WiSDom: A model-driven solitary death prevention system based on WiFi signals and real-time supervised training

Abstract

A primary concern of many elders is ‘solitary death,’ being found as a rancid corpse long after they had passed away. In numerous cases, bodies are left unattended for days, months, or even years. These unfortunate cases have increased every year and have become a major social problem in many nations. Current warning systems utilize sensors or smartwatches, which are often costly, ineffective, and uncomfortable. This paper proposes WiSDom, a model-driven solitary death prevention system based on WiFi signals and real-time supervised training. The proposed methodology utilizes WiFi's Channel State Information (CSI) for the primary activity identification estimation, represents the system by a discrete event state transition model, maps the estimated activities into the external events of the model, validates its estimation with the forthcoming events, and labels the validated samples for the supervised training of its clustering algorithm in real-time closed-loop. Through the experimental results, we show that the system effectively warns emergency cases and swiftly detects fatal situations.