The orbit and clock errors of GPS satellites are examined culminating in error distributions and Gaussian overbounds. The Precise Orbit and Clock (POC) data are then used to produce a Precise Point Position (PPP) solution with a maximum vertical error of ±1 meter. Flight test data are processed using the Single Difference Residual (SDR) method to detect and correct cycle-slips. The corrected carrier phase measurements are then used in PPP to determine the user's position. The results are compared with truth data to characterize the PPP solution accuracy performance.
POC data from the International GNSS Service (IGS), National Geospatial-intelligence Agency (NGA), and the Center for Orbit Determination in Europe (CODE), are used as the truth reference and compared to the broadcast GPS ephemerides obtained from the Scripps Orbit and Permanent Array Center (SOPAC). The error is then analyzed over the three year period from June 2005, through June 2008. Two hundred forty-seven events in which the error in orbit or clock exceed 25 m are identified. These events are classified as outliers. The outliers are compared against data from IGS station receivers to determine if the event is observable. Observable events are reclassified as anomalies. Thirteen anomalies are identified and tabularized to enable future researchers to test anomaly detection mechanisms against historical events. The table of anomalies represents an original contribution to existing research in this field.
The anomalies are removed from the data. The remaining data are analyzed statistically. Seven error metrics are analyzed. Error distribution and a Gaussian overbound are provided by error metric and satellite block type.
The error data are then differenced over time. Three Operational Scenarios (OSs) are analyzed. OS1 is the normal operating procedure in which each ephemeris is used as soon as it is received. OS2 uses only one ephemeris. In OS2 the ephemeris is used from the time it is received until it expires, or until the satellite health is set to unhealthy. OS3 uses each ephemeris as it becomes available but the transitions are made continuous. OS2 and OS3 require that the duration of the anomaly be extended. These data are tabularized separately. Distributions for the error data and Gaussian overbounds are provided. Distributions are provided by error metric, block type, and time interval.
Flight test data are preprocessed using the SDR method to detect and remove cycle-slips before performing a PPP solution. The SDR method requires five satellites and can detect a cycle-slip on one satellite. This is done by comparing the precise average velocity residual for each solution with one satellite removed. Removal of a cycle-slipped satellite removes the clock velocity error from the solution and allows the offending satellite to be identified.
A PPP solution is generated for the SDR processed data. The PPP solution provides errors of less than 1 m maximum vertical. Maximum horizontal errors are less than 0.35 m two-dimensional.