Risk-resilient GPS-based positioning, navigation, and timing using sensor fusion and multi-agent platforms

The Global Positioning System (GPS) has become a stealth utility for many applications, such as positioning and navigation for ground and aerial vehicles, and timing and synchronization for the modern power systems. However, GPS receivers suffer from various vulnerabilities, including measurement faults, satellite faults, and spoofing attacks. Measurement faults occur when GPS signals are blocked or reflected, especially by tall buildings in urban areas, whereas satellite faults are caused by anomalies in the broadcast satellite ephemeris or clock information. Spoofing attacks occur when counterfeit GPS signals are transmitted by malicious sources to overpower authentic ones. This dissertation aims to provide risk-resilient GPS-based positioning and timing against these vulnerabilities. We will not only detect and mitigate the vulnerabilities, but also quantify the risk associated with the estimated position and timing. For instance, claiming a 10-meter position accuracy does not suffice, unless accompanied by a certain risk, say 10^(-3). We will focus on two application scenarios: urban positioning in the presence of satellite and measurement faults, and timing in power systems that may suffer from spoofing. For vehicle positioning in urban environments, we first design a Graph-SLAM (Simultaneous Localization and Mapping) framework to simultaneously localize both the vehicle and the GPS satellites, by treating GPS satellites as landmarks. We then extend this Graph-SLAM framework from GPS-only to GPS-vision fusion; from a single receiver to multi-agent platform with Ultra-Wide Band (UWB) inner ranging. We design multi-fault detection and isolation algorithms for both GPS-vision fusion and multi-agent platforms, and quantitatively assess the risk of the resulting positioning solution. For timing in power systems, we perform time authentication by utilizing networked GPS receivers at power substations across the US. We cross-check the GPS signals for the presence of encrypted P(Y) codes, which serve as a unique signature, to detect spoofing. We further design a set-valued distributed Kalman Filter that mitigates spoofing and quantifies the timing risk