WiFi Amplitude and Phase-Based Respiratory Rate Monitoring

Contactless respiratory rate monitoring methods have shown significant potential for patient monitoring and home healthcare in recent years because they could supersede traditional wearable equipment, enabling non-contact monitoring. However, existing experiments are constrained to specific devices that are difficult to access in daily life and have strict limitations on the version of the system. Therefore, it would be highly desirable if the latest easily accessible low-cost IoT devices, such as Raspberry Pi, could be used for respiratory rate detection tasks. In this paper, we tackle this limitation by applying Raspberry Pi for respiratory rate detection and introducing the envelop-based preprocessing method. The envelop-based method enables human respiratory pattern extraction from both the amplitude and phase of WiFi channel state information(CSI). The combination of autocorrelation function of selected quality subcarrier then estimates the respiratory rate. Our experiment result indicates that the estimation accuracy from amplitude and phase reach 98.94% and 98.54%, respectively. Compared with the traditional preprocessing method based on the Savitzky-Golay filter, the enveloped-based method reaches 5.98% and 5.78% improvement in the accuracy of exploiting amplitude and phase information respectively, demonstrating the superiority and potential for further applications.