Partial Integer Ambiguity Resolution For RTK Positioning Using Multi-constellation GNSS Signals

This thesis focuses on the carrier phase partial ambiguity resolution and validation under the multi-constellation multi-frequency navigation system.First,regarding the problem that the atmospheric delay error cannot be canceled in the the Real Time Kinematic(RTK)positioning under medium long baseline scenario,we put the ionospheric and tropospheric delay as unknown parameters in the observation equations,and provide the Kalman filter process to resolve the equations.To study the relation of measurement noise and the satellite elevation,we make specific linear combinations of the observables which separate the measurement noise from the observations.Using this technique,we build the elevation weighting model for the satellites.The experiment test shows that in the pseudo-range positioning,using the elevation weighting model can improve the positioning accuracy.Second,we introduce several commonly used ambiguity resolution methods and the calculation of their success rate.We also studied the choice of the threshold of the ratio test,in order to control the failure rate of the ambiguity resolution.We proposed fitting functions to calculate the threshold of the ratio test,and provided the functions for 15 different tolerable failure rates and 66 different dimensions of ambiguity vector.Through the simulation test,it shows that the fitting functions can provide proper threshold of the ratio test: when the model is weak,it can control the failure rate tightly,while the model is strong,it can avoid the unnecessary false alarm rate.For the first time,we used the real data test to demonstrate that Fixed Failure-rate Ratio Test(FFRT)can avoid unnecessary false alarm rate,and improved the availability of high accuracy baseline solution with about 30%.Moreover,in order to control the conditional failure rate,we proposed the Fixed Conditional Failura-rate Ratio Test(FCFRT).Using the Monte Carlo simulation,we find the relation of the threshold of the ratio test and the conditional failure rate,and find the best fitting function.The simulation test and real data test demonstrate that FCFRT can control the conditional failure rate being lower than the tolerable value.Third,under the multi-constellation multi-frequency navigation system,in order to solve the problem that in the medium long baseline scenario the carrier phase ambiguity cannot be correctly fixed,we proposed a partial ambiguity resolution method–Two Step Success Rate(TSRC)method.This method selects the subset of the ambiguities using the precision gain number,aiming at maximize the precision gain number among all possible subsets.Through Monte Carlo simulation,we provide the expecting fix rate for the first step of TSRC,and analyzed the effect of using this expecting fix rate.The result shows that by maximizing the precision gain number,it can improve the availability of high accuracy baseline solution.Besides,to control the failure rate of TSRC,we give the relation among the tolerable failure rates for the first step,second step,and the total tolerable failure rate.In both steps,we use fitting functions to control the failure rate,and it can control the total failure rate under the tolerable value.Meanwhile,the fix rate is nearly 100%.Therefore,TSRC can provide high reliability and high availability of high accuracy baseline solution.Through the simulation test,we studied the influence of the key parameters in TSRC to the availability of baseline solutions.Based on the result of the simulation study,the optimal parameters were proposed.In the end,TSRC and classic PAR strategy Success Rate Criterion(SRC)were compared with respect to the availability and convergence time using data from both simulation and real world experiments.The simulation results show that under the multi-GNSS multi-frequency and medium long baseline scenarios,TSRC can significantly improve the availability of baseline solutions(up to 40%),compared to SRC.Under the short baseline or long baseline scenarios,the improvement of availability brought by TSRC to SRC is not significant(within 10%),due to the very strong or very weak models.The experiment results show that under the dual-GNSS dual-frequency and medium long baseline scenario,TSRC shortens the convergence time to decimeter level and centimeter level by 40-100 s and 10-70 s respectively,comparing to Full Ambiguity Resolution(FAR),and by 100-300 s and 100-400 s respectively,comparing to the classic partial ambiguity resolution method.Fourth,the GNSS constellations will be completed within the near future,and the number of broadcasting signals will be increased.Under this situation,the RTK positioning with short baseline will face the challenge such that the number of ambiguities are too many to resolve within a short time interval,as the computational burden increases dramatically with the increase of ambiguity number.Therefore,how to selectively resolve ambiguities within short time and keeping a high success rate is a crucial problem for a real time RTK application.This thesis first studied the relation of success rate and computation time with the dimension of ambiguity vector.Then with the aim of shortening the computation time and keeping a high success rate,we proposed a PAR strategy and the optimal subset sizes.Under the dual-GNSS dual-frequency and dual-GNSS triple frequency scenarios,the optimal subset size is 10.The processing results of data from several baseline experiments show that the success rate by the proposed PAR is almost identical to FAR,while the computation time is shortened by 5-6 times.In the end of the thesis,we summarized the findings and contributions,and gives the future research plan.