RF-Based Drone Detection Enhancement via a Generalized Denoising and Interference-Removal Framework
Ziqi Wang, Zihan Cao, Julan Xie, Wei Zhang, Zishu He
Abstract
Radio frequency-based (RF-based) detection methods are currently the main means of countering drones. However, these prevalent approaches frequently exhibit deficiencies in effectively addressing noise and interference, making them potentially unsuitable for application in realistic urban environments. This paper proposes a generalized RF signal-enhanced framework that explicitly addresses noise and interference. We decompose the RF signal into three components and uniformly integrate them into the proposed framework for decomposition. To accomplish this, three innovative loss functions and two appropriate neural networks are devised. To validate our framework, we create a real-world drone RF dataset sampled from urban surroundings, faithfully representing drone RF signals in real-world scenarios. Experimental results demonstrate that our framework exhibits satisfactory denoising and interferenceremoval performance, significantly improving the accuracy of multiple detection methods.