基于智能手机的室内标签定位技术研究

发布时间：2018-04-25 12:20

本文选题：室内标签定位 + 楼层定位　；参考：《南京大学》2016年博士论文

【摘要】：随着Internet的快速发展与移动智能设备的广泛普及,基于位置的应用及服务已经与人们的日常生活息息相关,GPS无法满足的室内位置需求出现了爆发式增长,迫切需要新的更有效的室内标签定位技术。目前室内标签定位场景下的应用,对定位的需求已经不仅仅满足于精度,对可靠性、可用性及成本等方面也有较高的要求。但室内场景的普遍复杂化、定位设施的广泛缺乏及定位需求的多样化,导致传统的室内定位技术难以满足要求。基于传统的室内标签定位技术,本文的研究思路是充分利用定位场景下丰富的智能手机感知和计算资源,不依赖额外的基础设施和技术维护人员,使定位系统在满足标签定位精度的同时提高可靠性、可用性和降低成本。本文提出了一个室内标签定位技术框架,然后对室内楼层定位问题和地铁定位问题进行了研究,还提出了一套通用的智能手机传感器校准方案。最后在框架基础上设计与实现了一个室内标签定位技术平台并开发了“位动校园”应用示例。具体而言,本文取得了以下创新性成果：·系统性的提出了一个基于智能手机、基础设施低依赖的面向室内标签定位的技术框架(SIFIS)。针对传统室内标签定位技术对基础设施依赖性较高的不足,提出了一个融合行为识别、群智感知计算和模式识别技术的新型室内标签定位技术框架。该框架的基础设施依赖性较低,使得该技术在满足精度的要求下有更好的可靠性、可用性和较低的成本。·基于SIFIS技术框架,针对室内楼层定位需求,仅利用智能手机的加速度、气压、地磁等感知能力,提出了基于机器学习的上下楼行为识别算法、基于轨迹匹配的相遇算法、基于动态时间规整(DTW)的标签定位等算法,实现了无基础设施支撑条件下的室内楼层定位。模拟实验和现场实验均表明,该方案不仅能满足定位精度,而且具有更好的可靠性、可用性和更低的成本。·基于SIFIS技术框架,针对地铁位置定位需求,仅利用智能手机的加速度、气压、地磁等感知能力,提出了基于状态机的地铁启停判定和开关门判定算法、基于动态时间规整(DTW)的标签定位等算法,实现了无基础设施支撑条件下的地铁位置定位。现场实验表明,该方案不仅能满足定位精度,而且具有更好的可靠性、可用性和更低的成本。·针对智能手机传感器误差的自然存在以及必定影响定位准确性的特征,提出了一个相对通用的传感器校准方案。该方案利用群智感知技术,提出了基于相遇识别的两两校准和基于最矮生成树的全局校准方案,校准过程透明于用户,为室内标签定位奠定了数据准确性的基础。气压传感器校准实验表明,该方案具有良好的实用性和通用性。·设计并初步实现了室内标签定位技术平台,并基于此平台实现了一个基于位置的社交应用示例。该应用基于定位平台,通过用户的位置描述用户的特征,依据特征进行交友推荐。体现了室内标签定位技术的可行性和广泛的应用前景。
[Abstract]:With the rapid development of Internet and the widespread popularity of mobile intelligent equipment, location based applications and services have been closely related to people's daily life. GPS can not meet the explosive growth of the indoor location demand, and a new and more effective indoor label positioning technology is urgently needed. The demand for positioning has not only been satisfied with accuracy, but also has high requirements on reliability, availability and cost. However, the general complexity of the indoor scene, the wide lack of location facilities and the diversification of the location requirements lead to the difficulty of meeting the requirements of the traditional indoor positioning technology. The research idea is to make full use of the rich intelligent mobile phone perception and computing resources under the positioning scene, and do not rely on additional infrastructure and technical maintenance personnel, so that the positioning system can improve the reliability, availability and reduce the cost while satisfying the positioning accuracy of the label. The problem of floor location and subway positioning is studied, and a set of universal smart phone sensor calibration scheme is put forward. At last, an indoor label positioning technology platform is designed and implemented on the basis of the framework and the application example of "mobile campus" is developed. In particular, the following innovative results are obtained:. A technical framework based on smart phone and low dependency on infrastructure (SIFIS) is proposed. In view of the high dependence of traditional indoor label positioning technology on infrastructure, a new indoor tag positioning technology framework is proposed for fusion behavior recognition, swarm intelligence perception calculation and mode recognition technology. The dependency of the framework is low, which makes the technology have better reliability, availability and lower cost under the requirement of precision. Based on the SIFIS technical framework, the performance of the indoor floor positioning needs only by using the intelligent mobile phone acceleration, pressure, and geomagnetism ability, and puts forward the behavior of up and down building based on machine learning. Recognition algorithm, track matching based meeting algorithm, based on dynamic time warping (DTW) label location algorithm, implemented the indoor floor positioning without infrastructure support. Both simulation experiment and field experiment show that the scheme not only satisfies the positioning accuracy, but also has better reliability, availability and lower cost. Based on the SIFIS technology framework, based on the acceleration, air pressure and geomagnetism of the smart phone, the subway start stop decision and the switch gate determination algorithm based on the state machine are proposed. Based on the dynamic time regularization (DTW), the subway location is determined without infrastructure support. Field experiments show that the scheme not only satisfies the positioning accuracy, but also has better reliability, availability and lower cost. In view of the natural existence of the sensor error and the characteristics that must affect the accuracy of the location, a relative universal sensor calibration scheme is proposed. The scheme uses group intelligence sensing technology. The 22 calibration based on meeting recognition and the global calibration scheme based on the shortest generation tree are proposed. The calibration process is transparent to the user, which lays the foundation for the accuracy of the indoor label location. The pressure sensor calibration experiment shows that the scheme has good practicability and generality. Based on this platform, an example of a location based social application is implemented. The application is based on the location platform, describes the user's features through the user's position, and recommends friends according to the features. It embodies the feasibility and wide application prospect of the indoor label positioning technology.

【学位授予单位】：南京大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN929.5

【相似文献】