Wau: A User Authentication System Based on Channel State Information and Deep Neural Networks

Abstract Recently, user authentication techniques have attracted more attention and played a more important role in human-computer interaction applications. In this paper, we propose a user authentication system based on CSI (channel state information) called Wau. This system works in a device-free way and has no privacy violation because it facilitates the crucial characteristics of CSI that the unique CSI variation from different user activity can be measured and recognized. We adopt deep neural networks to realize the authentication of 8 people in three environments, including a through-the-wall scenario. The highest recognition accuracy reaches up to 98%. These results validate the feasibility and effectiveness of the user authentication system using CSI and deep neural networks.