臺灣博碩士論文加值系統

基於WiFi的室內定位已經得到深入研究。此類解決方案中的一個基本問題是 WiFi 指紋庫收集。然而，由於現實世界的限制，有時禁止在所有採集點採集完整的指紋。為了緩解這個限制，我們考慮了 WiFi 指紋修復問題。這個問題不同於典型的圖像/影片修復，並且帶來一些挑戰。首先，採集場地並不總是像圖像中是矩形的。其次，修復點並不總是如圖像位於網格點（像素）上。第三，一個場域的內部區域不太可能與其他場域相同，難以從過去的經驗中學習。因此，從指紋庫本身學習，我們稱之為 zero-shot 或 in-map 學習，似乎是不可避免的。第四，WiFi信號傳播遵循不同於自然視覺的物理規律。但幸運的是，可以將多個 WiFi AP 視為一個多通道環境，從中可以利用豐富的相關性。通過探索 AP 間和 AP 內的相關性提出了zero-shot learning 模型，並且有一些新發現。

WiFi-based indoor positioning has been intensively studied. A fundamental issue in such solutions is WiFi fingerprints collection. However, due to real-world constraints, collecting full fingerprints at all intented points is sometimes forbidden. To relieve this limitation, this work considers the WiFi fingerprint inpainting problem. This problem is different from typical image/video inpainting and poses several challenges. First, the field map is not always rectangle as that in images. Second, the inpainted points are not always on grid points (pixels) as that in images. Third, a field's interior partitions are unlikely to be the same as others, making it difficult to learn from past experiences. Therefore, learning from a fingerprint map itself, which we call zero-shot or in-map learning, seems to be inevitable. Fourth, WiFi signal propagations follow physical laws that are different from natural visions. Fortunately, multiple WiFi APs can be regarded as a multi-channel environment, from which rich correlations may be exploited. This work proposes zero-shot learning models by exploring inter-AP and intra-AP correlations and several new findings are reported.

摘要 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3 Proposed Inpainting Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.1 InterAP (IAP) Inpainting Model . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Interand IntraAP (I2AP) Inpainting Model . . . . . . . . . . . . . . . . . . . 6
4 Experiment Results and Ablation Study . . . . . . . . . . . . . . . . . . . . . . . 9
4.1 Comparisons of Inpainting Errors . . . . . . . . . . . . . . . . . . . . . . . . 11
4.2 Dimension of Latent Vectors (dim(vi)) . . . . . . . . . . . . . . . . . . . . . . 11
4.3 AP Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.4 Number of Nearest Neighbors (k) . . . . . . . . . . . . . . . . . . . . . . . . 13
4.5 Positioning Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.6 Ablation Study of Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.7 Sample on single point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.8 Positioning analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

[1] F. Liu, J. Liu, Y. Yin, W. Wang, D. Hu, P. Chen, and Q. Niu, “Survey on WiFibased indoor
positioning techniques,” IET Communications, pp. 1372–1383, 2020.
[2] C. Yang and H.R. Shao, “WiFibased indoor positioning,” IEEE Communications Maga
zine, 2015.
[3] A. Correa, M. Barcelo, A. Morell, and J. L. Vicario, “A review of pedestrian indoor posi
tioning systems for mass market applications,” Sensors, 2017.
[4] H. Liu, H. Darabi, P. Banerjee, and J. Liu, “Survey of wireless indoor positioning tech
niques and systems,” IEEE Transactions on Systems, Man, and Cybernetics, Part C, pp.
1067–1080, 2007.
[5] X. Ge and Z. Qu, “Optimization WiFi indoor positioning KNN algorithm locationbased
fingerprint,” in IEEE International Conference on Software Engineering and Service Sci
ence, 2016, pp. 135–137.
[6] K. Sabanci, E. Yigit, D. Ustun, A. Toktas, and M. F. Aslan, “WiFi based indoor localization:
Application and comparison of machine learning algorithms,” in International Seminar/
Workshop on Direct and Inverse Problems of Electromagnetic and Acoustic Wave Theory,
2018.
[7] F. Zafari, A. Gkelias, and K. K. Leung, “A survey of indoor localization systems and
technologies,” IEEE Communications Surveys & Tutorials, pp. 2568–2599, 2019.
[8] Y. Sun, M. Liu, and M. Q.H. Meng, “WiFi signal strengthbased robot indoor localiza
tion,” in IEEE International Conference on Information and Automation, 2014.
[9] P. Mirowski, R. Palaniappan, and T. K. Ho, “Depth camera slam on a lowcost WiFi map
ping robot,” in IEEE International Conference on Technologies for Practical Robot Appli
cations, 2012.
[10] B. R. Mantha, C. C. Menassa, and V. R. Kamat, “Robotic data collection and simulation
for evaluation of building retrofit performance,” Automation in Construction, pp. 88–102,
2018.
[11] A. Rai, K. K. Chintalapudi, V. N. Padmanabhan, and R. Sen, “Zee: Zeroeffort crowd
sourcing for indoor localization,” in MobiCom, 2012, pp. 293–304.
[12] G. Shen, P. Z. Z. Chen, T. Moscibroda, and Y. Zhang, “Walkiemarkie: Indoor pathway
mapping made easy,” in USENIX Symposium on Networked Systems Design and Imple
mentation, 2013.
[13] H. Zou, Z. Chen, H. Jiang, and L. Xie, “Accurate indoor localization and tracking using
mobile phone inertial sensors, WiFi and iBeacon,” in IEEE International Symposium on
Inertial Sensors and Systems, 2017.
[14] Y. Zhuang, Z. Syed, and Y. Li, “Evaluation of two WiFi positioning systems based on au
tonomous crowdsourcing of handheld devices for indoor navigation,” IEEE Transactions
on Mobile Computing, pp. 1982–1995, 2015.
[15] Y. Zhuanga, Z. Syed, J. Georgy, and N. ElSheimy, “Autonomous smartphonebased WiFi
positioning system by using access points localization and crowdsourcing,” Pervasive and
Mobile Computing, pp. 118–136, 2015.
[16] B. Ciftler, A. Albaseer, N. Lasla, and M. Abdallah, “Federated learning for localization:
A privacypreserving crowdsourcing method,” arXiv:2001.01911, 2020.
[17] M. Laska, M. Laska, and R. Klamma, “Adaptive indoor area localization for perpetual
crowdsourced data collection,” Sensors, p. 1443, 2020.
[18] F. Gu, J. Blankenbach, K. Khoshelham, J. Grottke, and S. Valaee, “Zeefi: Zeroeffort floor
identification with deep learning for indoor localization,” in GlobeCom, 2019.
[19] G. G. Anagnostopoulos and A. Kalousis, “Proxyfaug: Proximitybased fingerprint aug
mentation,” arXiv:2102.02706, 2021.
[20] K. A. Nuaimi and H. Kamel, “A survey of indoor positioning systems and algorithms,” in
IEEE International Conference on Innovations in Information Technology, 2011.
[21] G. Liu, F. A. Reda, K. J. Shih, T.C. Wang, A. Tao, and B. Catanzaro, “Image inpainting
for irregular holes using partial convolutions,” in ECCV, 2018.
[22] M. Bertalmio, G. Sapiro, V. Caselles, and C. Ballester, “Image inpainting,” in Computer
Graphics and Interactive Techniques, 2000, pp. 417–424.
[23] U. Demir and G. Unal, “Patchbased image inpainting with generative adversarial net
works,” arXiv:1803.07422, 2018.
[24] R. A. Yeh, C. Chen, T. Y. Lim, A. G. Schwing, M. HasegawaJohnson, and M. N. Do,
“Semantic image inpainting with deep generative models,” in CVPR, 2017.
[25] S. A. Chavan and N. M. Choudhari, “Various approaches for video inpainting: A survey,”
in International Conference On Computing, Communication, Control And Automation,
2019.
[26] R. Xu, X. Li, B. Zhou, and C. C. Loy, “Deep flowguided video inpainting,” in CVPR,
2019.
[27] D. Kim, S. Woo, J.Y. Lee, and I. S. Kweon, “Deep video inpainting,” in CVPR, 2019.
[28] M. Richard and M. Y.S. Chang, “Fast digital image inpainting,” in Appeared in the Pro
ceedings of the International Conference on Visualization, Imaging and Image Processing,
2001.
[29] G. Mao, B. D. Anderson, and B. Fidan, “Path loss exponent estimation for wireless sensor
network localization,” Computer Networks, pp. 2467–2483, 2007.
[30] E. Assaf, Ahmad, S. Zaidi, S. Affes, and N. Kandil, “Robust annsbased WSN localization
in the presence of anisotropic signal attenuation,” IEEE Wireless Communications Letters,
pp. 504–507, 2016.
[31] A. Chatterjee, “A fletcher–reeves conjugate gradient neuralnetworkbased localization
algorithm for wireless sensor networks,” IEEE Transactions on Vehicular Technology, pp.
823–830, 2009.
[32] D. Niculescu and B. Nath, “Ad hoc positioning system (aps) using aoa,” in INFOCOM,
2003.
[33] S. He and S.H. G. Chan, “WiFi fingerprintbased indoor positioning: Recent advances
and comparisons,” IEEE Communications Surveys & Tutorials, pp. 466–490, 2016.
[34] C. Zhu, L. Xu, X.Y. Liu, and F. Qian, “Tensorgenerative adversarial network with two
dimensional sparse coding: Application to realtime indoor localization,” in IEEE Inter
national Conference on Communications, 2018.
[35] H. Zou, C.L. Chen, M. Li, Y. Z. J. Yang, L. Xie, and J. C. Spanos, “Adversarial learning
enabled automatic WiFi indoor radio map construction and adaptation with mobile robot,”
IEEE Internet of Things J., pp. 6946–6954, 2020.
[36] W. Njima, M. Chafii, A. Chorti, R. M. Shubair, and H. V. Poor, “Indoor localization us
ing data augmentation via selective generative adversarial networks,” IEEE Access, pp.
98 337–98 347, 2021.
[37] W.Y. Kim, S.H. Tae, and D.H. Seo, “Accesspoint centered windowbased radiomap
generation network,” Sensors, 2021.
[38] X. Han, L. Xue, Y. Xu, and Z. Liu, “A radio environment maps estimation algorithm based
on the pixel regression framework for underlay cognitive radio networks using incomplete
training data,” Sensors, 2020.
[39] A. BelmonteHernandez, G. HernandezPenaloza, D. M. Gutierrez, and F. Alvarez, “Re
current model for wireless indoor tracking and positioning recovering using generative
networks,” IEEE Sensors J., 2019.
[40] Y. Yang, P. Dai, H. Huang, M. Wang, and Y. Kuang, “A semisimulated rss fingerprint
construction for indoor wifi positioning,” Electronics, p. 1568, 2020.
[41] J. Yang, “Indoor localization system using dualfrequency bands and interpolation algo
rithm,” IEEE Internet of Things J., 2020.
[42] S.P. Kuo and Y.C. Tseng, “A scrambling method for fingerprint positioning based on
temporal diversity and spatial dependency,” IEEE Transactions on Knowledge and Data
Engineering, 2008.
[43] V. Moghtadaiee, S. A. Ghorashi, and M. Ghavami, “New reconstructed database for cost
reduction in indoor fingerprinting localization,” IEEE Access, pp. 104 462–104 477, 2019.
[44] J. Yoon, J. Jordon, and M. Schaar, “Gain: Missing data imputation using generative ad
versarial nets,” In Machine Learning Research, 2018.
[45] S. C.X. Li, B. Jiang, and B. Marlin, “Misgan: Learning from incomplete data with gen
erative adversarial networks,” arXiv:1902.09599, 2019.
[46] Rasmussen and C. Edward, “Gaussian processes in machine learning,” in Summer School
on Machine Learning, 2003.
[47] P. Isola, J.Y. Zhu, T. Zhou, and A. A. Efros, “Imagetoimage translation with conditional
adversarial networks,” in CVPR, 2017, pp. 1125–1134.
[48] D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv:1412.6980,
2014.
[49] W. Nelson, “Use of circular error probability in target detection,” Mitre Corp Bedford MA,
Tech. Rep., 1988.