foysalhaque/BeamSense

---
license: gpl-3.0
tags:
- wifi
- mimo
- wireless-sensing
- CSI
- Beamforming-Feedback-Information
pretty_name: BeamSense
size_categories:
- 100B<n<1T
---

# BeamSense: MU-MIMO Beamforming Feedback Dataset for Wi-Fi Sensing
## Dataset Summary

BeamSense is a large-scale MU-MIMO Wi-Fi sensing dataset built upon Beamforming Feedback Information (BFI) and Channel State Information (CSI) collected from commodity 802.11ac devices.
The dataset corresponds to the research presented in:

- **BeamSense: Rethinking Wireless Sensing with MU-MIMO Wi-Fi Beamforming Feedback (Computer Networks, 2025)**

- **BFA-Sense: Learning Beamforming Feedback Angles for Wi-Fi Sensing (PerCom Workshops, 2024)**

BeamSense contains synchronized BFI, BF angles, and CSI samples from multiple Wi-Fi stations (STAs) under different environments and orientations. It is designed for:

- Human activity recognition with Beamforming Feedback Information (BFI) and CSI
- Gesture recognition with Beamforming Feedback Information (BFI) and CSI
- Device-free sensing with BFI
- Cross-domain generalization
- Beamforming Feedback analysis
- Beamforming Feedback-based attacks
- Wi-Fi sensing algorithm development

## Dataset Contents

- Raw Beamforming Feedback Packet Captures: Contains BF report packets (BFI), and PHY metadata
- Synchronized CSI: Extracted from PCAPs using Nexmon-modified pipelines
- Synchronized BFI and CSI

## Multi-Environment, Multi-Day Collection:

- Includes data from: Office, Classroom, and Kitchen environments
- LOS and NLOS cases
- Multiple days of data collection
- 3 Different human subjects perform 20 different activities in multiple environments and orientations
- Spatially diverse data from multiple STAs

## How To Use

### Download the Dataset

Download the dataset with
```
hf download foysalhaque/BeamSense \
--repo-type dataset \
--local-dir your_directory/BeamSense_HF
```

You will have a directory structure like
```
your_directory/BeamSense_HF/
BeamSense_Dataset_aa
BeamSense_Dataset_ab
BeamSense_Dataset_ac
...
README.md
data.json
```

**These BeamSense_Dataset_aa, BeamSense_Dataset_ab, … files are split parts of one large ZIP file.**

Move into the download directory with ```cd your_directory/BeamSense_HF/``` and Combine all pieces into one ZIP with
``` cat BeamSense_Dataset_* > BeamSense_Dataset.zip ```. This produces a large file which is around 180 GB.

Extract the combined ZIP with ``` unzip BeamSense_Dataset.zip -d BeamSense_Data```

### MATLAB BFI Extraction

```
./Feedback_split_STAs.sh
cd BFI_Extraction
matlab -nojvm -nosplash -r "pcap_to_bfa; exit"
matlab -nojvm -nosplash -r "bfa_to_batches; exit"
```

### MATLAB CSI Extraction

```
cd CSI_Extraction
matlab -nojvm -nosplash -r "Extract_CSI; exit"
matlab -nojvm -nosplash -r "CSI_to_batches; exit"
```

## Citation

If you find our work and the dataset useful, please cite the following works:

```
@article{haque2025beamsense,
title={BeamSense: Rethinking wireless sensing with MU-MIMO Wi-Fi beamforming feedback},
author={Haque, Khandaker Foysal and Zhang, Milin and Meneghello, Francesca and Restuccia, Francesco},
journal={Computer Networks},
pages={111020},
year={2025},
publisher={Elsevier}
}
```
```
@inproceedings{haque2024bfa,
title={BFA-Sense: Learning Beamforming Feedback Angles for Wi-Fi Sensing},
author={Haque, Khandaker Foysal and Meneghello, Francesca and Restuccia, Francesco},
booktitle={2024 IEEE International Conference on Pervasive Computing and Communications Workshops},
pages={575--580},
year={2024},
organization={IEEE}
}
```

## Contact

For dataset questions or access issues:

K. Foysal Haque </br>
Northeastern University </br>
Email: [e-mail]

---
许可证: gpl-3.0
标签:
- Wi-Fi
- MU-MIMO
- 无线感知
- 信道状态信息（Channel State Information，CSI）
- 波束成形反馈信息（Beamforming-Feedback-Information，BFI）
美观名称: BeamSense
规模类别:
- 100B < n < 1T
---
# BeamSense：面向Wi-Fi感知的MU-MIMO波束成形反馈数据集

## 数据集概述

BeamSense是一款大规模MU-MIMO Wi-Fi感知数据集，基于从商用802.11ac设备采集的波束成形反馈信息（Beamforming Feedback Information，BFI）与信道状态信息（Channel State Information，CSI）构建而成。本数据集关联以下两项研究成果：

- **《BeamSense：利用MU-MIMO Wi-Fi波束成形反馈重新审视无线感知》（Computer Networks，2025）**

- **《BFA-Sense：面向Wi-Fi感知的波束成形反馈角度学习》（PerCom Workshops，2024）**

BeamSense包含不同环境与姿态下，来自多个Wi-Fi站点（STAs）的同步BFI、波束成形角度与CSI采样数据。本数据集的应用场景包括：

- 基于波束成形反馈信息（BFI）与信道状态信息（CSI）的人体活动识别
- 基于波束成形反馈信息（BFI）与信道状态信息（CSI）的手势识别
- 基于BFI的无设备感知
- 跨域泛化研究
- 波束成形反馈分析
- 基于波束成形反馈的攻击研究
- Wi-Fi感知算法开发

## 数据集内容

- 原始波束成形反馈数据包捕获文件：包含波束成形报告数据包（BFI）与PHY元数据
- 同步CSI：通过经Nexmon修改的数据处理流水线从PCAP文件中提取得到
- 同步BFI与CSI

## 多环境、多周期采集

- 涵盖办公室、教室与厨房三类环境
- 包含视距（Line-of-Sight，LOS）与非视距（Non-Line-of-Sight，NLOS）场景
- 多周期数据采集
- 3名不同受试者在多类环境与姿态下完成20种不同活动
- 来自多个STAs的空间多样化采样数据

## 使用指南

### 数据集下载

使用以下命令下载数据集：

hf download foysalhaque/BeamSense
--repo-type dataset
--local-dir your_directory/BeamSense_HF

你将得到如下目录结构：

your_directory/BeamSense_HF/
BeamSense_Dataset_aa
BeamSense_Dataset_ab
BeamSense_Dataset_ac
...
README.md
data.json

**注：BeamSense_Dataset_aa、BeamSense_Dataset_ab等文件均为单个大型ZIP压缩包的拆分分片。**

进入下载目录并执行以下命令将所有分片合并为一个ZIP文件：

cd your_directory/BeamSense_HF/
cat BeamSense_Dataset_* > BeamSense_Dataset.zip

合并后的文件大小约为180 GB。

使用以下命令解压合并后的ZIP文件：

unzip BeamSense_Dataset.zip -d BeamSense_Data

### MATLAB BFI提取

./Feedback_split_STAs.sh
cd BFI_Extraction
matlab -nojvm -nosplash -r "pcap_to_bfa; exit"
matlab -nojvm -nosplash -r "bfa_to_batches; exit"

### MATLAB CSI提取

cd CSI_Extraction
matlab -nojvm -nosplash -r "Extract_CSI; exit"
matlab -nojvm -nosplash -r "CSI_to_batches; exit"

## 引用说明

若您认为本研究与数据集对您的工作有所帮助，请引用以下文献：

@article{haque2025beamsense,
title={BeamSense: Rethinking wireless sensing with MU-MIMO Wi-Fi beamforming feedback},
author={Haque, Khandaker Foysal and Zhang, Milin and Meneghello, Francesca and Restuccia, Francesco},
journal={Computer Networks},
pages={111020},
year={2025},
publisher={Elsevier}
}

@inproceedings{haque2024bfa,
title={BFA-Sense: Learning Beamforming Feedback Angles for Wi-Fi Sensing},
author={Haque, Khandaker Foysal and Meneghello, Francesca and Restuccia, Francesco},
booktitle={2024 IEEE International Conference on Pervasive Computing and Communications Workshops},
pages={575--580},
year={2024},
organization={IEEE}
}

## 联系方式

若您对数据集有疑问或遇到访问相关问题，请联系：

K. Foysal Haque </br>
东北大学（Northeastern University）</br>
邮箱：[e-mail]