Jackknife ARAIM: Efficient GNSS Integrity Monitoring for Simultaneous Faults under Non-Gaussian Errors

TL;DR

This paper introduces a jackknife-based integrity monitoring method for GNSS that effectively detects multiple faults under non-Gaussian errors, offering improved accuracy and computational efficiency over traditional Gaussian-based approaches.

Contribution

The paper develops a novel jackknife ARAIM algorithm that handles non-Gaussian errors and multiple faults more efficiently than existing methods, with proven performance and real-world simulation validation.

Findings

Reduces vertical protection level below 45 m in GPS-only scenarios.

Maintains VPL below 40 m in GPS-Galileo dual-constellation scenarios.

Achieves up to 59.4% faster processing time compared to SS ARAIM.

Abstract

Legacy and advanced receiver autonomous integrity monitoring (RAIM/ARAIM) rely on Gaussian error models that can be overly conservative for real-world non-Gaussian errors. This paper proposes an extended jackknife detector capable of detecting multiple simultaneous faults with non-Gaussian nominal errors. Furthermore, an integrity monitoring algorithm, jackknife ARAIM, is developed by systematically exploiting the properties of the jackknife detector in the range domain. We prove that the proposed method has equivalent monitoring performance with the solution separation (SS) ARAIM, but is significantly computationally efficient for single-fault cases with non-Gaussian nominal errors, while maintaining similar efficiency to SS ARAIM for multiple-fault cases. The proposed method is examined in worldwide simulations, with the nominal measurement error simulated based on authentic experimental data, which reveals different findings in existing research. In a single Global Positioning System (GPS) constellation setting, the proposed method can reduce the 99.5 percentile vertical protection level (VPL) below 45 m, outperforming 50 m VPL produced by the ARAIM algorithm using Gaussian nominal error models. In GPS-Galileo dual-constellation setting, while these Gaussian-based ARAIM algorithms suffer VPL inflation over 60 m due to Galileo's heavy-tailed errors, the proposed method maintains VPL below 40 m, achieving over 92% normal operations for 35 m Vertical Alert Limit. Moreover, we tentatively implement the SS ARAIM using non-Gaussian overbounds and compare it with the proposed Jackknife ARAIM method regarding computation efficiency. The proposed method achieves up to 59.4% reduction in median processing time compared to SS ARAIM in single-constellation scenarios.