Optimal Independent Baseline Searching for Global GNSS Networks

An $n$-station continuously operated global navigation satellite system (GNSS) network contains $n\text{-}1$ independent baselines. Baseline structure is critical to positioning accuracy, and the final result is dependent on the baseline selection strategies. The baseline length and amount of common observations are the primary principles for baseline selection. However, there are few discussions about the optimal strategy to determine the independent baseline of a huge GNSS network. To enhance the performance of the multibaseline solution, a comparison is drawn between the conventional method and a weighting strategy. Observations from continuous stations distributed globally within the International GNSS Service (IGS) are explored. At first, two conventional principles for baseline selection are tested. Subsequently, a weighting scheme is developed to exploit these two strategies. The enhanced method improves nearly 10% external accuracy compared with the classical methods, which can be verified from the experiment on January 1, 2012. Lastly, the network experiment is extended to the whole year of 2012 to increase statistical significance. It is therefore revealed that the novel weighting strategy (WEIGHT), with an equal chance of two conventional strategies, mitigates 0.4%–3.0% three-dimensional (3D) coordinate error of the whole year. Also, an analysis of the probability of gross errors indicates that WEIGHT exhibits better performance. Unlike the conventional view, it is shown that a proper weight of OBS-MAX and SHORTEST could form a better coordinate calculation result and a lower gross error rate. In conclusion, these experiments suggest a proposed method that synthetically considers the length of total stations and the total number of observations, and it is verified that WEIGHT is a better choice for searching independent baselines.