Zebo Zhou, Yunzhong Shen and Bofeng Li
Doppler measurement is popularly applied to estimate the vehicle's velocity in GPS navigation and its stochastic characteristics attach a great importance to the quality of velocity estimate. In this contribution, firstly a single differenced (SD) Doppler observation is established with the known velocity. Secondly, the precision of un-differenced Doppler measurement is derived from SD observation model. Based on the model above, the stochastic characteristics of Doppler measurements are evaluated, including elevation-dependent standard deviation, cross correlation between-frequency as well as the time correlation of each frequency. The real Doppler data collected on ultra-short baselines using six types of receivers for a few hours are carried out to evaluate and analyze the Doppler stochastic characteristics. The results show that the precision of Doppler measurement is obviously dependent on satellite elevation angle. Doppler measurements between dual-frequency are evidently cross-correlated and time correlation is proven nonexistent for all GPS receivers. Based on these experiment results, an elevation angle dependent Doppler stochastic model is proposed and fitted with curve fitting tool of Matlab. At last, both static and kinematic tests are carried out to demonstrate the efficiency of our proposed stochastic model. The results indicate that the proposed Doppler stochastic model significantly improves the velocity accuracy and indirectly contributes to GPS positioning solutions
