Device Free Wireless Sensing based Human Activity Recognition Using Commercial Off-the-Shelf IoT Single-Board Computers

As South Korea faces a growing aging population, there’s an increasing demand for comprehensive elderly care. This paper introduces a cost-effective method using the ESP32 microcontroller for device-free wireless sensing of ubiquitous WiFi signals to recognize activities through radio frequency (RF) reflections. A deep learning model is trained on diverse labeled activities using channel state information (CSI) data, enabling activity recognition without extra sensors or complex hardware. This system improves caregiving efficiency, simplifies daily log maintenance, and respects elderly individuals’ privacy. Acquiring a robust recognition rate for human activities through the utilization of a compact IoT device underscores the versatility of ESP32, not only in activity recognition but also in its potential to proactively address underlying issues affecting the elderly, including depression and anxiety.