MobileRF: A Robust Device-Free Tracking System Based On a Hybrid Neural Network HMM Classifier

Area of Science:

• Computer Science
• Electrical Engineering
• Signal Processing

Background:

• Existing indoor tracking systems often rely on body-worn devices or tags, limiting their practicality.
• Received Signal Strength (RSS) from radio frequency (RF) transceivers offers a potential avenue for device-free localization.
• Environmental changes and signal fluctuations pose significant challenges for reliable RSS-based tracking.

Purpose of the Study:

• To develop and evaluate a novel device-free indoor tracking system utilizing RF signal characteristics.
• To accurately estimate a person's location and movement patterns without requiring any wearable sensors.
• To assess the system's capability for occupancy estimation and its suitability for elder care applications.

Main Methods:

• A system employing wall-mounted RF transceivers to monitor Received Signal Strength (RSS) variations.
• A hierarchical neural network hidden Markov model (NN-HMM) classifier to interpret signal changes and estimate location.
• Algorithm designed for robustness against environmental RSS mean shifts over extended periods.

Main Results:

• Achieved over 90% region-level classification accuracy in tracking individuals over an extended testing period.
• Demonstrated accurate estimation of movement patterns (e.g., standing vs. walking).
• Successfully estimated the number of people within different monitored regions.

Conclusions:

• The developed device-free indoor tracking system offers a promising solution for non-intrusive human localization and monitoring.
• The NN-HMM approach provides robust and accurate performance, even with environmental signal variations.
• The system has significant potential for applications in independent living and long-term senior care.