle at 10 knots creates Doppler shifts (frequency changes caused by motion) far beyond anything aerial radio faces.
| Metric | Value | Source |
|---|---|---|
| Nominal speed of sound in seawater | about 1,500 m/s | Qarabaqi and Stojanovic 2009 |
| Doppler impact on underwater acoustic links compared with aerial radio | 4 to 5 orders of magnitude greater | JAMSTEC Underwater Acoustic Technology Research Group 2026 |
| Two-way travel time over a 400 m link in 10 m deep water | about 0.5 s | Qarabaqi and Stojanovic 2009 |
| Extra delay of an echo whose path is 15 m longer than the direct path | 10 ms | Qarabaqi and Stojanovic 2009 |
| Channel coherence time implied by a 0.2 Hz Doppler spread (3 dB) | about 5 s | Qarabaqi and Stojanovic 2009 |
| Signal attenuation: acoustic vs radio vs optical (clear ocean water) | 0.1 to 4 dB/km vs 3.5 to 5 dB/m vs 0.39 dB/m | Schirripa Spagnolo et al., Sensors 2020 |
| Optical attenuation in turbid water | 11 dB/m | Schirripa Spagnolo et al., Sensors 2020 |
Context: the Qarabaqi and Stojanovic channel was measured in an 8 to 18 kHz band; the same paper reports that average received power can swing by several dB within a few seconds. Most recent available data: Qarabaqi and Stojanovic, 2009; Schirripa Spagnolo et al., 2020. The physics has not changed, but newer field datasets are scarce in open access.
2. The Rate-Range Trade-off
The defining statistic of this field is not a speed but a product: data rate multiplied by distance. Higher frequencies carry more bits but are absorbed quickly, so every extra kilometer is paid for in bandwidth. A 2008 Chinese lake study measured itself against an “internationally acknowledged” benchmark of 40 km·kbps and beat it with 9.09 kbps over 5 km, which works out to 9.09 × 5 = 45.45 km·kbps (Shen et al., Journal of Northwestern Polytechnical University 2008). Multi-transducer designs have since pushed that ceiling by an order of magnitude, as the table shows.
The same logic sets frequency choices. The 2024 smart-ocean survey by Xu, Irigoien and Alouini places 1 to 10 kHz signals on links of tens or even hundreds of kilometers at only tens to hundreds of bits per second, and 100 kHz to 1 MHz signals on links of tens to a few hundred meters. Milica Stojanovic, a leading researcher in the field, told IEEE Spectrum in 2017 that 11.5 kHz suits roughly 1 to 10 km, while 10 to 100 km links call for frequencies around 1 kHz.
| Metric | Value | Source |
|---|---|---|
| Typical data rate on links of 100 km or more | 1 kbps or less | Xu, Irigoien and Alouini, arXiv 2024 |
| Typical data rate on links of 5 to 100 km | about 10 kbps | Xu, Irigoien and Alouini, arXiv 2024 |
| Typical data rate on links of 1 to 5 km | about 20 kbps | Xu, Irigoien and Alouini, arXiv 2024 |
| Typical data rate on links of 1 km or less | up to 100 kbps | Xu, Irigoien and Alouini, arXiv 2024 |
| Rate-distance product cited as the international benchmark | 40 km·kbps | Shen et al., J. Northwestern Polytechnical University 2008 |
| OFDM lake test at 5 km (OFDM splits data across many parallel tones) | 9.09 kbps at a bit error rate of 10^-4 | Shen et al. 2008 |
| JAMSTEC multipath MIMO link (several transducers sending in parallel) | over 600 kbps·km, more than 10 times existing devices | JAMSTEC Underwater Acoustic Technology Research Group 2026 |
| Usable bandwidth: acoustic vs optical links | 1 to 100 kHz vs 150 MHz | Schirripa Spagnolo et al., Sensors 2020 |
Outlier note: the survey’s bands are typical values, not limits; the field records in section 3 sit at both extremes. Most recent available data for the 40 km·kbps benchmark: Shen et al., 2008.
3. Field Records and Sea Trials
Records split into two camps: distance and depth. A Chinese team reported 600 km of error-free transmission at only 37.5 bits per second, using QPSK modulation and an 8-element hydrophone array on a buoy over water averaging about 5.5 km deep (South China Morning Post 2025; details via Indian Defence Review 2025). At 1,500 m/s, each bit on that link arrives about 400 seconds after it leaves (600,000 m divided by 1,500 m/s), close to 7 minutes.
At the other end, JAMSTEC’s Urashima 8000 trial sent seafloor sonar data to a ship during the dive, letting the crew spot structures 2 to 5 m tall and adjust the plan mid-mission. Slant range, the straight-line distance between a vehicle at depth and the receiver, is what most trials report; a 2025 acoustic glider study reached more than 5 km of it (Zou and Sun, Sensors 2025).
| Metric | Value | Source |
|---|---|---|
| Longest error-free link reported (study in Acta Acustica) | 600 km at 37.5 bps | Northwestern Polytechnical University, reported by SCMP and Indian Defence Review 2025 |
| Sonar data rate from Urashima 8000 at 7,000 to 7,500 m depth (November 2025) | about 58 kbps maximum user data; modem about 80 kbps effective at 6,500 m | JAMSTEC press release 2026 |
| Link between two vehicles moving at 10 knots relative to each other, shallow water | over 40 kbps | JAMSTEC Underwater Acoustic Technology Research Group 2026 |
| Acoustic glider maximum slant range (8 to 16 kHz modem, sea area about 1,800 m deep) | 5,113 m | Zou and Sun, Sensors 2025 |
| JANUS fast-mode payload rates (BPSK, QPSK, 8-PSK) in lake and coastal tests, 11 to 415 m | 11.5, 23 and 34.5 kbps | Li and Zheng, IEEE Journal of Oceanic Engineering 2022 |
| Coastal test, BPSK payload: bit error rate and packet error rate | about 0.13 × 10^-3 and 0.8% | Li and Zheng 2022 |
| Fish-farm observatory link with a commercial cNODE MiniS modem (21 to 31 kHz) | up to 6.0 kbit/s over about 300 m | Bergh et al., Sensors 2024 |
| Packets lost in the observatory’s bench test at minimum acoustic power | 2 of 5,933 | Bergh et al., Sensors 2024 |
Context: the observatory study (Bergh et al., Sensors 2024) found background noise across 3 to 35 kHz and strong echo-sounder bursts at 35 to 55 kHz, right inside common modem bands; shipping is another major noise source, covered in our propeller cavitation noise statistics. The 600 km paper itself was not reachable, so its figures rest on press reports (see Data watch).
4. JANUS: The NATO Interoperability Baseline
The bottleneck for many operators is not speed but the fact that modems from different makers cannot understand each other. CMRE scientist João Alves described the situation as “wild west communications underwater” (IEEE Spectrum 2017). JANUS answers with a deliberately modest, robust signal that any modem can implement: an 80 bps baseline, in a band near 11.5 kHz chosen as a common meeting point. A STANAG is a NATO standardization agreement; this one is distributed without security markings, so civilian operators can use it too (Petroccia and Alves, NATO CMRE, OCEANS 2024).
Speed can be layered on top. In the Lehigh University fast-mode experiments, the JANUS structure moved to a 96 to 134 kHz band and carried payloads up to 34.5 kbps, about 431 times the baseline (34,500 divided by 80).
| Metric | Value | Source |
|---|---|---|
| JANUS baseline data rate | 80 bps | Li and Zheng, IEEE Journal of Oceanic Engineering 2022 |
| JANUS centre frequency and band | 11.52 kHz centre, 9.44 to 13.6 kHz band (4.16 kHz wide) | Li and Zheng 2022 |
| Frequency-hopping tone pairs and baseline packet size | 13 tone pairs; 64-bit packet encoded into 144 chips | Li and Zheng 2022 |
| Date JANUS was promulgated as NATO STANAG 4748 | March 24, 2017 | Petroccia and Alves, OCEANS 2024 |
| Open international standards for digital underwater acoustic communication, as of 2024 | 1 (JANUS is the only one) | Petroccia and Alves, OCEANS 2024 |
| Development time at CMRE before NATO recognition | about 10 years | Ocean News and Technology 2017 |
| Packet payload described at launch | 56 bits, followed by a redundancy check | IEEE Spectrum 2017 |
| Pause before a transmission so nearby devices can prepare | about 400 ms | IEEE Spectrum 2017 |
Context: the development history comes from Ocean News and Technology 2017. Most recent available data for the Spectrum figures: 2017; the 2024 CMRE paper notes a more recent STANAG revision.
5. Rated vs Delivered: Research and Commercial Modems
Headline rates are burst rates, meaning the speed during a packet, not what a user gets once gaps, decoding time and acknowledgments are counted. The WHOI Micro-Modem, a research workhorse, shows the gap clearly: its fastest packet type bursts at 5,388 bps but delivers 3,343 bps of single-channel throughput, about 62% (3,343 divided by 5,388), according to the WHOI Acoustic Communications Group.
Commercial spec sheets confirm the rate-range trade-off but list best-case “up to” figures, so we label them as vendor claims. On the LinkQuest model list, the UWM3000H reaches 6,000 m at 80 to 320 bps, under 1% of the UWM2200’s top rate (320 divided by 38,400 = 0.83%).
| Metric | Value | Source |
|---|---|---|
| WHOI Micro-Modem burst rate, PSK Rate 5 (fastest) vs FSK Rate 0 (most robust) | 5,388 bps vs 80 bps | WHOI Acoustic Communications Group 2026 |
| WHOI Micro-Modem measured throughput, Rate 5, single-channel vs four-channel receiver | 3,343 bps vs 1,622 bps | WHOI Acoustic Communications Group 2026 |
| Full send-and-acknowledge cycle at 1,500 m range | about 15.5 s | WHOI Acoustic Communications Group 2026 |
| LinkQuest UWM2200 range and data rate | up to 1,000 m at 19,200 to 38,400 bps | LinkQuest product list (vendor claim) |
| LinkQuest UWM3000H range and data rate | up to 6,000 m at 80 to 320 bps | LinkQuest product list (vendor claim) |
| Sonardyne Modem 6 data rate in its 20 to 34 kHz band | 200 to 9,000 bps; depth ratings to 5,000 m | Sonardyne product page (vendor claim) |
| EvoLogics S2C R modem family | data rates up to 33 kbps; working ranges up to 8,000 m | EvoLogics profile on Naval Technology (vendor claim) |
Context: vendor figures come from the Sonardyne Modem 6 page and the EvoLogics profile on Naval Technology. The LinkQuest list also includes a UWM10000 rated up to 10,000 m at 2,500 to 5,000 bps (vendor claim).
6. Power and Energy Budgets
Battery-powered sensors and vehicles live or die on the transmit budget. The WHOI Micromodem-2 needs about 30 W nominal to transmit but draws about 300 mW while receiving, a 100-fold gap (30 W divided by 0.3 W), according to the WHOI Micromodem-2 design paper. That is why sensor networks favor short, rare messages and deep sleep states.
Research designs push the floor down: the ITACA modem idled at 10 uW and used about 120 mW to transmit, though at only 1 kbps over 240 m. At the high end, manufacturers advertise transmitters of 100 W (Popoto, vendor claim 2019).
| Metric | Value | Source |
|---|---|---|
| WHOI Micromodem-2 transmit power for 185 dB re 1 uPa at 1 m (source level) | 30 W nominal, 8 to 48 W range | WHOI, Micromodem-2 paper 2010 |
| Micromodem-2 receive draw vs hibernate draw (Micromodem-1 receive: 158 mW) | about 300 mW vs 165 uW at 5 V | WHOI, Micromodem-2 paper 2010 |
| Popoto modem transmitter power and board size | 100 W on a 60 mm × 90 mm board set | Popoto (vendor claim) 2019 |
| ITACA low-power modem: standby vs receive vs transmit | 10 uW vs 24 mW vs about 120 mW | Sanchez et al., Sensors 2012 |
| Acoustic glider modem standby power | as low as 5 mW | Zou and Sun, Sensors 2025 |
| Typical acoustic transmission power in an acoustic, radio and optical comparison | 10 W | Schirripa Spagnolo et al., Sensors 2020 |
| Sonardyne Modem 6 (8307) battery life, alkaline vs lithium | 833 days vs 1,390 days | Sonardyne product page (vendor claim) |
Context: the ITACA paper reports that receive consumption in earlier research modems was generally higher than 200 mW. Most recent available data: WHOI Micromodem-2 paper, 2010; Sanchez et al., 2012. Newer WHOI power figures exist only as an image table on its site.
7. Market Size and Operational Deployments
Market estimates disagree by a factor of about 20: from USD 96.4 million to USD 1.95 billion for the same year (1.95 billion divided by 96.4 million = 20.2). The mid estimate from Global Market Insights is USD 727.3 million in 2024 (GMI 2025), with Teledyne and L3Harris holding around 20% between them. The spread tells you scope definitions differ, so quote any single figure with its publisher.
The most durable public deployment is NOAA’s DART tsunami network, where an acoustic link carries pressure data from a seafloor recorder to a surface buoy (NOAA National Data Buoy Center).
| Metric | Value | Source |
|---|---|---|
| Global market value 2024, forecast and growth | USD 727.3 million; about USD 1.5 billion by 2034 at 7.7% CAGR | Global Market Insights 2025 |
| Global market value 2024 and 2033 forecast | USD 1.95 billion; USD 3.96 billion by 2033 at 8.2% CAGR | SkyQuest via Global Information Inc. 2025 |
| Global market value 2024 and 2030 forecast | USD 96.4 million; USD 198.9 million by 2030 at 12.8% CAGR | Market Glass via Global Information Inc. 2025 |
| Shallow-water modem share and point-to-point link share, 2024 | over 40% and around 50% | Global Market Insights 2025 |
| U.S. share of the North American market, 2024 | about 80% | Global Market Insights 2025 |
| NOAA DART network size reached in March 2008 | 39 stations | NOAA National Data Buoy Center |
| DART seafloor sampling interval and standard-mode reporting | every 15 seconds; reports every 6 hours | NOAA National Data Buoy Center |
| DART stations maintained by NOAA as of 2020, by basin | 32 Pacific, 7 Atlantic | DOSITS 2021 |
Context: the alternative estimates come from the SkyQuest report listing and the Market Glass report listing; the basin split is from DOSITS (University of Rhode Island). Most recent available data for the DART counts: NOAA NDBC 2008 and DOSITS 2020.
Summary: Underwater Acoustic Modem Bitrate and Distance by the Numbers
| Metric | Value | Source |
|---|---|---|
| Nominal speed of sound in seawater | about 1,500 m/s | Qarabaqi and Stojanovic 2009 |
| One-way delay over 1 km | about 0.67 s | Derived: 1,000 m divided by 1,500 m/s |
| Acoustic vs radio attenuation in seawater | 0.1 to 4 dB/km vs 3.5 to 5 dB/m | Schirripa Spagnolo et al. 2020 |
| Typical rate on links of 100 km or more | 1 kbps or less | Xu, Irigoien and Alouini 2024 |
| Typical rate on links of 1 km or less | up to 100 kbps | Xu, Irigoien and Alouini 2024 |
| Long-standing rate-distance benchmark | 40 km·kbps | Shen et al. 2008 |
| Longest error-free link reported | 600 km at 37.5 bps | Northwestern Polytechnical University, reported 2025 |
| Sonar data from 7,000 to 7,500 m depth | about 58 kbps | JAMSTEC press release 2026 |
| Acoustic glider maximum slant range | 5,113 m | Zou and Sun 2025 |
| JANUS baseline rate vs fast-mode top payload rate | 80 bps vs 34.5 kbps | Li and Zheng 2022 |
| JANUS promulgated as STANAG 4748 | March 24, 2017 | Petroccia and Alves 2024 |
| WHOI Micro-Modem burst vs throughput (Rate 5) | 5,388 vs 3,343 bps | WHOI Acoustic Communications Group 2026 |
| Low-power modem standby draw | 10 uW | Sanchez et al. 2012 |
| LinkQuest UWM3000H range and rate | 6,000 m at 80 to 320 bps (vendor claim) | LinkQuest |
| Global market value 2024 (mid estimate) | USD 727.3 million | Global Market Insights 2025 |
| Range of 2024 market estimates | USD 96.4 million to USD 1.95 billion | Market Glass; SkyQuest 2025 |
Methodology and Sources
- Qarabaqi and Stojanovic, Statistical Modeling of a Shallow Water Acoustic Communication Channel (2009): sound speed, travel time, multipath delay and coherence time.
- JAMSTEC Underwater Acoustic Technology Research Group and JAMSTEC Urashima 8000 press release (May 15, 2026): Doppler, rate-distance product, moving-vehicle link and deep sonar transmission.
- Xu, Irigoien and Alouini, State-of-the-Art Underwater Vehicles and Technologies Enabling Smart Ocean, arXiv 2412.18667 (2024): distance and data rate bands, frequency bands.
- Shen et al., Journal of Northwestern Polytechnical University (2008): 40 km·kbps benchmark and 5 km lake test.
- Schirripa Spagnolo, Cozzella and Leccese, Underwater Optical Wireless Communications: Overview, Sensors 2020: acoustic, radio and optical comparison.
- South China Morning Post (May 24, 2025) and Indian Defence Review (May 25, 2025): reporting on the 600 km trial.
- Zou and Sun, An Acoustic Underwater Glider for the Real-Time Transmission of Observation Data, Sensors 2025: slant range and standby power.
- Bergh et al., A Modular Smart Ocean Observatory, Sensors 2024: cNODE MiniS link and packet loss.
- Li and Zheng, Experimental Evaluation of JANUS Fast Modes in Very High Acoustic Frequency Bands, IEEE Journal of Oceanic Engineering 2022: JANUS parameters and fast-mode results.
- Petroccia and Alves, The JANUS Underwater Communications Standard: From Promulgation to Present, OCEANS 2024: STANAG 4748 date and status.
- IEEE Spectrum, NATO Develops First Standardized Acoustic Signal for Underwater Communications (2017) and Ocean News and Technology (2017): JANUS launch details.
- WHOI Acoustic Communications Group, Micro-Modem and Gallimore et al., The WHOI Micromodem-2 (2010): burst rates, throughput and power.
- Sanchez et al., An Ultra-Low Power and Flexible Acoustic Modem Design, Sensors 2012: low-power modem figures.
- Vendor claims: LinkQuest model list, Sonardyne Modem 6, EvoLogics on Naval Technology, Popoto (2019).
- Market: Global Market Insights (January 2025), SkyQuest via Global Information Inc. (December 2025), Market Glass via Global Information Inc. (July 2025).
- Deployments: NOAA National Data Buoy Center, DART and DOSITS (2021).
- Data watch: the 600 km, 37.5 bps result comes from press reports of a study in Acta Acustica that we could not open, so treat it as reported rather than verified. Market estimates for 2024 range from USD 96.4 million (Market Glass) to USD 727.3 million (GMI) to USD 1.95 billion (SkyQuest), a roughly 20-fold spread driven by scope; the GMI page calls the 2024 figure both “valued at” and “surpassed” USD 727.3 million. The WHOI Micro-Modem page lists rates up to 5,400 bps, while the 2010 Micromodem-2 paper says 80 to 5,000 bps burst. Li and Zheng give the JANUS band as 4.1 kHz in their abstract and 4.16 kHz in their table. Several physics and power figures are older than three years (2008 to 2022) because open field data is scarce; the physics is stable, but newer modem power data would sharpen section 6. Vendor ranges and rates are best-case claims under favorable conditions, not measured averages.
Last updated: October 25, 2026. We update this roundup quarterly, and the next refresh is expected when the IEEE Journal of Oceanic Engineering publishes its next underwater communications issue and NATO CMRE reports on the latest STANAG 4748 revision.