This dataset contains laboratory measurements of passive Ultra-High-Frequency (UHF) RFID tags obtained using a commercial off-the-shelf (COTS) reader under controlled office conditions. The data was collected to support research in Angle-of-Arrival (AoA) estimation using a physics-informed Bayesian regression model. Measurements capture multiple replicas per transmission, across all four legal UHF RFID frequencies, and for multiple inter-antenna spacings and tag positions. Raw signal parameters were recorded, including received signal strength indicators (RSSI), carrier phase measurements, and derived phasors. The data was then aggregated to extract complex and geometric features used in model training and evaluation. File names encode key parameters such as timestamp, tag ID, carrier frequency, tag-to-antenna distance, inter-antenna spacing, and transmission power. Additional information regarding data processing, feature extraction, and model development can be found in the associated publication.

METHODOLOGICAL INFORMATION

1. Description of methods used for data collection or generation: Measurements were conducted using a commercial Zebra FX7500 UHF RFID reader with two AN480 antennas. Tags were placed at varying positions and distances relative to the reader under office-laboratory conditions. Each measurement included multiple replicas per reading session, and data was collected across all four legal UHF RFID carrier frequencies. The methodology is described in detail in: Nedal M. Benelmekki et al. “Physics-Informed Hierarchical Bayesian Modeling for Angle-of-Arrival Estimation with Commercial Off-The-Shelf RFID”. In: IEEE ICASSP 2026 (submitted).

2. Methods for processing the data: Raw RFID readings were parsed using Zebra’s Developer API and saved in structured .csv files. Subsequent processing included tag filtering, frequency fusion, replica averaging, and extraction of phase, magnitude, and geometry-based features. These features were used to train and evaluate a Bayesian regression model for AoA estimation.

3. Instrument- or software-specific information needed to interpret the data: The data files are provided in .csv format and can be interpreted using any standard spreadsheet or data analysis software. The custom data acquisition and processing scripts are available at the following repository: https://github.com/Wireless-Information-Networking/Bayesian-Enhanced-AoA-Estimator

4. Instruments, calibration, and standards information:

   • RFID Reader: Zebra FX7500 (fixed reader)
   • Antennas: 2x Zebra AN480
   • Cabling: RP-TNC RF coaxial cables
   • Power: 24V DC regulated supply
   • Connection: Ethernet (PoE optional)
   • Computer: Ubuntu-based PC for data acquisition
   • Tags: Commercial off-the-shelf UHF RFID passive tags (various models)
   • Calibration: No explicit calibration was applied to the reader; relative measurements and inferred phase differences are used instead.

5. Environmental or experimental conditions: All measurements were taken in a typical indoor office/lab setting with known multipath characteristics. Temperature was held constant at ~24–25°C. No external interference was injected. The experimental environment includes multipath reflection sources (e.g., walls, furniture), which makes the data more representative of real-world deployment.