Multimodal Data fusion for SRGPS antenna motion error reduction

Abstract

Antenna motion is a primary fault that degrades the integrity of shipboard relative GPS (SRGPS) systems, so we must investigate how to monitor and mitigate its impacts. Previous work proposed a single-baseline mode, but it had obvious problems, such as large motion errors, error measurement difficulties, and poor reliability. To solve these issues, we propose a multiple reference station architecture that deploys more than one reference station at different positions on ship. The observations are first translated to the ship reference point, and then integrated into a comprehensive measurement that accounts for the overall antenna motion errors. We consider two integrating methods: direct and weighted average methods. The direct average method is simple and intuitive, but the weighted average method more effectively reduced the overall antenna motion error by using the reciprocal standard deviation of historical measurements as weighting factors. All the antennas are mounted on the same ship body, so their correlations should also be considered when determining weighting factors. We used the ranking scores of the antennas derived using a graph-learning algorithm and fused them with the standard deviations to create new weighting factors. Finally, our experimental results demonstrated that the weighting factors based on the standard deviations and ranking scores reduced the overall antenna-motion error variance and improved the system integrity.