面向非视距与密集多径的复杂场景下无线定位机制与方法研究

The application of wireless localization technology is restricted greatly due to NLOS (non-line-of-sight) and dense multipath effect in complex environment. Therefore, there is of signifiance to inestigate the wireless localization mechanism and method in such complex scenario. The project focuses on three issues: (1) By studying a twofold characteristic on multipath amplitude and delay spread, to develop a bi-variable model based on amplitude and delay. Throught designing the measuring for the cluster sparsity of multipath signal, to propose a compressed sensing based DP (Direct Path) detection and estimation method. (2) By analyzing the statistical property of the localization measurements both in LOS (line-of-sight) and NLOS, to develop a mathematical expression of the NLOS status identification probability. Consider the time interval uncertainty of NLOS status, to propose a status identification strategy by employing the alterable interval analysis. (3) Aiming to the localization problem in severe NLOS scenario wherein the DP is undetectable, by designing removal algorithm of the multi-bounce signal and the optimization of the target possible location, to develop a novel localization mechanism which is based on the non-direct path. This project is of great value to provide theoretical guidance and practical support for the localization application in complex environment, such as accurate positioning and navigation, public safety and emergency rescue.

复杂场景下的无线定位技术受到非视距(Non-line-of-sight, NLOS)与密集多径现象的严重制约, 因此研究这种复杂场景下的无线定位机制与方法具有重要意义。本项目研究多径信号幅度和时延的关联特性，建立基于幅度-时延的双变量模型，通过研究多径簇稀疏性的度量方法，提出基于压缩感知理论的直射路径(Direct Path, DP)检测与估计方法；分析测距定位时LOS和NLOS测量数据的统计特性，构建NLOS状态鉴别概率的数学表述方法，针对NLOS状态的时域区间不确定性，提出一种基于可变区间分析方法的NLOS状态鉴别策略；针对严重NLOS场景下无法检测DP的问题，通过探寻障碍物阻挡效应的规避方法，设计多次反射路径滤除算法和目标可能位置优化判定算法，实现基于非直射路径的定位机制与方法。本课题研究成果能为复杂场景下精确定位导航、公共安全和应急救援等领域应用提供理论指导和实践支持。

非视距（NLOS）和密集多径的复杂环境是制约无线定位技术精确性和可靠性的难点问题。本项目针对定位信号被遮挡造成的非视距场景的定位算法及其实际应用的关键技术展开深入研究，提出新的研究思路和方法。项目主要完成了如下内容：(1) 严重遮挡非视距和部分方位信息缺失时的三维定位框架与方法；(2) 面向GPS信号中断等复杂非视距城区环境下的车辆状态估计算法研究；(3) 集成学习和回归框架下的融合定位与位置预测算法研究。项目研发过程中组织了多次城市道路场景中各种不同路面的测试实验，对移动车辆位置和运动状态估计与跟踪算法的性能进行了细致评估，与产业界的实际应用开发也取得了实质性进展，通过与深圳市麦谷科技有限公司进行合作，共同搭建车联网平台，本项目相关成果在私家车定位与轨迹采集过程中发挥了重要作用，目前已经获得大量的私家车轨迹原始数据，并进行了初步分析，为进一步研究复杂城市环境下的车辆位置准确获取、探讨基于位置服务的领域应用提供科学的理论基础和技术支持。