基于传感器数据融合的WiFi室内定位算法设计及应用研究

发布时间：2018-05-05 20:22

本文选题：WiFi + 传感器　；参考：《浙江大学》2016年硕士论文

【摘要】：WiFi基础设施在室内场所的普及和移动智能设备的迅猛发展给室内定位应用带来了新的活力。基于WiFi的室内定位方法由于受到室内环境的干扰以及移动智能设备的异构性等因素影响,定位精度误差较大。针对上述问题,本文利用室内WiFi信号和移动智能设备内置多传感器采集的数据思考设计提高室内定位精度的方法。首先,针对移动设备异构性及环境动态变化(室内环境布局变化及人员走动等)将导致WiFi信号强度不稳定的问题,利用不同WiFi接入点的WiFi信号强度之间的稳定大小关系设计了一种更加稳定的室内定位算法,该算法利用WiFi信号强度的潜在相似规律设计一种新的定位指纹特征向量,具有稳定和无需校验的优点；其次,分析了目前WiFi室内定位方法在多楼层环境下的问题与不足,提出一个基于楼层判别的多楼层室内定位方法,通过Fisher线性判别算法建立一个楼层判别模型,并利用加速度传感信息识别行人走路、上下电梯和上下楼梯的状态,预测楼层变化事件,并在六层大楼中实地定位测试,定位准确性平均提高了约6%；再次,又针对复杂的多楼层室内环境及现有指纹定位算法能耗较高的现状,提出一种混合多楼层室内定位算法,实际场景中试验结果表明,所设计的定位算法,不但能提高定位精度,而且只需更少的定位时间从而节省能耗；最后,在上述算法研究基础上,开发了一个基于实际场景的多楼层室内导航原型系统。
[Abstract]:The popularization of WiFi infrastructure and the rapid development of mobile intelligent devices bring new vitality to indoor positioning applications. Because of the interference of indoor environment and the heterogeneity of mobile intelligent equipment, the positioning accuracy of indoor positioning method based on WiFi is large. In view of the above problems, this paper designs a method to improve the indoor positioning accuracy by using the data collected by indoor WiFi signal and mobile intelligent device with multiple sensors. First of all, aiming at the heterogeneity of mobile devices and the dynamic change of environment (such as indoor environment layout change and people moving around etc.), it will lead to the instability of WiFi signal intensity. A more stable indoor localization algorithm is designed based on the stable relationship between the WiFi signal intensities of different WiFi access points, and a new fingerprint feature vector is designed based on the potential similarity of the WiFi signal intensity. It has the advantages of stability and no checking. Secondly, the problems and shortcomings of the current WiFi indoor positioning method in multi-floor environment are analyzed, and a multi-floor indoor location method based on floor discrimination is proposed. By using Fisher linear discriminant algorithm, a floor discriminant model is established, and the acceleration sensing information is used to identify the state of pedestrian walking, elevator and staircase, to predict the event of floor change, and to locate and test in a six-story building. Thirdly, aiming at the complex indoor environment and the high energy consumption of the existing fingerprint localization algorithm, a hybrid multi-story indoor location algorithm is proposed, and the experimental results show that, The designed localization algorithm can not only improve the positioning accuracy, but also save energy consumption with less positioning time. Finally, based on the above algorithm, a prototype system of multi-floor indoor navigation based on the actual scene is developed.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN92

【相似文献】