Terrestrial LTE and GSM coverage mapping from LEO
Spaceborne SDR receivers detect LTE and GSM downlink emissions beyond their intended service areas, providing an independent audit of operator coverage claims and flagging unplanned cross-border signal spillage.
Sensors
- HawkEye 360 Cluster: Three-satellite formation flying at roughly 575 km altitude; uses time-difference-of-arrival and frequency-difference-of-arrival across the cluster to geolocate emitters. Published geolocation accuracy is approximately 1 km CEP for cooperative signals; revisit at mid-latitudes is several times per day depending on cluster geometry.
- Spire LEMUR-2 SDR payload: Software-defined radio payloads hosted on 3U CubeSats in a constellation exceeding 100 satellites. Capable of recording raw IQ data across a programmable frequency range covering GSM (850–1900 MHz) and LTE bands. Revisit is high but non-uniform; detection is opportunistic rather than tasked stare.
- NORSAT-1: Norwegian Technology Demonstration satellite launched in 2017, carrying an AIS receiver and an experimental maritime broadband receiver. Its SDR payload demonstrated that LTE-band signals from coastal base stations are detectable from 600 km altitude, validating the basic propagation geometry for subsequent commercial programmes.
- Sentinel-1 (indirect reference): Not an RF monitor, but its published orbit geometry is useful for validating line-of-sight windows over target regions. Sentinel-1 SAR passes are sometimes used alongside RF data to correlate detected emissions with ground infrastructure visible in imagery.
Why a base station is visible from 600 km up
GSM and LTE base stations transmit on downlink frequencies between roughly 700 MHz and 2.6 GHz with typical effective isotropic radiated power values of 40 to 60 dBm. Those signals are designed to reach handsets a few kilometres away at ground level, but they propagate in all directions, including upward. A satellite in low Earth orbit sits well within the radio horizon for any base station it passes over, provided the geometry is line-of-sight, which at 500 to 600 km altitude it almost always is. The signal arriving at the satellite is weak, typically tens of decibels below the noise floor of a consumer receiver, but a sensitive SDR payload with a low-noise amplifier and coherent integration over a dwell period of several seconds can extract it reliably.
The physical limit is not sensitivity but geometry. A satellite moving at roughly 7.5 km per second crosses a base station's footprint in under two minutes, constraining integration time. Signals from dense urban areas, where many base stations transmit simultaneously on adjacent channels, create an interference floor that complicates single-emitter isolation. Rural areas with sparse base station density are actually easier to characterise from orbit, which is one reason the technique has practical value for regulators assessing remote-area coverage obligations.
What detection tells you, and what it does not
Detecting a downlink emission from orbit confirms that a base station is transmitting on a given frequency at a given time. It does not confirm that a handset on the ground can achieve a usable connection. The link budget from base station to satellite is entirely different from the link budget from base station to handset at ground level. A site may be detectable from orbit while still providing only marginal service to a user in a valley or inside a building. Conversely, a site that meets operator coverage obligations at ground level may be undetectable from orbit if it is low-power, heavily sectorised away from zenith, or temporarily off-air during the satellite pass.
This distinction matters enormously for regulatory applications. Orbital RF data is best used as a screening tool: it can flag areas where no emission is detected despite operator claims of coverage, prompting ground-truth investigation. It can also identify emissions in areas where no licensed coverage should exist, which is the cross-border spillage problem. It is not a substitute for drive-test surveys or crowdsourced signal-quality data when the question is whether subscribers receive adequate service.
Cross-border spillage and spectrum sovereignty
LTE and GSM signals do not stop at national borders. In regions where countries share a land border or where coastal base stations face across narrow straits, downlink emissions routinely propagate into adjacent jurisdictions. This creates two problems. First, a foreign operator's signal may interfere with a domestic licensee's allocation if both use overlapping frequency bands. Second, a government may have no independent means of verifying what foreign emissions are present on its licensed spectrum without conducting its own measurement campaign.
Spaceborne RF monitoring provides a persistent, independent audit that neither operator cooperation nor diplomatic goodwill is required to run. A single satellite pass over a border region produces a snapshot of which frequencies are active and, with multi-satellite TDOA geometry, where the emitters are located to within roughly 1 km. Repeated passes build a temporal picture: which sites are always on, which are intermittent, and whether new emitters have appeared since the last measurement cycle. For small island states and landlocked countries with limited spectrum-monitoring infrastructure, this is often the only practical way to conduct the audit at all.
Inferring coverage extent without inferring capacity
Mapping coverage extent from orbital detections requires a propagation model. The standard approach is to take each confirmed detection, assign it to a known or estimated base station location using TDOA geolocation or by matching the detected frequency and timing to a published network database, then apply a terrain-aware propagation model such as ITU-R P.1812 to estimate the ground-level service contour. The result is a modelled coverage polygon, not a measured one. Its accuracy depends on the quality of the terrain data, the assumed antenna pattern, and whether the base station's configuration matches the database entry.
Subscriber capacity is a different question entirely and orbital RF monitoring cannot answer it. Capacity depends on the number of active users, the modulation and coding scheme in use, backhaul constraints, and real-time network management decisions. None of those variables are encoded in the downlink signal in a form recoverable from a satellite receiver. Claiming otherwise would misrepresent the technique. What orbital data can do is confirm that a cell is active, estimate its approximate transmit power class from received signal strength corrected for path loss, and flag anomalies such as a site transmitting on a frequency inconsistent with its licence.
Practical limits a buyer should understand
Revisit is the binding constraint for most operational applications. Even a large LEO constellation of SDR satellites cannot provide continuous stare over a fixed point. HawkEye 360's published revisit is several passes per day at mid-latitudes, with dwell times of one to two minutes per pass. A base station that is switched off during those windows will not be detected, which means a single-day snapshot can produce false negatives. Building a reliable coverage map requires aggregating detections across multiple days and multiple passes.
Urban signal density creates a second limit. In a city with hundreds of base stations operating on adjacent LTE channels, the received signal at the satellite is a composite of many overlapping emissions. Separating individual emitters requires either very high frequency resolution, prior knowledge of the network's channel plan, or a multi-satellite formation with enough baseline to resolve emitters spatially. Single-satellite passes over dense urban cores produce useful aggregate occupancy data but poor emitter-level attribution. Honest use of the technique acknowledges this and reserves single-emitter geolocation claims for lower-density environments where the geometry supports it.
Satellize structures its RF analytics engagements around these limits, combining orbital detection data with licensed network databases and terrain models to produce coverage-gap reports that are explicit about confidence levels. The methodology is similar in spirit to the independent audit approach used in the Kingdom of Tonga crop-estimation programme: orbital data as the independent variable, ground reference as the calibration layer.
Typical figures
| Frequency coverage | 700 MHz to 2.6 GHz (covers principal GSM 850/900/1800/1900 and LTE bands; exact range is payload-dependent) |
| Geolocation accuracy (TDOA/FDOA) | Approximately 1 km CEP for HawkEye 360 cluster geometry; degrades to 3–5 km for single-satellite Doppler-only methods |
| Minimum detectable EIRP | Approximately 30–40 dBm at 600 km altitude with coherent integration; varies with payload noise figure and dwell time |
| Revisit (mid-latitudes) | 2–6 passes per day for multi-satellite constellations; dwell per pass typically 60–120 seconds over a fixed point |
| Spatial resolution of coverage map | Modelled ground contour at 100–500 m grid resolution, constrained by terrain model quality and propagation assumptions |
| Frequency resolution | Configurable via SDR; typical operational settings 10–100 kHz per bin for band occupancy surveys |
| Archive depth | HawkEye 360 data available from 2018; Spire SDR tasking records from approximately 2019 onward |
| Latency (detection to delivery) | Typically 2–24 hours after satellite downlink, depending on ground station contact and processing pipeline |
| Delivery formats | GeoJSON emitter point files, GeoTIFF coverage rasters, CSV detection logs, PDF regulatory audit reports |
Analytics Satellize can run
| Coverage-claim audit report | Orbital detection aggregated over 30-day window, matched against operator licence database, gaps flagged where no emission detected despite claimed coverage | PDF report with GeoJSON gap polygons, per-site detection confidence scores |
| Cross-border spillage map | TDOA geolocation of detected emitters, filtered to signals whose estimated source location falls outside the monitoring country's border | GIS layer (GeoPackage) of foreign emitter locations with frequency, estimated EIRP class, and detection frequency |
| Spectrum occupancy time series | Band-by-band power detection across each satellite pass, aggregated to show temporal occupancy percentage per licensed channel block | CSV time series and interactive dashboard; monthly summary report |
| Anomalous emitter alert | Automated comparison of detected frequencies against national licence register; unlicensed or out-of-band detections flagged for human review | Near-real-time alert feed (JSON webhook or email) with detection coordinates, frequency, and pass metadata |
| Modelled service-area contour | ITU-R P.1812 terrain-aware propagation model applied to confirmed emitter locations and estimated antenna parameters; outputs ground-level field-strength grid | GeoTIFF raster at 100 m resolution, with confidence band showing uncertainty from antenna-pattern assumptions |
| Baseline spectrum survey (new licence area) | Multi-pass orbital scan over target region prior to licence award, characterising existing emission environment to inform interference assessment | Spectrum environment report with frequency-geography heat maps and identified incumbent emitters |
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.