aselya9185/wifi-csi-human-activities

WiFi CSI Human Activities Dataset

This repository contains WiFi Channel State Information (CSI) measurements, intermediate processing outputs, extracted features, visualizations, and experimental results used for WiFi-based Human Activity Recognition (HAR).

The dataset accompanies the Master's thesis:

Assel Ussenova WiFi Sensing Through Digital Receive Beamforming and CSI MSc in ICT and Internet Engineering Università degli Studi di Roma Tor Vergata (2024/2025)

---

Overview

This dataset contains CSI measurements collected in indoor office environments and the outputs of a complete signal processing and machine learning pipeline for WiFi-based human activity recognition.

The monitored activities include:

• Empty room
• Sitting
• Standing
• Walking

The dataset can be used for research in:

• WiFi sensing
• Human Activity Recognition (HAR)
• RF sensing
• Channel State Information (CSI) analysis
• Digital receive beamforming
• Machine learning for wireless sensing

---

Dataset Structure

saved_csi_raw/
interpolation/
phase_processing/
reconstructed_csi/
csi_distance/
theta/
features/
plots/
results/

Folder Description

Folder	Description
saved_csi_raw	Original CSI measurements extracted from WiFi packets
interpolation	CSI after pilot subcarrier reconstruction
phase_processing	Phase after preprocessing and calibration
reconstructed_csi	CSI reconstructed after phase preprocessing
csi_distance	CSI distance metrics used for activity analysis
theta	Beamforming phase-angle outputs
features	Feature datasets used for machine learning experiments
plots	Generated figures and visualizations
results	Classification results, evaluation metrics, and experiment outputs

---

Processing Pipeline

saved_csi_raw
    ↓
interpolation
    ↓
phase_processing
    ↓
reconstructed_csi
    ↓
csi_distance / theta
    ↓
features
    ↓
results

---

Experimental Setup

The CSI measurements were collected using the Nexmon CSI extraction framework on an Asus RT-AC86U router operating in IEEE 802.11ac VHT 80 MHz mode.

Experiments were conducted in two indoor office environments. The monitored subject performed different activities, including sitting, standing, and walking along multiple paths, while CSI measurements were continuously recorded.

The complete experimental setup and methodology are described in the referenced publication.

---

Dataset Origin

The raw CSI measurements used in this repository originate from the experimental campaign presented in:

M. De Sanctis, R. Fallani, T. Rossi, E. Cianca, M. Ruggieri, and V. Poulkov, "WiFi Sensing Through Digital Receive Beamforming and CSI," 2025 IEEE 36th International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), 2025. DOI: 10.1109/PIMRC62392.2025.11274992

This repository contains the original CSI measurements together with additional preprocessing, phase calibration, beamforming analysis, feature extraction, visualizations, and classification results developed as part of the accompanying Master's thesis.

---

Source Code

The complete implementation used to generate and process the datasets is available on GitHub:

https://github.com/aselya9185/har-wifi-csi

---

Thesis

Assel Ussenova WiFi Sensing Through Digital Receive Beamforming and CSI Master's Thesis Università degli Studi di Roma Tor Vergata Academic Year 2024/2025

---

Citation

If you use this dataset in your research, please cite the original publication:

@inproceedings{desanctis2025wifi,
  title={WiFi Sensing Through Digital Receive Beamforming and CSI},
  author={De Sanctis, Mauro and Fallani, Rebecca and Rossi, Tommaso and Cianca, Ernestina and Ruggieri, Marina and Poulkov, Vladimir},
  booktitle={2025 IEEE 36th International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC)},
  year={2025},
  doi={10.1109/PIMRC62392.2025.11274992}
}

---

License

This dataset is released for research and educational purposes.

Please cite the original publication and acknowledge the accompanying Master's thesis when using this dataset in academic work.