During hazardous activities, continuous human motion monitoring can be a step forward towards increased safety standards. Many situations bring people like firefighters, soldiers or policemen inside indoor spaces where different factors may put their life at risk. The paper presents a multilevel sensor network architecture to monitor the movement of people in indoor spaces with poor visibility and aggressive environment (like fire smoke) and do not allow the use of video systems in good condition. Data collected from the accelerometers (components of primary network nodes) are analyzed in order to detect the distances traveled, the type of movement, activity, falls. For analysis and validation were used commercial accelerometers and relay nodes. Attempts were made with different sampling frequencies and positioning. The experimental results show the possibility of remaking the road map or the positioning inside a building with several levels. The system architecture (femtocell type) allows embedding sensors in protective equipment and wireless local communication. Key-Words: dead reckoning, activity recognition, motion recognition, local filer, body sensor network, accelerometer, monitoring, localization.
Download Full PDF Version (Non-Commercial Use)