Dependency-Aware Multi-Reference Fault Monitoring for Dynamic Positioning Under Common-Mode GNSS Degradation

Abstract

Two global navigation satellite system (GNSS) receivers using differential corrections can agree while supplying a shared position bias to a dynamic-positioning controller. We evaluate a dependency-aware multi-reference monitor (DA-MRIM) that combines source residuals, a configured GNSS-group test, temporal confirmation and recovery. Laser and hydroacoustic references provide evidence outside the GNSS group. Tests use two synthetic closed-loop plants with correlated receiver errors and asynchronous measurements. Snapshot comparators share the same dependency information; one updates decisions only on new acquisitions. Under a common-mode bias ramp, DA-MRIM reduces estimation root-mean-square error (RMSE) relative to this event-driven comparator by 0.0499 m on the linear plant and 0.0456 m on the nonlinear plant. Paired 95% intervals are 0.0325–0.0647 and 0.0316–0.0578 m. Compound-fault gains vary across the plants, and lower estimation error does not consistently reduce vessel excursion. A biased laser can cause repeated exclusion of healthy GNSS receivers. Coordinated monitoring, therefore, improves estimation for the tested common-mode fault, while fault attribution and timing remain important costs. Vessel-reference data and hardware testing are needed for operational assessment.