USRP Respiratory Data

This dataset presents human respiratory patterns collected using a Software-Defined Radio (SDR)-based Radio Frequency (RF) sensing system. The data was acquired using a USRP X310 device with VERT2450 omni-directional antennas and validated using a Vernier Go Direct respiration belt as ground truth. The dataset includes respiratory signals recorded from five healthy participants under controlled experimental conditions at three different orientation angles (45°, 90°, and 180°), simulating real-world positioning scenarios. Three breathing patterns were captured: Normal Respiration (NR) Fast Respiration (FR) Sleep Apnea Respiration (SAR) Each respiratory sample was recorded for 10 seconds. The raw Channel State Information (CSI) data was collected and subsequently preprocessed to produce structured datasets suitable for machine learning and signal processing applications. The final dataset consists of nine CSV files corresponding to different combinations of breathing patterns and angles. Each CSV file contains processed CSI data with 2000 features per sample, representing respiratory waveform characteristics. In total, the dataset includes: 5 participants 3 breathing patterns 3 orientation angles 225 experimental recordings 32,850 total samples across all files This dataset is suitable for: Machine learning-based respiratory classification RF sensing and wireless signal analysis Healthcare monitoring research Signal processing studies involving human physiological signals How to Cite?If you use this dataset, please cite: U. Saeed and S. A. Shah, “Human Respiratory Data Collection Using USRP SDR Device,” Zenodo, 2026. https://doi.org/10.5281/zenodo.19040427