Electronic warfare emitter geolocation from space
Commercial RF-geolocation satellites use time- and frequency-difference-of-arrival to locate ground-based radars and jammers without intercepting content. Position and frequency, not signals intelligence.
Sensors
- HawkEye 360 RF geolocation constellation: Clusters of three satellites flying in formation collect RF signals across roughly 150 MHz to 18 GHz. TDOA/FDOA processing yields geolocations with published accuracy of approximately 1 km CEP under good geometry; accuracy degrades significantly when only one cluster is in view or signal duration is under a few seconds.
- Kleos Space RF geolocation constellation: Four-satellite clusters in multiple orbital planes, targeting HF through X-band emissions. Revisit improves as the constellation grows; the company's published design aim is sub-kilometre geolocation accuracy with sufficient cluster geometry.
- Spire Global GNSS radio-occultation and AIS/ADS-B: Spire's 100-plus-satellite constellation is primarily atmospheric sounding and vessel/aircraft tracking, but its RF-monitoring payloads contribute signal-environment data. Not a geolocation instrument in the TDOA sense, but useful for corroborating activity patterns around emitter sites.
- VIIRS Day/Night Band (Suomi NPP, NOAA-20): Not an RF sensor, but VIIRS DNB at roughly 750 m resolution can detect the lighting signatures of active radar or EW installations operating at night, providing a complementary optical-infrared layer for site confirmation.
What TDOA and FDOA actually measure
Every radio emitter, whether a ground-surveillance radar, a GPS jammer, or an airspace-denial system, radiates energy that travels at the speed of light. When two or more satellites receive the same burst, they receive it at slightly different times. That time difference, multiplied by the speed of light, constrains the emitter to a hyperbolic surface in space. Add a second satellite pair and you get a second hyperbola. Their intersection is the emitter's position. This is time-difference-of-arrival, TDOA.
Frequency-difference-of-arrival, FDOA, exploits the Doppler shift imposed on the signal by each satellite's velocity relative to the emitter. Because the satellites move at different speeds along their formation track, the same transmission arrives at each spacecraft with a subtly different frequency. FDOA provides an independent geometric constraint that sharpens the fix, particularly when the satellites are close together and TDOA geometry alone is weak. HawkEye 360 has published that its clusters use both methods in combination.
What the public record says about accuracy, and where it breaks down
HawkEye 360 has stated publicly that its geolocation accuracy is approximately 1 km CEP under favourable conditions. That figure deserves unpacking. CEP, circular error probable, means half of all fixes fall within that radius. The other half do not. Accuracy is a function of the number of satellites in simultaneous view of the emitter, the baseline geometry between them, the signal duration, and the signal-to-noise ratio.
Short-duration signals are the hardest problem. A radar that transmits a one-second burst gives the constellation very little time to accumulate phase and timing data. Jammers that operate in short pulses, or that frequency-hop across wide bands, can produce fixes with errors of several kilometres or produce no fix at all. A continuous-wave emitter, by contrast, gives the system time to integrate and yields the best accuracy. Buyers should ask vendors for accuracy distributions across their actual signal library, not headline CEP figures.
Revisit is a separate constraint. A small cluster constellation does not provide persistent surveillance. Depending on orbital configuration and target latitude, a given ground location may be in view of a capable cluster for only minutes per day. That is enough to detect and locate a persistent emitter, but an adversary who transmits only when no satellite is overhead can, in principle, avoid detection. This is not a hypothetical concern.
Emitter identification: frequency is not a fingerprint
Commercial RF geolocation delivers a position, a time stamp, and a frequency or frequency range. It does not deliver the signal's content, modulation details, or pulse-repetition interval at the resolution that classified SIGINT systems achieve. This distinction matters legally and operationally.
Identification relies on matching the observed frequency and, where the signal duration allows, basic modulation characteristics against a signal library. If an S-band emission at a known frequency appears repeatedly at the same coordinates, and that frequency matches a documented air-search radar type, an analyst can make a reasonable attribution. But frequency alone is not conclusive. Civilian weather radars, ship navigation radars, and military surveillance systems can share overlapping frequency allocations. Confident identification requires corroboration from other sources, optical imagery of the site, known order-of-battle data, or pattern-of-life analysis across multiple passes.
The ITU Radio Regulations define frequency allocations publicly, and the published technical characteristics of many radar families are available in open literature. That library-matching process is where analytic rigour lives.
Operational applications a government buyer can realistically expect
The most straightforward application is persistent emitter mapping: building a georeferenced catalogue of active RF sources in a region of interest, updated each time a satellite cluster passes. Over weeks, this reveals which sites are active, at what times, and whether new emitters have appeared. A radar that was absent last month and is present this month is an intelligence datum worth having, even if the fix is only accurate to 1.5 km.
Jamming detection is a second application with a strong public record. GPS jamming events have been geolocated by commercial RF satellites and corroborated against reported aviation disruptions, particularly in the eastern Mediterranean and conflict zones. The jamming source does not need to cooperate; it just needs to transmit.
Maritime and airspace denial monitoring follows naturally. A coastal defence radar that begins operating in a previously quiet sector, or a mobile jammer that relocates between passes, produces a pattern that TDOA analysis can track over time. The geolocation accuracy of 1 km or worse is usually sufficient to cue higher-resolution optical or SAR collection for site confirmation.
Satellize can fuse RF geolocation detections from licensed commercial feeds with open Sentinel-1 SAR imagery and optical tasking to produce site-level activity reports, giving clients a corroborated picture rather than a single-sensor fix.
The boundary with classified SIGINT, and why it matters
Commercial RF geolocation is not SIGINT in the legal or technical sense used by signals-intelligence agencies. It does not capture, decode, or retain signal content. It measures the physical properties of propagation: arrival time and Doppler shift. This keeps it on the right side of most national legal frameworks governing interception, and it means the product can be shared more freely within allied or coalition structures than classified intercepts.
That boundary is also a capability ceiling. If a buyer needs to know what a radar operator said on a voice channel, or to recover the waveform parameters needed to build an electronic countermeasure, commercial RF geolocation cannot provide that. The two capabilities are complementary, not substitutes. Understanding which questions each can answer prevents both over-reliance and underuse.
Typical figures
| Geolocation accuracy (CEP, favourable geometry) | Approximately 1 km, per HawkEye 360 published figures; degrades to several km with poor geometry or short signal duration |
| Frequency coverage | Approximately 150 MHz to 18 GHz (HawkEye 360 published range); HF through X-band for Kleos |
| Revisit per target location | Minutes to low single-digit hours depending on constellation size, orbital plane distribution, and target latitude; not persistent |
| Minimum detectable signal duration | Seconds; sub-second pulses may not produce a usable fix; continuous-wave emitters yield best accuracy |
| Latency from collection to product | Typically hours for processed geolocation data; near-real-time pipelines exist for priority tasking but are constellation-dependent |
| Deliverable format | Georeferenced point features (lat/lon, frequency, time stamp, accuracy ellipse); GeoJSON, KML, or shapefile typical |
| Archive depth | HawkEye 360 commercial operations began 2019; Kleos 2020 onwards; historical tasking coverage is sparse compared with optical archives |
| Content captured | Position, frequency, time, basic signal characteristics only. No signal content, no voice, no data payload |
Analytics Satellize can run
| Emitter activity catalogue | TDOA/FDOA geolocation aggregated across multiple passes; frequency-band classification against ITU allocation tables | Georeferenced point layer with emitter frequency, first-detected date, last-detected date, and activity frequency; updated per new collection pass |
| New emitter appearance alert | Change detection against baseline emitter catalogue; threshold on position and frequency novelty | Alert report with location, frequency, time of first detection, and nearest known site from open-source order-of-battle data |
| Jamming source geolocation | TDOA/FDOA fix on broadband noise or swept-frequency emission; corroboration against reported GNSS disruption events | Single-event or time-series geolocation report with accuracy ellipse and confidence assessment |
| Pattern-of-life emitter analysis | Time-series aggregation of detections by site; operational hours, duty cycle estimation, correlation with known exercise or conflict calendars | Site-level activity timeline chart and summary report; GIS layer with temporal heat map |
| Multi-sensor site confirmation | RF geolocation fix used to cue Sentinel-1 SAR or optical tasking; site-level cross-check for physical infrastructure consistent with emitter type | Fused report combining RF fix, SAR backscatter change, and optical annotation; confidence-rated site assessment |
| Frequency-band environment map | Aggregation of all detections within a defined area and time window; binned by ITU frequency allocation | Regional RF environment summary: which bands are active, where, and at what density; delivered as tabular report and GIS polygon layer |
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.