由於在室內環境下，全球衛星導航系統(Global Navigation Satellite System, GNSS)的訊號嚴重衰減或多路徑干擾，導致在建築物內無法提供有效的定位資訊。在大型室內場所，望著眼花撩亂的指標，不知身在何處。因此，本論文為了提供使用者更完整的定位服務，研究室內定位測試系統。本論文使用了無線網路(Wi-Fi)環境特徵演算法(Fingerprinting Algorithm)和地磁匹配(Magnetic Matching)，利用整合兩種不同的訊號源以提升定位精度。若上述兩種方法無法提供有效資訊時，為了提供使用者替代的定位方法，本論文亦研究行人航位推算(Pedestrian Dead Reckoning, PDR)，此方法只需利用手機內建的感測器，偵測步數、步伐和轉向即可推算出行人軌跡。
隨著無線網路(Wi-Fi)的設備普及，亦是現在智慧型手機必備的功能，其優勢在覆蓋範圍大，易於安裝且成本低，故使用無線網路當作環境特徵演算法的訊號源。同樣地，建築物結構會對地球磁場造成干擾，可利用此干擾作為磁場特徵以幫助定位。基於智慧型手機應用的發展，慣性感測定位技術也隨之興起，透過手機內建的加速度計、陀螺儀等慣性元件和地磁計偵測步數、步伐和轉向，和已知的初始位置推估出使用者的行進軌跡。而環境特徵演算法則是利用使用者接收到的訊號強度指示和其預先建好的資料庫進行特徵比對，進而推估出使用者的可能位置，本論文討論了加權最近鄰居演算法(K-Weighted Nearest Neighbors, KWNN)及利用動態時間校正(Dynamic Time Warping, DTW)進行不同資料長度的磁場特徵比對，再利用無線網路定位的標準差(Standard Deviation, STD)作為搜尋半徑找到地磁匹配的結果以提升定位精度和限制地磁匹配的定位誤差。本論文調整了校正階段資料庫的儲存方法以增強環境特徵演算法的定位結果，在大型賣場68%的定位誤差減少了9公尺。相反地，行人航位推算不需事先建立資料庫，利用手機內建的感測器之讀數偵測步數、步伐和轉向。由於慣性感測器本身有偏差和雜訊的干擾，本論文使用低通濾波器和互補濾波器校正誤差。
本論文在國立成功大學航太系館及大型賣場進行實驗實際呈現整合無線網路環境特徵比對演算法和地磁匹配之結果、行人航位推算之結果。結果呈現出無線網路的環境特徵演算法平均定位誤差為3.5公尺，地磁匹配平均定位誤差為6.7公尺，整合兩者的平均定位誤差為2.9公尺，以及行人航位推算的閉合水準控制在2%內。

To provide complete positioning services, it is essential to develop indoor positioning systems. Due to the popularity of Wi-Fi and mobile devices, this thesis presents an analysis of an indoor positioning system, that uses the built-in sensors of a smartphone with combination of IEEE 802.11 Wi-Fi fingerprinting and magnetic matching (MM). The main reason for using Wi-Fi signals as fingerprints is that Wi-Fi access points are commonly distributed in indoor environments and are the basic devices in smartphones. On the other hand, concrete building frames cause perturbations in the indoor magnetic field that creates a unique magnetic distribution in every building. We can thus make use of this unique magnetic distribution through a specific algorithm to acquire more accurate indoor positioning results. In general, MM results might have small errors on some occasions and might also suffer from significant mismatches. Wi-Fi fingerprinting and MM have similar concept, the former is a point-by-point matching approach, and the latter is based on profile matching, where by utilizing the Wi-Fi standard deviation as the searching space to limit the positioning errors of MM, the positioning results of Wi-Fi fingerprinting can be improved.
If neither Wi-Fi fingerprinting nor MM is available, this thesis also explores another positioning method for indoor environment. The PDR (Pedestrian Dead Reckoning) estimates the user trajectory by utilizing the built-in sensors of a smartphone to calculate the step detection, the step length, and orientation without the pre-recorded building database.
Finally, the positioning results for the integration of Wi-Fi fingerprinting and MM are demonstrated in the Department of Aeronautics and Astronautics (DAA) building at National Cheng Kung University (NCKU) and in the warehouse test field. The positioning results for PDR are conducted in the DAA at NCKU and in the warehouse. Based on the results of this thesis, the average positioning error for Wi-Fi fingerprinting alone is 3.5 meters and for MM alone is 6.7 meters. Therefore, the average positioning error for the integrated Wi-Fi fingerprinting and the MM is 2.9 meters. In addition, the relative error of PDR can be estimated by 2% of the trajectory.

[1]
M. A. Youssef, A. Agrawala, A. Udaya Shankar, "WLAN Location Determination Via Clustering and Probability Distributions," in Proceeding of the First IEEE International Conference on Pervasive Computing and Communications, 2003.
[2]
L.T. Hsu, W.M. Tsai, S.S. Jan, "Development of a Real Time Indoor Location Based Service Test Bed," in ION GNSS, Portland, OR, September 2010, pp. 1175-1183.
[3]
B. Li, Y. Wang, H.K. Lee, A. Dempster, C. Rizos, "Method for Yielding a Database of Location Fingerprinting in WLAN," in IEEE Proc.-Commun, 2005.
[4]
Y. Li, Y. Zhuang, H. Lan, P. Zhuang, X. Niu and N. EI-Sheimy, "WiFi-Aided Magnetic Matching for Indoor Navigation with Consumer Portable Devices," in micromachines, 2015, 6, 747-764.
[5]
D.F. Silva, G.E.A.P.A Batista, "Speeding Up All-Pairwise Dynamic Time Warping Matrix Calculation," in SIAM, 2015.
[6]
K. Kloch, P. Lukowicz, C. Fischer, "Collaborative PDR Localization with Mobile Phones," in IEEE Sensors Journal, 2011.
[7]
W.H. Kang, Y.N. Han, "Smart PDR: Smartphone-Based Pedestrian Dead Reckoning for Indoor Localization," in IEEE Sensors Journal, 2015.
[8]
H.T. Hou and C.S. Lee, "Using Kinect Technology to Help Rehabilitation for Patients with Frozen Shoulders," Master's thesis, National Taipei University of Nursing and Health Science, 2014.
[9]
X. Li, J. Wang, T. Li, "Seamless positioning and navigation by using Geo-referenced images and multi-sensor data," in MDPI sensors, 2013.
[10]
G.A. ten Holt, M.J.T. Reinders, E.A. Hendriks, "Multi-Dimensional Dynamic Time Warping for Gesture Recognition," in ASCI, 2007.
[11]
W. Song, J.W. Lee, B.S. Lee, H. Schulzrinne, "Finding 9-1-1 callers in tall buildings," in Proceeding of IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks 2014.
[12]
Y. Chon, E. Talipov and H. Cha, "Autonomous Management of Everyday Places for a Personalized Location Provider," in IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 42, no. 4, pp. 518-531, July 2012.
[13]
A. Rai, K.K. Chintalapudi, V.N. Padmanabhan, R. Sen, "Zee: Zero-Effort Crowdsourcing for Indoor Localization," in MobiCom, Istanbul, Turkey, August 2012, pp. 293–304.
[14]
H. Ookura, H. Yamamoto, K. Yamazaki, "Development and evaluation of walking path estimation system using sensors of Android device and vector map matching," in The International Conference on Information Network (ICOIN), 2012.
[15]
H.H. Lee, S. Choi, M.J. Lee, "Step Detection Robust against the Dynamics of Smartphones," in Sensors 15(10): 27230.
[16]
S.L. Altmann, "Rotations, Quaternions, and Double Groups," Oxford, New York: Courier Corporation, 2005.
[17]
J.B. Kuipers,, "Quaternions and Rotation Sequences: A primer with Applications to Orbits, Aerospace, and Virtual Reality," New Jersey: Princeton Univ Pr, 1999.
[18]
P.G. Savage, "Strapdown Analytics: Part 1," Strapdown Associates, 2000.
[19]
V. Renaudin and C. Combettes, "Magnetic, acceleration fields and gyroscope quaternion (MAGYQ)-based attitude estimation with smartphone sensors for indoor pedestrian navigation," in Sensors 14(12), 2014.
[20]
R.G. Valenti, I. Dryanovski and J. Xiao, "Keeping a Good Attitude: A Quaternion-based Orientation Filter for IMUs and MARGs," in Sensors 15(8), 2015.
[21]
H. Weinberg, "AN-602: Using the ADXL202 in Pedometer and Personal Navigation Applications," in Analog Devices, USA, 2012.
[22]
W. M. Y. W. Bejuri, M. M. Mohamad and R. Z. R. M. Radzi, "Offline Beacon Selection-Based RSSI Fingerprinting for Location-Aware Shopping Assistance: A Preliminary Result," in D. Barbucha, N. T. Nguyen and J. Batubara (ed.), New Trends in Intelligent Information and Database Systems, 2015
[23]
R. Mustafa, J.H. Kim, J. Kelly, Ro Le, J.H. Kim, W.E. Alexander, "Wireless Sensor Network Application for Cost Effective Environment Monitoring," in IEEE Southeastcon, 2009.
[24]
P. Bahl and V. N. Padmanabhan, "RADAR: an in-building RF-based user location and tracking system," in Proceedings IEEE INFOCOM Conference on Computer Communications. Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies (Cat. No.00CH37064), 2000.
[25]
A. Sabatini, "Quaternion-based extended Kalman filter for determining orientation by inertial and magnetic sensing," in IEEE Transactions on Biomedical Engineering 53(7), 2006.
[26]
B. Durak, "A Classification Algorithm Using Mahalanobis Distance Clustering of Data with Application on Biomedical Data sets," Master's thesis, Middle East Technical University, 2011.
[27]
Z. Li, C. Liu, J. Gao, and X. Li, "An Improved WiFi/PDR Intergration System Using an Adaptive and Robust Filter for Indoor Localization," in ISPRS International Journal of Geo-Information 5(12)., 2016.
[28]
K. Kircher, "Low-Pass Filter: The Basics," Retrieved November 30, 2016, from http://www.kircherelectronics.com/blog/index.php/11-android/sensors/8-low-pass-filter-the-basics.
[29]
H. Wang, "Sensor Calibration," Retrieved April 22, 2017, from https://hom-wang.gitbooks.io/quadcopter/content/05_SensorCorr.html
[30]
LEGENDS2K, "Mathematical reason behind gimbal lock in Euler angles," Retrieved January 10, 2017, from https://sundaram.wordpress.com/2013/03/08/mathematical-reason-behind-gimbal-lock-in-euler-angles/.
[31]
L. Fang, P.J. Antsaklis, L. Montestruque, M.B. Mcmickell, M. Lemmon, Y. Sun, H. Fang, I. Koutroulis, M. Haenggi, M. Xie, and X. Xie , "Design of a Wireless Assisted Pedestrain Dead Reckoning System - The NavMote Experience" in IEEE Transactions on Instrumentation and Measurement 54(6), 2005.
[32]
E.H. Shin, "Estimation Techiques for Low-Cost Inertial Navigation" Ph.D. dissertation, University of Calgary, 2005.
[33]
S.S. Jan, P.Y. Huang, D.S. De Lorenzo and Ivy Tseng, "Smartphone-based hybrid indoor positioning system with integration of Wi-Fi fingerprinting and magnetic matching" GNSS ION, 2016.