一种针对设备异质性的室内定位方法  

Indoor Localization Method for Device Heterogeneity

作　　者：曾衍华 单志龙[1,2] 陈之彧 ZENG Yan-hua;SHAN Zhi-long;CHEN Zhi-yu(School of Computer Science,South China Normal University,Guangzhou 510631,China;School of Distance Education,South China Normal University,Guangzhou 510631,China)

机构地区：[1]华南师范大学计算机学院,广州510631 [2]华南师范大学网络教育学院,广州510631

出　　处：《小型微型计算机系统》2023年第11期2464-2470,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(62192711)资助;广州市科技计划项目(201904010195)资助。

摘　　要：针对室内环境下Wi-Fi信号波动及在线定位阶段移动设备异质性而导致指纹定位算法精度不高的问题,本文考虑了用户移动轨迹上一系列接收信号强度指示(Received Signal Strength Indicator,RSSI)指纹之间的相关性,并提出了一种基于深度神经网络的室内定位算法.该算法在离线阶段,针对部分手机难以采集大量RSSI指纹的问题,通过数据增强扩充指纹库,减少指纹采集工作量;然后使用轨迹生成算法生成大量轨迹数据,训练基于卷积神经网络和循环神经网络的定位模型.在线阶段将RSSI指纹转换为差分矩阵以缓解设备异质性问题,并结合指纹的隐含特征以及模型上一时刻预测的位置进行定位.在不同移动终端设备上的实验结果表明,该算法可以有效缓解设备异质性的影响,提高定位精度.Aiming at the problem of unsatisfactory accuracy of fingerprint localization algorithm due to Wi-Fi signal fluctuations in indoor environments and heterogeneous mobile devices in the online localization stage,this paper considers the correlation between a series of RSSI fingerprints on the user's movement trajectory,and a deep neural network based indoor localization algorithm is proposed.In the offline phase,the algorithm enlarges the fingerprint database by data augmentation to reduce the fingerprint collection workload,and then numerous trajectories which are generated by trajectory generation algorithm is used to train the localization model based on convolutional neural network and recurrent neural network.In the online phase,the RSSI fingerprint is converted into a difference matrix to alleviate the problem of device heterogeneity,and the latent features of the fingerprints is combined with the location predicted by the model at the last moment to get current position.Experimental results on different mobile terminal devices show that the proposed algorithm can effectively mitigate the effects of device heterogeneity and improve localization accuracy.

关 键 词：室内定位 设备异质性 接收信号强度指示 卷积神经网络 循环神经网络 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]