轻量扩展的射频指纹地图构造方法  

A Scalable Lightweight Radio Fingerprint Map Construction Method

作　　者：刘文远[1,2] 刘慧香 温丽云 王林 

机构地区：[1]燕山大学信息科学与工程学院,秦皇岛066004 [2]河北省计算机虚拟技术与系统集成重点实验室,秦皇岛066004 [3]河北省质量技术监督局信息中心,石家庄050091

出　　处：《电子与信息学报》2018年第2期306-313,共8页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(61672448;61772453);河北省留学归国人员择优资助项目(CL201625)~~

摘　　要：随着群智感知和机器学习的融合,基于射频指纹的室内定位技术引起研究者的广泛关注。然而现有工作存在指纹地图构建阶段开销过大形成的可扩展性和实时性瓶颈问题。针对这一问题,该文提出一个新颖的轻量可扩展指纹地图构造方法(FFIL)。在指纹构建阶段,将整个室内环境划分为多个环路快速分割地图并获取射频指纹;在指纹匹配阶段,首先计算AP与目标点间的距离,然后选择与圆环半径最相似的环路上的参考点一一匹配;在定位阶段,采用等高线聚类算法来提高定位精度。通过真实数据驱动的大量仿真和实验证明,FFIL能减小指纹地图构建的开销,同时提高定位精度和系统实时性。Fingerprint-based indoor localization technology is attracted extensive attention of researchers with the fusion of crowd-sensing and machine learning. However, existing approaches have the bottleneck of scalability and instantaneity caused by high radio map construction effort. Focusing on this issue, this paper proposes a novel and scalable lightweight radio map construction method, named FFIL. In the fingerprint construction phase, the whole indoor environment is divided into multi-loop to segment map rapidly and fingerprint data are obtained. In the fingerprint matching phase, the distance is calculated from Access Point （AP） to target firstly, and then the reference point is selected on the loop with most similar with the circle radius to match fingerprint data one by one. In the localization phase, contour-based clustering algorithm is used to improve the positioning accuracy. Abundant simulations and experiments are driven by real data show that FFIL can reduce the overhead of constructing radio fingerprint map and improve the positioning accuracy and the real-time performance of system simultaneously.

关 键 词：室内定位 指纹 WIFI 等高线 定位精度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]