Phenotype Matching: RF Sensor-Based Indoor Subject Identification with Wearable Sensor Assistance.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Human-Computer Interaction

Background:

• RF sensor-based indoor monitoring faces challenges in accurate subject identification.
• Existing machine learning methods require extensive user-specific data collection and training.
• Privacy concerns often limit the use of traditional indoor monitoring systems.

Purpose of the Study:

• To develop a novel, customization-free approach for subject identification in RF sensor-based indoor monitoring.
• To leverage comparable physiological signals from RF sensors and wearable devices for identity matching.
• To enable scalable, privacy-preserving, long-term behavioral monitoring in multi-subject environments.

Main Methods:

• Integrating wearable sensors with RF sensors to collect physiological time-series signals (phenotypes).
• Developing a phenotype matching algorithm based on signal similarity between RF and wearable sensor data.
• Validating the system with various wearable sensors: PPG, accelerometers, respiratory belts, and pedometers.
• Testing the system in both stationary and motion scenarios for real-time identity resolution.

Main Results:

• Demonstrated the feasibility of phenotype matching between RF sensors and common wearable devices.
• Achieved accurate real-time subject identity resolution for both stationary and moving subjects.
• The proposed method is customization-free, eliminating the need for user-specific training.
• Successfully bridged contactless radar sensing and user-attached wearables for enhanced monitoring.

Conclusions:

• The integrated RF and wearable sensor approach offers a robust solution for subject identification.
• This method enables privacy-preserving, scalable indoor monitoring without compromising user identity.
• The customization-free nature significantly reduces implementation barriers for new users and environments.