Drone detection by microphone: Joanneum Research listens to the airspace passively

The typical hum of a drone is barely perceptible to the human ear – a microphone array, by contrast, captures it precisely. This is the basis of drone detection at Joanneum Research: the researchers are developing systems that “listen in” on the airspace without emitting any signals themselves. Artificial intelligence teaches the technology to distinguish sound patterns so that unauthorised flying objects are reliably detected.

For several weeks, unknown drones have increasingly been reported from several European countries. Northern and eastern Europe in particular recorded incidents in which airspace had to be temporarily closed and flights diverted. Where many of these aircraft come from remains unclear; in some cases they are suspected of being linked to reconnaissance or disruption operations.

The more such incidents occur, the clearer it becomes how much depends on reliable and early detection of unmanned flying objects. At Joanneum Research, the Intelligent Acoustic Solutions research group has many years of experience in the acoustic detection of drones – a method that has several advantages over other detection technologies.

Project leader Martin Blass describes the difference: “Unlike active systems such as radar, the acoustic method works passively – it emits no signals itself and therefore remains undetected. This makes it particularly valuable in security-critical scenarios”. Development currently focuses on hemispherical 3D microphone arrays that can also locate drones precisely in altitude. AI ensures that the system keeps improving: trained with the acoustic signatures of different drone types, the models reliably recognise characteristic sound patterns – even amid strong background noise.

“Microphone arrays record sound sources from different directions, and the position of a drone can be determined from the differences in the signals’ propagation times.” This is how Franz Graf explains it; he heads the research group and invented the so-called “tunnel ears” (acoustic accident detection in tunnels – AKUT).

“Unlike active systems such as radar, the acoustic method works passively – it emits no signals itself and therefore remains undetected. This makes it particularly valuable in security-critical scenarios.”
Project leader Martin Blass

Drone detection with distributed sensor nodes: the UAS-Check Point project

The first long-term successes are followed by the next development step: the UAS-Check Point project starts in December 2025. Planned are distributed, multimodal sensor nodes mounted as so-called checkpoints on buildings, masts or vehicles and running energy-autonomously on batteries. Each node contains modules for audio, video, global navigation satellite system (GNSS) and Remote ID, as well as a processing unit (edge computing). This allows unmanned flying objects in the so-called U-space to be verified with multiple sensors.

The advantages of this approach at a glance:

• Wide-area surveillance: thanks to distributed sensors, drone movements can be captured over a large area and tracked over longer distances.
• Robustness: because the decentralised systems are energy-autonomous, they do not depend on central infrastructure and work even when power and communication networks fail.
• Tactical security: acoustic sensors work passively, are small and easy to camouflage – potential attackers therefore cannot detect them.

Participants in the UAS-Check Point project are the AIT Austrian Institute of Technology, Airlabs Austria, Frequentis AG and Austro Control, as well as the drone manufacturers Twins GmbH and Flightec GmbH. Funding comes from the Austrian Research Promotion Agency (FFG), awarded in the Take Off 2024 programme.

The foundation of all these developments is Joanneum Research’s know-how in acoustic signal processing and AI-based analysis. “We combine physical understanding with data-driven intelligence”, Blass emphasises. “This is how we create systems that not only hear, but also understand what they hear.”