An improved inertial/wifi/magnetic fusion structure for indoor navigation
An improved inertial/wifi/magnetic fusion structure for indoor navigation
复制标题
DOI：
10.1016/j.inffus.2016.06.004
复制
发表时间：
2017-03-01
影响因子：
18.6
通讯作者：
El-Sheimy, Naser
中科院分区：
文献类型：
--
作者：
Li, You;Zhuang, Yuan;El-Sheimy, Naser
本文提出了一种航位推算（DR）/WiFi指纹识别/磁匹配（MM）集成结构，该结构使用消费者便携式设备和现有WiFi基础设施中的现成传感器。该结构相对于先前的DR/WiFi/MM融合结构的一个关键改进是引入了基于不同技术之间的交互的三级质量控制（QC）机制。在QC级别#1上，应用了多个标准来过滤每个单独技术中的错误或不可靠的测量结果。然后，在第2级，使用基于阈值的方法通过调查EKF新息序列来自动设置WiFi结果的权重。最后，在级别#3上，DR/WiFi结果被用于限制MM搜索空间，并进而降低失配率和计算负载。所提出的结构减少了均方根（RMS）的位置误差在13.3至55.2%的范围内，在步行实验与两个智能手机，在四个运动条件下，在两个室内环境。此外，所提出的结构降低了失配率（即，与先前的DR/WiFi/MM集成结构相比，匹配到地理上位于距离真实位置超过15米的不正确点）的比率超过75.0%。© 2016 Elsevier B. V.版权所有。
This paper proposes a dead-reckoning (DR)/WiFi fingerprinting/magnetic matching (MM) integration structure that uses off-the-shelf sensors in consumer portable devices and existing WiFi infrastructures. One key improvement of this structure over previous DR/WiFi/MM fusion structures is the introduction of a three-level quality-control (QC) mechanism based on the interaction between different techniques. On QC Level #1, several criteria are applied to filter out blunders or unreliable measurements in each separate technology. Then, on Level #2, a threshold-based approach is used to set the weight of WiFi results automatically through the investigation of the EKF innovation sequence. Finally, on Level #3, DR/WiFi results are utilized to limit the MM search space and in turn reduce both mismatch rate and computational load. The proposed structure reduced the root mean square (RMS) of position errors in the range of 13.3 to 55.2% in walking experiments with two smartphones, under four motion conditions, and in two indoor environments. Furthermore, the proposed structure reduced the rate of mismatches (i.e., matching to an incorrect point that is geographically located over 15 m away from the true position) rate by over 75.0% when compared with previous DR/WiFi/MM integration structures. 2016 Elsevier B.V. All rights reserved.