BDS/GPS stochastic model refinement and assessment using satellite elevation angle and SNR

Abstract

Satellite elevation angle and Signal-to-Noise Ratio (SNR) are usually used as measurement quality indicators for global navigation satellite system (GNSS) measurements. The relationship of quality indicators and accuracies of measurements can be expressed as stochastic models. To model the relationship for Beidou navigation satellite system (BDS) and global position system (GPS), five basic stochastic models are presented from satellite elevation angle and SNR. Also, coefficients of these models are refined. It’s found that SNR stochastic models with same coefficients can’t treat all measurements from BDS and GPS. Moreover, stochastic models with an additive constant could model the relationship better. The performance of the five models are tested, independent and combined, in BDS/GPS precise positioning. The results show that refined stochastic models could improve the success rate of integer ambiguity single-epoch solution 8% comparing to empirical models. Models with an additive constant could improve the success rate 10% comparing to models without additive constants. SNR model with an additive constant performs better in performance for integer ambiguity resolution, especially for low elevation satellite or combined system. Using stochastic models with an additive constant, ratios of posteriori and prior variances are closer to 1 in precise positioning. Therefore, for the used receivers, we suggest to choose refined stochastic models with an additive constant, and give priority to SNR model. Here, a refinement and assessment method is proposed to derive proper stochastic models for GNSS data processing, taking into account the differences between navigation satellite systems (e.g. BDS and GPS) and stochastic models.