中文摘要：

随着对WiFi信号研究的不断深入,人们发现信道状态信息的收集是利用WiFi探测人体动作的工作基础。文章首先根据国内外相关研究总结得出WiFi探测动作系统的基本结构,尽管这些研究论文中的除噪、特征提取和分类方法不一样,但共同点是都需要获取信道状态信息。文章介绍了信道状态信息的定义和表达方式,然后将信道状态信息与传统的接收信号强度指示进行对比,发现信道状态信息更有利于细粒度的动作识别,进一步阐述采用普通的商业WiFi收发设备来获取信道状态信息的工作机制,分析信道状态信息数据包中包含的各种参数的意义,以及如何在时域中使用信道状态信息。然后使用三个应用实例介绍人体动作对信道状态信息影响的方式,通过数学建模来有效利用信道状态信息并提高识别人体动作的精确度。根据实验中所遇到的异常问题,分析异常成因,给出一种识别并消除异常的算法。最后总结出目前研究成果中抗干扰能力差、对实验环境的要求太过苛刻等缺点,对未来研究方向提出新的工作思路。

英文摘要：

With the deepening of WiFi signal research, it is found that the collection of channel state information is the basis of using WiFi to detect human action. In this paper, the definition and expression of the channel state information are given. Then, the channel state information is compared with the traditional signal strength indicator. It is found that the channel state information is more suitable for fine-grained action recognition. The paper describes working mechanism of the channel state information to be obtainedwith the common commercial WiFi transceiver and analyzes the meaning of the various parameters contained in the channel state information packet.The paper also states the method of using the channel state information in the time domain. Through mathematical modeling, the channel state information is effectively utilized and the accuracy of recognition of human motion is improved. According to the outlier problem in the experiment, the paper analyzes the cause of the outlier, gives an algorithm to identify and eliminate the outlier.