Abnormal Behavior Detection with Fuzzy Clustering for Elderly Care

Abstract

Home care for the elders who live alone is considered in this research. We focus on the movements of the elders at home. The RFID technology is used to collect the movement data first. Active RFID tags are deployed in the home environment. The elder carries a reader that can detect the signals sent from the tags in real time. The collected signals give us the movements of the elder at home. Clustering analysis is then utilized to build a personal behavior model for each elder based on these collected RFID signals/data. Here Fuzzy C-Means is chosen. This is different from our previous work [1] which used K-Means for clustering. The reason is that Fuzzy C-Means can provide a better representation of the distribution of the data. After the behavior model is built, any incoming datum that falls outside the model is considered abnormal. In this paper, we also discuss the criterion settings for issuing an alarm. Extensive experiments have been done and the results are presented. The experimental results demonstrate the usefulness of the system.