INFLUENCE OF TRAVEL DIRECTION ON GPS ACCURACY FOR VEHICLE TRACKING

Abstract

The influence of travel direction on GPS dynamic accuracy for vehicle tracking is discussed in two sections. The first section investigates the influence of travel direction on GPS accuracy due to the GPS satellite sky distribution. GPS dilution of precision (DOP) was calculated based on GPS satellite geometry at a variety of locations and different mask angle settings. Results show a significant difference between north DOP and east DOP in a mid-latitude area. A clear trend of the 24 h average ratio of the north DOP to the east DOP was found related to latitudes and mask angle settings. Cross-track dilution of precision (XDOP) is defined as the GPS DOP perpendicular to the travel direction. The influence of the GPS satellite geometry on GPS accuracy was mapped into the vehicle platform frame to derive the XDOP, and accordingly to derive the influence of travel direction on the GPS dynamic accuracy. Results showed that the XDOP increased as the course over ground (COG) changed from 0° to 90°. Considering that a regression line fitting through GPS data may be referenced as the true path for calculating GPS errors, the second section reviews methods for fitting linear models. The most commonly used approach for linear fitting is least-square regression that minimizes the sum square of vertical offsets, rather than perpendicular offsets. This approach can result in a potential model fitting error, which was found to be dependent on the direction of travel and the dynamic accuracy of the tested GPS receiver when this approach was used to generate the referenced true path for calculating GPS cross-track errors. Our results showed that the fitting error reached its maximum when the tested vehicle was traveling in the N-S (or S-N) direction and decreased when the travel direction moved away from the N-S direction.