Experimental Study of Posture Detection Using Purely Passive Magneto-Inductive Tags

Authors

Henry Schulten and Armin Wittneben

Reference

IEEE Wireless Communications and Networking Conference (WCNC), Austin, TX, USA, pp. 6, 2022-04-10, to appear.

[BibTeX, LaTeX, and HTML Reference]

Abstract

We designed a wireless and wearable posture detection system for the human body, which uses four measuring coils (anchors) placed on the torso and additional purely passive resonant coils (tags) placed on each limb and the head. Due to low-frequency magneto-inductive coupling between all coils, each different sensor topology, as a result of the user’s posture, leads to a different and possibly unique set of input impedances at the anchors. Learning the relationship between each posture and its corresponding impedance pattern via supervised classifiers thus enables low-cost and low-complexity posture detection. Previously, this concept was only investigated via simulation, which showed promising results. We now take the next step by presenting the first experimental study of this concept. As such, we propose a simple design for testing purposes and use it to assess the system’s real world posture detection capabilities for 14 different everyday postures. Moreover, we examine the system’s robustness against additional disturbances which would occur when using cheaper and simpler measurement equipment. Lastly, we investigate the impact of each individual anchor on the full system performance. Overall, the proposed system classifies all test cases correctly using the direct measurements and also proves to be resistant against additional Gaussian disturbances.

This publication is currently not available for download.