针对设备差异性问题的增量式室内定位方法  

Incremental Indoor Localization for Device Diversity Issues

作　　者：夏俊 刘军发 蒋鑫龙[1,2,3] 陈益强 XIA Jun;LIU Jun-fa;JIANG Xin-long;CHEN Yi-qiang(University of Chinese Academy of Sciences,Beijing 100049,China;Research Center for Ubiquitous Computing Systems,Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190,China;Beijing Key Laboratory of Mobile Computing and Pervasive Device,Beijing 100190,China)

机构地区：[1]中国科学院大学,北京100049 [2]中国科学院计算技术研究所泛在计算系统研究中心,北京100190 [3]北京市移动计算与新型终端重点实验室,北京100190

出　　处：《计算机科学》2018年第10期69-77,共9页Computer Science

基　　金：国家自然科学基金面上项目:面向可穿戴用户行为识别的增量学习方法研究(61572471);国家自然科学基金面上项目:融合多元传播模型和指纹模型的免标定室内定位方法研究(61472399);国家自然科学基金面上项目:递归深度学习网络的多极限环神经动力学模型及方法研究(61572004);广东省科技计划项目:面向健康监护的新型智能贴件关键技术研发及产业化(2015B010105001)资助

摘　　要：随着WLAN的普及,基于RSS(Received Signal Strength)的室内定位方法逐渐成为研究与应用的热点。其中,基于指纹的定位方法已成为主流,此类方法的特点之一在于要求离线训练数据与在线测试数据满足独立同分布,然而,在实际环境中,现有的指纹定位方法或系统存在以下3个问题:1)不同终端设备的无线通讯硬件存在差异性,训练数据和测试数据的采集设备之间的差异性将严重影响定位精度;2)环境中的无线信号呈现高动态性,采集的数据存在显著的时效性,因此由训练数据得到的模型的定位性能将随着时间的推移不断下降;3)传统增量式定位模型需要大量的标定数据,不具有实际可用性。为解决以上问题,提出了一种针对设备差异性问题的增量式室内定位方法,利用终端在持续定位服务中采集的无标记数据来实时更新定位模型。实验表明,在实际蓝牙定位数据集上,相比于传统的定位模型方法,所提方法的整体定位精度更高,误差距离为3～5m时,其优势更为明显;同时,该方法具有时效优势,能够长时间保持有效定位。With the rapid development of Wireless Local Area Network(WLAN),the Received Signal Strength(RSS)based indoor localization becomes a hot area in research and application fields.Among various kinds of up-to-date indoor localization methods,fingerprint based methods are most widely used because of its good performance,and one feature of those methods is that the accuracy is determined by the identification of the training and the testing dataset.Howe-ver,in practical applications,there are three problems in existing fingerprint based methods or system.Firstly,the loca-lization error caused by device variance is a severe problem.Secondly,the wireless data are changing as the time passed,leading to the reduction of the prediction accuracy.Thirdly,traditional fingerprint based methods or system cannot avoid the dependency on a large amount of labeled data to keep effective positioning performance,which usually involves high cost in labor and time.To solve these problems,this paper proposed a incremental indoor localization method for device diversity issues,which keeps real-time update by training uncalibrated data that collected in localization.Experimental results show that the proposed method can increase the precision of overall indoor localization system,especially when error distance is between 3 and 5 meters.What’s more,this method possesses good advantage of timeliness compared with traditional indoor localization method on real BLE dataset.

关 键 词：室内定位 设备差异性 增量学习 极速学习机 物联网 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]