CAEFi：基于卷积自编码器降维的信道状态信息指纹室内定位方法

王旭东; 刘帅; 吴楠

doi:10.11999/JEIT210663

CAEFi：基于卷积自编码器降维的信道状态信息指纹室内定位方法

doi: 10.11999/JEIT210663

大连海事大学信息科学技术学院大连 116000

基金项目: 国家自然基金(61371091)

详细信息

作者简介:
王旭东：男，1967年生，博士，教授，研究方向为MIMO无线通信、空间调制、光无线通信及定位

刘帅：男，1996年生，硕士生，研究方向为无线室内定位

吴楠：男，1979年生，博士，副教授，研究方向为现代移动无线通信系统、可见光通信系统(包括MIMO、OFDM、信道编码、协作通信、自组织网络)等

通讯作者:
吴楠　alfred.nan.wu@gmail.com

中图分类号: TN929.1
计量
- 文章访问数: 1050
- HTML全文浏览量: 612
- PDF下载量: 135
- 被引次数: 0
出版历程
- 收稿日期: 2021-07-02
- 修回日期: 2021-10-28
- 网络出版日期: 2021-11-10
- 刊出日期: 2022-08-17

CAEFI: Channel State Information Fingerprint Indoor Location Method Using Convolutional Autoencoder for Dimension Reduction

Information Science Technology College, Dalian Maritime University, Dalian 116000, China

Funds: The National Natural Science Foundation of China (61371091)

摘要

摘要: 针对提高Wi-Fi指纹室内定位技术性能，该文首先提出一种基于卷积神经网络(CNN)的信道状态信息(CSI)指纹室内定位方法。该方法在离线阶段联合CSI幅度差和相位差信息对CNN模型进行训练。在廊厅和实验室两种不同室内定位场景进行了定位实验，分别获得了25 cm和48 cm的平均定位误差；然后，在此基础上重点针对提高基于CNN的CSI室内定位时效性，引入卷积自编码器(CAE)实现CSI的降维处理，在保证原始定位方法精度的前提下，定位时间提高了40%，同时将内存消耗降低到原算法的1/15，实验结果验证了所提算法的有效性。
- 室内指纹定位 /
- 信道状态信息 /
- 卷积神经网络 /
- 卷积自编码器
Abstract: In order to improve the performance of Wi-Fi fingerprint indoor positioning technology, a method based on Convolutional Neural Networks (CNN) for Channel State Information (CSI) fingerprint indoor positioning is first proposed. This method combines the CSI amplitude difference and phase difference information to train the CNN model in the offline stage. Positioning experiments are carried out in two different indoor positioning scenarios in the gallery and the laboratory, and the average positioning errors of 25 cm and 48 cm are obtained respectively; Then, on this basis, the focus is on improving the timeliness of CNN-based CSI indoor positioning. The Convolutional AutoEncoder (CAE) is introduced to realize the dimensionality reduction processing of CSI. Under the premise of ensuring the accuracy of the original positioning method, the positioning time is increased by 40% and the memory consumption is reduced to 1/15 of the original algorithm. The experimental results verify the effectiveness of the proposed algorithm.
- Indoor fingerprint positioning /
- Channel State Information(CSI) /
- Convolutional Neural Network(CNN) /
- Convolutional AutoEncoder(CAE)

HTML全文

图 1 CNNFi系统结构

下载: 全尺寸图片幻灯片

图 2 CNNFi误差收敛情况

下载: 全尺寸图片幻灯片

图 3 卷积自编码网络结构

下载: 全尺寸图片幻灯片

图 4 CAEFi系统结构

下载: 全尺寸图片幻灯片

图 5 廊厅场景

下载: 全尺寸图片幻灯片

图 6 实验室场景

下载: 全尺寸图片幻灯片

图 7 廊厅误差累计分布图

下载: 全尺寸图片幻灯片

图 8 实验室误差累计分布图

下载: 全尺寸图片幻灯片

图 9 幅度与幅度差的平均定位误差

下载: 全尺寸图片幻灯片

图 10 相位差的差值对比图

下载: 全尺寸图片幻灯片

图 11 CSI降维平均定位误差

下载: 全尺寸图片幻灯片

图 12 不同维度的平均定位误差

下载: 全尺寸图片幻灯片

图 13 不同测试数据个数的平均定位误差

下载: 全尺寸图片幻灯片

表 1 CNN网络参数

网络层	参数	输出维度
输入层	训练数据	(30,30,3,m)
2维卷积层1	Conv 2D,f_s=5,s=1,padding=same	(30,30,16,m)
2维卷积层2	Conv 2D,f_s=5,s=1,padding=same	(30,30,16,m)
2维卷积层3	Conv 2D,f_s=2,s=2,padding=valid	(15,15,32,m)
2维卷积层4	Conv 2D,f_s=5,s=1,padding=same	(15,15,32,m)
平坦层	K=7200	(7200,m)
全连接层1	K=1024	(1024,m)
全连接层2	K=512	(512,m)
输出层	K=N_rp	(N_rp,m)

下载: 导出CSV

表 2 CAE网络参数

网络层	参数	输出维度
输入层	训练数据	(90,1,m)
1维卷积层1	Conv 1D,f_s=5,s=1,padding=same	(90,64,m)
1维池化层1	Maxpool 1D,s=2,padding=same	(45,64,m)
1维卷积层2	Conv 1D,f_s=5,s=1,padding=same	(45,32,m)
1维池化层2	Maxpool 1D,s=2,padding=same	(23,32,m)
1维卷积层3	Conv 1D,f_s=5,s=1,padding=same	(23,32,m)
1维池化层3	Maxpool 1D,s=2,padding=same	(12,32,m)
1维卷积层4	Conv 1D,f_s=5,s=1,padding=same	(12,12,m)
1维卷积层5	Conv 1D,f_s=3,s=1,padding=valid	(10,10,m)
1维上采样层1	Upsampool 1D,size=2	(20,10,m)
1维卷积层6	Conv 1D,f_s=9,s=1,padding=valid	(12,12,m)
1维上采样层2	Upsampool 1D,size=2	(24,12,m)
1维卷积层7	Conv 1D,f_s=2,s=1,padding=valid	(23,23,m)
1维上采样层3	Upsampool 1D,size=2	(46,23,m)
1维卷积层8	Conv 1D,f_s=2,s=1,padding=valid	(45,32,m)
1维上采样层4	Upsampool 1D,size=2	(90,32,m)
1维卷积层9	Conv 1D,f_s=5s=1,padding=same	(90,64,m)
输出层	Conv 1D,f_s=5,s=1,padding=same	(90,1,m)

下载: 导出CSV

表 3 廊厅定位误差

定位算法	平均误差(m)	标准差(m)
CNNFi combine	0.25	0.58
CNNFi single	0.73	1.24
CiFi	1.09	1.28
DeepFi	0.99	1.62
PhaseFi	1.10	1.53
ImageFi	0.93	1.34

下载: 导出CSV

表 4 实验室定位误差

定位算法	平均误差(m)	标准差(m)
CNNFi combine AP=2	0.48	0.86
CNNFi combine AP=1	0.72	0.86
CNNFi single	1.15	1.22
CiFi	1.36	1.14
DeepFi	1.45	1.21
PhaseFi	1.71	1.43
ImageFi	1.67	1.47

下载: 导出CSV

表 5 3种神经网络的训练参数

神经网络	可训练参数	非训练参数	总参数
CNNFi	7944414	3264	7947678
Auto Encoder	16121	0	16121
CAE-CNN	490494	1344	491838

下载: 导出CSV

表 6 两种定位方法的在线定位时间与内存大小

定位方法	Matlab运行时间(s)	Python运行时间(s)	总运行时间(s)	内存大小 (MB)
CNNFi	0.1269	0.1399	0.2668	91
CAEFi	0.0062	0.1598	0.1660	5.7898

下载: 导出CSV

参考文献(28)

[1]	ALI M U, HUR S, and PARK Y. Poster abstract: IoT enabled Wi-Fi indoor positioning system using raster maps[C]. The 18th ACM/IEEE International Conference on Information Processing in Sensor Networks, Montreal, Canada, 2019: 327–328.
[2]	REZAZADEH J, SUBRAMANIAN R, SANDRASEGARAN K, et al. Novel iBeacon placement for indoor positioning in IoT[J]. IEEE Sensors Journal, 2018, 18(24): 10240–10247. doi: 10.1109/JSEN.2018.2875037
[3]	TORRES-SOSPEDRA J, JIMÉNEZ A R, MOREIRA A, et al. Off-line evaluation of mobile-centric indoor positioning systems: The experiences from the 2017 IPIN competition[J]. Sensors, 2018, 18(2): 487. doi: 10.3390/s18020487
[4]	ZHANG Daqiang, ZHAO Shengjie, YANG L T, et al. NextMe: Localization using cellular traces in internet of things[J]. IEEE Transactions on Industrial Informatics, 2015, 11(2): 302–312. doi: 10.1109/TII.2015.2389656
[5]	PAK J M, AHN C K, SHMALIY Y S, et al. Improving reliability of particle filter-based localization in wireless sensor networks via hybrid particle/FIR filtering[J]. IEEE Transactions on Industrial Informatics, 2015, 11(5): 1089–1098. doi: 10.1109/TII.2015.2462771
[6]	ZHU Yuke, MOTTAGHI R, KOLVE E, et al. Target-driven visual navigation in indoor scenes using deep reinforcement learning[C]. 2017 IEEE International Conference on Robotics and Automation (ICRA), Singapore, 2017: 3357–3364.
[7]	DARDARI D, CLOSAS P, and, DJURIĆ P M. Indoor tracking: Theory, methods, and technologies[J]. IEEE Transactions on Vehicular Technology, 2015, 64(4): 1263–1278. doi: 10.1109/TVT.2015.2403868
[8]	WISANMONGKOL J, KLINKUSOOM L, SANPECHUDA T, et al. Multipath mitigation for RSSI-based Bluetooth low energy localization[C]. The 19th International Symposium on Communications and Information Technologies (ISCIT), Ho Chi Minh City, Vietnam, 2019: 47–51.
[9]	ABBAS M, ELHAMSHARY M, RIZK H, et al. WiDeep: WiFi-based accurate and robust indoor localization system using deep learning[C]. 2019 IEEE International Conference on Pervasive Computing and Communications, Kyoto, Japan, 2019: 1–10.
[10]	FENG Daquan, WANG Chunqi, HE Chunlong, et al. Kalman-filter-based integration of IMU and UWB for high-accuracy indoor positioning and navigation[J]. IEEE Internet of Things Journal, 2020, 7(4): 3133–3146. doi: 10.1109/JIOT.2020.2965115
[11]	PEREKADAN V, MUKHERJEE T, BANERJEE C, et al. RF-MSiP: Radio frequency multi-source indoor positioning[C]. 2019 IEEE International Conference on Big Data (Big Data), Los Angeles, USA, 2019: 5259–5268.
[12]	许浩, 王旭东, 吴楠. 基于卷积神经网络的室内可见光指纹定位方法[J]. 激光与光电子学进展, 2021, 58(17): 1706008. XU Hao, WANG Xudong, and WU Nan. Indoor visible light fingerprint positioning scheme based on convolution neural network[J]. Laser &Optoelectronics Progress, 2021, 58(17): 1706008.
[13]	SINGH V, AGGARWAL G, and UJWAL B V S. Ensemble based real-time indoor localization using stray WiFi signal[C]. 2018 IEEE International Conference on Consumer Electronics (ICCE), Las Vegas, USA, 2018: 1–5.
[14]	KHALAJMEHRABADI A, GATSIS N, and AKOPIAN D. Modern WLAN fingerprinting indoor positioning methods and deployment challenges[J]. IEEE Communications Surveys & Tutorials, 2017, 19(3): 1974–2002. doi: 10.1109/COMST.2017.2671454
[15]	ACHROUFENE A, AMIRAT Y, and CHIBANI A. RSS-based indoor localization using belief function theory[J]. IEEE Transactions on Automation Science and Engineering, 2019, 16(3): 1163–1180. doi: 10.1109/TASE.2018.2873800
[16]	SOHAN A A, ALI M, FAIROOZ F, et al. Indoor positioning techniques using RSSI from wireless devices[C]. The 22nd International Conference on Computer and Information Technology (ICCIT), Dhaka, Bangladesh, 2019: 1–6.
[17]	BAHL P and PADMANABHAN V N. RADAR: An in-building RF-based user location and tracking system[C]. IEEE INFOCOM 2000. Conference on Computer Communications. Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies, Tel Aviv, Israel, 2000: 775–784.
[18]	YOUSSEF M and ASHOK A. The Horus location determination system[J]. Wireless Networks, 2008, 14(3): 357–374. doi: 10.1007/s11276-006-0725-7
[19]	HOANG M T, YUEN B, DONG Xiaodai, et al. Recurrent neural networks for accurate RSSI indoor localization[J]. IEEE Internet of Things Journal, 2019, 6(6): 10639–10651. doi: 10.1109/JIOT.2019.2940368
[20]	WANG Xuyu, GAO Lingjun, and MAO Shiwen. CSI phase fingerprinting for indoor localization with a deep learning approach[J]. IEEE Internet of Things Journal, 2016, 3(6): 1113–1123. doi: 10.1109/JIOT.2016.2558659
[21]	HALPERIN D, HU Wenjun, SHETH A, et al. Predictable 802.11 packet delivery from wireless channel measurements[J]. ACM SIGCOMM Computer Communication Review, 2010, 40(4): 159–170. doi: 10.1145/1851275.1851203
[22]	XIE Yaxiong, LI Zhenjiang, and LI Mo. Precise power delay profiling with commodity Wi-Fi[J]. IEEE Transactions on Mobile Computing, 2019, 18(6): 1342–1355. doi: 10.1109/TMC.2018.2860991
[23]	WU Kaishun, XIAO Jiang, YI Youwen, et al. CSI-based indoor localization[J]. IEEE Transactions on Parallel and Distributed Systems, 2013, 24(7): 1300–1309. doi: 10.1109/TPDS.2012.214
[24]	WANG Xuyu, GAO Lingjun, MAO Shiwen, et al. DeepFi: Deep learning for indoor fingerprinting using channel state information[C]. 2015 IEEE Wireless Communications and Networking Conference, New Orleans, USA, 2015: 1666–1671.
[25]	WANG Xuyu, WANG Xiangyu, and MAO Shiwen. CiFi: Deep convolutional neural networks for indoor localization with 5 GHz Wi-Fi[C]. Proceedings of 2017 IEEE International Conference on Communications (ICC), Paris, France, 2017: 1–6.
[26]	江小平, 王妙羽, 丁昊, 等. 基于信道状态信息幅值-相位的被动式室内指纹定位[J]. 电子与信息学报, 2020, 42(5): 1165–1171. doi: 10.11999/JEIT180871 JIANG Xiaoping, WANG Miaoyu, DING Hao, et al. Passive fingerprint indoor positioning based on CSI amplitude-phase[J]. Journal of Electronics &Information Technology, 2020, 42(5): 1165–1171. doi: 10.11999/JEIT180871
[27]	LI Haihan, ZENG Xiangsheng, LI Yunzhou, et al. Convolutional neural networks based indoor Wi-Fi localization with a novel kind of CSI images[J]. China Communications, 2019, 16(9): 250–260. doi: 10.23919/JCC.2019.09.019
[28]	SHI Shuyu, SIGG S, CHEN Lin, et al. Accurate location tracking from CSI-based passive device-free probabilistic fingerprinting[J]. IEEE Transactions on Vehicular Technology, 2018, 67(6): 5217–5230. doi: 10.1109/TVT.2018.2810307