An uncrewed aircraft in silhouette against an open sky

Detect drones that don’t want to be seen.

A mesh of passive sensors, built from commercially available parts, that hears drones, fuses the contacts, and triangulates them.

01 · The problem

Today’s systems only see the aircraft that announce themselves.

Remote ID and ADS-B assume a compliant operator. The threat broadcasts nothing. That is the gap.

Cost asymmetry
A drone can cost a few hundred euros. The interceptors used against them can cost far more per shot.
Qualitative. Figures vary by system.
Airport incursions
Drone sightings closed London Gatwick for about 33 hours in December 2018, affecting around 140,000 passengers.
Rising exposure
Incidents near airports, borders, and critical sites keep rising. The barrier to entry stays low.
Qualitative. We avoid inventing precise counts.

02 · How it works

Listen. Fuse. Locate.

LISTEN

Signatures it cannot hide

Each node passively picks up rotor sound and the RF of control and video links. A quiet drone still moves air, and still talks to its pilot.

FUSE

Cross-cue and friend or foe

Contacts combine across the mesh. A cooperative drone broadcasts Remote ID. Its absence around a track is itself a signal.

LOCATE

Triangulate, honestly

Enough bearings give a fix with a clear uncertainty ellipse. Fewer nodes give a coarser fix, and the system says so.

IllustrativeNODE ANODE BNODE CFIX ± σ
Fig. 01 · Bearing triangulation and uncertainty ellipse
IllustrativeACOUSTICRotor / motor harmonicsRF 2.4 / 5.8 GHzControl + video downlinkREMOTE-IDBroadcast presence / absenceGNSSNode time + position referenceFUSE
Fig. 02 · Passive sensor layers
IllustrativeBRAINMESH · P2P
Fig. 03 · Mesh relay topology
03 · The node

Commodity parts. Emits nothing.

Built entirely from commercially available parts. Affordable to field at scale, and built to be attritable. It only listens, and never transmits.

ComputeQuad-core ARM64 single-board computer (Linux)
AcousticUSB or I2S MEMS microphone array
RFSDR receiver, 2.4 and 5.8 GHz
Remote IDESP32 OpenDroneID receiver
Timing and positionu-blox GNSS (time and node fix)
EmissionNone. Receive only, no active radar.
BOM targetAbout €150 per node
PowerBattery or solar capable, low draw
PassiveAttritableCommodity HWField serviceableOpen schema
Close-up of a circuit board, representing commodity node hardware
IllustrativeBRAINMESH · P2P
Losing a node lowers resolution. It never blinds the mesh.
04 · The mesh

Scales with nodes. Fails gracefully.

  • / /Coverage scales linearly. More nodes, more overlap, tighter fixes.
  • / /Nodes cue each other. A weak hit becomes a confident track when neighbors agree.
  • / /No single point of blindness. Lose a node and the mesh reroutes. Resolution drops, awareness does not.
05 · Honest capability

Systems that overclaim get people hurt.

Detection is probabilistic and bounded by physics. We treat that as engineering, not marketing.

  • 01Every track carries a confidence.
  • 02Uncertainty is shown, never hidden.
  • 03We state where the sensing ends.
06 · Roadmap and extensibility

New sensors plug into one shared schema.

Each new layer publishes to the same fusion schema the acoustic and RF layers already use.

Layer · Seismic

Ground vibration

Geophone sensing for low, close passes.

Layer · Magnetometer

Magnetic anomaly

Motor and battery signatures as a corroborating cue.

Layer · PIR

Passive infrared

Low-power thermal triggers that wake the richer sensors.

Layer · Camera

Visual cueing

Slew to cue optical confirmation on an existing track.

07 · Contact

Talk to us about the uncooperative airspace.

Built for the NATO DIANA application. For a briefing or evaluation, get in touch.

Triangle Mesh is an independent effort. This page uses no NATO or DIANA marks and implies no endorsement, affiliation, or selection. All diagrams are our own technical illustrations, not screenshots of live detections.