Acoustic drone detection
Acoustic sensors identify a drone by the sound of its rotors and motors. The method is inexpensive and entirely passive, with short range and a strong sensitivity to ambient noise.
Key facts
- Acoustic detection identifies an aircraft from the sound of its propellers and motors, matched against a library of known signatures.
- The sensors only listen, so nothing is transmitted and the method is generally available to private operators.
- Useful detection ranges are typically a few hundred metres, which is short compared with radar or RF and limits the method to close-in coverage.
- Performance falls quickly as ambient noise rises, so industrial sites, road corridors and urban settings are difficult environments.
- Acoustic detection works on the aircraft itself rather than its transmissions, so an autonomous flight with no radio link is still detected within range.
- Sensor cost per unit is low, which makes dense arrays practical where the area to be covered is small and well defined.
How it works
An aircraft with spinning rotors makes a characteristic noise. Acoustic detection uses one or more microphones to pick that up and compares it against a library of signatures for known aircraft types.
Nothing is transmitted, so the method raises none of the authorisation questions that apply to mitigation, and it is generally available to private operators.
Classification is based on the pattern of the sound rather than its loudness. That allows a system to separate a small multirotor from a passing vehicle or a lawnmower, within the limits of the environment it is working in.
Range and environment
Two things determine whether acoustic detection is worth deploying at a particular site, and both are physical.
The first is range. Useful detection is typically a few hundred metres for a small aircraft in quiet conditions. That is short in comparison with radar or RF, and it means acoustic sensors provide close-in coverage rather than early warning.
The second is ambient noise. Performance drops quickly as background sound rises, and it drops most in the frequency ranges the method relies on. Industrial plant, road traffic, generators and wind all reduce it. The practical consequence is that many of the sites with the strongest security case are also the sites where this method works least well.
Where it is a good fit
Acoustic detection suits a small, bounded, comparatively quiet area where the sensor can be close to the space being protected. A correctional exercise yard is a good example: the area is defined, the distances are short, and a delivery has to come close to the ground to be useful.
Unit cost is low enough that a dense array is practical in those settings, which partly offsets the short range.
Where it fits in a layered system
Acoustic detection is rarely the primary sensor at a large site. It works well as an additional layer where the geometry suits it, and it covers one case RF sensing misses, since an aircraft flying autonomously with no radio link is still audible within range.
Used alongside radar for area coverage and cameras for visual confirmation, it contributes close-in detection at low cost. On its own, at a large or noisy site, it will disappoint.
Related work
- Site assessment & design Vulnerability assessment, RF survey, sensor siting and coverage modelling.
- Detection system integration Bringing sensors, command-and-control and existing security systems together.
- Passive physical barriers Netting, mesh and cages, where no spectrum or operating authority applies.
FAQ
▸What range should we expect?
Typically a few hundred metres for a small aircraft in reasonably quiet conditions, and less where there is traffic, machinery or wind. Quoted figures are usually measured in quiet conditions against a specific aircraft, so they should be treated as an upper bound rather than a planning number.
▸Is acoustic detection useful at an industrial site?
Less than it first appears. A refinery, a substation or a plant room is noisy in exactly the frequency ranges the method depends on. Acoustic sensors are better suited to quieter and more contained areas, such as a correctional yard or a section of perimeter away from plant.
▸Can it distinguish a drone from other noise?
Modern systems classify on rotor and motor signatures rather than volume, and they do this reasonably well against known aircraft in suitable conditions. False alarm rates are a fair question to put to any supplier, and the answer should come from a test in an environment resembling your own site.
▸Does it work at night or in bad weather?
Darkness makes no difference, which is an advantage over cameras. Wind and rain both reduce performance, and heavy wind in particular can make the method unreliable.