@article{MF1E0088F,
title = "Suppressing the Acoustic Effects of UAV Propellers through Deep Learning-Based Active Noise Cancellation",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2025",
issn = "1226-4717",
doi = "10.7840/kics.2025.50.4.535",
author = "Faisal Ayub Khan, Soo Young Shin",
keywords = "Deep Learning, Unmanned Aerial Vehicles (UAVs), Convolutional Neural Network (CNN), Adaptive ANC, Neural Network (NN), Real-time Noise Mapping",
abstract = "This study presents a deep learning-based Active Noise Cancellation (ANC) system for reducing UAV propeller noise using a Convolutional Neural Network (CNN) model. The proposed system effectively minimizes noise in real-time by extracting key audio features such as amplitude, phase, and frequency components, generating and calculating inverse feature values to construct precise anti-noise signals. This approach enables destructive interference, significantly reducing the propeller noise. The model achieved high-performance metrics, including 94.5% accuracy, 93.2% precision, 96.1% recall, and a loss value of 0.115, demonstrating its efficacy in noise cancellation. Deployed on an Nvidia Jetson NX, the ANC system integrates high-quality microphones and strategically placed speakers on a UAV platform, allowing for real- time noise analysis and anti-noise generation. Indoor and outdoor tests validated a substantial reduction in propeller noise up to 36 dB, highlighting the model’ s robustness and potential for quieter UAV operation in noise-sensitive settings."
}