Real-world Radar and LTE Signals Dataset Collected Over-the-air in Shared CBRS Band
Download our dataset:
Please use the below link to download the datasets:Real-world Radar and LTE Signals Dataset Collected Over-the-air in Shared CBRS Band
This dataset was used for the paper "Detection of Co-existing RF Signals in CBRS using ML: Dataset and API-based Collection Testbed" to be published in IEEE Communications Magazine (Data Sets for Machine Learning in Wireless Communications and Networks). Please use this link to download the paper. Any use of this dataset which results in an academic publication or other publication which includes a bibliography should include a citation to our paper. Here is the reference for our work:
Magazine version: PDF
C. Tassie, V. Chaudhary , A. Gaber, N. Soltani, M. Belgiovine, M. Loehning, V. Kotzsch, C. Schroeder and K. R. Chowdhury, "Detection of Co-existing RF Signals in CBRS using ML: Dataset and API-based Collection Testbed,” IEEE Communications Magazine, May 2023.
Abstract:
Opening up of the CBRS band for the secondary users' transmissions poses challenges in the protection of incumbent radar users from co-channel interference. The use of Machine Learning algorithms for addressing these challenges requires representative real-world datasets.This dataset contains overlapping radar and LTE signals captures over-the-air in the shared CBRS band using an experimental testbed composed of software defined radios in RF anechoic chamber. We collected this dataset using our open-source RF data recording API tool.Dataset Description
This dataset contains simultaneous transmissions of LTE signals of different bandwidth and test models, and type 1 radar generated by National Institute of Standards and Technology’s radar waveform generator. The collected samples are captured in overlapping and non-overlapping scenarios for a wide SINR range of 15 − 35 dB. The collected samplesare saved in SigMF format with a corresponding SigMF meta file of the same name. The meta file contains the parameters of the data collection such as sampling rate, number of samples and center frequency. The resulting dataset is 55.5GB containing 5640 samples, each with a frame duration 40ms and collected at 30.72MHz sampling rate. Details of these datasets are provided in the paper.Funding Agency: NSF
Grant Number: CCRI RFDataFactory project CNS # 2120447
Fig. 1: Our proposed API-based CBRS dataset collection testbed, which configures the Tx SDRs for transmitting LTE and radar signals OTA, and the Rx SDR for collecting and saving IQ samples in SigMF format. Thereafter, spectrograms are created and fed to ML model (e.g., YOLO) for detection and localization of the interfering LTE and radar signals in the CBRS band.