基于干扰链路检测的多信道无线层析成像定位技术研究

Radio Tomographic Imaging (RTI) is a technology for imaging the attenuation caused by physical objects in wireless networks through radio transmissions. It has become one of the critical techniques for indoor localization issues due to its flexibility in handling different environments under smoke, fog and haze. RTI can also be applied in emergencies,rescuing operations,security breaches. However, so far most of the existing methods for RTI use single channel signal to detect received signal strength and deal with all measured data from all links in the wireless networks. This causes high computational complexity especially when the numbers of nodes and pixels are large.What is more important, the accuracy of localization cannot be adjusted adaptively based on fixed pixel size and regularization method for ill-posed inverse problem.Under complex environments with rich multipath, it is hard to correctly detect the shadowed link using RSS of single channel signal.False alarm and missing detection will lead to high error of localization. Our proposal focuses on two research directions: first, we will develop a measuring method based on multi-channel signal, and detect interference links in time domain and spatial domain detection according to clustered feature of real attenuated links. Second, we propose to use a novel localization algorithm, which will get coarse-grained position estimation with partial links and get a fine-grained position estimation by difference features among attenuated links.Our proposed techniques fundamentally change the view of the traditional RTI with measuring method and localization algorithm, reducing localization error and computational complexity so as to solve basic problems of RTI under multipath environments.

无线层析成像技术是通过测量无线网络节点间信号的阴影衰落来形成探测区域 内障碍物的截面图像，该技术能应用于烟、雾、霾等特殊环境,可用于紧急救援、安全监控的目标定位。现有无线层析成像定位系统的一般思路是采用单信道信号探测、利用所有链路建立测量方程，用正则化方法求解不适定反问题重建图像定位目标，复杂多径环境下，单一信道的接收信号强度变化难以准确反映目标是否遮挡链路，虚警和漏检导致大的定位误差，成像算法需要运用全部链路测量数据导致逆问题求解随节点数增加计算复杂度巨增，同时定位精度受限于像素大小。本课题将从两方面突破，一是探索多信道信号探测技术，采用时空域检测和聚类特性来识别干扰链路，二是提出新的定位算法，实现基于部分链路快速低复杂度目标位置近似估计和基于链路衰减差异的精确定位。本课题从测量机制和定位算法上改变传统无线层析成像方法，降低定位误差和计算复杂度，解决该技术在多径环境下目标定位的基本问题。

快速无损人体探测在安全监控、紧急救援、人类行为科学研究领域具有重要意义，红外和光学成像工作环境受限，穿墙雷达作用距离受限、成本高。无线层析成像(RTI)技术通过部署无线网络测量节点间信号的阴影衰落来形成探测区域内障碍物的截面图像，能应用于烟、雾、霾等特殊环境。本项目通过理论分析、仿真研究、系统研制、试验测试等工作，开展多信道信号探测技术研究，采用时空域检测和聚类特性来识别干扰链路降低多径干扰。取得的主要研究成果包括：（1）提出基于多信道探测的无线层析成像定位系统体系结构和接收信号强度检测算法；（2）分析了多径信号对RTI性能的影响并提出了基于时空域干扰链路检测的RTI增强方法；（3）提出了基于超宽带信号的RTI方法提高RTI在多径环境中的鲁棒性；（4）提出了基于绕射模型和粒子滤波的贝叶斯RTI方法提高目标跟踪精度；（5）提出了RTI中基于RSS测量的目标识别方法。构建了20个节点的无线层析成像实验系统，开展了室外及室内办公环境、穿墙场景下的目标定位跟踪实验，成像时间小于1s，跟踪精度室外0.1m，室内0.16m。发表论文19篇，申请发明专利7项，获授权3项。培养学生8名，持续开展国际合作及国内外学术交流，成果将在消防救援、室内导航以及智能医疗等领域进行推广应用。