Comprehensive Measurement-Based Evaluation of Posture Detection from Ultra Low Power UWB Signals


Robert Heyn and Armin Wittneben


IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Virtual Conference, Sept. 2021.

[BibTeX, LaTeX, and HTML Reference]


Due to the ever-increasing life expectancy, extending the autonomy of elderly people is of great social importance. Posture detection can be a crucial element here. An example of particular significance is fall prevention, whereby critical postures are detected before a fall occurs - in any environment. An on-body posture detection system based on wireless signaling is particularly attractive because it is lightweight, not obstructive, and promotes a ubiquitous use as it does not require any off-body infrastructure. Due to the manifold of human physiques and living environments (i.e. signal propagation and multipath conditions), posture detection based on wireless signals is very challenging. The feasibility of such a system for an extensive set of postures has not been demonstrated and appropriate comprehensive wideband on-body matrix channel measurements are missing. This paper tries to fill this void. We first define an extensive set of postures related to fall prevention. For these 43 postures we perform a large-scale measurement campaign of (18x18) channel impulse response (CIR) matrices between 18 on-body nodes. In order to make the results as representative as possible, we intentionally include variation in each posture and consider various test subjects and indoor environments. The posture detection performance is evaluated for two metrics: (i) total energy of the CIR (ultra low complexity), and (ii) magnitude of CIR (very low complexity). We investigate suitable choices of carrier frequency and bandwidth, and demonstrate that ultra low transmit power wireless posture detection is feasible under real-world constraints for an extensive set of postures.


posture detection, UWB, WBAN, fall prevention

This publication is currently not available for download.