Abstract
Smartphones are extensively used for communication, business, study,
entertainment, and other purposes in everyone’s daily life.
Unfortunately, using the smartphone for prolonged periods causes several
problems. The development of a complicated cluster of clinical symptoms
known as “text neck syndrome” may be linked to the improper usage of
personal devices, especially mobile phones. In addition, typical
postures while using mobile phone devices can cause musculoskeletal
problems. Various technologies are being considered to keep track of
health and identify problems unobtrusively. This paper employs
software-defined radio (SDR) based RF sensing and machine learning (ML)
algorithms to develop a testbed for detecting text neck syndrome and
classifying healthy and unhealthy postures. Specifically, fine-grained
orthogonal frequency division multiplex (OFDM) samples are leveraged for
channel state information (CSI) acquisition for detecting neck tilt
angles while using the mobile phone. For classification purposes, the ML
algorithms are used, and their performance in terms of prediction time,
training time, and accuracy is assessed. The performance evaluation
results of the testbed validated that this platform can faithfully
detect and classify healthy and unhealthy postures with maximum accuracy
of 99.9% with fine kth-nearest neighbors (KNN). The developed testbed
has a considerable clinical impact on improving human health.