Undifferenced and uncombined GNSS approach for absolute and relative POD of LEO satellites in formation flying

Low Earth orbit (LEO) satellites are widely used in space missions such as satellite gravimetry, radio occultation, Earth monitoring, and in formation flying. Precise orbit determination (POD), in either absolute or relative modes, is an essential prerequisite for these missions. The onboard collected signals of the global navigation satellite system (GNSS) are used for the POD of LEO satellites. Typically, the Ionosphere-free (IF) combination of these signals is used in the absolute LEO POD, which has some disadvantages. Firstly, the observation information is wasted in constructing IF observation. Secondly, although different IF combinations can be constructed in multi-frequency scenarios, they may be correlated. Thirdly, integer ambiguity resolution (IAR) based on the IF observations can only be achieved with precise external products, which has limited availability in space. Concerning relative POD, the most classical method is the double-differenced (DD) model with IAR, which also has drawbacks. Firstly, strict common-view GNSS satellites are needed, which is not guaranteed in the complex space environment with the high dynamic of the satellites. Secondly, the opportunity to impose dynamic constraints on the eliminated parameters is lost during differencing. Therefore, in this contribution, we focus on the use of undifferenced uncombined (UDUC) observations and propose a new POD model. The UDUC POD model has several advantages. Firstly, the common-view GNSS satellites are only used to form the DD ambiguities for IAR; therefore, there is no need for the external satellite phase bias (SPB) products. Secondly, the model shows flexibility in multi-frequency scenarios. Thirdly, and more importantly, as precise GNSS orbits and clocks products are used, the model can be used for both absolute and relative LEO POD. Based on onboard GPS observations of a formation flying mission, comprising two closely spaced LEO satellites working in a formation, we explored the performance improvement of the proposed model over traditional models. Two conclusions can be drawn. Firstly, the proposed model's performance in absolute POD is improved by more than 20% compared with the classical IF POD method, when computing their differences with reference orbits. Secondly, for relative POD, it is proposed that the consistency between the model and the reference orbit is within 1.5 mm, proving that the method can serve for formation flying missions. The above results show that the proposed UDUC POD method can achieve higher POD accuracy than the traditional methods.