PWiG: A Phase-based Wireless Gesture Recognition System

With the development of the artificial intelligence, people hope that the advancement of the science and technology will continuously bring more convenience to their daily life. Therefore, gesture recognition has drawn more attentions. In the previous research, people paid their attentions on the aspect of the wearable devices and computer graphic. Although the recognition accuracy of previous methods is satisfactory under line-of-sight, the costs are very high and they require the users to be equipped with wearable sensors, which hinder their wide deployment. The development of wireless network provides convenience for solving this problem. Different gestures can be identified only according to the fluctuation of RSSI of wireless signals. But the intrinsic limitations of RSSI make it difficult to be feasible. Fortunately, CSI-based gesture recognition systems overcome the shortcomings above. In this paper, we propose a Phase-based Wireless Gesture recognition system PWiG, an accurate device-free gesture recognition approach. Compared with other existing gesture recognition systems, PWiG only needs commercial Wi-Fi devices to identify different gestures. PWiG uses the more sensitive phase information. Moreover, it reduces the hardware cost and greatly simplifies the process of the gesture extraction only by calculating the variance values. In addition, PWiG meets the standard of gesture recognition. According to the experimental results, PWiG achieves an average recognition accuracy of 94% in line-of-sight scenario and 89% in none-line-of-sight scenario.