臺灣博碩士論文加值系統

在現今時代，全球衛星導航系統(Global Navigation Satellite System, GNSS)普遍運用在許多生活中的電子產品內，然而因為城鎮都市高樓聳立的環境，衛星定位系統受到的干擾相當嚴重。而造成定位系統測量干擾的主要原因之一，就是多路徑效應。

本論文利用長短期記憶循環神經網路來改善多路徑效應誤差干擾以及偵測訊號異常，期許能在定位結果異常出現前偵測出訊號因多路徑效應或人為干擾造成的誤差。由於造成訊號干擾的變數有很多，因此本文透過多個輸入模型進行訓練，分析出真實準確資料與具有多路徑效應干擾誤差資料的模擬預測線，透過預測線與誤差資料交叉比對，觀察模型預測準確度，找出最佳模型，並且實現異常偵測，期許能透過此模型提早偵測及預防多路徑干擾或人為干擾訊號。


關鍵字：全球衛星導航系統、長短期記憶循環神經網路、多路徑效應、多輸入模型、異常偵測

In modern times, the Global Navigation Satellite System (GNSS) is widely used in many electronic products in daily life. However, due to the environment of high-rise buildings in cities and towns, the satellite positioning system is severely interfered with. One of the main causes of measurement interference in positioning systems is the multipath effect.

This thesis uses long-short-term memory recurrent neural network to improve multipath error interference and detect signal anomalies, hoping to detect signal errors caused by multipath effects or artificial interference before abnormal positioning results occur. Since there are many variables that cause signal interference, this paper uses multiple input models for training to analyze the real and accurate data and the simulated prediction line with multipath interference error data. Through the cross-comparison between the prediction line and the error data, observe the prediction accuracy of the model, searching the best model, and realize anomaly detection. It is expected to detect and prevent multipath interference or jamming signals in advance through this model.


Keywords：GPS, Long Short-Term Memory, Multipath, Deep learning, Multiple input model

摘要 I
Abstract II
目錄 III
圖次 V
表次 VII
第一章、緒論 1
1.1 前言 1
1.2 研究目的與動機 1
1.3 文獻回顧 2
1.4 本文架構 2
第二章、衛星導航觀測基本原理 3
2.1 GPS觀測量簡介 3
2.2 電碼相位觀測量(Pseudo-range measurement) 3
2.3 載波相位觀測量(Carrier phase measurement) 4
2.4 衛星軌道誤差 5
2.5 衛星時鐘誤差 5
2.6 接收機時鐘誤差 5
2.7 對流層延遲誤差 5
2.8 電離層延遲誤差 6
2.9 多路徑效應 6
第三章、欺騙訊號與異常偵測 8
3.1 欺騙訊號 8
3.2 公開式欺騙訊號 8
3.3 隱藏式欺騙訊號 8
3.4 異常類型 10
3.5 異常偵測方法 10
3.6 異常偵測的應用 11
第四章、神經網路與深度學習架構 12
4.1 神經元 12
4.2 神經網路架構 12
4.3 多輸入神經網路 13
4.4 深度學習 13
4.5 循環神經網路 14
4.6 長短期記憶 15
第五章、實驗方法與分析 18
5.1 輸入資料描述 18
5.2 調整設定模型參數 25
5.3 模型預測結果與分析 25
5.4 異常偵測結果與分析 37
5.5 結果 54
第六章、結論與未來展望 55
6.1 結論 55
6.2 未來展望 57
參考資料 58

[1]E. Basoglu、M.Yigit、B.Erol (2017) Evaluation of GPS Spoofing Impact on UAV Navigation System IEEE Aerospace Conference Proceedings, pp. 1-8,
[2]A.Chapman, (2017) "GPS Spoofing"
[3]Silvio Semanjski , Ivana Semanjski, Wim De Wilde and Alain Muls (2020) "Use of Supervised Machine Learning for GNSS Signal Spoofing Detection with Validation on Real-World Meaconing and Spoofing Data"
[4] Taro Suzuki and Yoshiharu Amano (2021) "NLOS Multipath Classification of GNSS Signal Correlation Output Using Machine Learning"
[5] Louis H. Estey and Charles M. Meertens. (1999).TEQC: The Multi-Purpose Toolkit for GPS/GLONASS Data. GPS Solutions, Vol. 3, No. 1, pp. 42-49
[6]Psiaki、Humphreys(2016 )GNSS Spoofing and Detection.
[7]Alex Sherstinsky (2020), Fundamentals of Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) Network
[8]Sepp Hochreiter, Jürgen Schmidhuber, “Long Short-Term Memory”, in Neural
Computation 9(8):1735-80, December 1997.
[9]Understanding LSTM Networks:http://colah.github.io/posts/2015-08-Understanding-LSTMs/.
[10]Phan Q.H. , Tan S.L. , McLoughlin I. (2013). GPS multipath mitigation: A nonlinear regression approach. GPS Solut.
[11]Phan Q.H. , Tan S.L. , McLoughlin I, Vu D.L. (2013). A unified framework for GPS code and carrier-phase multipath mitigation using support vector regression. Adv. Artif. Neural Syst.
[12]莊智清 and 黃國興, 電子導航. 全華科技, 2003.
[13]Z. Liao, Y. Yu, and B. Chen, "Anomaly detection in GPS data based on visual analytics," in 2010 IEEE Symposium on Visual Analytics Science and Technology, 2010, pp. 51-58: IEEE.
[14]X. Shen and S. Agrawal, "Kernel Density Estimation for An Anomaly Based Intrusion Detection System," in MLMTA, 2006.
[15]E. Kaplan and C. Hegarty, Understanding GPS: principles and applications. Artech house, 2005.
[16]K. C. Zeng et al., "All your GPS are belong to us: Towards stealthy manipulation of road navigation systems," in 27th USENIX Security Symposium (USENIX Security 18), 2018.