基于全向信息特征的设备无关的人员被动定位问题研究

The passive and device-free detection and localization of human based on the wireless signal act as an important and foundational role in the elderly health care and smart home applications and research fields. Approaches based on the WiFi signal characteristics catch great attention because of relatively easier implementation and lower cost. However, existing solutions mainly adopt Received Signal Strength Indicator (RSSI) for human detection and location, which suffers from dramatic performance degradation such as low accuracy and limited detection direction due to multipath fading and temporal dynamics. In consideration of multipath effect and channel response of the PHY layer, this project is based on the WiFi signal characteristics to launch a systematic and deep study for omnidirectional passive human detection in terms of fundamental theories, technology and methods, theoretical analysis and experimental verification. This study aims at launching the following topics: (1) the identification and quantification of signal's sensitivity based on the presence of human body;(2) the collection of reliable signal characteristic based on multidimensional channel response characteristics; (3) the balance decision of multidimensional signal fingerprint and passive localization for multi-user.The objective is to provide high precision accuracy, high reliability of omnidirectional passive human detection and localization theory model and the method. This study brings forth new approaches for omnidirectional passive human detection, offers the theoretical and technological support for multiple indoor localization applications of wireless channel detection technology.

在老人健康关照及智慧家庭等应用与研究领域中，基于无线信号的设备无关的人体被动检测与定位具有重要的基础性作用。基于WiFi信号特征的方法因其相对容易实施及成本较低受到了高度重视。现有解决方案主要依据MAC层的RSSI，存在精准度低、检测方向受限等问题。本项目基于WiFi的全向信息特征，综合考虑PHY层的多径功率特性与信道响应特性，从理论模型、技术方法、理论分析和实验验证等四个方面开展设备无关的室内人体被动全向检测与定位研究。主要研究内容包括：⑴ 信号对人体存在敏感度的可识别和量化；⑵ 基于多维信道响应特征的可靠信号特征采集；⑶ 面向多用户的多维信号指纹平衡决策和被动定位。本项目以提供高精准度、高可靠的人的全向被动检测与定位理论模型及方法为目标，为全向被动检测问题提供新的思路与方法，为无线信道检测技术的应用提供理论基础与技术支撑，具有显著的理论意义和应用价值。

本项目针对在老人健康关照及智慧家庭等一系列应用领域中，基于无线信号的设备无关的人体被动全向检测与定位存在数据处理难、定位不精确等现状，研究了基于信道频率响应的人体存在敏感度的可识别性与量化方法，解决了室内环境中非同步性人体存在全向检测问题；研究了面向复杂信道环境下基于深度学习理论的多维信道可靠信号特征采集方法，提出并实现了高可靠性、高精确度、低成本的信号特征采集解决方案；研究了基于生成对抗网络理论和多层次分类框架的多用户高维信号指纹库建立和指纹分类方法，提出了低时延、高精度的被动定位解决方案；基于以上的研究成果，开展融合WiFi、ZigBee及蓝牙等多种通信设备的老人健康关照原型系统和实验平台的示范应用研究，实现对居家老人的全面安全监测和照护。本项目的研究成果能够为物联网产业相关应用提供基础理论和技术支撑，具有显著的理论意义和应用价值。