TSNetIQ: High-Resolution DOA Estimation of UAVs Using Microphone Arrays

Authors:
Open Access
Publisher website
Google Scholar
Abstract
With the rapid development of unmanned aerial vehicle (UAV) technology and the rise of the low-altitude economy, the accurate tracking of UAVs has become a critical challenge. This paper considers a deep learning-based localization scheme that combines microphone arrays for audio source reception. The microphone array is utilized to capture sound source reception from various angles. The proposed TSNetIQ combines elaborately designed Transformer and convolutional neural networks (CNN) modules, and the raw in-phase (I) and quadrature (Q) components of the audio signals are used as input data. Hence, the direction of arrival (DOA) estimation is treated as a regression problem. Experiments are conducted to evaluate the proposed method under different signal-to-noise ratios (SNRs), sampling frequencies, and array configurations. The results demonstrate that TSNetIQ can effectively estimate the direction of the sound source, outperforming conventional architectures trained with the same dataset. This study offers superior accuracy and robustness for real-time sound source localization in UAV applications under dynamic scenarios.
Funding Information
  • Natural Science Foundation of the Jiangsu Higher Education Institutions of China (23KJA510004, 2023020402010590)
  • Key Research and Development Program of Wuhan (23KJA510004, 2023020402010590)

This publication has 34 references indexed in Scilit: