Dataset of WiFi-based Environment-independent In-baggage Object Identification System

Description:

The dataset of environment-independent in-baggage object identification system leveraging low-cost WiFi. The dataset contains the extracted CSI features from 14 representative in-baggage objects of 4 different materials. The experiments are conducted in 3 different office environments with different sizes. We hope this dataset will help researchers to reproduce the former work of in-baggage object identification through WiFi sensing.  

Dataset Format: 

.mat files

Section 1: Device Configuration: 

• Transmitter: Aaronia HyperLOG 7060 direction antenna with a Dell Inspiron 3910 desktop for control.  

• Receiver: Hawking HD9DP orthogonal antennas with a Dell Inspiron 3910 desktop for control

• NIC: Atheros QCA9590. The configuration and installation guide of CSI tool can be found at https://wands.sg/research/wifi/AtherosCSI/

• WiFi Packet Rate: 1000 pkts/s 

Section 2: Data Format

We provide the CSI features through .mat files. The details are shown in the following:

• 14 different objects made of 4 different materials are included in 3 different environments and 3 different days.

• Each object is tested for 60 seconds and repeated for 3 times. 

• The dataset file name is presented as "Object_Number". The detailed information are:

  • Object: The object we involved in the experiment (e.g., book, laptop)

  • Number: The number of repeats.  

Section 3: Experimental Setups

There are 3 different office experiment setups for our data collection. The detailed setups are shown in the paper. For the objects, we involve 14 types of objects made of 4 different materials.  

• Environments: 

  • 3 different environments are involved, including 3 office environments with the size of 15 ft × 13 ft, 16 ft × 12 ft, 28 ft × 23 ft, respectively. 

  • For each room environment, data is collected on different days and with different furniture settings (i.e., 2 desks and 2 chairs are moved at least 3 ft. )

• Representative objects: 

  • Data is collected using 14 representative objects of 4 different materials including fiber: book, magazine, newspaper; metal: thermal cup, laptop; cotton/polyester: cotton T-shirts (×2), cotton T-shirts (×4), hoodie, polyester T-shirts, polyester pants; water: 1L bottle with 1L water, 1L bottle with 500ml water, 500ml bottle with 500ml water. 

Section 4: Data Description

For our data organization, we separate the data files into different folders based on different days and different environments. Under these folders, data are further distributed in terms of different objects and repeat times.  All the files are .mat files, which can be directly read for further applications.  

• Features of CSI amplitude: We calculate 7 different types of statistical features, including mean, variance, median, skewness, kurtosis, interquartile range and range, and polarization feature from CSI amplitude. Particularly, we calculate the features for all 56 subcarriers with different operating frequencies and responses to the target object.  

• Features of CSI phase: For the features of CSI phase, the same features with CSI amplitude are extracted and stored in the dataset.  

Section 6: Citations

If your work is related to our work, please cite our papers as follows. 

https://ieeexplore.ieee.org/document/9637801

Shi, Cong, Tianming Zhao, Yucheng Xie, Tianfang Zhang, Yan Wang, Xiaonan Guo, and Yingying Chen. "Environment-independent in-baggage object identification using wifi signals." In 2021 IEEE 18th International Conference on Mobile Ad Hoc and Smart Systems (MASS), pp. 71-79. IEEE, 2021.