Gait recognition using wifi signals

By Wen Wang, Alex X. Liu, Muhammad Shahzad. Ubiquitous Computing, 2016

534 citations recorded by the source corpus.

In this paper, we propose WifiU, which uses commercial WiFi devices to capture fine-grained gait patterns to recognize humans. The intuition is that due to the differences in gaits of different people, the WiFi signal reflected by a walking human generates unique variations in the Channel State Information (CSI) on the WiFi receiver. To profile human movement using CSI, we use signal processing techniques to generate spectrograms from CSI measurements so that the resulting spectrograms are similar to those generated by specifically designed Doppler radars. To extract features from spectrograms that best characterize the walking pattern, we perform autocorrelation on the torso reflection to remove imperfection in spectrograms. We evaluated WifiU on a dataset with 2,800 gait instances collected from 50 human subjects walking in a room with an area of 50 square meters. Experimental results show that WifiU achieves top-1, top-2, and top-3 recognition accuracies of 79.28%, 89.52%, and 93.05%, respectively.

Subfields: Biomedical Engineering, Electrical and Electronic Engineering, Signal Processing

View Paper: https://www.semanticscholar.org/paper/d601f7736f9e16449fa2d2c6ce6171bca4b05b10

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

1. Jun Liu
2. Qiuhong Ke
3. Hossein Rahmani
4. Zehua Sun
5. Mingmin Zhao
6. Antonio Torralba
7. Kun Qian
8. Gang Zhou

In Electrical and Electronic Engineering

1. Zheng Yang
2. Yunhao Liu
3. Kun Qian
4. Gang Zhou
5. Chenshu Wu
6. Yongsen Ma
7. Shuangquan Wang
8. Guidong Zhang

In Signal Processing

1. Hang Zhao
2. Kun Qian
3. Yunhao Liu
4. Zheng Yang
5. Chenshu Wu
6. Jun Liu
7. Yonglong Tian
8. Antonio Torralba

Citations in the corpus

• RF-based 3D skeletons (2018, background): “Different papers localize the people in the environment [2, 17], monitor their walking speed [15, 31], track their chest motion to extract breathing and heartbeats [3, 39, 41], or track the arm motion to identify a particular gesture [21, 23].”
• Human Action Recognition From Various Data Modalities: A Review (2020, background): “[341] proposed WifiU, generating spectrograms from CSI measurements and extracting features for capturing finegrained activity patterns based on commercial off-the-shelf WiFi devices.”
• Cross-Domain WiFi Sensing with Channel State Information: A Survey (2022, methodology): “While the works in [27, 118] directly estimate the velocity of different body parts, including torso contour speed, the torso speed, and the leg speed, with the upper contour torso reflection method [86] and percentile method [73].”
• Zero-Effort Cross-Domain Gesture Recognition with Wi-Fi (2019, background): “…extract various parameters of signals reflected or shadowed by human, including DFS [26, 32, 44], ToF [2–4, 21], AoA / AoD [2 , On the human side, existing model-based works only tracks coarse human motion status, such as location [4, 41], velocity [26, 32], gait [43, 49] and figure [2, 19].”
• Widar3.0: Zero-Effort Cross-Domain Gesture Recognition With Wi-Fi (2021, unknown): “On the human side, existing model-based works only tracks coarse human motion status, such as location [21], [28], velocity [17], [20], gait [30], [31] and figure [24], [32].”
• Rapid (2017, methodology): “As pointed out in [26], when user walks in a different path with respect to WiFi transmitter/receiver (Tx/Rx), the perceived CSI features (such as Doppler frequency) change even if the user retains the same gait with the same walking speed.”
• Wireless Sensing for Human Activity: A Survey (2020, methodology): “Different from the previous studies [16], [69], the proposed system adopts a robust semi-supervised learning approach for estimation of the number of participants, which can be easily extended to a new environment.”
• Through-Wall Human Pose Estimation Using Radio Signals (2018, background): “Finally, some prior papers have explored human identification using wireless signals [49, 43, 18].”
• A Review on Machine Learning and Deep Learning Perspectives of IDS for IoT: Recent Updates, Security Issues, and Challenges (2020, background): “Therefore, multiple nodes are needed to forward message in the network [8–10].”
• Towards 3D human pose construction using wifi (2020, background): “WiFi Sensing: Recently, a considerable number of investigations are conducted to make use ofWiFi devices to enable various sensing tasks, including indoor localization and tracking [5, 22, 25, 33, 39, 42, 48–50, 52, 53, 65], walking speed estimation [17, 43], breathing and heartbeats inference [6, 5”
• WiFi Sensing with Channel State Information (2019, background): “For example, CSI amplitude variations in the time domain have different patterns for different humans, activities, gestures, and so on, which can be used for human presence detection [3, 24, 67, 73, 75, 83, 112, 114, 121, 148, 149, 152], fall detection [32, 68, 92, 135, 137], motion detection [23, 2”