Spaceborne RF spectrum monitoring and interference detection
Satellites carrying wideband software-defined radio receivers can detect, characterise and geolocate interference to GPS, satcom and radar bands from low Earth orbit. This page explains how TDOA/FDOA geolocation works, what accuracy is achievable, and where the method hits its limits.
Sensors
- Hawkeye 360 (HawkEye 360 Inc.): Clusters of three formation-flying microsatellites in 550 km SSO carrying wideband SDR receivers covering roughly 144 MHz to 15 GHz. TDOA/FDOA geolocation from a single cluster pass; published position accuracy of 1–3 km CEP depending on signal bandwidth and cluster baseline. Revisit improves with each cluster addition; the company had six clusters on orbit by 2023.
- Kleos Space: Four-satellite formation clusters in SSO at approximately 510–580 km altitude, SDR payloads targeting VHF/UHF and L-band. Designed for maritime and military emitter attribution. Geolocation via TDOA across the formation; published accuracy in the low-kilometre range. Multiple clusters planned to reduce revisit below 30 minutes over priority regions.
- Spire Global (GNSS-R and RO payloads): Over 100 LEO satellites carrying GNSS receivers. Primarily used for radio occultation and GNSS reflectometry, but the constellation's dense GNSS signal monitoring can flag anomalous signal environments consistent with spoofing or jamming events along occultation tracks. Not a dedicated interference monitor but a complementary data layer.
- LeoLabs (space situational awareness radar): Ground-based phased-array radars rather than spaceborne, but directly complementary: LeoLabs tracks the RF-monitoring satellites themselves, providing precise orbital state vectors that are essential inputs for accurate TDOA/FDOA geolocation calculations. Tracks objects down to roughly 2 cm in LEO.
What a wideband receiver in orbit actually measures
A software-defined radio payload has no fixed filter bank. It digitises a broad slice of spectrum, sometimes several gigahertz wide, and defers the channel selection to software. From 550 km altitude, the receiver sees every emitter within its antenna footprint simultaneously, which at LEO is a ground swath of several hundred kilometres. That is the fundamental advantage over a ground monitor: one pass covers territory that would require dozens of fixed stations.
The raw measurement is a power spectral density snapshot, time-stamped to nanosecond precision using an onboard GNSS clock. When the same signal is received by two or more satellites in a formation, the difference in arrival times (TDOA) and the difference in Doppler shift (FDOA) between platforms encodes the emitter's position. Neither measurement alone is sufficient for a unique fix; together they constrain the solution to a small ellipse on the ground. The size of that ellipse depends on three things: the baseline separation between satellites, the bandwidth of the signal being measured, and the quality of the timing reference.
The geometry of TDOA and FDOA, and why accuracy has a floor
TDOA accuracy scales with signal bandwidth. A narrowband CW tone at 1 kHz bandwidth gives a timing resolution on the order of milliseconds, translating to position errors of hundreds of kilometres. A wideband spread-spectrum signal at 10 MHz bandwidth pushes timing resolution below a microsecond, shrinking the position error to single-digit kilometres. GPS L1 C/A, at 1.023 MHz chipping rate, sits in an awkward middle ground: enough structure to detect jamming, not enough bandwidth for precise geolocation from a single cluster pass.
FDOA adds a second hyperbolic surface. The intersection of the TDOA and FDOA hyperbolas gives a point fix rather than a line, but the geometry degrades when the satellite velocity vectors are nearly parallel to the line between the satellites and the emitter. Formation-flying clusters are designed to avoid this by maintaining a baseline of tens to hundreds of kilometres in the along-track and cross-track directions simultaneously. Hawkeye 360 has published geolocation accuracy figures of 1–3 km CEP for signals with sufficient bandwidth; Kleos has cited similar figures. Below about 500 m accuracy, you need either a very wide signal bandwidth, a very long baseline, or multiple passes with precise orbit knowledge, and even then atmospheric and multipath effects impose a practical floor.
What interference looks like from orbit, and what it does not
GPS jamming is the most commonly detected category. Cheap vehicle-mounted jammers operating around 1.575 GHz (L1) produce a detectable elevation in the noise floor across the band. A spaceborne receiver overhead during a pass will see the elevated power, can characterise the jammer's centre frequency and modulation type, and with a formation can geolocate it. The same physics applies to satcom uplink interference, where an earth station transmitting on the wrong frequency or with excessive power bleeds into adjacent satellite slots, a problem the ITU Radiocommunication Bureau documents routinely.
The honest caveat: a single cluster pass over a given location lasts roughly 5–8 minutes. If a jammer operates intermittently, the probability of intercept per pass is low. Revisit rates improve with more clusters, but even with six clusters in the same orbital shell, the revisit over a fixed mid-latitude point is measured in hours, not minutes. Spoofing, where a false GPS signal is broadcast at the correct frequency and power, is harder still: the spoofer's signal looks like a legitimate GPS satellite to a power-only detector. Detecting spoofing from orbit typically requires correlation against known satellite ephemeris, not just power measurement.
The ITU framework that gives detections legal weight
The International Telecommunication Union's Radio Regulations assign frequency bands to services and require member states to coordinate transmissions. Article 15 of the Radio Regulations prohibits harmful interference; Article 16 obliges administrations to investigate and suppress it. When a spaceborne detection is submitted as evidence, it enters this regulatory process. The ITU's Radiocommunication Bureau maintains the Master International Frequency Register, against which a detected emitter can be cross-referenced to identify whether it holds a coordination record.
This is where geolocation accuracy becomes a legal question, not just a technical one. A 3 km CEP ellipse placed over a city contains many possible emitters. Enforcement action requires narrowing the fix further, typically by combining the satellite detection with direction-finding from ground assets or with a second satellite pass from a different geometry. Spaceborne RF monitoring is best understood as a cueing layer: it tells you something is wrong, in which general area, and at what frequency. The attribution step still requires corroborating evidence.
Sovereign spectrum enforcement: the operational picture
National spectrum regulators and defence ministries are the primary buyers of this data. The use cases split into two broad categories. The first is persistent monitoring of a country's own licensed bands, to detect unlicensed or malfunctioning transmitters causing interference to critical services. The second is external threat detection: identifying GPS jammers near military exercises, detecting uplink interference to government satcom, or tracking illicit maritime radio traffic.
A government operating its own RF-monitoring payload, or purchasing data from Hawkeye 360 or Kleos under a national licence, can integrate detections with its existing signals intelligence infrastructure. The data product is typically a time-stamped detection report giving frequency, estimated power, modulation class and a geolocation ellipse with stated confidence. Satellize works with governments building sovereign space programmes and can advise on how commercial RF-monitoring data integrates with national spectrum management workflows, including the regulatory filings required under ITU procedures. The analytic output is not a replacement for a national SIGINT capability; it is a persistent, politically neutral audit layer that operates regardless of cloud cover or terrain.
Where the method is reliable and where it is not
RF spectrum monitoring from LEO is reliable for detecting persistent, high-power emitters in well-characterised bands. Truck-mounted GPS jammers, rogue satcom uplinks, and poorly coordinated radar systems all produce signatures that a wideband SDR payload will capture on a pass. The method is less reliable for low-power, short-duration or frequency-agile emitters. A jammer that transmits for 30 seconds every hour will be missed by most passes. A frequency-hopping system requires either a very fast scan rate or prior knowledge of the hop sequence.
Archive depth is also a constraint. Unlike optical or SAR imagery, RF monitoring data is not routinely archived as a global public resource. Commercial operators hold their own archives, and access is by licence. This limits retrospective analysis. The field is also evolving quickly: as more clusters are launched by Hawkeye 360, Kleos and potential new entrants, revisit rates will improve and the probability-of-intercept problem will diminish. But for now, anyone planning an enforcement operation should treat spaceborne RF monitoring as a high-confidence detector with a meaningful duty-cycle gap, not a continuous surveillance system.
Typical figures
| Frequency coverage (Hawkeye 360) | Approximately 144 MHz to 15 GHz (published mission scope); specific sub-band sensitivity varies by payload generation |
| Orbital altitude | 510–580 km SSO (Kleos); ~550 km SSO (Hawkeye 360) |
| Geolocation accuracy (CEP) | 1–3 km for wideband signals with adequate SNR; degrades to tens of km for narrowband CW tones |
| Single-pass dwell time over a ground point | Approximately 5–8 minutes per cluster pass |
| Revisit rate (Hawkeye 360, six clusters, 2023) | Multiple passes per day over most latitudes; exact revisit depends on target latitude and cluster phasing |
| Minimum detectable power | Not publicly specified in absolute terms; detection is SNR-dependent and varies with emitter bandwidth and antenna gain |
| Geolocation method | TDOA and FDOA from formation-flying cluster; requires simultaneous reception by at least two satellites |
| Timing reference | Onboard GNSS-disciplined clock; nanosecond-level synchronisation between formation members |
| Data latency (detection to report) | Hours for standard delivery; near-real-time downlink possible over ground station contacts |
| Archive depth | Operator-held; Hawkeye 360 operational since 2018, Kleos since 2020. No open public archive. |
Analytics Satellize can run
| Interference event detection report | Power spectral density anomaly detection against a baseline noise floor for the band; threshold exceedance flagging | Time-stamped PDF or JSON report per event: frequency, estimated bandwidth, modulation class, pass geometry |
| Geolocation ellipse for detected emitter | TDOA/FDOA solution using formation satellite positions and precise timing; ellipse size reflects GDOP at the solution | GeoJSON polygon with CEP confidence level, delivered per event or as a batch after each orbital pass |
| Persistent emitter activity timeline | Multi-pass aggregation of detections at consistent frequency and approximate location; temporal pattern analysis | Activity chart showing detection frequency, time-of-day patterns and estimated duty cycle over a defined period |
| ITU cross-reference check | Comparison of detected emitter frequency and estimated location against the ITU Master International Frequency Register | Annotated report flagging whether a coordination record exists for the detected emission; input to regulatory filing |
| GPS jamming density map | Aggregation of L1/L2 band power anomalies across multiple passes; spatial binning to produce a heatmap of jammer activity | Raster or vector heatmap layer (GeoTIFF or GeoJSON) suitable for import into a national spectrum management GIS |
| Satcom uplink interference attribution | Frequency and polarisation matching of detected uplink against licensed earth station database; geolocation comparison | Candidate earth station shortlist with confidence scores, formatted for submission to ITU Radiocommunication Bureau |
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.