Plils：一种通过子区域聚类利用低成本无线芯片实现的实用室内定位系统。

Plils: A Practical Indoor Localization System through Less Expensive Wireless Chips via Subregion Clustering.

作者信息

Li Xiaolong, Yang Yifu, Cai Jun, Deng Yun, Yang Junfeng, Zhou Xinmin, Tan Lina

机构信息

Key Laboratory of Hunan Province for New Retail Virtual Reality Technology, Hunan University of Commerce, Changsha 410205, China.

Mobile E-business Collaborative Innovation Center of Hunan Province, Hunan University of Commerce, Changsha 410205, China.

出版信息

Sensors (Basel). 2018 Jan 12;18(1):205. doi: 10.3390/s18010205.

DOI:10.3390/s18010205

PMID:29329226

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC5795826/

Abstract

Reducing costs is a pragmatic method for promoting the widespread usage of indoor localization technology. Conventional indoor localization systems (ILSs) exploit relatively expensive wireless chips to measure received signal strength for positioning. Our work is based on a cheap and widely-used commercial off-the-shelf (COTS) wireless chip, i.e., the Nordic Semiconductor nRF24LE1, which has only several output power levels, and proposes a new power level based-ILS, called Plils. The localization procedure incorporates two phases: an offline training phase and an online localization phase. In the offline training phase, a self-organizing map (SOM) is utilized for dividing a target area into subregions, wherein their grids in the same subregion have similar fingerprints. In the online localization phase, the support vector machine (SVM) and back propagation (BP) neural network methods are adopted to identify which subregion a tagged object is located in, and calculate its exact location, respectively. The reasonable value for has been discussed as well. Our experiments show that Plils achieves 75 cm accuracy on average, and is robust to indoor obstacles.

摘要

降低成本是促进室内定位技术广泛应用的一种务实方法。传统的室内定位系统（ILS）利用相对昂贵的无线芯片来测量接收信号强度以进行定位。我们的工作基于一种廉价且广泛使用的商用现货（COTS）无线芯片，即 Nordic Semiconductor nRF24LE1，它只有几个输出功率级别，并提出了一种新的基于功率级别的 ILS，称为 Plils。定位过程包括两个阶段：离线训练阶段和在线定位阶段。在离线训练阶段，利用自组织映射（SOM）将目标区域划分为子区域，其中同一子区域内的网格具有相似的指纹。在在线定位阶段，分别采用支持向量机（SVM）和反向传播（BP）神经网络方法来识别标签对象位于哪个子区域，并计算其精确位置。还讨论了合理的值。我们的实验表明，Plils 平均精度达到 75 厘米，并且对室内障碍物具有鲁棒性。