Analyzing resident life cycle based on short-time cross-correlation

The main purpose of smart environment (SE) is to assist human daily lives by analyzing and recognizing related information. Usually, equipments such as video cameras and voice recorders are useful to capture the information. However, the data often contain unnecessary information related to privacy. To solve this problem, it is important to construct SE by using simple sensors like motion sensor, pressure sensor, and so on. To validate the potentiality of simple sensors for analyzing and recognizing activities, in this paper, we focus on infrared sensors. We try to inspect the resident's activity regularity based on sensor data obtained in several ordinary Japanese houses. To obtain more concrete results, we introduce the concept short-time cross-correlation (STCC). Similar to short-time Fourier transform, STCC calculates the cross-correlation in a sliding window. From the results we can see “when” and “how” a resident's activity is regular or irregular. Thus, STCC can be a useful tool for analyzing the resident life styles and for constructing privacy preserving SE in general.