基于PWNN和粒子滤波的室内定位跟踪系统设计与实现

发布时间：2018-03-19 01:38

本文选题：Wi-Fi　切入点：室内定位　出处：《电子科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：基于Wi-Fi(Wireless Fidelity)的位置指纹定位技术优势很多,比如简单方便而且精度也十分理想。尤其是硬件方面Wi-Fi设备已经得到广泛应用,因此成为本文研究的出发点。本文依托北京综合交通一体化出行服务关键技术研究与应用示范项目(子课题:重点区域行人导引信息服务技术研究与示范应用),研究交通枢纽中定位技术的设计和实现,其中主要对交通枢纽中的位置指纹定位技术进行设计和改进。首先,本文对当前的位置指纹定位技术进行了较为系统的研究,在模式匹配上对各种相似度(欧式距离、余弦相似度、杰卡德相似系数、皮尔逊相关系数等)进行了技术对比,选择了皮尔逊相关系数和欧式距离相结合的算法。利用Wi-Fi信号强度形成指纹相关性,设计了一种基于PWNN(Nearest Neighbors based on Pearson correlation coefficient and Distance-weighted)模式匹配的位置指纹定位算法,在局部区域测试中不仅提升了定位精度,而且有效解决定位效率的问题,与其他传统的定位算法例如WKNN(Weighted K-nearest Neighbor)相比较,整体定位误差降低了近30%。其次,为了解决运动状态下定位跟踪的问题,对比分析了不同滤波技术(高斯滤波、粒子滤波、卡尔曼滤波),选择对粒子滤波进行改进。通过融合终端传感器信息,利用改进的粒子滤波器PF(Particle Filter)对位置进行跟踪,并结合指纹模式匹配算法进行切换式定位。通过实际测试表明该算法可以排除定位逻辑错误,降低定位误差,提高定位算法的可行性。最后,基于PWNN和改进粒子滤波算法,本文研究并实现了PWNN-PF室内定位跟踪系统,尤其是详细阐述了客户端和服务端各模块的设计和实现。并通过大量的实验表明,该定位系统有效降低了定位误差,尤其在终端运动状态下可以达到跟踪定位的效果,进一步提高定位实用性。
[Abstract]:There are many advantages of location fingerprint location based on Wi-Fi(Wireless Fidelity, such as simplicity, convenience and high precision. Especially, Wi-Fi devices have been widely used in hardware. Therefore, it becomes the starting point of this study. This paper relies on the key technology research and application demonstration project of integrated travel service in Beijing (sub-topic: research and demonstration application of pedestrian guidance information service technology in key areas. This paper studies the design and implementation of positioning technology in transportation hub. Among them, the location fingerprint location technology in transportation hub is mainly designed and improved. Firstly, this paper makes a systematic study on the current location fingerprint location technology, and makes a variety of similarity (Euclidean distance) in pattern matching. CoSine similarity, Jackard similarity coefficient, Pearson correlation coefficient, etc.) are compared. The algorithm of combining Pearson correlation coefficient with European distance is selected. The fingerprint correlation is formed by using Wi-Fi signal intensity. A location fingerprint location algorithm based on PWNN(Nearest Neighbors based on Pearson correlation coefficient and Distance-weighted pattern matching is designed. It not only improves the location accuracy but also solves the problem of location efficiency effectively in local area testing. Compared with other traditional localization algorithms such as WKNN(Weighted K-nearest neighbor, the overall positioning error is reduced by nearly 30 percent. Secondly, in order to solve the problem of locating and tracking in moving state, different filtering techniques (Gao Si filter, particle filter) are compared and analyzed. Kalman filter, choose to improve the particle filter. Through the fusion of terminal sensor information, using the improved particle filter PF(Particle filter to track the position, Finally, based on PWNN and improved particle filter algorithm, the algorithm can eliminate the location logic errors, reduce the location error, and improve the feasibility of the localization algorithm, finally, based on the PWNN and the improved particle filter algorithm, the algorithm can eliminate the location logic errors, reduce the location error and improve the feasibility of the location algorithm. In this paper, the PWNN-PF indoor positioning and tracking system is studied and implemented, especially the design and implementation of the client and server modules are described in detail, and a large number of experiments show that the positioning system can effectively reduce the positioning error. Especially in the state of terminal motion, it can achieve the effect of tracking and positioning, and further improve the practicability of location.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN92

【参考文献】