Fault Detection and Correction Using Observation Domain Optimization for GNSS Applications

Abstract

Global Navigation Satellite System (GNSS) is ubiquitously used and integrated into a variety of applications that require accurate and reliable positioning, navigation, and timing (PNT). The rapid growth in research and development into autonomous and semi-autonomous land and aerial vehicle platforms in recent years has redefined industry standards for accurate and reliable PNT. To ensure the integrity of a PNT solution, effective fault detection and exclusion/correction (FDE/C) is needed. Least-squares residuals (LSR) and solution separation (SS) are two well-known receiver autonomous integrity monitoring (RAIM) methods. LSR is computationally efficient but is not applicable, nor is theoretically correct, in scenarios where multiple faulty observations are present. While SS is effective for detecting and isolating multiple faulty observations at a time, it has high computational complexity, hence not suitable for most real-time applications. Other existing fault classifier methods lack the industry required performance due to either data generalization and/or high computational complexity. A novel scalable multi-fault detection and correction method is presented here. As demonstrated by our analysis and test results based on both simulated and real data, the proposed method outperforms LSR providing a more accurate PNT solution and is 80% more computationally efficient than SS under nominal multi-constellation scenarios with 30 or more satellites used in the position estimation.