Abstract: Human activity recognition is an important research topic in pervasive computing. With user activity recognized, intelligent assistive service and personalization service can be offered to the user. The development of miniaturized low-cost and low-power computing devices makes sensing-based activity recognition a hot topic, which has attracted growing amount of attention in recent years. This paper proposes a new approach to extract features pertaining user based on UWB sensor data. It adopts the na?ve Bayes model for classifying human activity. The experimental results verify the proposed approach.

Key words: activity recognition, Ultra-Wideband（UWB） sensor, na?ve Bayes

摘要： 行为识别是普适计算的一个重要研究内容，通过识别用户行为，可以为用户提供智能辅助和个性化服务。随着微型低功耗传感器的发展，基于传感器技术的行为识别已经成为一个研究热点。采用UWB（Ultra-
Wideband，超宽带）传感器获取用户数据，提出一种针对用户行为识别的特征提取方法，采用朴素贝叶斯分类算法实现对行为的分类，并通过实验验证了该方法的有效性。

关键词: 行为识别, 超宽带传感器, 朴素贝叶斯