WiFi-Based Real-Time Calibration-Free Passive Human Motion Detection.

Area of Science:

• Computer Science
• Electrical Engineering
• Signal Processing

Background:

• Wireless local area network (WLAN) technology enables device-free passive human detection for smart applications.
• Existing methods often require dense transmitter-receiver links or extensive offline training, limiting rapid deployment and robustness.
• Indoor fine-grained human motion detection using physical layer information is an emerging research area.

Purpose of the Study:

• To develop a novel, fine-grained, real-time, calibration-free device-free passive human motion detection system.
• To achieve a system independent of indoor scenarios and prior calibration or normal profiles.
• To investigate the sensitivity of physical layer information (amplitude and phase) to human motion.

Main Methods:

• Investigated the sensitivity of amplitude and phase features to human motion, identifying phase as more sensitive, especially for slow movements.
• Developed two novel schemes: short-term averaged variance ratio (SVR) and long-term averaged variance ratio (LVR) for lightweight and robust detection.
• Implemented the system using commercial WiFi devices and evaluated performance in multipath-rich indoor environments.

Main Results:

• Phase information demonstrated higher sensitivity to human motion compared to amplitude.
• The proposed SVR and LVR schemes provide lightweight and robust device-free passive human motion detection.
• Experimental evaluation in typical indoor scenarios confirmed high detection rates and low false positive rates.

Conclusions:

• The developed calibration-free approach offers a practical solution for real-time device-free passive human motion detection.
• The system's independence from specific indoor environments and prior calibration enhances its deployability and robustness.
• This technique holds significant potential for various ubiquitous smart applications leveraging WLAN infrastructure.