Abstract

Global Navigation Satellite Systems (GNSS) provide reliable location tracking for vehicles, but their accuracy can degrade in challenging environments such as urban canyons or tunnels. Traditional methods struggle to maintain precision under multipath interference and signal obstruction. To address this, a deep assisted attention mechanism is proposed, enhancing GNSS tracking by dynamically weighting input signals based on their relevance. The method integrates deep learning and attention modules to filter noise and amplify critical features from the GNSS data. Experimental results on real-world datasets show a significant improvement in tracking accuracy, with a reduction in position error from 15 meters to 3 meters under challenging conditions. Additionally, signal loss recovery improved by 40%, further enhancing the system''''s reliability. These results demonstrate the model''''s potential to significantly enhance vehicle tracking in harsh environments.