View the Docs!
Recent studies have developed different sensing applications like human activity recognition (HAR) using WiFi channel state information (CSI) information. However, they usually use high and different sampling rates of CSI, which is impractical and will hurt the communication performance. Besides, different sensing tasks or applications may require different minimum/best sampling rates due to different movement speeds and highest frequency of the activities. E.g, [1] suggests CSI sampling rate should be chosen as 800Hz for HAR, while [2] uses 500Hz and [3] uses 30Hz; [4] chooses 100Hz for indoor crowd counting; [5] exploits 200Hz for sign language recognition; etc. Therefore, it’s interesting to explore different applications’ dependence on sampling rates.
Resources: Dataset: NTU-Fi (HAR, Human ID), UT-HAR, Widar (hand gesture recognition), SignFi, WiAR, Exposing the CSI, RFDataFactory, and other datasets.
[1] Wang, Wei, et al. "Understanding and modeling of wifi signal based human activity recognition." MobiCom (2015).
[2] Yang, Jianfei, et al. "EfficientFi: Toward large-scale lightweight WiFi sensing via CSI compression." IEEE Internet of Things Journal (2022).
[3] Guo, Linlin, et al. "Wiar: A public dataset for wifi-based activity recognition." IEEE Access (2019).
[4] Hou, Huawei, et al. "DASECount: Domain-Agnostic Sample-Efficient Wireless Indoor Crowd Counting via Few-Shot Learning." IEEE Internet of Things Journal (2022).
[5] Ma, Yongsen, et al. "SignFi: Sign language recognition using WiFi." ACM IMWUT (2018).