198 / 2023-09-12 09:46:19

Smart phone GNSS data quality analysis and stochastic model establishment

Smartphone,Data quality analysis,SNR model,Parameter fitting,Positioning accuracy

As commonly used electronic devices, mobile smart terminals such as smart phones have functions such as location sharing and navigation and positioning, which greatly facilitates people's daily life. Therefore, this paper takes the two common smart phones on the market Xiaomi Mi 8 and Huawei Mate 20 as data acquisition terminals, analyzes the observation data quality of different satellite systems from the perspectives of SNR and single errand pseudorange residuals, and then determines the parameters of the SNR model according to the difference in the pseudorange observation quality of different satellite systems on the smartphone side, and compares and analyzes them with the height angle stochastic model.

In terms of raw data quality, the SNR distribution of the Xiaomi Mi 8 data is slightly better than that of the Huawei Mate 20. The latter's SNR data are all integers. This may be related to the navigation chip data collection and operating system processing mechanisms of different smartphones. Moreover, the single-satellite SNR of the two mobile phones shows periodic enhancement and weakening in the time series. This is related to the linearly polarized antenna used inside the mobile phone, which makes it less capable of suppressing multipath effects. Through analysis, the pseudorange accuracy of the BDS B1I and GPS systems seen by the two mobile phones is relatively close, while the GLONASS pseudorange accuracy is the worst. The L5 frequency pseudorange accuracy of GPS is slightly higher than the L1 frequency. The pseudo-range accuracy saw by Mi 8 is slightly better than that of Huawei Mate 20. Except for the GPS L5 frequency, the pseudorange residuals show a stronger correlation with the SNR than the altitude angle. The GPS L5 frequency may be due to the weak anti-interference ability. The pseudorange residuals are closely related to the SNR and altitude. It is difficult to show correlation between angles.

Regarding the parameters of the SNR model, it is known from experiments that the relationship between the pseudorange residual and the SNR of Xiaomi 8 mobile phone is stronger than that of Huawei Mate 20. Therefore, when performing parameter estimation, the results fitted by Xiaomi 8 are more consistent with the normal distribution. According to the 3-sigma principle of median error, the fitting effect is better, while the fitting result of BDS B1I is slightly worse than that of GPS L1 and GLONASS G1 frequencies. The fitting values of different system frequencies are like the accuracy of the pseudorange residuals, and the fitting values of the parameters of GLONASS with the lowest pseudorange accuracy are larger.

In the pseudo-range single point positioning results, the weighting results of the SNR model using parameters fitted by different systems are better than the altitude angle model, and the positioning accuracy in the plane and elevation directions is about 2.37m and 8.38m respectively. Compared with the height angle model, the angle model is improved by 10.23% and 9.11% respectively, showing that the SNR fixed weight model is reasonable and effective.