Push-to-talk VHF/UHF radio geolocation from space
Short-duration VHF/UHF push-to-talk radios are hard to fix from orbit precisely because they are brief and unpredictable. Multi-satellite TDOA/FDOA geometry can locate them, but only when the intercept probability problem is solved first.
Sensors
- HawkEye 360 RF cluster: Three-satellite formation flying at roughly 575 km altitude, spaced to produce TDOA and FDOA baselines across VHF and UHF bands (approximately 100 MHz to 3 GHz). Published geolocation accuracy for sustained emitters is in the range of 0.5–3 km CEP depending on geometry; short-duration emitters degrade that figure. Revisit at any given point is typically several times per day but is not uniform.
- Kleos Space RF: Cluster-based formation of four satellites per scouting mission, operating in similar VHF/UHF ranges. Designed specifically for short-duration emitter collection; published cluster separation is tuned to produce usable TDOA baselines for signals as brief as a few hundred milliseconds. Coverage is global but revisit cadence per location depends on orbital slot and cluster count.
- Spire SDR payload: Software-defined radio payloads hosted on Spire's LEMUR-2 bus, covering portions of VHF and UHF. The large constellation (100-plus satellites) improves probability of intercept for unpredictable emitters but individual satellites are not formation-flown, so standalone TDOA geolocation requires cross-cueing with other assets or relies on FDOA alone, which reduces fix accuracy.
- Generic hosted-payload SDR (opportunistic): Several commercial operators offer hosted SDR capacity on LEO buses without dedicated formation flying. These can collect signal presence and frequency data but cannot independently produce a confident TDOA geolocation fix for a two-second PTT burst without coincident passes from at least two other receivers with known, stable baselines.
Why a two-second burst is a hard target
Conventional VHF and UHF push-to-talk radios transmit in bursts that typically last between one and ten seconds. A satellite at 550 km altitude crosses a ground point in roughly eight to twelve minutes, so the orbital geometry for a single pass is favourable. The problem is not geometry. It is time. A PTT user who transmits once, unpredictably, during a six-hour window has a low probability of being overhead any given satellite at the moment of transmission. With a single satellite, that probability can be well below five percent per pass.
The solution the commercial RF-monitoring industry has converged on is cluster formation flying: two or more satellites separated by tens to hundreds of kilometres, flying coordinated passes so that any transmission within the footprint is received by all cluster members simultaneously. That simultaneity is what makes time-difference-of-arrival (TDOA) and frequency-difference-of-arrival (FDOA) geolocation possible for signals too brief to track across sequential passes.
TDOA and FDOA: what the geometry actually requires
TDOA measures the difference in signal arrival time at two receivers. For a two-satellite baseline at LEO altitudes, one TDOA measurement constrains the emitter to a hyperbolic surface. A second baseline from a third satellite collapses that to a point, or more precisely to an error ellipse whose size depends on baseline length, signal bandwidth, and signal-to-noise ratio. HawkEye 360's published literature describes three-satellite cluster geometries designed to produce two independent TDOA measurements per pass. For a narrowband VHF signal with a bandwidth of 12.5 kHz (the standard channel spacing for analogue FM PTT in many jurisdictions), timing resolution is limited by the inverse of that bandwidth, which bounds TDOA accuracy to roughly 80 microseconds and translates to a position uncertainty on the order of tens of kilometres from TDOA alone.
FDOA partially compensates. The Doppler shift on a received signal differs between two satellites moving at slightly different velocities relative to a stationary ground emitter. FDOA adds an independent geometric constraint and, critically, does not require the signal to be long, only detectable. Combining TDOA and FDOA from a three-satellite cluster, HawkEye 360 has published geolocation accuracies of approximately 0.5 to 3 km CEP for emitters with sufficient signal duration and SNR. Short PTT bursts at the lower end of that duration range, or at range from the cluster boresight, will sit toward the worse end of that accuracy band.
Probability of intercept: the honest arithmetic
Probability of intercept (POI) is the fraction of transmissions that a constellation will actually capture. It depends on three things: how often the emitter transmits, how long each transmission lasts, and how much of the day a cluster footprint covers the area of interest. For a PTT user transmitting ten times a day for three seconds each, the total on-air time is thirty seconds out of 86,400. A single three-satellite cluster with a footprint dwell of roughly eight minutes per pass, making perhaps four passes per day over a given location, covers about thirty-two minutes of the day. The naive POI for that scenario is around 0.037 percent per transmission, or roughly a one-in-three chance of capturing at least one transmission across a full day of monitoring.
Stacking multiple clusters, as Kleos Space does across its scouting missions, or cross-cueing with Spire's larger constellation to identify when a target is transmitting and then tasking a formation cluster, materially improves POI. But there is no honest way to guarantee intercept of a low-duty-cycle emitter within a single day. Persistent collection over multiple days is the realistic operational model. Buyers should ask vendors for POI estimates specific to their target area's latitude, the number of active clusters, and the assumed transmission duty cycle.
What the fix actually tells you, and what it does not
A confirmed TDOA/FDOA fix places the emitter within an error ellipse, typically elongated along the satellite track direction because cross-track TDOA geometry is stronger than along-track. At best-case geometry with a three-satellite HawkEye 360 cluster, that ellipse might be 0.5 km by 2 km. At poor geometry, it can be 5 km by 20 km. That is enough to identify a district, a road corridor, or a coastline segment, but not a specific building.
The fix also tells you the carrier frequency and, for analogue FM, sometimes the audio content if the collection system is configured for demodulation. It does not tell you who is transmitting unless correlated with frequency licensing databases or prior signal intelligence. Repeated fixes on the same frequency and signal fingerprint over multiple days can establish a pattern of life, which is often more operationally useful than a single precise location.
Operational limits worth stating plainly
Cloud cover is irrelevant to RF collection, which is one genuine advantage over optical or SAR tasking for this application. Ionospheric absorption affects the VHF band below roughly 30 MHz but has negligible impact on the 100 MHz to 500 MHz range where most PTT radios operate. Terrain matters: a transmitter in a deep valley or beneath a dense forest canopy will have a reduced effective radiated power toward the satellite, reducing SNR and degrading or preventing detection.
Frequency agility is a countermeasure. A radio operator who hops channels between transmissions will produce fixes on different frequencies, complicating signal association. Spread-spectrum or frequency-hopping waveforms are largely invisible to current commercial RF-monitoring constellations, which are optimised for narrowband analogue and digital PMR446, MURS, and similar conventional PTT standards. Encrypted digital PTT (DMR, P25, TETRA) can still be geolocated by its RF envelope even when the content is inaccessible.
Satellize can structure a collection and analysis tasking plan around a client's specific area of interest and frequency range of concern. The Tonga crop-estimation programme is a different analytics domain, but the underlying approach of defining a precise analytical question before selecting sensors applies equally here. Speak to the team about which commercial RF constellation has active clusters scheduled over your region of interest in the next thirty days.
Typical figures
| Frequency coverage (typical commercial RF LEO) | ~100 MHz to 3 GHz; VHF/UHF PTT bands (136–174 MHz, 400–512 MHz) well within range |
| Geolocation accuracy (TDOA/FDOA, three-satellite cluster) | 0.5–3 km CEP for sustained emitters; 3–20 km CEP for sub-three-second bursts at poor geometry |
| Minimum detectable signal duration | Approximately 100–300 ms for FDOA alone; ~1 s preferred for combined TDOA/FDOA fix |
| Revisit per location (single cluster) | Typically 2–6 passes per day depending on latitude; non-uniform |
| Probability of intercept (low-duty-cycle PTT, single cluster, single day) | Highly variable; can be below 5% per transmission for infrequent emitters |
| Geolocation error ellipse orientation | Elongated along satellite track; cross-track accuracy stronger than along-track |
| Latency from collection to fix delivery | Hours to sub-day for near-real-time pipelines; archive delivery within 24 h typical |
| Archive depth (HawkEye 360) | Operational since 2018; historical collection available through commercial licensing |
| Delivery formats | GeoJSON emitter fix reports, KML overlays, time-stamped signal event logs, API feed |
Analytics Satellize can run
| Single-event emitter fix | TDOA/FDOA from three-satellite cluster pass; hyperbolic intersection with Doppler constraint | GeoJSON point with error ellipse semi-axes, timestamp, carrier frequency, and confidence score |
| Multi-day pattern-of-life report | Temporal clustering of repeated fixes on matching frequency and signal fingerprint across successive passes | PDF report with fix map, transmission timeline, and inferred activity rhythm |
| Probability-of-intercept forecast | Orbital propagation of available cluster assets over defined area of interest; duty-cycle modelling against client-supplied transmission assumptions | Collection planning brief showing expected POI per day for a 30-day window |
| Frequency occupancy context layer | Aggregated signal presence counts per channel across all passes, independent of geolocation quality | GIS raster layer showing detected channel activity density by grid cell |
| Signal fingerprint association | Comparison of modulation type, symbol rate, and spectral mask across multiple intercepts to link transmissions to a common device | Association table linking fix events to probable common emitter, with confidence rating |
| Terrain-adjusted detection probability map | Line-of-sight modelling from satellite ephemeris against digital elevation model to flag areas where low-elevation transmitters will be shadow-masked | GeoTIFF mask layer overlaid on area of interest, identifying collection blind spots |
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.