Title: Utilizing Rational Gaussian Wavelet Neural Networks for the Identification and Detection of Multiple UAVs

Abstract: Safeguarding both civilian and military assets requires effective unmanned aerial vehicle (UAV) detection capabilities. This study introduces an economical detection framework that relies on acoustic signals captured by microphones. The audio data undergoes a processing workflow featuring interpretable, adaptive feature extraction tools known as rational Gaussian wavelets. These specialized wavelet transformations are integrated into a compact neural network, which is trained jointly to identify and categorize UAVs using the extracted features. The resulting approach yields a machine learning model that is physically interpretable and capable not only of classifying individual drones but also of recognizing UAV swarms. The efficacy of this method is validated through datasets gathered from both indoor studio settings and challenging outdoor environments with significant noise. The findings indicate that the proposed technique surpasses conventional machine learning methods in detecting and classifying both single UAVs and drone swarms, all while maintaining strong interpretability. To ensure reproducibility, the code for this implementation has been made publicly accessible.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC