xEARBY XEROTH AI

The xEAR system

Different sensors.
Connected evidence.

xEAR combines bioacoustic and radio-frequency sensing with local coordination, analysis and retained recordings. Each node has a distinct job.

The product family

ACOUSTIC NODE

Record the living world.

A solar-powered design captures audible bird calls and ultrasonic bat calls, with local storage while recordings await transfer. The selected design uses 48,000 samples/s audible and nominal 384,000 samples/s mono PCM16 ultrasonic acquisition.

Hourly mesh transfers bring candidate recordings to a local hub for species inference. Original candidates, including unknowns and failed classifications, require server archive and independent cloud backup.

Selected development design: node v2, design B. Firmware and physical qualification remain open.

RF NODE

Discover emissions.

Coordinated receive windows investigate transient radio activity across planned bands. Multiple stations observing the same emission can support localisation using differences in arrival time, with calibrated receiver delays and independent timing.

Discovery must include unfamiliar activity. Scanning leaves gaps: one receiver cannot watch every band simultaneously, and narrowband signals can make timing ambiguous.

Architecture exploration. Common reception, timing and localisation performance require field validation.

RF RE-ID NODE

Keep the waveform.

A separate receive-only recorder preserves selected original complex RF samples and receiver context for transmitter fingerprint research. Its initial profile targets selected 868 MHz LoRa channels.

Coordinated retention keeps useful receiver views and records uncertainty, calibration and data loss. The baseline does not provide precision timing for RF localisation.

Architecture exploration. A research dataset is not a validated transmitter identity service.

SHARED HUB

Coordinate the network.

The local hub connects field nodes, runs acoustic inference and coordinates transfer and retention. Recordings and results can be forwarded when connectivity returns, within available storage and power.

The RF sensing receivers are distinct from the radio backhaul that carries data between nodes and hub. Combining sensing roles at one position requires its own hardware, power and interference checks.

Selected development design: hub v1, design B. Integration and physical qualification remain open.

The observation journey

Capture. Connect.
Interpret. Preserve.

Keep the evidence behind the result.

A useful event record ties the original signal to its time, receiver, processing and uncertainty. Detection and classification can be reviewed against that evidence; localisation and re-ID add hypotheses that must be tested.

Power, storage, receiver bandwidth and connectivity are finite. Missing observations must remain visible rather than appearing as an absence of activity.

Planning a system

Does every node do every job?

No. Acoustic, RF localisation and RF re-ID nodes are distinct designs. A site can combine them around a shared hub, but each sensing role needs the appropriate receiver and qualification.

What determines coverage?

Terrain, vegetation, mounting, weather, animal calls, transmitter characteristics and the receiving band all matter. Planning positions do not establish acoustic or radio detection ranges.

Does RF localisation use transmitted coordinates?

The RF node explores localisation from received emissions and calibrated timing. It does not depend on drone Remote ID or transmitter-reported coordinates.

What can re-ID establish?

Re-ID investigates repeat encounters with the same animal or transmitter. Similar signals alone do not prove identity. Both acoustic and RF re-ID require labelled evaluation across changing conditions.

Begin with a landscape

Explore a connected landscape.

See how acoustic and RF sensing can work together in a staged system.

Explore the example system