An Open Dataset for Spectro-Temporal RF identification
About
Sharing a large, labelled datasets of RF emissions is critical to scientific research. It allows researchers to collaborate, enables access to valuable data to people who are unable to acquire the necessary equipment, and stimulates the development of new techniques, models, and DL architectures for RF research. In the absence of a large, curated dataset for spectro-temporal identification of different RF technologies, we are making our dataset SPREAD available to the research community. SPREAD is complemented with the supporting API for the reproduction of the iterative learning approach, as well as the expansion of the dataset with new RF technologies.
Content
We provide the dataset with three sizes for different uses depending on the application.
Small (~32GB): Image data and labels to train and evaluate Deep Learning model for Spectro-Temporal RF Identification of five RF classes: Wi-Fi, Bluetooth, ZigBee, Lightbridge, and XPD. Data and labels collected from the anechoic chamber are also included. Please click the following button to start downloading the data.
Medium (>1.2TB): Recordings of raw RF samples collected from over-the-air experiments and in the anechoic chamber. The recordings are associated with image data and labels without RF augmentation in SPREAD small dataset.
Large (>10TB): Recordings of raw RF samples associated with all data and labels (including recordings from RF augmentation) in SPREAD small dataset.
If you wish to get the Medium or Large dataset, please prepare a sufficiently large storage drive to be sent to us and let us know (You will be responsible for the drive and the shipping).
The supporting API is available here. Note: The API supports the Medium and Large dataset, as well as new recorded data for the reproduction and expansion of Spectro-temporal RF Identification datasets. For more details about the supported functionality, please refer to our paper.
For any use of this project for research, academic publications, or other publications which include a bibliography, please include the following citations (BIBTEX):
H. N. Nguyen, M. Vomvas, T. Vo-Huu, and G. Noubir, “WRIST: Wideband, Real-time, Spectro-Temporal RF Identification System using Deep Learning”, IEEE Transactions on Mobile Computing, 2023.
H. N. Nguyen, M. Vomvas, T. Vo-Huu, and G. Noubir, “Wideband, real-time spectro-temporal rf identification”, in Proceedings of the 19th ACM International Symposium on Mobility Management and Wireless Access, 2021, pp. 77–86.
H. N. Nguyen, M. Vomvas, T. Vo-Huu, and G. Noubir, "SPREAD: An Open Dataset for Spectro-Temporal RF Identification", 2023, https://sprite.ccs.neu.edu/datasets/SPREAD/.