Unmanned aerial vehicle RF control link detection from orbit
Spaceborne RF receivers can intercept UAS command-and-control and video downlink transmissions, but the physics is unforgiving. This page sets out what is genuinely detectable, under what conditions, and where the method runs out of road.
Sensors
- HawkEye 360 Cluster: Three-satellite formation flying in LEO at roughly 575 km altitude, using time-difference-of-arrival and frequency-difference-of-arrival across the cluster to geolocate emitters. Geolocation accuracy is typically 0.5 to 3 km CEP depending on geometry and signal duration. Covers 100 MHz to 18 GHz, encompassing 2.4 GHz and 5.8 GHz ISM bands used by most consumer and commercial drones.
- Spire Global SDR Payload: Software-defined radio payloads aboard Spire's 100-plus satellite LEMUR constellation. Programmable frequency coverage allows targeting of specific UAS bands including 2.4 GHz, 5.8 GHz, and 900 MHz. Revisit over a fixed point can be as frequent as every 30 to 90 minutes with full constellation scheduling, though dwell time per pass is short.
- Kleos Space RF Reconnaissance: Cluster-based TDOA/FDOA geolocation constellation operating in LEO. Designed for persistent maritime and land emitter monitoring across VHF to Ku-band. Geolocation accuracy comparable to HawkEye 360 at roughly 1 to 5 km CEP. Relevant for longer-duration or higher-power UAS datalinks rather than short-burst consumer links.
- NOAA/NESDIS VIIRS (indirect): Not an RF sensor, but included for completeness: VIIRS nighttime light detection at 375 m resolution can occasionally reveal persistent illuminated UAS operations at night, providing a corroborating layer rather than a primary detection method.
Why the link budget is the whole argument
A typical consumer drone control link transmits at 20 to 30 dBm (100 to 1000 mW) in the 2.4 GHz or 5.8 GHz ISM band. By the time that signal travels 500 to 600 km to a LEO satellite, free-space path loss alone is around 160 to 165 dB. The satellite receiver must then detect a signal that has lost roughly sixteen orders of magnitude of power before it arrives. That is the central problem.
A spaceborne SDR with a modest antenna gain of perhaps 5 to 10 dBi and a noise figure of 3 to 5 dB will have a noise floor somewhere around minus 100 to minus 110 dBm in a 1 MHz bandwidth. A 30 dBm drone transmitter at 600 km range arrives at roughly minus 130 to minus 135 dBm. The gap is real. Detection is not impossible, but it requires signal integration over multiple seconds, a quiet spectral environment, and ideally a directional antenna the satellite does not have. Honest assessment: most consumer drone links are below the practical detection floor of current spaceborne receivers operating in the ISM bands.
The cases where detection becomes plausible
Three conditions shift the probability meaningfully. First, higher transmit power. Military and tactical UAS datalinks, including video downlinks in the C-band or Ku-band, can transmit at watts rather than milliwatts, and some use directional antennas that increase effective radiated power further. A 5 W C-band link is a qualitatively different target from a 100 mW 2.4 GHz hobby controller.
Second, signal duration. TDOA/FDOA geolocation requires the emitter to be on for long enough to collect a coherent measurement across the cluster baseline. Consumer drones in active flight transmit continuously, which helps. A brief frequency-hopping burst from a tactical link is much harder to catch.
Third, spectral isolation. ISM bands at 2.4 GHz are extraordinarily congested. Wi-Fi, Bluetooth, and other emitters create a noise floor that makes individual drone links nearly indistinguishable from orbit. Dedicated UAS frequency allocations, such as the 5030 to 5091 MHz band designated by ITU for UAS control, are quieter and offer better prospects for detection and identification, though spaceborne receivers must be specifically tuned to them.
Geolocation accuracy and what it means operationally
Even when a signal is detected, the geolocation product carries substantial uncertainty. TDOA/FDOA with a three-satellite cluster at LEO altitudes typically achieves 0.5 to 3 km CEP for a cooperative, stationary emitter. A drone moving at 10 to 20 m/s during the measurement window introduces additional position smear. The resulting uncertainty ellipse may cover several square kilometres.
That accuracy is adequate for some applications: identifying that a UAS is operating within a broad exclusion zone, flagging activity over a sensitive facility at regional scale, or cueing ground-based direction-finding assets to a search area. It is not adequate for tracking an individual drone's flight path in real time, or for any application requiring sub-100-metre position knowledge. The satellite pass is also a snapshot, typically 30 to 90 seconds of useful geometry, so the operational picture is episodic rather than continuous.
Video downlinks are a harder problem than control links
The video downlink from a drone is often the signal of most intelligence interest. Consumer systems use analogue or digital video at 5.8 GHz; more capable systems use proprietary compressed digital formats. Receiving and decoding the video content from orbit would require not just detection but a coherent demodulation chain matched to the specific waveform, and sufficient signal-to-noise ratio to reconstruct frames. No publicly documented spaceborne RF system claims this capability for consumer or tactical drone video.
Detection of the carrier, confirming that a video downlink is active, is more tractable than decoding it. Energy detection in the 5.8 GHz band can flag the presence of a broadband emission consistent with a video link, without recovering content. This is the realistic ceiling for current systems.
Where the method fits in a broader counter-UAS picture
Spaceborne RF monitoring is most credibly positioned as a wide-area, low-cadence cueing layer rather than a primary detection system. A government operating a sensitive facility or border zone cannot rely on satellite RF passes to provide real-time UAS awareness. Ground-based radar, acoustic sensors, and dedicated RF direction-finding equipment remain the primary tools for that task.
What satellite RF can offer is pattern-of-life analysis at scale: identifying zones where UAS RF activity is recurring, correlating activity with known operator locations, or detecting the emergence of new UAS operating areas in regions with limited ground-based monitoring. In ungoverned or remote terrain, where no ground sensor network exists, even a probabilistic satellite detection has intelligence value. Satellize treats this as an analytical layer to be combined with other data, not as a standalone alert system. Clients considering this capability should request a sensitivity analysis for their specific frequency bands and geographic area before committing to an operational design.
The ITU's allocation of 5030 to 5091 MHz for UAS control links, and ongoing regulatory work on remote identification (Remote ID) broadcast protocols, may gradually improve the detection picture. Remote ID requires drones to broadcast position and identity over Wi-Fi or Bluetooth, and while those signals face the same link-budget constraints, standardised waveforms are easier to detect and classify than proprietary ones.
Typical figures
| Primary frequency bands of interest | 2.4 GHz ISM (control), 5.8 GHz ISM (video/control), 900 MHz ISM (long-range control), 5030–5091 MHz (ITU UAS allocation), C-band and Ku-band (tactical datalinks) |
| Spaceborne receiver altitude | Typically 500–600 km LEO for HawkEye 360 and Kleos clusters |
| Free-space path loss at 600 km, 2.4 GHz | Approximately 162 dB |
| Practical minimum detectable transmit power (2.4/5.8 GHz ISM) | Estimated 1–5 W ERP under favourable conditions; most consumer drones at 0.1–1 W are at or below the detection floor |
| Geolocation accuracy (TDOA/FDOA cluster) | 0.5–3 km CEP for stationary or slow-moving emitters; degrades with target speed and short dwell time |
| Revisit frequency | 30–90 minutes over a fixed point with full Spire constellation scheduling; 2–6 passes per day typical for HawkEye 360 and Kleos |
| Dwell time per pass | 30–90 seconds of useful geometry per satellite pass |
| Spectral resolution | Configurable on SDR payloads; typically 1 kHz to 1 MHz bandwidth bins depending on mode |
| Archive depth | Varies by operator; HawkEye 360 and Kleos maintain commercial archives from constellation commissioning (2019–2020 onwards) |
| Delivery format | Geolocation event files (CSV/GeoJSON), spectrum occupancy plots, time-stamped RF detection logs |
Analytics Satellize can run
| UAS RF activity zone map | Energy detection and spectral occupancy analysis across target frequency bands; clustering of detection events by location and time | GIS polygon layer showing zones of recurring UAS-band RF activity, updated per satellite pass |
| Emitter geolocation event report | TDOA/FDOA processing on cluster observations; uncertainty ellipse computed from baseline geometry and signal duration | Time-stamped point or ellipse shapefile with CEP estimate and confidence level, delivered within hours of pass |
| Pattern-of-life activity timeline | Temporal aggregation of detection events over days to weeks; frequency and timing analysis to distinguish persistent from transient operators | Activity timeline chart and summary report, weekly cadence |
| Band occupancy baseline and anomaly flag | Statistical baseline of spectral occupancy in target bands over a defined area; z-score or threshold alerting for occupancy departures | Automated alert (email or API push) when occupancy in monitored bands exceeds baseline by a configurable threshold |
| Sensitivity feasibility assessment | Link budget modelling using published receiver characteristics and target transmit power assumptions; comparison against detection floor for client-specified scenario | Written technical assessment with detection probability estimates for specified drone types, frequencies, and geographic area |
| Remote ID broadcast detection trial | SDR tasking in 2.4 GHz Wi-Fi and Bluetooth bands during known UAS operation periods; waveform classification against published Remote ID frame formats | Trial report with detection rate, false alarm rate, and geolocation accuracy against ground truth |
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.