Once the integrity concept and the local error models are derived, the next part of the project consisted of the proof of concept (PoC).
The PoC, also called experimentation, consists of a set of different test cases to assess the performance of the proposed solution (see Test Cases post for more information). The main conclusions derived for Rail are:
Protection Levels and Availability Performance
- For Open Sky environments, the number of subsets employed in the simulations (single SV faults and constellation faults) are enough to cope with the integrity risk without exploding in terms of extra subsets with the corresponding extra computational time.
- For Urban environment, the computed PLs are increased to cover the errors. The reduced number of satellites caused by the urban conditions and the virtual FDE employed to cope with the local effects produce a reduction in the availability.
- The applications with the most stringent requirements suffer greatly from this loss of satellites with the impossibility to find enough subsets to reach the monitoring probability of the ARAIM algorithm
Sensitivity Analysis
- Higher exposure time values led to poorer filter convergence and occasional spikes, while lower values improved filtering.
- Lower Time To Alert (TTA) values resulted in less filtering and brief divergences in HPL calculation, whereas higher values enhanced filter convergence.
- Increasing Probability of Non-Line of Sight (PNLOS) or adjusting parameters like the threshold of monitoring probability significantly increased CPU time and the number of required Kalman filters.
FDE Test and Assessment of Integrity
- The FDE does not significantly impact accuracy and PL statistics in Open Sky environments.
- However, in urban environments, the application of FDE increases protection levels due to the rejection of more satellites. This increase in protection levels is accompanied by a decrease in errors, effectively suppressing Misleading Information Events, making FDE crucial for urban settings.
- The FDE also reduces the maximum error value during error ramps. Although it does not eliminate all Misleading Information Events, it keeps errors below the alert limit.
Based on these findings, it is recommended to:
- Include additional constellations to increase the number of satellites available in urban environments. More satellites would be required to improve the ARAIM solution separation method, reducing the PLs at the expense of an additional computational load on the algorithm.
- Minimize the PNLOS as much as possible by enhancing the Local Effects model, covering the NLOS events with multipath models, and optimizing the FDE process. The exploration and usage of advanced NLOS rejection techniques are recommended to enhance NLOS detection and mitigate their effects in the algorithm.
- Consider the trade-off between IR and monitoring probability when configuring the system to avoid reducing availability or increasing computational time.
