A Post-Processing Methodology for Improving Kinematic Consistency in GNSS-Based Running Trajectories

The use of Global Navigation Satellite Systems (GNSS) for recording sports trajectories has increased significantly, particularly in endurance activities such as running. Although widely adopted, consumer-grade GNSS devices produce kinematically inconsistent measurements due to positional noise and motion-induced artifacts, thereby limiting the accuracy of derived metrics such as speed and slope. This paper presents a post-processing methodology to improve measurement accuracy by integrating sequential filtering, temporal interpolation, and outlier detection. It was validated using 36 trajectories of repeated running laps along a predefined track path, recorded under controlled conditions with eight consumer-grade GNSS devices, and evaluated against the corresponding ground-truth trajectory. Results indicate substantial improvements in kinematic consistency, including a 90.6% reduction in slope root mean square error (RMSE) and a 34.6% decrease in out-of-range points. Consequently, the accuracy of speed and slope signals also improved. The proposed method provides a practical and scalable solution for improving the reliability of GNSS-derived running metrics for performance analysis and planning.