無線定位(Wireless Localization)的技術隨著通訊的發展，受到越來越多人的關注，由於接收到的訊號資訊不同，分析方式也不同，根據基地台傳送訊號到達的角度(Angle of Arrival,\ AOA)，再利用三角測量技術(Triangulation)或是最小平方法(Least Square method)達到估計用戶端位置。

本篇論文，我們提出深度學習演算法來處理定位估計。在深度學習演算法中使用的是深度神經網路(Deep Neural Network,\ DNN)，針對二維平面空間進行分類，三角測量技術與最小平方法主要的問題就是當多個基地台(Base Stations, \ BS)接收到的角度是相近的就容易造成線性估計的誤差增加，而深度學習演算法透過學習特徵的方式使用非線性系統分類進而改善定位估計的效能。但在高密度類別的情況下，深度神經網路分類的正確率是有限的，因此透過自動編碼器(Auto Encoder,\ AE)結合深度神經網路改善此問題，自動編碼器的概念像是根據學習資料的結構特徵，先區分相同類型的數據，再將這些數據根據特徵細分，這有助於減輕後續深度神經網路估計用戶端位置的負擔，最後在結果分析與比較中，可以看出此方法在穩定性和精確性都有不錯性能。

Thanks to the development of modern communication technologies,
localization of end-user with wireless base stations has drawn more
and more attention. Subject to the received signals to be utilized,
the employed localization method is different. According to the
angle-of-arrival (AOA) of the signal received by the base station
equipped with multiple antennas, conventional methods such as
triangulation and least squares methods can be used to determine the
position of the end-user.

In this thesis, we study the approach of deep learning methods. Here, the deep neural network (DNN) is revised to perform spatial classification on a two-dimensional plane. The main problem of triangulation and least square method is that when the angles of arrival at multiple base stations are similar, a large localization error will increase. By contrast, the deep learning methods improve the performance of position estimation at learning features through non-linear system classification.

However, in the case of high-density classification categories, the accuracy of the conventional neural network method is limited. Therefore, the autoencoder is combined with the neural network to solve this problem.

The autoencoder acts like a group of spatial filters, decomposing the input into multiple smaller spatial sub-regions. The range of the input covered by each spatial sub-region is narrower than that of the original input ,and hence, the distribution of the input is better centralized for classification. From simulation analysis and performance comparison, the autoencoder
method achieves better performance than the LS and triangulation
methods and lower complexity than the DNN method.

參 考 文 獻
[1] S. Aditya, A. F. Molisch, and H. M. Behairy, “A survey on the im- pact of multipath on wideband time-of-arrival based localization,” Proceedings of the IEEE, vol. 106, no. 7, pp. 1183–1203, 2018.
[2] K. F. Taraktaş, O. Ceylan, and B. Yağcı, “Received signal strength based localization in sensor network,” in 2010 IEEE 18th Signal Processing and Communications Applications Conference, 2010, pp. 804–807.
[3] A. Pages-Zamora, J. Vidal, and D. Brooks, “Closed-form solution for positioning based on angle of arrival measurements,” in The 13th IEEE International Symposium on Personal, Indoor and Mobile Ra- dio Communications, vol. 4, 2002, pp. 1522–1526 vol.4.
[4] M. Nur-A-Alam and M. M. Haque, “A least square approach for tdoa/aoa wireless location in wcdma system,” in 2008 11th Interna- tional Conference on Computer and Information Technology, 2008, pp. 686–690.
[5] N. A. A. Elhag, I. M. Osman, A. A. Yassin, and T. B. Ahmed, “Angle of arrival estimation in smart antenna using music method for wideband wireless communication,” in 2013 INTERNATIONAL CONFERENCE ON COMPUTING, ELECTRICAL AND ELEC- TRONIC ENGINEERING (ICCEEE), 2013, pp. 69–73.
[6] J. Maire, A. Ziad, J. Borgnino, and F. Martin, “Comparison be- tween atmospheric turbulence models by angle-of-arrival covari- ance measurements,” Monthly Notices of the Royal Astronomical Society, vol. 386, no. 2, pp. 1064–1068, 2008.
[7] C. de Moustier, “Empirical uncertainty in angle of arrival estima- tion for bathymetric sidescan sonars,” in OCEANS’11 MTS/IEEE KONA, 2011, pp. 1–4.

[8] J. He, “A new 2-d triangulation optimization algorithm,” in 2010 Third International Conference on Knowledge Discovery and Data Mining, Jan 2010, pp. 284–287.
[9] Y. Lei, X.-x. Feng, C.-b. Zhu, and B.-b. Li, “Triangulation location fusion algorithm for 2d sensor network,” in Proceedings of 2011 International Conference on Computer Science and Network Tech- nology, vol. 2, 2011, pp. 889–893.
[10] M. Zane, M. Rupp, and S. Schwarz, “Performance investigation of angle of arrival based localization,” in WSA 2020; 24th Interna- tional ITG Workshop on Smart Antennas, 2020, pp. 1–4.
[11] S. Kay, Fundamentals of Statistical Signal Processing: Detection theory, ser. Fundamentals of Statistical Si. Prentice-Hall PTR, 1998. [Online]. Available: https://books.google.com.tw/books?id= vA9LAQAAIAAJ
[12] M. H. Mughal, M. J. Khokhar, and M. Shahzad, “Assisting uav lo- calization via deep contextual image matching,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sens- ing, vol. 14, pp. 2445–2457, 2021.
[13] B. Labinghisa and D. M. Lee, “A deep learning based scene recogni- tion algorithm for indoor localization,” in 2021 International Con- ference on Artificial Intelligence in Information and Communica- tion (ICAIIC), 2021, pp. 167–170.
[14] Y. Yu, W. Wang, J. Luo, and P. Feng, “Localization based stereo speech separation using deep networks,” in 2015 IEEE Interna- tional Conference on Digital Signal Processing (DSP), 2015, pp. 153–157.
[15] N. H. Nguyen and K. Doğançay, “Closed-form algebraic solutions for angle-of-arrival source localization with bayesian priors,” IEEE Transactions on Wireless Communications, vol. 18, no. 8, pp. 3827– 3842, 2019.

[16] X. Wang, X. Wang, and S. Mao, “Cifi: Deep convolutional neural networks for indoor localization with 5 ghz wi-fi,” in 2017 IEEE International Conference on Communications (ICC), 2017, pp. 1– 6.
[17] M. Zhang, K. Xu, W. Xie, J. Xu, M. Shen, and L. Chen, “Multi- l-shape array division based 2-d localization for incoherently dis- tributed sources in airborne massive mimo systems,” in 2019 IEEE 19th International Conference on Communication Technol- ogy (ICCT), 2019, pp. 430–435.
[18] W. Feng, Z. Chen, and Y. Fu, “Autoencoder classification algorithm based on swam intelligence optimization,” in 2018 17th Interna- tional Symposium on Distributed Computing and Applications for Business Engineering and Science (DCABES), 2018, pp. 238–241.
[19] Y. Liu, C.-f. Zhou, and Y.-w. Chen, “Weight initialization of feed- forward neural networks by means of partial least squares,” in 2006 International Conference on Machine Learning and Cybernetics, 2006, pp. 3119–3122.
[20] H. Yuanhua and L. Hongsheng, “A non-line-of-sight error mitiga- tion method in wireless location system,” in 2009 2nd IEEE Inter- national Conference on Computer Science and Information Tech- nology, 2009, pp. 282–286.
[21] M. Rupp and S. Schwarz, “An ls localisation method for massive mimo transmission systems,” in ICASSP 2019 - 2019 IEEE Inter- national Conference on Acoustics, Speech and Signal Processing (ICASSP), 2019, pp. 4375–4379.
[22] V. Tiwari, C. Pandey, A. Dwivedi, and V. Yadav, “Image classifi- cation using deep neural network,” in 2020 2nd International Con- ference on Advances in Computing, Communication Control and Networking (ICACCCN), 2020, pp. 730–733.
[23] J. Shijie, W. Ping, J. Peiyi, and H. Siping, “Research on data aug- mentation for image classification based on convolution neural net-

works,” in 2017 Chinese Automation Congress (CAC), 2017, pp. 4165–4170.
[24] Q. Malik, Functional Corroboration of a Digital Multilayer Neural Network. Michigan State University. Department of Electrical Engineering, 1995. [Online]. Available: https://books.google.com. tw/books?id=uzm4_GpqG5AC
[25] S. Wang, J. Lin, J. Lei, and J. Yang, “Integration of kpca and paral- lel coordinates for visualizing classification,” in 2011 International Conference on Computer Science and Service System (CSSS), 2011, pp. 3294–3297.
[26] S. Xu, “Optimal sensor placement for target localization using hy- brid rss, aoa and toa measurements,” IEEE Communications Let- ters, vol. 24, no. 9, pp. 1966–1970, 2020.