WiFi-based identity recognition is predominant because of its noninvasive and ubiquitous advantages. However, existing approaches show slow training speed and limited applicability. In this article, we propose a lightweight deep learning model, named as lightweight WiFi-based identification (LW-WiID), to address these technical challenges. LW-WiID reconstructs original data of channel state information into frequency energy graph, which contains not only the temporal feature of the gait but also the spatial feature among subcarriers, ensuring the accuracy of identity recognition. Furthermore, a novel Balloon mechanism is designed to achieve the lightweight. Through information integration crossing both layers and channels, the Balloon mechanism effectively reduces the number of model parameters. Experimental results demonstrate that LW-WiID achieves an accuracy of 99.7% on a 50-person gait data set while the model size is compressed to 5.53% of the existing identity recognition approaches with the same accuracy.

In recent years, Wi-Fi sensing technology has become an emerging research direction of human–computer interaction due to its advantages of low cost, contactless, illumination insensitivity, and privacy preservation. At present, Wi-Fi sensing research has been expanded from target location to action recognition and identity recognition, among others. This paper summarizes and analyzes the research of Wi-Fi sensing technology in human identity recognition. Firstly, we overview the history of Wi-Fi sensing technology, compare it with traditional identity-recognition technologies and other wireless sensing technologies, and highlight its advantages for identity recognition. Secondly, we introduce the steps of the Wi-Fi sensing process in detail, including data acquisition, data pre-processing, feature extraction, and identity classification. After that, we review state-of-the-art approaches using Wi-Fi sensing for single- and multi-target identity recognition. In particular, three kinds of approaches (pattern-based, model-based, and deep learning-based) for single-target identity recognition and two kinds of approaches (direct recognition and separated recognition) for multi-target identity recognition are introduced and analyzed. Finally, future research directions are discussed, which include transfer learning, improved multi-target recognition, and unified dataset construction.

Wi-Fi-based contactless activity recognition is of great importance to computer-human interaction, accounting for convenience concerns. However, it remains challenging to recognize activities from multiple users due to the multi-path distortion and disruption of Wi-Fi signals. In this paper, we propose a highly universal framework, namely WISDOM, for Wi-Fi based multi-user activity recognition. Specifically, we first leverage an existing model to identify the number of users from the input Wi-Fi signals. Then, we develop a sub-carrier correlation and inversion based sorting algorithm to extract the signal for each user. Finally, we design a neural network, i.e., WISDOM-Net, which is built on a bidirectional gated recurrent unit network incorporated with the attention mechanism and the one dimension convolutional neural network, to recognize the corresponding user activities. Experimental results show that our proposed WISDOM-Net outperforms the existing baselines on both the public and our own datasets. In particular, WISDOM-Net can reach an average recognition accuracy of up to 98.19% and 90.77% in 2-user and 3-user scenarios, respectively.

In the fields of intelligent security and human-computer interaction, the rapid development of non-contact identity recognition technology based on Wi-Fi signals has shown promising application potential. To address the significant decrease in recognition accuracy in close-range scenarios, an close-range non-contact identity recognition method named WiDoor is proposed. During the data collection phase, the Fresnel propagation model is utilized by WiDoor to optimize the deployment layout of the receiving antennas. Gait information is reconstructed from the multiple antennas to enable the acquisition of more rich gait features. In the identity recognition stage, WiDoor employs a lightweight model that combines self-attention mechanisms with multi-scale convolutional neural networks. This combination effectively enhances the model’s capability to capture key features while significantly reducing computational complexity and maintaining a high recognition accuracy. Experimental results show that WiDoor achieves a recognition accuracy of up to 99.3% on an expanded dataset that includes 10 participants, with a distance of 1 meter between the receiving and transmitting ends, and the parameter quantity of the built-in model is only 2% of the compared model with the same accuracy, offering a significant advantages over similar methods. Additionally, the model can achieve a high-precision recognition across different distances between the transmitter and the receiver using a limited number of samples, showing strong robustness of the model.