Corresponding Author

Jian Gao

Document Type



Nowadays, more and more elderly people cannot take care of themselves, and feel uncomfortable in daily activities. Smart home systems can help to improve daily life of elderly people. A smart home can bring residents a more comfortable living environment by recognizing the daily activities automatically. In this paper, in order to improve the accuracy of activity recognition in smart homes, we conduct some improvements in data preprocess and recognition phase, and more importantly, a novel sensor segmentation method and a modified KNN algorithm are proposed. The segmentation algorithm employs segment sensor data into fragments based on predefined activity knowledge, and then the proposed modified KNN algorithm uses center distances as a measure for classification. We also conduct comprehensive experiments, and the results demonstrate that the proposed method outperforms the other classifiers.