Between-satellite ambiguity resolution based on preliminary GNSS orbit and clock information using a globally applied ambiguity clustering strategy

The use of undifferenced (UD) processing schemes of GNSS measurements is becoming more and more popular for the generation of global network solutions (GNSS orbits and clock products) within the GNSS community. As opposed to classical processing schemes, which are based on a two-step approach where the orbits (generally, the contributions to the observation geometry) are estimated in a double-difference (DD) scheme while leaving the estimation of the corresponding clock information (and other linear terms) to a second, independent UD procedure where the orbits are introduced as known, the newer designs combine both parts into a single, compact processing scheme. Although this offers a higher flexibility, some challenges arise from the handling of the many parameters, as well as from the implementation of robust ambiguity resolution (AR) strategies. The latter could lead to a prohibitive computational time for a growing size of the network due to the large amount of ambiguity parameters. To overcome that issue, we propose a new UD-AR strategy that adapts the DD-AR approach. This is accomplished by carefully inspecting the real-valued ambiguities in a stand-alone step, where the DD-AR information is explicitly considered through the use of ambiguity clusters. As a result, the preliminary satellite orbits and clock corrections are modified to become consistent with the integer-cycle property of the carrier phase ambiguities, allowing to resolve them as integer numbers in a computationally inexpensive station-wise parallelization. This strategy is introduced and explained in detail. Moreover, it is shown that the GPS and Galileo solutions generated by this procedure are at a competitive level compared to classical DD-based solutions.