Fall Detection Using Wi-Fi Channel State Information

The deployment of Wi-Fi Channel State Information (CSI) for non-invasive fall detection offers a promising avenue for enhancing elderly care without the constraints of wearable devices. This study highlights the potential of CSI in fall detection applications and demonstrates how it can be integrated into intelligent healthcare systems without compromising senior citizens' privacy or comfort. The proposed framework focuses on extracting amplitude data from CSI signals, which are then processed and segmented to identify fall-related patterns. Multiple Machine Learning (ML) and Deep Learning (DL) models were trained and evaluated for their performance in accurately detecting falls. Experimental results showed that the Long Short-Term Memory (LSTM) model excelled by displaying high accuracy and proving its proficiency in managing the time-series data characteristic of CSI.