Integrity architecture for carrier-phase differential GNSS system using Solution Separation RAIM

Abstract

Carrier-phase Differential GNSS (CDGNSS), a navigation system providing cm-level positioning accuracy, is receiving significant attention in safety-critical systems such as autonomous driving cars. The key to achieving cm-level positioning accuracy in CDGNSS is the cycle ambiguity resolution process. However, faults in measurements can lead to more frequent incorrect ambiguity resolutions, which may result in significant biases in position errors. To ensure the integrity of CDGNSS, it is essential to calculate a protection level that bounds the position errors resulting from incorrect ambiguity resolutions. The calculation of this protection level in CDGNSS is challenging due to the non-linear nature of the cycle ambiguity resolution process. This thesis proposes an integrity architecture for CDGNSS utilizing Solution Separation-Receiver Autonomous Integrity Monitoring (SS-RAIM), which enables the limitation of the maximum size of position error caused by undetected faults. The fault modes threatening the integrity of CDGNSS are identified, and SS-RAIM-based fault detection monitors are designed to detect these faults. For the fault detection monitors, the probability distributions of fault detection statistics, taking ambiguity resolution into account, were derived. Based on this, the monitors were designed to meet the system's continuity requirements. Additionally, a method was proposed to ensure that the protection level reliably bounds the actual position error. This method assumes that any incorrect ambiguity resolution always leads to integrity loss and utilizes an overbounding technique. However, due to this conservative approach, an initial time is required to use the precise navigation solutions. Consequently, a measurement-based ambiguity validation method was proposed to reduce this initial time. The use of a measurement-based ambiguity validation method leads to changes in the position error distribution, necessitating a revised approach to calculating the protection level. The performance evaluation of the proposed system demonstrated that its protection level is significantly improved compared to existing code-based systems. Furthermore, the results show that the calculated protection level always bounds the actual position error.