AISHub Open AIS Data Sharing Network

It is crowd-sourced infrastructure with an unusual business logic: hundreds of hobbyists and developers run shore-based AIS receivers, pool what their antennas hear, and everyone reads back one merged global feed. The scale is real, not aspirational. At research time in August 2026 the network counted 1,338 stations online and 287 offline out of 1,622-1,625 registered, spread across 83 countries, and reported 100,680 distinct vessels seen in the trailing 24 hours.

Each vessel record is a compact identity-plus-kinematics bundle. The MMSI anchors the row; latitude and longitude arrive in decimal degrees; course over ground and speed over ground describe the transit; true heading and rate of turn refine it; the navigational-status code says whether the hull is under way, anchored or moored. Voyage context rides along too - reported IMO number, name, call sign, ship-type code, the A/B/C/D dimension offsets, static draught, declared destination and the broadcast ETA.

Two things matter when you size this record against commercial AIS vendors. First, the geometry is terrestrial: coverage is excellent where volunteers have antennas and patchy where they do not, which is precisely why the station layer ships with the vessel layer. Second, the aggregate moves constantly, so the interesting analyses - anchorage density, port-approach queues, fleet behaviour - come from persistence, not from a single pull. Datadory handles both: the feed arrives normalised, and your history starts accumulating from the first delivery.

The verified sample below is a station-status record - the contributor-side half of the schema:

record_type      : station_status
station_id       : 2007
country          : vn
location         : Ho Chi Minh
latitude         : 10.77
longitude        : 106.77
last_update_unix : 1787320802

Read it as the provenance spine of the whole dataset: every vessel fix in the aggregate reaches the hub through a station shaped like this one, pinned to a country and a coastline, stamped with its last update. Vessel rows join to that geography through position and time, and station rows carry per-site activity counts that tell you how much traffic a given stretch of water is actually hearing. Get a sample of this dataset and the same two-surface schema lands cut to whichever waters, fleet list or freshness window you name.

Roughly twenty fields define the record set, and they divide cleanly: nine identity and voyage fields say which ship, seven say where and how it moves, the dimension quartet and draught say what kind of hull, and a separate station layer documents the receiving network itself. Field names were checked against the network's published documentation during research and are marked verified; the response envelope adds a top-level error flag so failed queries announce themselves rather than returning silence.

The table below is the confirmed core. Anything beyond it - the raw-encoding extras, the full station attribute set - is listed as an on-request note instead of being padded with guessed columns.

Geography - 83 countries field stations, and the footprint hugs coastlines: straits, port approaches and anchorages light up where hobbyist receivers cluster. That makes the network strong exactly where port and supply-chain work happens - the last fifty miles - and weaker in mid-ocean, where satellite AIS normally takes over.

Temporal - fixes are live, each carrying the timestamp of the vessel's last received report, and a maximum-fix-age filter keeps stale positions out of a pull. Research observed between 100,537 and 100,680 distinct vessels inside a rolling 24-hour window. There is no published historical archive, so trend lines start accruing from your first delivery rather than reaching back years.

Granularity - one snapshot per vessel per pull, filterable by bounding box, by named vessel list or by age of fix, with station-status records reporting per-site activity beside them. That is the grain port analysts actually need: not "the Singapore Strait is busy" but "this MMSI was here, doing this, twelve minutes ago."

API, files, or your warehouse. Daily, weekly, or hourly.

Your cadence is decoupled from the rhythm of the antenna network underneath - take a snapshot when you need one, or keep a warehouse table current so newly sighted vessels diff cleanly into yesterday's rows. Deliveries arrive normalised to the field dictionary above with the two-surface structure intact: vessel records and station records typed separately, so joins against your own fleet list, port geography or berth schedule hold up without parsing raw payloads. Name the waters, the vessels and the freshness window, and the sample returns in exactly the schema shown above.

• Port operations and berth planning - watch anchorage occupancy and approach queues build in near real time, and size waiting time from position history rather than anecdote.
• Trade-flow and commodity analysts - proxy export and import momentum by counting hull movements through chokepoints, weeks before customs statistics publish.
• Fleet monitoring and chartering desks - track named vessels by MMSI or IMO number, with speed, course, draught and destination attached to every fix.
• Machine-learning teams - train trajectory prediction, port-call estimation and anomaly models on position streams that carry navigational status and rate of turn, not just lat/lon dots.
• Journalists and academics - study shipping behaviour, sanctions-evasion patterns and dark-fleet gaps with a citable, community-built source of record.
• Supply-chain visibility teams - fuse vessel ETA broadcasts with inland logistics planning instead of waiting on carrier notifications.

Developers and data-product builders get a live position stream with clean machine-readable fields to prototype against - see the developers builders use cases page. Data scientists and ML engineers get labelled kinematics (status codes, rate of turn, draught) that most free position dumps strip out - detailed on the data scientists use cases page. Competitive intelligence and product teams get movement-level evidence for market and route questions - covered under competitive intel product teams use cases. Market researchers and consultants get the station-network statistics themselves as a coverage dataset - see market researchers use cases.

Within marine ports and services data, this record owns the cooperative live-position layer. Its nearest neighbour, the VT Explorer AIS Data API, is the slice's commercial counterpart: credit-based live positions plus voyage data, vessel particulars and port calls covering 300,000+ vessels tracked daily. Weigh the trade-offs side by side in the vs VT Explorer AIS Data API comparison.

Static geography pairs cleanly with moving hulls: Natural Earth World Ports adds a public layer of 1,081 world-port points, BTS Principal Ports adds 150 U.S. port polygons with CY 2023 tonnage, and World Bank Container Port Traffic supplies the throughput series to contextualise what those anchored hulls are waiting for. If your question is "which vessels are off this coast right now, and what are they doing", this is the record; if it is "how many containers passed through the terminal last year", start with the throughput indicators instead.

Three things worth knowing upfront. First, this is a present-tense dataset: there is no published historical archive of past positions, so longitudinal studies begin accruing from your first delivery. Second, quality varies by volunteer station - receiver placement, antenna height and local interference all show up in the fix stream, and the station layer lets you see which sites sit behind any given stretch of water. Third, the field dictionary above is the confirmed core; raw-encoding extras and the full station attribute set confirm with your sample rather than appearing here speculatively. Name the bounding box, the fleet list and the freshness window, and the sample returns cut to fit.