Wi-Actigraph: Wi-Fi CSI Sensing for Sleep Disturbances in the Care of Older People

In recent years, considerable effort has been directed toward unobtrusive sensing solutions for continuous in-home monitoring. Older adults increasingly suffer from disrupted sleep due to comorbid conditions, which affect their quality of life. Unobtrusive radio frequency (RF) sensing offers a promising solution for in-home sleep disturbance monitoring to aid in early detection and data continuity. Research to date has focused on vital sign extraction and monitoring of sleep stages rather than sleep disturbances in older adults using Wi-Fi channel state information (CSI). By drawing on concepts from sleep science, this article addresses this gap by implementing a novel Wi-Fi CSI sensing system to monitor sleep-disordered breathing and disturbances in the context of care of older people. We implement our system in a realistic sleeping environment and conduct a series of experiments to collect CSI data and measure different sleep parameters, such as vital signs, sleep-disordered breathing, and sleep disturbance movements, such as leg restlessness and confusional arousals. In terms of signal processing, we propose a novel level-dependent wavelet coefficient thresholding targeting coefficient scales of interest due to the sparse nature of the resulting transform. Finally, we extract vital signs, disordered breathing, and movement from wavelet-based features. The results obtained by our proposed system illustrate the effectiveness of wavelet analysis in detecting sleep disturbance events due to its robust time-frequency localization.