Thin Film Reconfigurable Intelligent Surface Assisted Device-Free Fall Detection

Fall detection is essential for safeguarding vulnerable
populations, such as the elderly. We design and implement
a low-complexity reconfigurable intelligent surface (RIS)-assisted
fall detection system. The manufactured thin film 4X1 RIS can
effectively control the scattering of the dominant harmonics of
the captured radio frequency (RF) carrier wave by steering
their beams to a desired direction ranging from 230° to 310°
in elevation. With beam steering, the RIS enables device-free
fall detection by conducting dynamic elevation beam-scanning
in an indoor space and analyzing variations in received signal
strength (RSS) caused by human presence and motion, without
requiring the person to carry any sensors or electronic devices.
We design a posture recognition-based fall detection algorithm.
Experimental results show that using a simple support vector machine
(SVM) classifier can achieve 92.50% accuracy in significant
posture recognition that includes standing, sitting, and lying. Five
voluntary participants were used for cross-validation. Compared
with a conventional sensor node pair-based solution with a static
beam, our proposed dynamic beam scanning technique improves
accuracy by 11.83% using SVM. The results indicate that the
RIS and the proposed algorithm have the potential to facilitate
fall detection applications for assisted living.