An Investigation Of Key Technologies Related To Combining BDS-2 And BDS-3 Observations In Data Processing

Relying on the characteristics of all-weather,global-coverage,high-accuracy and continuous services in Global Satellite Navigation Systems(GNSS),global users can obtain a three-dimensional and high-precision Positioning Navigation and Timing service(PNT).Therefore,an essential role of GNSS is presented in constructing national security,economic and other fields.China Bei Dou satellite system(BDS),with independent intellectual property rights,is developing with a strategy,called“three steps”,namely verification system(BDS-1),regional service system(BDS-2),and global coverage service system(BDS-3).Typically,high-quality satellite observations,and high-accuracy orbit and clock products are the prerequisites of global navigation and positioning services as the essential performances of system.Due to the limitations of the development process,a significant gap,compared with GPS and other mature systems,should be paid more attention.In this paper,an investigation of key technologies related to combining BDS-2 and BDS-3observations in data processing is typically analyzed and discussed,including the spatial geometry,rapidly high-precision estimation,improved orbit and clock products,and biases in observations.Accordingly,main contents,like stations selection using GDOP(Geometry Dilution of Precision)values of orbit determination,the improved strategy for station distribution by the combination of map grid zooming and genetic algorithm,ultra-rapid precise orbit determination based on DOP(Dilution of Precision)values,improved methods for BDS satellites ultra-rapid predicted clock offsets,analyses of bias parameters for BDS-2 and BDS-3 observation integrated solution are included.Hence,based on key technologies related to combining BDS-2 and BDS-3observations in data processing,main researches and its corresponding results are summarized as follows:(1)for the requirements of orbit and clock products in accuracy and timeliness,estimated by GNSS Analyses Centre(AC),and the uneven distribution of global tracking stations,especially for the estimation of rapid and ultra-rapid satellites products,a strategy to optimize and select the stations is proposed based on the GDOP values of orbit determination equations.Numbers of minimum and optimal stations for precise orbit determination are theoretically derived.According to the criterion of the minimum GDOP values in orbit determination,the stations distributions are firstly output by the global map grip for the minimum number of stations.Then,an iteration solution for the final station distribution of orbit determination is designed to improvethe data processing ability of AC for the ultra-rapid products.Results of orbit determination experiments suggest that the selected stations can obtain the improvements of accuracy and time consuming,compared with the solution of all stations,with-10% and 50%,respectively.Consequently,the proposed model can indirectly take the orbit accuracy and timeliness,and the tradeoff into consideration.(2)to further accelerate the station selection,a method is proposed by the combination of global map grip zooming and genetic algorithm.In this research,the OEDOP value is firstly defined with the accuracy of orbit and Earth rotation parameters(ERP)based on the relationship between OEDOP values and estimated parameters.Then,a strategy to optimize the spatial geometry for the estimation of multi-GNSS orbit and ERP products is designed based on the tradeoff between rapidly estimating stations distribution,and the characteristic of stations and parameters accuracy.According to different OEDOP factors(threshold values),5%,10%,15%,and 20%,experiments are indicated that the efficient of orbit determination solution can be improved with 1.96,3.32,5.27 and 6.17 times,in the condition of reducing accuracies of orbit and ERP(polar directions)by 0.33-9.92 cm and 5.77-41.53 uas,respectively.Additionally,the schemes of i GMAS tracking station distribution are also analyzed by simulation experiments to compare the availability of new strategy and exhaustive method.With the new locations of additional 17 stations based on initial 13 stations,results are shown that the time consuming can be decreased with 68 times as constructing additional 3 stations.(3)because of the lower accuracy of ultra-rapid observed orbit,especially for the last parts(3h lengths),a correction model for orbit determination is discussed in aspect of DOP values.With the aid of Akaike Information Criterion(AIC),a high-accuracy model for DOP values prediction is constructed and selected.Meanwhile,the accuracy of ultra-rapid observed orbit(for the last period)can be significant improved with the correction model by the function of DOP values in orbit determination.Moreover,the arc length of observations and the types of correction models are discussed in terms of the impacts on orbit corrections.According to experiments of orbit determination,it is suggested that the orbit accuracy can be improved with 12.35%-22.02% for the last 3 hours of ultra-rapid observed orbit.(4)considering the accuracy disadvantages of predicted clock offsets for BDS ultra-rapid products,improved models for BDS satellites clock offsets prediction are studied by the integrated processing of BDS-2/BDS-3 clocks.Firstly,a preprocessingstrategy for clock offset series not only using the combination of frequency data and Baarda algorithm is proposed,while the Tikhonov regularization algorithm is used to denoise the original series.According to results of experiments for clock offset predictions,it is shown that the accuracy of one-day clock offsets prediction can be improved with 1.0%-15.2% and 23.2%-31.9% for BDS-2 and BDS-3,respectively.Secondly,the the inter-satellite correlation between BDS-2 and BDS-3 are considered in the estimation of clock offset models.Thirdly,the PLS(partial least squares)and BPNN(Back Propagation Neural Networks)are used to reduce the model residuals.Based on the results of clock offsets prediction experiments,the improvements with30.7%-47.3% and 49.9%-59.3% for 18-hours prediction can be obtained for BDS-2and BDS-3,respectively.Additionally,for the BDS-2/BDS-3 satellite ultra-rapid orbit products,the constraints of clock offset products is introduced in orbit determination.Meanwhile,the optimal length of observed orbits for estimating the initial states for BDS orbit predictions are discussed.(5)to improve the compatibility of BDS-2/BDS-3 satellite observations,the inter-satellite bias(ISB)between BDS-2 and BDS-3 satellites is analyzed based on the combined orbit determination solutions.Firstly,to improve the accuracy and reliability of ISB estimation,the singular value decomposition is introduced to utilize the observation information between adjacent epochs.Secondly,the characteristics of ISB parameters are analyzed from the aspects of BDS-2/BDS-3/GPS integrated orbit determination,in which ISBs are significantly presented and related with the types of receivers for the different signals of BDS-3 and BDS-2.Thirdly,the correlation factors between ISB parameters and orbit parameters are calculated.With the estimated ISB series,the short-term prediction of ISB series is established by spectral analysis.According to the ultra-rapid orbit determination experiments by introduction of predicted ISB series,the orbit accuracy can be improved with –0.4-1.0cm and0.8-4.1cm for BDS-2 and BDS-3 orbit one-day overlapping,respectively.(6)to mitigate the difference between BDS-2 and BDS-3 observations,the improved model for BDS observations code bias is studied.Meanwhile,the code multipath delays of BDS observations are compared and analyzed for different stations.To correct the BDS code bias,the traditional method is mainly used as a two-step strategy,namely first multipath delay and then code bias.Therefore,an one-step strategy is proposed in this research to model and eliminate the code bias and multipath delay based on the BDS-2 and BDS-3 satellites integrated processing.In theimproved strategy,the Least Square(LS)combined with Autoregressive(AR)(LS+AR)are selected to estimate the model coefficients of code bias and construct the multipath delay with one solution,respectively.Moreover,the inter-satellite correlations between BDS-2 and BDS-3 satellites are extracted to improve the weight matrix in the estimation of model coefficients.According to the experiments,the one-step strategy can enhance the precision of B1 I and B3 I single-frequency PPP on E,N and U directions with 2.9%-31.6%,3.3%-21.4% and 0.2%-69.2%,and 2.8%-14.3%,0.0%-20.9% and 1.5%-15.3%,respectively.