GPS-IDS: An Anomaly-based GPS Spoofing Attack Detection Framework for Autonomous Vehicles

自动驾驶汽车（AVs）在导航时严重依赖传感器和通信网络，如全球定位系统（GPS）。先前的研究表明，像GPS这样的网络容易受到欺骗和干扰等网络攻击，从而导致导航错误和系统故障等严重风险。随着AVs的广泛部署，这些威胁预计将加剧，因此检测和缓解此类攻击至关重要。本文提出了GPS入侵检测系统（GPS-IDS），这是一种基于异常行为分析（ABA）的入侵检测框架，用于检测AVs上的GPS欺骗攻击。该框架使用一种新颖的基于物理的车辆行为模型，其中将GPS导航模型集成到传统的动态自行车模型中，以准确地表示AV行为。使用机器学习分析从此行为模型中提取的时间特征，以检测正常和异常的导航行为。 GPS-IDS框架的性能在AV-GPS数据集上进行了评估，该数据集由团队使用AV测试平台收集而成，并已公开发布供全球研究社区使用。据我们所知，该数据集是其类别中的第一个，将成为解决此类安全挑战的有用资源。

GPS networks are vulnerable to cyber-attacks like spoofing and jamming, which pose serious risks to the navigation of autonomous vehicles. This paper proposes a GPS Intrusion Detection System (GPS-IDS) to detect GPS spoofing attacks on AVs.

The GPS-IDS framework uses an Anomaly Behavior Analysis (ABA)-based intrusion detection approach, which integrates a GPS navigation model into a dynamic bicycle model to accurately represent AV behavior. Machine learning is used to analyze temporal features derived from this behavior model to detect abnormal navigation behavior.

The paper evaluates the performance of the GPS-IDS framework on the AV-GPS-Dataset, which is a real-world dataset collected by the team using an AV testbed. The dataset has been publicly released for the global research community. The framework's novelty lies in its integration of physics-based vehicle behavior modeling and machine learning for intrusion detection. The paper also highlights the need for further research into improving the robustness of AVs against cyber-attacks.

Recent related studies include 'A Survey of Cybersecurity in Autonomous Vehicles' and 'A Comprehensive Survey on Security of Autonomous Vehicles'.