DroneRF2025

At present, we have only uploaded a portion of the dataset. After receiving the paper, we will upload the entire dataset.To support the performance verification of the ESC-FSUAV framework in the few-shot Unmanned Aerial Vehicle Radio Frequency (UAV RF) identification task, this paper specifically collects and constructs the DroneRF2025 dataset. It aims to address the core issues of scarce real-environment RF data and poor interference adaptability in few-shot scenarios. The dataset is collected on a school playground, a typical real open scene, and fully incorporates natural electromagnetic noise and interference signals from surrounding electronic devices (such as WiFi and Bluetooth). This ensures the acquired data can fully retain the complex electromagnetic environment of UAVs in practical applications, avoiding the disconnection between data in the ideal laboratory environment and real scenarios.The dataset not only includes background interference signals but also covers 6 types of typical UAV models. These models include mainstream consumer-grade and professional-grade ones like DJI Mavic2, Mavic3, Tello, Air3, and Phantom3, as well as the non-DJI model Lingkongzhe QQLRC X11. This can meet the demand for model diversity in few-shot tasks. Table 1 presents the detailed device information of the 6 UAV types.Table 2 shows the detailed parameter information for collecting the DroneRF2025 dataset. In this study, professional equipment such as the 4051B high-performance microwave spectrum analyzer and HyperLOG 3080 X antenna are used, and experimental variables are strictly controlled through a standardized process to achieve stable collection of real RF signals.Ultimately, the dataset provides a high-quality and highly reproducible data source for the few-shot UAV RF identification task of the ESC-FSUAV framework. It not only ensures the reliability of evaluating the model's identification performance in real interference environments but also lays a unified data foundation for comparative experiments of different few-shot UAV RF identification methods.