The complete system, from a signal in the sky to the map on your screen

Aircraft announce themselves

Most aircraft carry a transponder that broadcasts a short radio message about twice a second. This is called ADS-B. Each message contains the aircraft's unique 24-bit identity code, and usually its GPS position, altitude, speed and heading, plus its flight number when the crew has set one. The aircraft is effectively calling out "I am here" to anyone within range, and the broadcast is open and unencrypted.

Because it is an open signal, anyone with a suitable antenna can receive it. Everything else is built on top of that one fact.

Small receivers on the ground listen in

A receiver is usually nothing more than a single-board computer, a small radio dongle, and an antenna near a window or on a roof. Software on it decodes the messages into a stream of aircraft reports. One receiver can typically hear aircraft out to a few hundred kilometres, limited mainly by the curve of the earth.

No single receiver sees the whole sky, so PlaneTrack combines many. Some coverage comes from a broad pool of receivers shared across the wider tracking community, and some comes from people who point their own receiver directly at PlaneTrack. The direct feeders are what push coverage into places the shared pool cannot reach, and that part is open to anyone.

Two ways a receiver sends its data

A contributor does not need a second antenna or a spare computer. They simply send a copy of what their receiver already decodes. There are two ways to do it, and both arrive at the same place:

Either way, the receiver keeps doing everything it already does. PlaneTrack is just one more destination it sends to.

Everything flows into one stream

Here is the central idea: no matter where a message comes from, it lands in a single high-speed stream. From that moment, one source is treated exactly like any other. Nothing is a special case.

   direct submit        shared pool         raw stream + timing
        |                   |                       |
        v                   v                       v
   +--------------------------------------------------------------+
   |               one real-time message stream                   |
   +--------------------------------------------------------------+
                              |
                              v
              aggregator  (one live record per aircraft)
                              |
                 +-----------------------------+
                 v                             v
          current position               position history
                 |                             |
                 +-------------+---------------+
                               v
              enrichment, routing and estimation
                               |
                               v
                 live map  .  flight pages  .  API

An aggregator keeps a single up-to-date record for each aircraft, keyed by its identity code. If several receivers hear the same aircraft, they resolve to the same record and the aggregator simply keeps the freshest details. Two things are then saved: the aircraft's current state, which is what you see moving on the map, and its position history, which powers tracks and replays.

Locating aircraft that send no position

Not every aircraft transmits its GPS position. Older or lighter aircraft may send only a short reply with their identity and altitude, but no location. On their own, these are invisible on a position map. Multilateration, or MLAT, solves this.

When the same silent aircraft is heard by several receivers at once, each notes the exact instant it heard the message, down to fractions of a microsecond. Radio travels at the speed of light, so a receiver slightly further away hears it slightly later. By comparing those tiny timing differences across receivers, the position can be calculated by geometry, which is the same principle as GPS worked in reverse.

Working out what the aircraft is

A raw signal only carries a 24-bit code and a flight number. To turn that into "an Airbus A350 operated by an airline, registered somewhere," PlaneTrack matches the code against aviation registries: large public databases that link each aircraft's code to its registration, model, operator and category. That is how the map knows to draw a jet differently from a helicopter or a light aircraft, and how a flight gets an airline name.

Where a photograph of that exact airframe exists, it is shown too. When it does not, PlaneTrack falls back to a clean profile of the aircraft type so there is always something recognisable to look at.

Working out where it is going

A live position does not say where a flight started or where it is headed. PlaneTrack works that out by combining several kinds of evidence, and cross-checking them so a single bad source cannot mislead the map:

• The flight's own track. Where the aircraft was first seen climbing away, and where it is descending toward, are strong clues to the true origin and destination.
• Published schedules. Known airline timetables give expected departure and arrival airports and times for a given flight number.
• Observed patterns. The same flight number flown day after day builds a picture of its usual route, which helps when other sources are missing or disagree.

These are weighed against each other. If the schedule says one destination but the aircraft is clearly descending somewhere else, the observed track wins. The result is a best-estimate route, with scheduled and estimated times where they are known, and an honest "predicted" label when the route is inferred rather than certain.

Filling the gaps beyond the horizon

Ground receivers cannot cover open oceans, deserts or polar routes, so aircraft regularly fly out of range for hours. Rather than have them vanish from the map, PlaneTrack estimates where they are while they are dark.

Using the aircraft's last known position, speed and intended route, it projects the most likely path forward along the natural great-circle line between origin and destination, adjusted for forecast winds aloft so the estimate does not drift. These estimated positions are clearly marked as such, and are replaced by real data the instant the aircraft comes back within range of any receiver.

Turning a firehose into smooth motion

Positions arrive as discrete updates, not a continuous glide. If the map only jumped whenever a new message landed, aircraft would stutter across the screen. Instead, the map animates smoothly between updates by carrying each aircraft forward along its known heading and speed, then correcting gently as the next real position arrives. The effect is fluid movement that still stays true to the data.

On top of that sit the things that make the map useful at a glance: colouring aircraft by altitude or speed, filtering by airline or type, highlighting emergencies, and drawing a selected aircraft's full path as a single connected line, with the detected portion, the assumed portion and any prediction shown as one continuous track that follows the curve of the earth.

Every position is remembered

The live map is only the present moment. Behind it, every position is stored, which turns the system from a live view into a searchable record. That memory powers several things:

• Flight tracks and replay. Scrub back through a flight, or replay the whole sky at a chosen moment in the past.
• Per-flight and per-aircraft pages. A page for each flight number and each airframe, showing its recent legs and history.
• Stitching legs together. The same aircraft often flies several legs a day under different flight numbers; the history links them so a track shows the current flight, not an accidental blend of the last one.

Opening the data up

Everything the map shows is also available to build on. Developers can query live and historical data through a metered API, and pull larger one-off exports for analysis. Businesses that need volume or guarantees can go further with commercial access. The same clean, merged, enriched data that draws the map is what flows through these products, so what you build on matches what you see.

Keeping it honest and online

A live system is only as good as its weakest quiet failure, so PlaneTrack watches itself. Automated checks continuously confirm that data is not just present but fresh, that every stage of the pipeline is running, and that storage has room. When something stalls, it is restarted automatically and an alert is raised, so gaps are measured in minutes rather than discovered days later.

The system also publishes where it is blind: a live view of busy airports with little or no reception. That honesty is practical, because it tells the community exactly where one new receiver would do the most good.

What contributors get back

Running a receiver costs a few watts of power and a corner of a windowsill. In return, every contributor gets full Premium membership for free while their data is flowing. Their contribution is credited automatically by a key unique to them, and their account shows a live view of their feed: messages per day, distinct aircraft seen, and a running history. Premium stays active as long as data keeps arriving, pausing only after a long silence and resuming on its own when data returns.

That is the whole loop: open signals, gathered by many small receivers, merged and made sense of centrally, and given back as a live map, a searchable history and an open data feed, with the people who extend coverage sharing in the result.

Antenna to map, in seconds