臺灣博碩士論文加值系統

室內定位是一個被廣泛應用的技術，諸如購物導覽、居家照護或智慧倉儲…等，均有使用到室內定位的技術。目前室內定位正處於百家爭鳴的狀態，基於Wi-Fi的Received Signal Strength Indicator (RSSI) 定位法便是屬於室內定位的其中一種技術。由於其運作只需仰賴Wi-Fi Access Point (AP) 發出的訊號，具有成本低廉且容易佈置的優勢而被廣泛使用。然而這個技術的實現所仰賴的RSSI容易受到環境的干擾而產生波動，致使定位精度下降。
因此，本篇論文提出了一種運用RSSI、延遲(Delay)、吞吐量(Throughput) 與錯誤率(Error rate) 進行室內定位的方法。這個方法先在多個位置上採集並紀錄各個Wi-Fi AP的訊號強度、延遲、吞吐量與錯誤率作為建立指紋庫之用，然後讓Long short-term memory (LSTM) 使用這些資料進行訓練找出其特徵值。而後，只要在這個區域內掃描該位置上的各AP的訊號強度與其他資訊，再比對所建之指紋庫後便能定位出目前的位置。

Indoor positioning is a widely adopted technology used in various places such as shopping guides, home care, and smart warehouse. At present, indoor positioning is in a highly competitive arena, and the Wi-Fi-based received signal strength indicator (RSSI) positioning method, which is one of the indoor positioning technologies, is widely adopted because its operation relies on the signals sent by Wi-Fi access point (AP) and it is inexpensive and handy. However, this technology relies on RSSI, which is subjected to environment influence, resulting in a decrease in positioning accuracy.
Therefore, this paper proposes a method that uses RSSI, delay, throughput, and error rate for indoor positioning. This method first collects and records the signal strength, delay, throughput, and error rate of each Wi-Fi AP in various locations to build a fingerprint database, and further uses long short-term memory to drill the characteristic value. Moreover, the current location can be located after comparing the fingerprint database as long as the signal strength and other information of each AP in the area are scanned.

中文摘要 I
Abstract II
誌謝 III
目錄 IV
圖目錄 V
表目錄 VI
第一章　緒論 1
第一節　研究背景與動機 1
第二節　研究方法 2
第三節　論文架構 3
第二章　文獻探討 4
第一節　基於RSSI的室內定位法 4
第二節　長短期記憶網路 LSTM 6
第三節　文獻探討 10
第三章　多維特徵室內定位法與實驗工具 12
第一節　Best Wi-Fi 12
第二節　RSSI模擬訊號產生模組 14
第三節　二維空間RSSI模擬訊號產生模組 14
第四節　基於多維特徵的室內定位系統 16
第四章　實驗結果 18
第一節　一維特徵定位 18
第二節　多維特徵定位 19
結論 20
參考文獻 21

[1] GIANT group at the Institute for New Imaging Technology. " K-nearest neighbours, " Internet: http://indoorlocplatform.uji.es/methods/knn/, April 18, 2019.
[2] Weixing Xue, Weining Qiu, Xianghong Hua, and Kegen Yu, " Improved Wi-Fi RSSI Measurement for Indoor Localization, " IEEE Sensors Journal, vol.4, no.pp. 2224-2230, 2017.
[3] 陳輝, 施文鵬. " 一種基於混合濾波和冪指數映射wifi室內定位方法. " China. CN107071902A, August 18, 2017.
[4] Aiguo Zhang, Ying Yuan, Qunyong Wu, Shunzhi Zhu, and Jian Deng, " Wireless Localization Based on RSSI Fingerprint Feature Vector, " International Journal of Distributed Sensor Networks, vol.2015, no.,pp. 1-7, 2015.
[5] 薛衛星, 邱衛寧, 花向紅, 吳幫與蔡建民, " RSSI信號特徵值對 WiFi室內定位精度的影響分析, " 測繪地理信息, 41卷(4), 23-26.
[6] Xiaolong Xu, Yu Tang, and Shanchang Li, " Indoor Localization based on Hybrid Wi-Fi Hotspots, " in proc. of 2017 International Conference on Indoor Positioning and Indoor Navigation (IPIN), Sapporo, Japan, 18–21 September 2017, pp. 1–8.,
[7] Dishashree Gupta. " Fundamentals of Deep Learning – Introduction to Recurrent Neural Networks. "Internet: https://www.analyticsvidhya.com/blog/2017/12/introduction-to-recurrent-neural-networks/, March 26, 2019.
[8] Christopher Olah. " Understanding LSTM Networks. " Internet: https://colah.github.io/posts/2015-08-Understanding-LSTMs/, March 28, 2019.