移动计算平台的室内定位系统研究及优化加速

发布时间：2018-06-15 06:17

本文选题：室内定位系统 + 粒子滤波　；参考：《浙江大学》2016年博士论文

【摘要】：近年来,基于位置的服务在人们生活中的地位越发重要,由于全球定位系统无法提供准确的室内位置信息,室内定位系统的研究成了关注的热点。本文的研究以室内定位系统为课题,在充分调研研究现状后针对现有室内定位系统的不足之处进行研究探索。本研究以拥有高普及度的智能移动设备为平台,设计了一个多模式、低成本、高精度的混合室内定位系统,其中最大的创新点在于多模式,即在室内定位系统的使用过程中智能设备的握持模式在系统支持范围内能够随意进行改变,提升了室内定位系统的实用性。本文设计的室内定位系统还具有低计算复杂度、低部署成本的优势,适于应用在计算与能量资源有限的移动平台上并进行推广。在路径长度超过一千米的实验中,对实验者最后位置的定位误差仅为1.36m。针对室内定位系统中常用的粒子滤波算法,本文还针对其高硬件资源开销、高计算复杂度的特性,提出了新的优化加速算法,能够大幅降低该算法在应用过程中的存储资源开销和计算处理时间,通过理论分析、仿真和基于可现场编程门阵列的原型验证展示了新设计算法的高硬件资源利用率和实时处理能力。在算法的实例分析中,本文设计的低存储开销算法能够在不影响系统滤波精度的前提下将存储器开销降低49.69%,高实时性的算法能够在分布式粒子滤波系统线性加速比的基础上将实时处理性能进一步提高38.62%。
[Abstract]:In recent years, location-based services have become more and more important in people's lives. Because GPS can not provide accurate indoor location information, the research of indoor positioning system has become a hot topic. In this paper, the indoor positioning system as a subject, after a full study of the status quo of the existing indoor positioning system deficiencies are explored. On the platform of intelligent mobile devices with high popularity, this study designed a multi-mode, low-cost, high-precision hybrid indoor positioning system, in which the most innovative point is multi-mode. That is, in the use of indoor positioning system, the holding mode of intelligent equipment can be changed at will within the scope of system support, which improves the practicability of indoor positioning system. The indoor positioning system designed in this paper also has the advantages of low computational complexity and low deployment cost, and is suitable for application on mobile platforms with limited computing and energy resources. In the experiment where the path length is more than one kilometer, the location error of the final position of the experimenter is only 1.36 m. Aiming at the particle filter algorithm commonly used in indoor positioning system, this paper also proposes a new optimization acceleration algorithm for its high hardware resource overhead and high computational complexity. The algorithm can greatly reduce the cost of storage resources and computing processing time in the process of application, through theoretical analysis, Simulation and prototype verification based on field programmable gate array demonstrate the high hardware resource utilization and real-time processing capability of the new algorithm. In the example analysis of the algorithm, The low memory overhead algorithm designed in this paper can reduce the memory cost by 49.69 without affecting the filtering accuracy of the system. The high real-time algorithm can be processed in real time on the basis of linear speedup of distributed particle filter system. The performance is further improved by 38.62.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN92;TN713

【相似文献】