Life pattern sensor with non-intrusive appliance monitoring

This paper presents a life pattern sensor with a new residents' behavior monitoring method for a health care system. The life pattern sensor detects operations of appliances from electrical current generated by their operating, and identifies patterns of daily residents' behavior based on the correlation of operating appliances. The sensor reduces the system cost by avoiding installation of massive sensor nodes and keeps residents' privacy without intrusion of the house, in comparison with legacy systems like the video-based health care system. The sensor is implemented by utilizing an algorithm with the wavelet transform method and installed in five real houses for a couple of weeks. Accuracy of the identifying appliance detection and the result of the life pattern estimation were evaluated through the field test.