Author (s)
Andrea Toma, Niccolò Cecchinato, Carlo Drioli, Gian Luca Foresti, Giovanni Ferrin
Affiliation
Department of Mathematics, Computer Science and Physics, University of Udine, Italy
Publication date
2021
Abstract
A novel radio-frequency (RF)-assisted algorithm for acoustic recognition and localization of unmanned aerial vehicles (UAVs) in a scenario with small size microphone array sensors is investigated where the multi-channel processing of acoustic signals is assisted by RF power patterns analysis. The propellers of the unidentified drone generate noise that can be used to obtain a number of cues on it, as drones with different size, weight, or mechanical characteristics produce different acoustic signals. Specifically, in this work the spectral signature of the acoustic signal detected by a multi-channel microphone array is used to recognize the drone. Furthermore, RF signals are emitted by Wi-Fi antennas and received signal strength (RSS) is measured to assist the acoustic localization. Both direction of arrival (DOA) and distance from the acoustic source can be predicted. A solution is proposed where a four-stage convolutional neural network (CNN) performs drone recognition through its acoustic spectral signature and produces the RF-assisted acoustic localization through intrinsic feature extraction, fusion of the RF and acoustic features, and regression. Applications are anti-UAV monitoring strategies from flying UAVs against illegal use of UAVs and external UAV attacks. A centralized architecture is proposed for data acquisition and streaming from multiple aerial nodes. A 19 channels spherical microphone array named Zylia is employed. To analyse the current state of this research, the experiments are presented with a description of the results.
Full paper
https://www.sto.nato.int/publications/STO%20Meeting%20Proceedings/STO-MP-IST-190/MP-IST-190-34.pdf
Keywords
ambisonic spatial room impulse responses (SRIRs), higher-order ambisonics (HOA), artificial reverberation, microphone array spacing, perceptual audio evaluation, three-dimensional sound reproduction, binaural and multichannel listening tests