Local Statistical Testing in Quality Control of GNSS Observations

Abstract

The paper discusses GNSS statistical-testing methods for detection and identifications of observation outliers at the estimation "current" epoch, defined here as local testing. First, detection methods are discussed including testing the probability distribution (likelihood) of the residuals, testing using dynamic control limits of the range, the mean and standard deviation of the residuals. A method is proposed examining the difference between the mean and the median of the residuals. To identify the satellites with faulty measurements, several methods were investigated, including checking the likelihood of the residuals at the present epoch, and in a time series. A test is presented utilizing the control limits of the residuals' moving range for each satellite. Testing of the proposed methods was carried out using only GPS phase measurements in the kinematic mode. Results show that the proposed methods are efficient for detection and identification of large errors/outliers. However, the performance degrades with error values less than 5 cycles and when using small significance levels.