A Review of Acoustic Techniques for Drone Detection

The rapid increase in drone usage has raised serious security concerns, as these devices can be exploited for surveillance, smuggling, or hostile attacks. Developing effective and affordable detection systems is therefore a critical need. This paper reviews recent research on acoustic-based drone detection, emphasizing its potential as a low-cost, portable, and passive alternative to radar and optical systems. The review synthesizes methodologies across the literature, including data acquisition through microphone arrays, feature extraction techniques such as Mel-Frequency Cepstral Coefficients (MFCCs) and related spectral features, and classification approaches ranging from traditional Support Vector Machines (SVMs) to advanced neural networks. Comparative analysis highlights that SVMs achieve strong accuracy with small datasets and low computational demands, while convolutional neural networks (CNNs) excel in handling large, noisy datasets but require greater resources. Key findings also show that microphone array configurations combined with beamforming significantly improve detection range and noise suppression compared to single-microphone systems. The review concludes that while acoustic techniques demonstrate strong promise, challenges remain in scalability, dataset generalizability, and computational efficiency. Addressing these limitations will be essential for advancing acoustic drone detection into reliable, real-world security applications.