A Hybrid IoT Fall Detection System Using Wearable Sensors and WiFi Channel State Information

This paper presents a robust and scalable fall detection framework that integrates wearable inertial sensors with WiFi Channel State Information (CSI) in a pipelined architecture suitable for Internet of Things (IoT) environments. Unlike conventional dual-modality systems that process sensor and ambient signals in parallel, the proposed approach employs a sequential decision strategy. The wearable sensor module continuously monitors body motion using an XGBoost classifier and triggers the WiFi-based verification module only upon detecting potential falls. The WiFi stream, powered by a Support Vector Machine (SVM) classifier, analyzes surrounding signal disturbances to confirm or reject the initial prediction. This conditional activation mechanism significantly reduces false alarms while maintaining high detection sensitivity. Experimental results in the Sensors and WIFI datasets demonstrate that the pipelined system achieves an overall accuracy of 99.5%, with a substantial reduction in the false alarm rate to 0.05%. The framework is designed for real-time deployment in smart home and healthcare settings, offering an effective trade-off between responsiveness, privacy preservation, and computational efficiency.