Research On Multi-frequency GNSS Precise Orbit Determination And Low Earth Orbit Satellite Enhancement

Currently,the global navigation satellite system(GNSS)is developing rapidly.Following the operation of the full constellation of GPS and GLONASS,BDS and Galileo will soon complete the deployment of the whole constellation.Modernized GNSS will transmit the navigation signals at least three frequencies.There are many ground stations receiving the triple-frequency(TF)signal transmitted by an increasing number of satellites,which brings new opportunities and challenges to GNSS's precise data processing.GNSS enhanced by the low earth orbit(LEO)satellite has become a current research hotspot.To improve the orbit accuracy of navigation satellite,this thesis focuses on the key issues of GNSS data quality analysis and control,multifrequency satellite precise orbit determination(POD),uncombined(UC)POD model refinement,LEO/GNSS combined POD.The main contribution and conclusions are as follows:1.BDS-3 observation data quality and the spaceborne atomic performance are evaluated.The signal noise ratio(SNR),pseudorange noise and multipath error of new and old signals of BDS-3 satellites are analyzed and compared with BDS-2/GPS/Galileo based on iGMAS data.The SNR of old signals of BDS-3 is slightly better than BDS-2,while for the new signals the SNR of B1C/L1/E1 is lower than other frequencies and GPS satellites are the best.The noise and multipath is comparable among GNSS.On the basis of analyzing the system noise of POD clock products,automatic batch processing is implemented for BDS/Galileo/GPS clocks data preprocessing and atomic clock performance evaluation.The overall performance of BDS-3 rubidium clock is better than that of BDS-2 satellites.Most indicators of the BDS-3 hydrogen clock are comparable to Galileo hydrogen clock.The frequency stability with the integrating time of 10000 s is 2.49E-14 and 2.32E-14 for the BDS-3 rubidium and hydrogen clocks,with the integrating time of 86400 s is 8.64E-15 and 5.28E-15,correspondingly2.The resolution of the datum problem of GNSS POD is summarized.On one hand,the coordinate datum needs to be given to fix the network in a certain reference frame for ensuring with the no-net-rotation and no-net-scale condition;on the other hand,the time datum and equipment delay datum are given to resolve the rank deficit problem of GNSS observation equation.On this basis,the ionospheric-free(IF)/UC multi-frequency multi-GNSS full-rank observation equation with the clock datum being IF combination or the first frequency is deduced and an example is carried out.It shows that the orbit determination results are basically the same when using different clock datum or observation model.Using the strategy of the first frequency as the clock datum can simultaneously obtain separable inter-code bias products,but it may increase the instability for the estimation of the bias.3.UC POD strategy needs to estimate the huge amount of ionospheric delay epoch parameters,which leads to the calculation time-consuming.An improved parameter elimination and recover algorithm is proposed.The ionospheric delay is eliminated timely once a group of station-satelliteepoch observations is overlaid,where there is no matrix inversion for the elimination of an ionospheric parameter.Experimental results show that the new algorithm can improve the calculation efficiency by several time or even tens of times in one iteration of least square adjustment,comparing to the conventional epoch parameter elimination strategy.4.The dual frequency(DF)undifferenced IF/UC POD model and its ambiguity resolution(AR)method are studied.It is theoretically proved that when using the “wide lane(WL)-narrow lane(NL)” AR strategy,the difference of the two re-parameterized observation equations can be eliminated and the equivalent AR results can be obtained.POD experiments with GPS and BDS-3 shows that the accuracy between the two strategies is comparable and the difference of POD products are compared from multiple aspects.The difference of orbits and clocks for GPS is within sub-mm and 1 ps,and for BDS-3 is within mm and 1-10 ps.5.The TF undifferenced IF/UC POD model as well as its AR method are proposed,and the contribution of the third frequency observations for GNSS POD is assessed.Aiming at the problem of existing time-variant carrier phase bias for GPS satellites,the bias is divided into time-variant and-invariant components and the TF POD model is deduced.The TF AR method of doubledifference(DD)strategy suitable for POD is developed.The GPS IIF,BDS-2 and Galileo satellites that can transmit TF signals are used to validate the new model.The improvement of POD products accuracy with the third frequency observations added is little(less than 5%).Meanwhile,the improvement of GPS TF POD vs.L1/L2 DF POD can reach about 10%,since L5 has a higher chip rate and signal power than L2 frequency.6.The estimation of the third frequency bias as well as the AR optimization strategy are studied.The TF UC POD model(not)considering the time-variant component of carrier phase bias from the receiver is deduced.Experiments show that the component can be basically neglected.The parameter estimation strategy of station-satellite integration/separation for the third frequency bias is discussed.Experiments show that the POD results are comparable for the two strategies,but the separation strategy is recommended in the case of network solution to reduce the estimated parameters.The AR method that WL DD ambiguity is derived by using UC float ambiguities instead of the Melbourne–Wübbena(MW)strategy is proposed.The WL residual distribution shows that the new method has higher precision than the conventional MW strategy.The update strategy of TF AR is improved.Only using the frequency 1/2 DF IF combination to update the ambiguity information does not consider the fixed information from the third frequency.The “matrix-solution” AR update strategy is developed.7.Estimation of satellite antenna phase center offset(PCO)using UC model is proposed.First the DF data verification is carried out.With GPS data of the whole year of 2018,the test results show that the PCO values obtained by the new method(after IF combination)are comparable to these of using the traditional IF method.The internal check accuracy of POD results using the new PCO product are basically the same.Aiming at the problem of serious correlation between the two frequency PCOs,a prior constraint method using IGS antenna product(IF combination)is proposed.The results show that the improved method reduces the STD of PCOs by about 20% and 60% in horizontal and vertical directions.And the new PCO product can achieve the superior POD accuracy.In view of the fact that the antenna file released by IGS only deal with the PCO from two frequencies,a new method that only the third frequency PCO is estimated by using the TF UC POD model is developed.GPS test results show that the POD results using the derived L5 PCO are better than using the L2 PCO,which is a further step for the refinement of TF GNSS model.8.The experiment of LEO enhanced GNSS POD with real and simulated data is studied.Based on three LEO satellites and 20 regional ground stations,the influence of LEO difference in case of different number of stations on LEO/GNSS POD is assessed.Results indicate that the difference of LEO orbit height and inclination angle has little effect on POD accuracy,but with a small number of stations,the influence of different two LEOs on POD accuracy can reach cm level.In view of the current status of the regional monitoring stations in China as well as the ease of deployment and management of miniaturized LEO constellation,a constellation consisting of 24 LEOs is simulated to enhance BDS-3.Simulation results show that the combined POD with LEOs can improve the POD accuracy of various BDS-3 satellites up to cm level.