Wi-Fi CSI Dataset for Person Presence Detection Using ESP32
Description
This dataset contains 17,008 Wi-Fi Channel State Information (CSI) and RSSI records collected using ESP32 hardware for person-presence detection in a single indoor room. The approximately 21.47 MB dataset was collected across three sessions between October 24 and 25, 2026. It includes four labels: empty, present, moving, and unlabeled. The CSV files contain timestamps, receiver and session identifiers, signal strength, Wi-Fi channel information, and numerical CSI measurements. The dataset is intended for machine-learning research on occupancy detection, human presence classification, and wireless signal analysis. The all-zero `len` column was removed from the cleaned dataset. The dataset is limited to one room, and approximately 14.8% of records are unlabeled.
Files
Steps to reproduce
1. Set up the ESP32 hardware to transmit and receive Wi-Fi packets and capture Channel State Information (CSI). 2. Connect the receiving ESP32 to a computer through a serial interface. 3. Run the Python-based data collection script with the appropriate serial port and baud rate. 4. Keep Room 1 empty during the initial calibration period. 5. Record CSI and RSSI measurements while assigning the labels empty, present, moving, or unlabeled according to the room's occupancy state. 6. Repeat the recording procedure across the three available sessions. 7. Merge the session CSV files into a single dataset and retain the original recordings for traceability. 8. Create a cleaned CSV by removing the len column, which contained only zero values, without modifying the remaining recorded values.