Cross-correlation method for acoustic detection of small unmanned aerial vehicles

The availability of Unmanned Aerial System (UAS) to consumers has increased in the recent years, with it came the potential for negligent or nefarious misuse of them. Stevens Institute of Technology has built a passive acoustic system for low flying aircraft detection, the application of the developed principles and algorithms for UAS acoustic detection and tracking is presented in this paper. The application of the developed principles and algorithms for UAS acoustic detection and tracking is presented in this paper. Several experiments were conducted aiming to establish the characteristics of the emitted noise of UAVs of various sizes while airborne and demonstrate the processing required to detect and find the direction toward the source. The vehicles tested included popular quadrotors: DJI Phantom 2 Vision + , 3DR Solo, DJI Inspire 1 as well as larger semi-professional vehicles: Freefly Alta 6, DJI S1000. The small array of 16 microphones was used for data collection in the tests near local NJ airport. Acoustic signatures of the tested UAS were collected for stationary and flying UAS. We applied the algorithm for detection and direction finding based on fusing time difference of arrival (TDOA) estimates computed by finding peaks in the output Generalized Cross-Correlation (GCC) function. The cross-correlation signal process provided UAS detection and bearing for distances up to 250m while the spectrograms did not reveal acoustic UAS signatures at that distance. This work is being supported by DHS’s S&T Directorate.