Research On BDS Medium/Long-range Quickly Ambiguity Resolution And Regional Error Correction

Since the end of 2012, the China BeiDou Navigation Satellite System (BDS) is providing the regional navigation, positioning and timing services to Asis-Pacific regions users. It is the first regional satellite navigation system that capable of transmitting the triple frequency signals in the world. Many domestic and overseas scientific research institutes and companies have established lots of permanent tracking stations to collect GNSS satellite signals. Especially, some Chinese province and city level Continously Operational Reference Stations Systems (CORS) are begin to update the receivers and antennas to promote the commercial application of BDS. Moreover, all these CORS datasets need to be processed with network RTK software.Currently, the BDS heterogeneous constellations are consisted with five Geostationary Earth Orbit (GEO), five Inclined Geosynchronous Orbit (IGSO) and four Medium Earth Orbit (MEO) satellites and the presence of non-ignorable atmosphere delays. All these reasons result in the difficulty of narrowlane (NL) AR for BDS medium/long-range baselines. Thus, the users can not acquire the high-precision BDS positioning and navigation services.It is meaningful to investigate the BDS medium/long-range relative positioning and regional error correction method. This paper will research the cycle slip detection and repair, BDS multipath mitigation, reference station NL AR and the regional error correction method to promote the BDS network RTK techniques.1. A cycle slip detection and repair method for dual-frequency carrier phase measurements is proposed for the CORS data pre-processing. The method jointly uses a double-differenced (DD) geometry-free (GF) combination and ionospheric-free observation corrected for the computed geometrical distance (IF_OMC) to estimate the cycle slips The DD GF combination, which is only affected by the ionospheric residual, can be used to detect cycle slips with high reliability except for special pairs such as (77k:60k, k ∈ Z) on GPS f1/f2 frequencies. The accuracy of IF_OMC is mainly affected by broadcast ephemeris orbit error. As the triple differencing can greatly decease the influence of orbit error, the IF_OMC has the high reliability and precision to detect the cycle slips. The detection principle of the IF_OMC observable is such that there is a large discontinuity related to the previous epoch when cycle slips occur at the present epoch. The disadvantages of these two combinations can be overcome by employing the proposed detection method. The cycle slip pairs (77k: 60k, k∈Z) has no effect on the GF combination, while a change of 14.65m is derived from GPS f1/f2 observations using the IF_OMC algorithm. When the special cycle slip pairs (60k:77k, k∈Z) that immune to the IF_OMC algorithm are happened in the dual frequency observations, a discontinuous change of -7.38 m would be detected with the GF combination. The orbit error will affect the accuracy of IF_OMC method. However, if the IF_OMC combinations in consecutive epoch are computed with the same ephemeris, the orbit error’s affection would be greatly decreased when forming the triple differencing observations. Therefore, cycle slips in dual-frequency observations can be correctly and uniquely determined using DD GF and IF_OMC equations. The proposed method was verified by adding simulated cycle slips in observations collected from the CORS network under a quiet ionosphere and shown to be effective. Moreover, the method was assessed with observations made during intense ionospheric activity, which generated extensive cycle slips. The results show that the algorithm could detect and repair all cycle slips apart from two exceptions relating to long data gaps.2. We investigate the characteristics of orbital repeat period, multipath errors and multipath mitigation for BDS GEO, IGSO and MEO satellites. The movement of GEO satellites is not strictly motionless, the variations of latitude are approximately ±1.5° for both GEO C04 and C05 satellites. The orbital repeat period of GEO satellites are aproximately one sidereal day. The multipath signals from GEO satellites are not a constant but are changing slowly. Due to the low-frequency variability, the multipath signals from GEO satellites cannot be mitigated with a short observation period. The IGSO C06 and C07 satellites show a variation range from 10° to approximately 90° in elevations and about 40° in the azimuths. Its multipath signals are changing rapidly than GEO satellites. The IGSO satellite’s orbital repeat periods are closely to the GEO satellites. The high-frequency variations in multipath signals are showed in the MEO satellites. Additionally, its orbital repeat periods are about seven sidereal day. We argue that multipath errors can be mitigated using models extracted from a previous period since there is a strong agreement correlation in the multipath errors between subsequent periods. A multipath mitigation method, which is based on observation domain filtering and considers the repeat period of every satellite independently, was applied. The application of the proposed method shows that BDS positioning precision was improved by about 56%,48% and 48% to 1.90mm,1.28mm and 4.37mm in the north, east and up components.3. We propose a method which uses the single-differenced (SD) multipath mitigated geometry-free and ionosphere-free (GFIF) combination for ambiguity determination. Because of the unknown systematic errors and special satellite constellations in the BDS along with the low-accuracy of broadcast ephemries, it is difficult to quickly and reliably determine the ambiguity over long-range baselines in CORS network. This study seeks to improve the effectiveness and reliability of BDS ambiguity resolution (AR) by GFIF combination and multipath mitigation algorithm. The GFIF combination composed with three-frequency signals is free of distance-dependent errors and can be used to determine the narrow lane ambiguity. The presence of multipath errors means that not all ambiguities can be correctly achieved by rounding the averaged GFIF ambiguity series. A multipath model of the SD GFIF combination from the previous period is established for each individual satellite. This model is subtracted from the SD GFIF combination for the current day to remove the effects of multipath errors. Using three triangle networks with lengths of approximately 120km,80km and 50km, we demonstrate that the proposed method improves the AR performance. The ambiguity averaged first fixing time (FFT) is typically less than 1800 s for IGSO and MEO satellites and less than 2000 s for the ～42° elevation GEO C02 satellite. However, it is more time-consuming for the low-elevation GEO satellites C04 (～18°) and C05 (～28°). Kalman filtering is used to estimate the troposphere delays and two unfixed ambiguities by employing the ionosphere-free observations of all ambiguity-fixed/unfixed satellites. The experimental results show that only tens of seconds are required for AR in around 90 km baselines. Moreover, This estimation method with ambiguity-fixed satellite observations can be used to facilitate the GPS AR. The experimental results indicate that 22% effectiveness can be achieved for GPS AR.4. A new regional error correction method has been proposed to assist the roving station AR. Firstly, the independent baseline in a CORS are selected out to fix the ambiguities. Secondly, we can transfer the DD ionosphere-free combination residuals to the zero-differenced (ZD) residuals. Then, the error corrections of roving station can be fitted with these ZD residuals. The observations for three consecutive days in three triangles with averaged length ～100km and the three roving stations that away from the main reference 59km,50km and 57 km are carried out to investigate the performance of this method. All the datasets are processed in kinematic mode in every one hour, the averaged FFT for these three examples are below 142s. Moreover,we arrived out another actually kinematic experiment to illustrate the good performance of this error correction method. The distance between roving station and reference is about 54km. It shows that around 160s is needed for AR in this test. Both the static and kinematic experiments can be used to demonstrate the effectiveness of our proposed method.