A novel approach for high speed wireless pre-fall detection multisensory system

• Published in: Conference proceedings : Midwest Symposium on Circuits and Systems pp. 857 - 859
• Main Authors: Lopez-Yunez, Alfredo, Vasquez, Diana, Palacio, Luis A., Tiwari, Nikhil, Suryadevara, Vinay Kumar, Anandwala, Mobin, Rizkalla, Maher
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2014
• Subjects: Acceleration; Accelerometers; hardware; MatLab software; microcontroller; safety device; Sensor systems; sensors; Software; Wireless communication; Wireless sensor networks
• ISBN: 9781479941346; 1479941344
• ISSN: 1548-3746
• DOI: 10.1109/MWSCAS.2014.6908550

Enhancing the living standards of elderly people is important, and providing them with proper care at the time of emergency is necessary. Unintentional falls often result in an emergency situation for seniors. A novel approach based on efficient pre-fall detection is used to develop a system with smart wireless sensors that provide monitoring and protection for seniors and patients from a fall occurrence. This paper demonstrates an efficient, low cost and high speed sensory system for monitoring and detecting falls before occurring. The system is designed to fit in medical settings where communication between individual patients and control monitoring room is provided to enhance patient safety. The sensory network considers data collected from tri-axial accelerometers to detect and classify the type of fall that is about to happen. Pre-fall detection and classification is done through an integrated logic that constantly evaluates filtered acceleration data from the sensors and by calculating the tilt angle of the body. The fall decision suggested from the system will lead to deployment of a safety device to protect sensitive human body areas in elders such as hips, the brain, and the neck. The approach presented here was implemented with 10 subjects for several fall and normal activity scenarios. A Freescale ZSTAR3 system with wireless tri-axial accelerometers and ZigBee network (2.4GHZ) capability was used. Software was written to filter the white noise, while detecting the desirable "fall" signals, and making decision about the pre-fall detection and classification. The tilt angle and phase shift between the pulse wave (prior to the fall event) and the pulse magnitude were analyzed to diagnose the characteristics of the fall. The classification determined the direction of the fall occurrence. Results presented here show that this approach allows enough time for deployment of an integrated protection system.