Design of an integrity support message for offline advanced RAIM

Kurzfassung

The modernization of GPS along with the emergence of new GNSS constellations opens new opportunities to redesign traditional Receiver Autonomous Integrity Monitoring (RAIM) in order to target more demanding navigation requirements. The evolution from legacy to Advanced RAIM will became a reality within the next years thanks to measurement redundancy that will guarantee navigation integrity, continuity, and accuracy on a global scale. In order for ARAIM users to evaluate these performance metrics, inputs from ground must be encapsulated within the Integrity Support Message (ISM). The first set of parameters broadcast through this message defines the individual satellite and constellation fault rates which reflect GNSS operational commitments. The second set provides the necessary parameters to create an integrity and accuracy bound for satellite unfaulted ranging errors which need to be assessed through GNSS performance characterization. In response to this need, this research focuses on the design of an ISM covering GNSS performance monitoring, error correlation analysis, sample independence, and overbounding theory. This dissertation presents a methodology to make use of the currently deployed Multi-GNSS EXperiment (MGEX) ground infrastructures to emulate the architecture of a future Air Navigation Service Provider (ANSP) ground network. The main scope of this technique is the establishment of a security layer between orbit and clock products and ISM generation. It guarantees that no fabricated errors are introduced at the same time that no integrity events are overlooked due to data unavailability. Using this monitor, GPS and Galileo service history are analyzed providing a comprehensive ephemeris and clock error characterization. A novelty introduced in this work is the time-dependent analysis which exposes the high correlation that inherently affects GNSS Signal-in-Space Range Error (SISRE). Based on an estimation variance study, this dissertation presents an analytical methodology to determine the time between effective independent samples. Results show that GPS and Galileo exhibit significantly different correlation behavior in that the European constellation is less affected by it. Based on Bayesian inference, this work proves that an analytical expression of the error Cumulative Distribution Function (CDF) as a function of the number of independent samples can be derived. In order to account for the impact of sample correlation on the error bounds, this work determines the factor by which the overbounding distribution needs to be inflated. This factor is inversely proportional to the number of independent samples representing the higher confidence that can be placed in the estimation as more independent data are collected. The fact Galileo range error is less correlated in time than GPS implies that shorter monitoring periods are needed to characterize the nominal performance of the European GNSS. This dissertation presents a modification of the error accuracy and integrity models in order to create more efficient and equally safe bounds. Based on empirical evidence, this work proposes the partition of error distributions in two sections; a quasi-Gaussian core and a flat tail distribution with large error magnitudes. Both distributions are individually bounded by Gaussian functions which are combined to create a weighted Multi Gaussian (MG) overbound. Unlike the current Single Gaussian (SG) bound, results show that the MG methodology provides the flexibility to bound large tail errors without sacrificing the narrow core. In order to incorporate the MG bound in the current ARAIM architecture, this dissertation modifies the currently used pair-bound theory proving that it still is a safe overbound in the position domain after convolution. This thesis carries out a modification of the current Multiple Hypothesis Solution Separation (MHSS) baseline algorithm defined by the US-EU Working Group C (WGC). ARAIM simulations show that a significant enhancement on service availability can be achieved with the inclusion of MG bounding within the user algorithm. Finally, this work presents three different ISM designs for incorporating the necessary parameters for the users to perform MG overbounds. Out of these three dissemination options, an optimal design is recommended allowing the ISM generator full flexibility to exploit core-tail partition.