A mobile receiver WiFi-CSI approach for fall detection of construction workers

This study introduces a novel fall detection method for construction workers that uses WiFi Channel State Information (CSI) with mobile smartphone receivers, which addresses the high incidence of fall-related injuries at construction sites. The innovative approach utilizes Doppler frequency shift features captured through mobile receivers, which adapt to dynamic construction environments where workers continuously move, overcoming limitations of conventional static configurations. Our framework extracts characteristic CSI patterns from WiFi signals and employs an improved deep learning model to classify falls and common construction activities. Experimental validation demonstrates robust performance with accuracy exceeding 93% across various distances and orientations. The mobile receiver design significantly enhances spatial adaptability while providing a non-invasive, privacy-preserving, and cost-effective solution that can be readily deployed using existing WiFi infrastructure and workers’ smartphones for construction site safety monitoring. • Proposes a mobile WIFI-CSI receiver for fall detection. • Improves adaptability and coverage in changing construction sites. • Introduces refined data processing and a deep learning model to reduce noise. • Achieves over 93% accuracy in multiple test conditions.