Document Type
Thesis - Open Access
Award Date
2024
Degree Name
Master of Science (MS)
Department / School
Electrical Engineering and Computer Science
First Advisor
Kwanghee Won
Second Advisor
Young Chang
Abstract
As autonomous vehicles (AVs) become integral to modern transportation, their susceptibility to cyber-attacks, particularly GPS spoofing, presents a serious security threat. This study addresses these challenges by applying a suite of deep learning models to enhance the detection of anomalous GPS signals. Focusing on autoencoder-based architectures, the proposed models such as long short-term memory-based variational autoencoder (LSTM-VAE), LSTM-based autoencoder (LSTM-AE), multilayer perceptron-based variational autoencoder (MLP-VAE), MLP-based Autoencoder (MLPAE), Stacked LSTM-based variational autoencoder (Stacked-LSTM-VAE), stacked LSTM-based autoencoder (Stacked-LSTM-AE), memory-augmented-LSTM-VAE (Mem-LSTM-VAE), and time-series-anomaly-detection-generative-adversarial-networks (TadGAN) were trained exclusively on authentic GPS data. This unsupervised learning approach which used for the above-mentioned models enables the models to accurately identify deviations from normal patterns, thereby enhancing their ability to detect spoofed signals through detailed reconstruction error analysis. In this thesis a comparative analysis highlights the differing strengths of these models in distinguishing between authentic and spoofed signals, with notable performance exhibited by the architectures of MLP-AE and Stacked-LSTM. The MLP-AE model achieved a detection accuracy of 95.40%, precision of 93.09%, and a ROC-AUC score of 94.35%. Similarly, the deeper learning structure of the Stacked-LSTM models demonstrated high efficacy in handling complex and noisy data, critical for high-stakes applications such as AV navigation. These results emphasize the potential of combining autoencoder-based models with MLP and Stacked-LSTM architectures to strengthen AV security against GPS spoofing attacks. By demonstrating the effectiveness of unsupervised learning in anomaly detection, this study presents a robust framework for improving the security of GPS-dependent technologies, laying the foundation for future advancements in safeguarding AV systems against evolving cyber threats.
Publisher
South Dakota State University
Recommended Citation
Mirzakhaninafchi, Hasan, "Comparative Analysis of Deep Learning-Based Anomaly Detection Models for GPS Spoofing Detection" (2024). Electronic Theses and Dissertations. 1351.
https://openprairie.sdstate.edu/etd2/1351
Rights Statement
