Non-Line-of-Sight Hand Gesture Recognition Using Wifi Channel State Information

In this study, we propose a hand gesture recognition method using Wi-Fi Channel State Information (CSI) obtained from an ESP32 device, a low-cost IoT module equipped with a single antenna. In smart homes and IoT environments, touchless interfaces are increasingly required for operating appliances and embedded devices. Because many IoT modules employ single-antenna Wi-Fi chips such as the ESP32, it is important to verify whether reliable gesture recognition can be achieved under such hardware constraints. Although single-antenna systems lack the spatial diversity of MIMO devices, the proposed preprocessing and learning pipeline achieved accuracies above 90%, demonstrating that reliable gesture recognition is feasible even with this hardware limitation.