False-alert discrimination for 406 MHz distress beacons from LEO
More than 97% of 406 MHz EPIRB and PLB activations processed by Cospas-Sarsat are false alerts. Spaceborne RF receivers can cross-check signal fingerprints, registration records and positional context to help rescue coordination centres triage faster.
Sensors
- Cospas-Sarsat MEOSAR payloads (GPS, GLONASS, Galileo): 406 MHz receivers hosted on MEO navigation satellites. MEOSAR provides near-global, near-instantaneous coverage and can deliver a location solution within minutes via TDOA/FDOA processing. Published system documentation reports position accuracy of 5 km or better for beacons without integrated GNSS, and sub-100 m when the beacon encodes an internal GNSS fix in its message.
- Cospas-Sarsat LEOSAR payloads (NOAA, MetOp, Sarsat LEO hosts): Legacy 406 MHz receivers on polar LEO satellites. Coverage per pass is limited to roughly 4,000 km swath; a beacon may wait up to 90 minutes for a pass at low latitudes. Useful for archive comparison but MEOSAR has largely superseded it for latency-critical work.
- Spire STRATOS RF payload: Software-defined radio receivers on Spire's LEO constellation of over 100 satellites. Capable of recording raw IQ data in the 400 MHz band. Revisit at any given point is typically under 30 minutes globally. Enables independent signal capture outside the Cospas-Sarsat processing chain, useful for waveform-level fingerprinting.
- AIS receivers (spaceborne): VHF AIS on LEO satellites provides vessel identity and position. Cross-referencing a beacon's registered MMSI against live AIS tracks is one of the fastest consistency checks available. Latency from commercial spaceborne AIS to a data feed is typically under 10 minutes.
- ADS-B receivers (spaceborne): For aviation ELTs and PLBs associated with aircraft, spaceborne ADS-B provides ICAO hex code and position. Confirming that a registered aircraft is airborne, stationary on the ground, or absent from ADS-B altogether is a direct discriminant for genuine versus accidental activation.
Why 97 percent false is not an acceptable baseline
Cospas-Sarsat's own published system performance reports have consistently shown that false alerts account for more than 97 percent of all 406 MHz activations processed globally. The causes are well documented: accidental triggering during vessel maintenance, hydrostatic release failures, battery replacement without proper isolation, and deliberate misuse. Each false alert consumes real resources at a rescue coordination centre and, more critically, dilutes the attention paid to genuine distress.
The problem is not detection. The Cospas-Sarsat network is exceptionally good at detecting 406 MHz signals. The problem is classification. A signal that arrives at a local user terminal looks identical whether it came from a yacht sinking in the Tasman Sea or from a beacon knocked off a shelf in a chandlery. The classification work has to happen downstream, and it has to happen fast.
What the 406 MHz message actually contains
Every compliant 406 MHz beacon transmits a 112-bit digital message at 1 W, nominally every 50 seconds. That message encodes a country code, a unique beacon identifier, and optionally an encoded internal GNSS position and a vessel or aircraft identifier. Beacons with an integrated GNSS fix encode position to roughly 100 m. Those without rely on the Cospas-Sarsat network's TDOA and FDOA processing across multiple satellites to derive a position, which for MEOSAR is typically within 5 km.
The 15-digit hexadecimal beacon ID links to a national registration database entry that should contain the owner's name, contact details, vessel name, MMSI and home port. In practice, a significant fraction of registered beacons have outdated or incomplete records. That gap is itself a discriminant: a beacon with no registration, or one registered to a vessel that was sold years ago, carries a different prior probability of genuine distress than one with a current, verified record attached to a vessel that filed a voyage plan.
RF fingerprinting: the signal knows more than the message
Beyond the encoded message, the raw waveform carries information that the Cospas-Sarsat standard does not specify. Carrier frequency offset relative to 406.028 MHz, phase noise characteristics, modulation accuracy and power ramp profile all vary between beacon manufacturers and individual units. A beacon that has been triggered accidentally in a workshop will often show a stable, stationary Doppler profile. A beacon on a vessel moving at 10 knots in distress will show a Doppler rate consistent with that motion.
Spire's STRATOS payload, which captures raw IQ data rather than decoded messages only, allows this kind of waveform-level analysis. Comparing the observed frequency offset and Doppler trajectory against the registered vessel's last known AIS position and speed provides a kinematic consistency check that the standard Cospas-Sarsat processing chain does not perform. This is not infallible. A vessel that has lost propulsion and is drifting will show a low Doppler rate that superficially resembles a stationary false alert. Honest analysis requires stating that ambiguity and flagging it for human review rather than auto-dismissing.
Kinematic consistency and the multi-source cross-check
The most operationally useful discriminants combine several independent data streams. First, the beacon's encoded or network-derived position is compared against the registered vessel's last AIS fix. If the vessel's AIS shows it berthed in a marina 200 km from the beacon position, that is a strong false-alert indicator, though not conclusive: AIS can be switched off. Second, the beacon identifier is checked against the national database. Third, if a contact number is on record, a voice call can confirm or deny distress within minutes. Fourth, for aircraft, ADS-B position and flight status provide an equivalent check.
Where spaceborne AIS and ADS-B receivers add value is in the speed and geographic completeness of that cross-check. A rescue coordination centre covering a remote ocean area may have no terrestrial AIS coverage. A LEO constellation with AIS payloads can provide a vessel position fix from the same orbital pass that detected the beacon, reducing the cross-check to a single data pull rather than a multi-agency phone chain.
The limit of all these checks is that they are probabilistic. A vessel in genuine distress may have a lapsed registration, a switched-off AIS transponder and a stationary Doppler profile because it has already stopped moving. The output of any automated discrimination system must be a confidence score and a recommended action, not a binary suppress-or-alert decision.
Where the analysis sits within a rescue coordination workflow
Cospas-Sarsat alert messages are delivered to national mission control centres and then to rescue coordination centres, typically within minutes for MEOSAR. The discrimination analysis described here is designed to arrive at the rescue coordination centre alongside or shortly after the Cospas-Sarsat alert, not to replace it. The goal is to give the duty officer a structured summary: beacon ID, registration status, last known vessel position from AIS, kinematic consistency score, and a recommended first action.
Satellize's RF analytics capability is built to ingest Cospas-Sarsat alert feeds, cross-reference against spaceborne AIS and ADS-B data, and return a structured confidence assessment. The architecture is the same one applied in the Tonga crop-estimation programme in the sense that it fuses multiple open and commercial data streams into a single analytical output, though the operational domain is entirely different. The deliverable for a rescue coordination centre is a triage report, not a map layer.
Latency is the binding constraint. A discrimination product that arrives 45 minutes after the alert has limited value if the duty officer has already dispatched a helicopter. The target for any operationally useful system is to deliver the cross-check within the first 10 minutes of alert receipt, which is achievable with pre-positioned data pipelines but requires deliberate engineering, not an ad hoc query.
Honest limits of the method
Several failure modes are worth stating plainly. Beacons without integrated GNSS rely on network-derived positions that carry 5 km uncertainty, which is enough to place a beacon ambiguously between a marina and open water. Registration databases in many countries are poorly maintained; Cospas-Sarsat's own documentation acknowledges this as a systemic problem. RF fingerprinting from a single LEO pass gives one Doppler arc, not a continuous track, so kinematic inference is limited. And a small number of genuine distress events will score poorly on every discriminant simply because the circumstances are unusual.
None of this argues against doing the analysis. It argues against treating the output as a decision rather than as evidence. The operational value is in reducing the number of alerts that require a full SAR response from, say, 100 percent to something closer to the genuine-distress rate, while ensuring that every high-confidence genuine alert is escalated immediately.
Typical figures
| Operating frequency | 406.028 MHz (±2 kHz per Cospas-Sarsat specification) |
| Beacon position accuracy (MEOSAR, no internal GNSS) | Typically within 5 km; sub-100 m when beacon encodes internal GNSS fix |
| Alert latency (MEOSAR) | Near-instantaneous coverage; alert delivery typically within 5 minutes of activation |
| Alert latency (LEOSAR) | Up to 90 minutes at low latitudes pending satellite pass |
| Spire STRATOS revisit (global average) | Under 30 minutes at any point globally across 100+ satellite constellation |
| Spaceborne AIS cross-check latency | Typically under 10 minutes from commercial LEO AIS feed |
| Beacon message length | 112 bits, transmitted at 1 W, nominally every 50 seconds |
| Published false-alert rate | Exceeds 97% of all activations per Cospas-Sarsat system performance reports |
| RF fingerprint discriminants | Carrier frequency offset, Doppler rate, phase noise, modulation accuracy, power ramp profile |
| Cross-check data sources | Cospas-Sarsat alert feed, national beacon registration database, spaceborne AIS, spaceborne ADS-B |
Analytics Satellize can run
| Registration consistency score | Automated lookup of 15-digit beacon hex ID against national and international Cospas-Sarsat registration databases; flags missing, expired or mismatched records | Structured alert annotation delivered alongside RCC triage report |
| Kinematic consistency assessment | Comparison of beacon-derived or encoded position against vessel's last spaceborne AIS fix; Doppler rate analysis from raw IQ capture where available via Spire STRATOS | Confidence score with supporting position delta and speed estimate, included in triage report |
| ADS-B cross-check for aviation PLBs and ELTs | ICAO hex code lookup against spaceborne ADS-B feed; determination of whether registered aircraft is airborne, on ground, or absent from coverage | Flight status annotation in triage report, flagged for immediate escalation if aircraft absent from ADS-B in remote area |
| Multi-source false-alert probability estimate | Bayesian fusion of registration status, kinematic consistency, AIS/ADS-B cross-check and historical false-alert base rate; outputs posterior probability of genuine distress | Single probability figure with confidence interval and recommended first action, delivered as structured JSON feed or PDF triage report |
| Repeat-offender beacon flagging | Time-series analysis of activation history for a given beacon ID; identification of beacons with multiple prior false alerts in the Cospas-Sarsat archive | Beacon history summary appended to triage report; persistent watchlist layer for RCC dashboard |
| Coverage gap identification for cross-check data | Analysis of spaceborne AIS and ADS-B pass geometry relative to beacon position; flags cases where no independent positional confirmation is available within the triage window | Data-gap annotation in triage report, prompting duty officer to weight RF evidence more heavily |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.