室内多目标的被动定位方法研究

With the rapid development of the wireless network, more and more location based services appear in the indoor environment, which greatly promote the research of wireless indoor location technologies. Device-free localization (DFL) technologies are useful in applications where people being tracked cannot be expected to participate actively in the localization process. This may be the case because they are intentionally evading the system, or because they are physically unable, or because they do not want to be inconvenienced by wearing a device. In this project multiple device-free targets localization method based on radio tomographic imaging (RTI) is explored using Received Signal Strength (RSS) in wireless sensor network. Since the relation between RSS and distance is very complex in indoor environment, due to multipath effects and other phenomena. Firstly, the communication channel characteristics are studied to design a time division communication scheme for single wireless link. Then, we establish a hybrid RTI model to formulate the relationship between RSS dynamics and shadow attenuation on pixels, which combines the fading and enhancing of link RSS caused by the appearance of targets. Considering the influence of different target postures on link RSS, we build an equivalent measurement model to modify the weight matrix and gain a more accurate RTI model. Taking advantage of the sparse characteristics of the weight matrix and shadow attenuation on pixels, the Orthogonal Complementary Matching Pursuit (OCMP) algorithm is adopted to solve the RTI ill-posed problem, reconstruction efficiency. Finally, after constructing the image, we utilize the biggest confidence algorithm to estimate the number of targets and determine their location by clustering. This project provides a solid theoretical foundation for accurate and real-time indoor multiple device-free objects localization. And the result can be applied to emergency rescue, intrusion detection, smart homes automation and low cost surveillance.

室内免携带设备的多目标定位不需要目标配合即能获取位置信息，具有不可估量的应用价值和社会意义。室内环境下的通信环境具有多径效应等复杂因素，使得免携带设备的定位技术面临挑战。本项目以无线传感节点获取的接收信号强度为研究对象，分析室内场景下信道统计特性，建立室内免携带设备的多目标定位理论。首先设计无线传感节点的单链路分时通信方案，分析信道环境并实现节点的有效布局；探究目标出现引起的阴影衰落和多径变化，构造RSS混合模型体现其衰落和增强；继而结合人体等效测量模型，改进权重矩阵，建立带有目标姿态因子的无线层析成像模型；然后研究阴影衰落像素的可压缩稀疏表示方法，利用权重矩阵的稀疏性，探索高效的正交补空间匹配追踪重构算法；最后增加虚拟像素点，提高RTI成像分辨率，先利用最大置信度算法估计多目标的个数,再进行聚类实现多目标的精确定位。本项目的研究可广泛应用于紧急救援，智能家庭和安防等领域。

室内免携带设备定位是被动定位的关键技术之一，利用目标引起的无线射频信号强度的变化进行定位，成本低覆盖率高，应用范围广。目前免携带设备定位技术的研究还处于实验室阶段，复杂多径环境的建模，精确的多目标定位等诸多难题还亟待研究突破。本项目以无线传感节点获取的接收信号强度为研究对象，考虑目标的人体姿态特征及室内多径环境，将其应用于室内无线信号强度的衰落建模，并建立室内免携带设备的多目标定位理论。首先搭建了免携带定位的数据采集硬件环境，设计无线传感节点的单链路分时通信方案，分析信道环境并实现节点的有效布局；进行无线层析成像法及信号衰减权重模型分析，探究无线层析成像方法实现免携带定位的原理；研究了基于压缩感知的多目标指纹定位方法，包括离线训练阶段和在线定位阶段；提出了基于菲涅尔理论的RTI定位算法，改变了传统的室内免携带设备定位中人体定位目标作为圆柱体模型的模式；在无线层析成像(RTI)的基础上提出了基于双重构的定位算法，并提出补空间稀疏度自适应匹配重构算法，将目标位置转化为稀疏信号重构问题完成定位；最后提出单目标射频地图到双目标射频地图的建模算法。被动定位是在不需要目标携带任何设备的情况下获取位置信息，因而本研究在紧急救援、安防、智能家居和医院病人检测等室内应用场合应用前景广泛。