Study on Improving GPS Measurement Accuracy

Abstract

By analyzing a number of GPS data, a principle was found that GPS single point location data was composed of position excursion and normal distribution. In order to separate the position excursion from normal distribution, normality elimination method was advanced. First, calculated average latitude and longitude coordinate of GPS data as central coordinate of normal distribution. Then, calculated confidence interval [-sigma, sigma] and removed excursion points outside of it. Finally, calculate a radius of the circle that ensure 70~80% points falling in the circle, namely defined threshold lambda as 0.7~0.8. By reason that the property of 70~80% points was able to take the place of all points, the method could be feasible and the result was obvious. Aimed at the location error of GPS single point measurement was large, a new method of two points relative location measurement was put forward which used two GPS receivers to acquire location data separately and simultaneously at two endpoints of a certain distance. The experiment results indicated that when threshold lambda was 0.7, the previous measurement error of single point location experiment was above 10m. However, after using the new method, the location error was about 1.2m. It was turned out that the latter improved measurement accuracy greatly