Wet-Ra

Diaper wetness monitoring is essential in various situations (e.g., babies and patients) to guarantee hygiene and avoid embarrassment. Existing diaper wetness monitoring methods include indicator lines, special sensors, and RFID, which require modifications on every diaper piece and cannot be easily checked under visual occlusions (e.g., trousers). In this paper, we introduce Wet-Ra, a contactless, ubiquitous, and user-friendly diaper wetness monitoring system based on RF signals. To extract informative features for wetness detection from RF signals, we construct Continuous-Radio-Snapshot and build corresponding signal representations that capture the distinct patterns of diapers of different wetness levels. We refine the signal representation by eliminating multi-path interference from the environment and mitigating the smearing effect with wavelet multisynchrosqueezing transform. To expand the usability of Wet-Ra, we build a transferable model that yields robust detection results in diversified environments and for new users. We conduct extensive experiments to evaluate Wet-Ra with 47 volunteers in 7 different rooms with three off-the-shelf diaper brands. Experiment results confirm that Wet-Ra can accurately identify diaper wetness in the real environment.