Abstract:
The two key factors in achieving high-precision positioning results from GPS carrier phase observations are the data differencing technique and the ambiguity resolution process. A double differencing technique has been widely used to reduce biases in GPS observations. However, un-modeled biases can persist, and these reduce the number of ambiguity-fixed solutions, especially in GPS kinematic positioning mode. Therefore, noisy or unwanted GPS satellites must be identified and removed from the data processing step. This requires users to edit or filter noisy data in the processing step. Data filtering is essentially a trial-and-error process, with users reprocessing the data until satisfactory ambiguity-fixed solutions are obtained. Often, data filtering is a time-consuming process that requires the skills of an experienced user. In this research, a genetic algorithm (GA) was developed to overcome the aforementioned problem. This GA is used to optimize the selection of GPS satellites in order to improve the number of ambiguity-fixed solutions in GPS kinematic positioning mode. Experimental results demonstrate that the GA can enhance the number of ambiguity-fixed solutions without user-assistance in the data processing step. However, the experiment did not use practical kinematic data. Therefore, further investigations were conducted to improve the ability of the GA. The sliding windows technique was integrated with the GA to enhance the number of ambiguity-fixed solutions by varying the finite-length selection windows in the data processing step. This new proposed methodology was tested with raw GPS kinematic observation data from a practical mobile-mapping system. The results indicate that the proposed method produces considerably more ambiguity-fixed solutions than either the GA alone or the standard GPS kinematic post-processing. The experimental results clearly demonstrate that the proposed methods can be used to improve the ambiguity-fixed solutions in GPS kinematic positioning mode without the need for user-assistance in the data processing step.