Monitoring Respiratory Motion With Wi-Fi CSI: Characterizing Performance and the BreatheSmart Algorithm.

Area of Science:

• Biomedical Engineering
• Wireless Communications
• Artificial Intelligence

Background:

• Respiratory motion analysis is crucial for diagnosing health disorders.
• Wireless, contactless monitoring offers advantages over traditional methods.
• Evaluating Wi-Fi-based respiratory monitoring technologies requires standardized assessment techniques.

Purpose of the Study:

• To demonstrate and assess the feasibility of a Wi-Fi-based system for monitoring human respiratory motion.
• To develop and evaluate a novel deep-learning algorithm for extracting respiratory information from Wi-Fi Channel State Information (CSI).
• To establish a replicable assessment technique for quantifying the performance of such systems.

Main Methods:

• Implementation of an end-to-end system using commercial off-the-shelf (COTS) Wi-Fi devices.
• Development of a deep-learning algorithm processing CSI amplitude and phase data.
• Conducting extensive laboratory experiments to assess system performance under various conditions.

Main Results:

• The proposed system achieved high accuracy in classifying respiratory patterns (99.54%) and rates (98.69%) even in degraded RF channels.
• Demonstrated the system's capability in detecting and classifying respiratory motions through comprehensive data acquisition.
• Validated the effectiveness of the developed assessment technique for similar technologies.

Conclusions:

• Wi-Fi CSI-based monitoring is a feasible, accurate, and low-cost approach for contactless respiratory monitoring.
• The developed deep-learning algorithm effectively extracts respiratory information from Wi-Fi signals.
• Understanding system limitations and evaluation methods is key for practical deployment of Wi-Fi respiratory monitoring.