AISAR Detect analyses the electromagnetic spectrum without emitting a single signal. It identifies the control link, the video downlink and the telemetry of a drone, classifies the model, and locates both aircraft and operator. No emissions. No spectrum authorisation. Nothing for anyone to detect.

Covers 400 MHz – 6 GHz — every commercial drone band. Identifies controller links, video transmission, telemetry and proprietary protocols including DJI OcuSync and FHSS families. Reception only: invisible to counter-detection, no authorisation required anywhere in the EU.
A three-level convolutional network answers three questions in sequence: is it a drone, what class of transmitter, what specific model. 37 drone classes from RF fingerprint alone, including DJI, Autel, FPV and custom-built platforms. Models trained on operational recordings, not laboratory captures.
Three or more nodes correlate time and frequency of arrival to produce meter-class geolocation, with cross-node correlation giving velocity and heading. Coverage scales linearly: five nodes cover 25+ km² with overlapping sectors.
Every node classifies locally on NVIDIA Jetson AGX Orin. No cloud dependency, no backhaul, no latency, no data leaving the site unless you choose. Pipeline: FFT → Spectrum Environment Monitor → CFAR detection → CNN classification.
The node learns the baseline electromagnetic environment of the specific site, then flags what does not belong — cutting false positives sharply compared with fixed-threshold detection. The baseline adapts as the site changes.
Bistatic and multistatic geometry using ambient Wi-Fi, GPS, 5G and broadcast signals reflected off aerial objects, extending passive coverage toward targets that transmit nothing.
Technical specifications depend on configuration, deployment geometry and site conditions, and are confirmed per project.
| CAPABILITY | AISAR | FIXED RF SENSORS | PROGRAMME-GRADE SYSTEMS |
|---|---|---|---|
| Passive RF detection | Yes | Yes | Yes |
| Reflected signal analysis | Yes | No | Partial |
| Mesh-native scaling | Yes | No | Custom integration |
| AI classification, 37 classes | Yes | Limited | Yes |
| Edge processing, no cloud | Yes | No | Yes |
| Digital twin development | Yes | No | No |
| Deployable in minutes | Yes | Fixed site | Hours to days |
| Own radar layers across four bands | Yes | No | Partial |
No other platform combines passive RF, mesh geolocation, four own radar bands, digital-twin engineering and edge AI in a single architecture.
Talk to our team about deploying AISAR Detect at your site.