CALS

Channel State Information Auto-Labeling System for Large-scale Deep Learning-based Wi-Fi Sensing

Abstract: Wi-Fi Sensing, which uses Wi-Fi technology to sense the surrounding environments, has strong potentials in a variety of sensing applications. Recently several advanced deep learning-based solutions using CSI (Channel State Information) data have achieved high performance, but it is still difficult to use in practice without explicit data collection, which requires expensive adaptation efforts for model retraining. In this study, we propose a Channel State Information Automatic Labeling System (CALS) that automatically collects and labels training CSI data for deep learning-based Wi-Fi sensing systems. The proposed system allows the CSI data collection process to efficiently collect labeled CSI for labeling for supervised learning using computer vision technologies such as object detection algorithms. We built a prototype of CALS to demonstrate its efficiency and collected data to train deep learning models for detecting the presence of a person in an indoor environment, showing to achieve an accuracy of over 90% with the auto-labeled data sets generated by CALS.

CALS Flowchart

CALS Architecture

• There are two versions included: a CSV version that works with pre-extracted CSI data in CSV format, and a DB version that collects CSI data directly from the Raspberry Pi extractor and stores it in a MySQL database for further process.

---

Prerequisites

• Python 3.x
• Nexmon CSI Extractor Raspberry Pi B3+/B4(Wi-Fi chip: bcm43455c0)
• Install ByteTrack on the server
• MySQL server

---

Installation

Server

1. Clone the repository

git clone https://github.com/INCLab/CALS.git

2. Install the required dependencies

pip install -r requirements.txt

3. Change to the project directory

cd bytetrack

4. Install the ByteTrack

• Note: If you want to use another CV module, you can install the desired module. And additionally, columns of output should be modified as follows: time frame_index label

Extractor

1. Clone the repository

git clone https://github.com/INCLab/CALS.git

2. Install the Nexmon CSI Extractor

---

Usage - CSV

1. Time syncronization

After installing Nexmon firmware on Raspberry Pi, synchronize time with the camera server.

Method: Network Time Protocol (NTP)
One of the easiest ways to synchronize time between two devices is to use the Network Time Protocol (NTP). This protocol allows devices to synchronize their clocks over a network connection.

2. Collect CSI data & record video

• Change to the extractor directory on the Raspberry Pi extractor:

cd csi

• Enter the following command on the Raspberry Pi extractor:

python3 extract_csi_csv.py

Note
※ Must be done under sudo privileges
※ If you want to stop CSI extraction, press the keyboard 's'

3. Vision module process

Generate a MOT output by following the instructions in the ByteTrack repository.

4. CSI labeling

Set the time, sampling rate on cfg.py and enter the following command on the server:

python csi_labeling.py

---

Referenced Projects

This project takes inspiration from the following two open-source projects:

1. Nexmon: The Nexmon project provides firmware patches for collecting CSI on Broadcom Wi-Fi chips. For more information about this project, please visit the Nexmon GitHub repository.

2. ByteTrack: ByteTrack offers an efficient algorithm for real-time multi-object tracking. For more information about this project, please visit the ByteTrack GitHub repository.