Detection of Fall-Related Body Postures from WBAN Signals

Abstract

Falls are a major health risk for elderly people living independently. In contrast to fall detection systems, which raise an alarm after the fall has occurred, fall prevention aims to avoid fall-related injuries by initiating countermeasures before the fall occurs. This requires early and fast detection of critical body postures which may lead to a fall. In this work, we evaluate the feasibility of posture detection based on ultra wideband (UWB) communication links between wireless on-body nodes. Our proposed wireless body area network (WBAN) features a hierarchical architecture with low-complexity agent nodes, medium-complexity relays and a processing unit, which can be located off-body. We demonstrate the feasibility of the concept by performing posture detection on measurements from 18 on-body antennas. Our approach proves to be robust towards external disturbances and performs well even under low requirements for synchronization. © 2020 IEEE.