Home / Technology

EKHO's Detection Architecture

Distributed acoustic sensors, cable-connected to a local processing unit, the EKHO Core, which identifies the threat and alerts a supervision console, without relying on a radio signal or a cloud connection.

Distributed acoustic sensing

Each EKHO sensor carries a microphone array in continuous passive listening. Several sensors are positioned around the perimeter of the site to be monitored, covering the whole area instead of a single observation point.

Local signal processing

Captured signals are processed locally, on the EKHO Core, using signal-processing techniques and classification models that separate a drone's signature from ambient noise. Detection latency is 2 to 5 seconds.

EnvironmentRangeLimiting factor
Rural, low noise120 – 150 mWind, wildlife
Suburban80 – 110 mTraffic, human activity
Noisy industrial site60 – 90 mMachinery, ventilation
Limits
  • Range depends directly on the site's ambient noise: these figures are measured ranges, not guarantees.
  • Functional prototype: these specifications reflect the system as validated in a representative environment, currently undergoing field validation campaigns.

Sensor fusion and localisation

When several sensors detect the same source, comparing their observations estimates the origin of an acoustic signature through triangulation, producing a direction, an azimuth, rather than a simple binary alert.

Plan of an industrial site with eight EKHO sensors around the perimeter (C-01 to C-08): sensors C-04 and C-05 detect a drone by triangulation, azimuths 096° and 042°, estimated distance 150 meters, 97% confidence

Local infrastructure: sensors, EKHO Core, console

Sensors report to the EKHO Core over a wired RJ45 Ethernet connection. The EKHO Core in turn reports to the supervision console over the site's local network (LAN). Every link in this chain stays on the customer's own infrastructure.

Sensorsacoustic
RJ45
EKHO Corelocal processing
LAN
Consolesupervision
  • No cloud
  • No Wi-Fi
  • No internet connection required
Architecture V1
  • Outdoor housingIP66 target
  • Operating temperature-20°C to +55°C
  • Power & dataPoE 802.3af, ~15 W
  • Detectionmicrophone array, passive acoustic
  • Connectivitywired Ethernet RJ45, no Wi-Fi
  • Processinglocal, on the EKHO Core
  • Detection latency2 to 5 seconds
  • Outputalert + azimuth estimate
  • Networkmulti-sensor deployment
  • Supervisionlocal console, client's network
  • Data leaving the sitealerts & signatures only, no cloud

No cloud dependency

Analysis happens on site, on the EKHO Core unit cable-connected to the sensors. No cloud, no Wi-Fi and no internet connection are required for the detection function: a deliberate architectural choice built around edge processing, for data sovereignty and simpler deployment on sensitive sites.

No continuous raw audio transmission

Beyond the sensor itself, the architecture is designed to transmit only the information needed for detection and localisation (alerts, azimuth, signatures) rather than continuous raw audio, in line with a local-processing, network-frugal approach.

Exploded view of the EKHO sensor: vented housing, microphone array, processing board, RJ45 network module and wall-mount arm
Vented housing, microphone array, processing board, network module and mount arm.
EKHO control post screen: f0 frequency, harmonics, detection score, sound level, real-time spectrogram and alert log
Go further

Acoustic Drone Detection

The principle of acoustic detection, explained in detail.

Learn more about acoustic drone detection →

Company

EKHO Technologies, a French deeptech company.

Discover EKHO Technologies →

Contact

A technical question about EKHO's architecture?

Talk to EKHO Technologies →