Summary

The real-time health monitoring system is a promising body area network application to enhance the safety of fire fighters when they are working in harsh and dangerous environment. Except for monitoring the physiological status of the fire fighters, on-body monitoring network can be also regarded as a candidate solution of motion detection and classification. In this paper, a novel Support Vector Machine (SVM) classifier has been implemented using RF signals as classification features. The classifier is capable of detecting and classifying seven frequently appeared motions of fire fighters including standing, walking, running, lying, crawling, climbing and running up stairs. The average true classification rate of our classifier reaches 87.9175% and the effects of different human motions and sensor locations have been analyzed by plotting Receiver Operating Characteristics (ROC) curves.

Dynamic scenario on the 3rd floor of Atwater Kent Laboratory at WPI.

Principal Investigators

• Kaveh Pahlavan (CWINS)

Research Staff (CWINS)

• Yishuang Geng
• Jin Chen

Sponsor

UCLA

Duration

Started Spring 2013