Equipment
Perimeter acoustic arrays
Unauthorized drones over critical infrastructure — substations, reservoirs, public sites — increasingly slip past RF and radar. Their rotors and motors carry an acoustic signature that doesn’t, but at standoff that signature sits inside a soundscape of wind, traffic, wildlife, and the site’s own machinery. The job is to surface it against an environment that is, most of the time, simply itself.
At standoff, a single microphone hears a drone the way it hears everything else: quietly, intermittently, mixed in. The signature is real, but on any one channel it sits near the noise floor and beside sounds that come and go on their own. The combination across the array is where the drone separates from the day.
Each microphone has well-established signal-processing tools: time-frequency analysis, cepstral features, array processing. EchoSignal doesn’t replace them — it runs them as domain plugins on top of the general engine, then models normal across all of them together. Different methods, combined into one perimeter baseline.
Time–frequency representations capture propeller harmonics and their drift under maneuver — the kind of structure flat-threshold detectors miss in a busy soundscape.
Beamforming and inter-channel coherence separate a moving acoustic source from clutter that lives only at one mic. Time-difference-of-arrival across elements gives an approximate direction for response.
Short-time energy, zero-crossing rate, spectral centroid — cheap features that run on every frame and add a second view alongside the spectrogram, so the combined baseline is sharper.
Read one at a time, each microphone is mostly hearing weather, traffic, and itself. The meaningful anomaly is a coherent acoustic event across multiple elements — a pattern no single channel, and no operator listening to one stream, can reliably hold in view.
That cross-array coherent event is exactly what EchoSignal surfaces — and brings to your security team to verify.
On a perimeter acoustic array deployed around critical infrastructure, occasional unauthorized overflights had to be distinguished from continuous ambient noise — weather, on-site machinery, traffic, and wildlife. By modeling normal across the combined acoustic channels of the array together, EchoSignal surfaced the cross-channel coherent events that matched drone-like signatures, and brought them to the site’s security team for verification.
Outcome figures are placeholders, to be filled from your records once cleared for publication.
Bring your own acoustic data — from your own site, in your own environment — and see what the combined baseline surfaces. The best proof runs on your own data.