Title: Sleep behaviour monitoring based on the probability density model

Authors: Ying Liu; Zhiyang Cao; Mengyuan Hu

Addresses: School of Electronic Information and Engineering, Liaoning Technical University, Daan 131300, China ' School of Electronic Information and Engineering, Liaoning Technical University, Huludao 12500, China ' Luxshare Precision Industry Co., Ltd. Yulin 719000, China

Abstract: Behavioural changes in the human body during sleep are an important reflection of sleep quality. Most of the existing methods for the recognition of sleep behaviour are based on the magnitude or phase of CSI, but such features often suffer from low differentiation and poor stability. To address the above problems, a sleep behaviour monitoring system based on the CSI amplitude probability density model is proposed on the theoretical basis of wireless channel fading characteristics. First, CSI amplitudes from different antennas and carriers are collected as sense base signals using spatial diversity and frequency diversity techniques of commercial Wi-Fi devices. Next, the raw data are preprocessed and probability densities of the different subcarrier amplitudes are obtained to segment the action using Gaussian fitting. Finally, a support vector machine classifier is constructed to classify and recognise different sleep postures, and a back-propagation neural network is also used to recognise related actions. The results show that the recognition accuracy of this method is higher than that of the traditional method in the case of small samples, which proves the robustness of the feature model.

Keywords: channel state information; probability density model; sleep posture recognition; action recognition.

DOI: 10.1504/IJSNET.2025.148201

International Journal of Sensor Networks, 2025 Vol.48 No.4, pp.227 - 240

Received: 05 Jul 2024
Accepted: 21 May 2025
Published online: 29 Aug 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article