Real-Time CSI-Based Wireless Imaging for Human-Machine Interaction

In human-machine interaction systems, device-free and privacy-aware activity monitoring and gesture recognition is of great interest. In this framework, this work introduces a multi-step wireless imaging approach based on channel state information (CSI) and machine learning for the user's presence detection, localization, identification, and gesture recognition by using commodity Wi-Fi signals. The preliminary experimental validation in a real environment shows an excellent classification accuracy of 98%.