*3.4. DGNSS/DBA Combined Kinematic Vehicle Positioning Results*

In this section, two kinematic vehicle experiments were designed to evaluate the dynamic positioning performance of the single-frequency low-cost NEO-M8T receiver and the BMP280 barometer DGNSS/DBA combined positioning algorithm.

The site photo of the mobile station equipment of the kinematic vehicle experiment is shown in Figure 11. The platform contained four multisystem dual-frequency geodetic GNSS antennas which were connected to Trimble Net R9 receivers to obtain the high precision reference value. The single-frequency, low-cost u-blox NEO-M8T receiver, a patch antenna and a BMP280 barometer were laid on the roof of the car. Two sets of kinematic vehicle data were collected, the first set was located in an open urban environment and the second set was in a complex urban environment.

**Figure 11.** The mobile station hardware equipment site for kinematic vehicle experiment, including four geodetic GNSS receivers and antennas, NovAtel SPAN-FSAS GNSS/INS system, low-cost NEO-M8T receiver with a patch antenna, and a BMP280 barometer: (**a**) the experimental vehicle; (**b**) equipment setup diagram.

#### 3.4.1. Open Urban Environment

This experiment was collected near the industrial park in Caidian District, Wuhan, China, on 20 November 2020, with a data duration of about 80 min. The area has an open urban environment with less shading and good data quality. Figure 12 shows the kinematic vehicle experiment test scene and test trajectory in the open urban environment. Figure 13

shows the number of GNSS visible satellites and the sequence of PDOP values during the kinematic vehicle experiment.

**Figure 12.** Kinematic vehicle experiment in the open urban environment: (**a**) test scene; (**b**) test trajectory.

As can be seen in Figure 13, the number of observable satellites of GPS and BDS fluctuated greatly for the kinematic vehicle, resulting in a significant increase in PDOP values compared to a static environment. The addition of DBA observations significantly improved the satellite geometric spatial distribution and reduced the PDOP values. In the open urban environment kinematic vehicle experiment, the comparison of the low-cost BMP280 barometer DBA altitude with the Trimble NET R9 reference altitude is shown in Figure 14. It can be seen that the BMP280 barometer DBA altitude had a high consistency with the reference altitude with an RMSE of 2.10 m. This value can be used to set the a priori weight matrix in the DGNSS/DBA combined positioning process.

**Figure 13.** Observation values of kinematic vehicle experiment in the open urban environment: (**a**) the number of GNSS visible satellites; (**b**) the sequence of PDOP values.

**Figure 14.** Comparison of the low-cost BMP280 barometer DBA altitude with the Trimble NET R9 reference altitude during the kinematic vehicle experimental in the open urban environment.

The RMSE statistical results and deviation sequence in the N/E/U directions for both DGNSS and DGNSS/DBA modes are given in Table 6 and Figure 15, respectively. The RMSE values of single GPS and single BDS in the N/E/U directions were 1.20/1.32/3.18 m and 1.58/2.11/5.02 m for DGNSS mode, respectively. The GPS/DBA, BDS/DBA, and GPS+BDS/DBA modes improved the RMSE by 40% to 60% in the U direction and increased slightly in the N and E directions, and the DBA altitude showed a good constraint effect.

**Table 6.** The RMSE of bias in the N/E/U directions for DGNSS and DGNSS/DBA modes during the kinematic vehicle experiment in the open urban environment.


**Figure 15.** The deviation sequence diagram in the N/E/U directions during the kinematic vehicle experiment in the open urban environment: (**a**) DGNSS positioning mode; (**b**) DGNSS/DBA positioning mode.

#### 3.4.2. Complex Urban Environment

This experiment was collected in Wuchang District, Wuhan, China, on November 20, 2020, with a valid data duration of about 1 h. The area is a complex urban environment, and the occlusion is relatively serious. Figure 16 shows kinematic vehicle experiment test scene and test trajectory in the complex urban environment. Figure 17 shows the number of visible GNSS satellites and the sequence of PDOP values during the kinematic vehicle experiment, and it can be seen that compared with the open urban environment, the number of GNSS visible satellites of the kinematic vehicle in the urban environment was significantly lower and the PDOP values became larger.

**Figure 16.** Kinematic vehicle experiment in the complex urban environment: (**a**) indicative test scene; (**b**) test route.

Figure 18 shows the DBA altitude results of the low-cost BMP280 barometer during the kinematic vehicle experiment in the complex urban environment compared to the Trimble Net R9 reference altitude. It can be seen that the DBA altitude in the complex urban environment was also very consistent with the reference altitude with an RMSE of 2.19 m, which is approximately the same as the RMSE results of the DBA altitude in the open urban environment in Section 3.4.1. It indicates that the DBA altitude accuracy was reliable and stable in different environments, and could assist GNSS to improve the positioning accuracy in the complex urban environment.

**Figure 17.** Observation values of kinematic vehicle experiment in the complex urban environment: (**a**) number of GNSS visible satellites; (**b**) sequence of PDOP values.

**Figure 18.** Comparison of the low-cost BMP280 barometer DBA altitude with the Trimble NET R9 reference altitude during the kinematic vehicle experiment in the complex urban environment.

The RMSE statistics and deviation sequence of the single-frequency, low-cost u-blox NEO-M8T and the geodetic Trimble Net R9 receiver for both DGNSS and DGNSS/DBA modes in the N/E/U directions are given in Table 7 and Figures 19 and 20. The RMSE of the low-cost u-blox NEO-M8T receiver with single GPS and single BDS in DGNSS mode were 4.33/4.69/8.35 m and 4.91/6.91/19.48 m in the N/E/U directions, respectively, and the RMSE of the GPS+BDS dual system was 3.28/5.23/8.91 m. The RMSE statistics of the geodetic Trimble Net R9 receiver in N/E direction in the complex urban environment was maintained at submeter level; the U direction was relatively poor, with RMSE not exceeding 2.7 m. The difference of RMSE between the two GNSS receivers in 3D directions was determined by their hardware performance. Compared with DGNSS mode, the RMSE of single GPS, single BDS, and GPS+BDS dual system in DGNSS/DBA mode of low-cost NEO-M8T receiver slightly worsened in the N and E directions, while the RMSE in U direction could be improved by 50% to 80%, and this improvement ratio is higher than that in the open urban environment. Similarly, for the geodetic Trimble Net R9 receiver, the RMSE of the DGNSS/DBA combination remains the same in the N/E directions compared to DGNSS mode, and the RMSE in the U direction could be improved by 30% to 60%, which verifies the advantage of DBA altitude in assisting DGNSS positioning.

**Table 7.** The RMSE of bias in the N/E/U directions for low-cost u-blox NEO-M8T and geodetic Trimble Net R9 receivers in the complex urban environment.


**Figure 19.** The deviation sequence of low−cost NEO−M8T receiver in the N/E/U directions during the kinematic vehicle experiment in the complex urban environment: (**a**) DGNSS positioning mode; (**b**) DGNSS/DBA positioning mode.

**Figure 20.** The deviation sequence of geodetic Trimble Net R9 receiver in the N/E/U directions during the kinematic vehicle experiment in the complex urban environment: (**a**) DGNSS positioning mode; (**b**) DGNSS/DBA positioning mode.

It can also be seen that when comparing Tables 3–5 for the static experiment with Tables 6 and 7 for the kinematic experiments, a low-cost receiver with only BDS signals provided lower RMSE values than GPS in static experiments, which was mainly due to BDS having more observation satellites and the PDOP value being lower than GPS at this time. In open and complex urban kinematic experiments, as shown in Figures 13 and 17, BDS had very large fluctuations for the number of observable satellites and a larger average PDOP value (3.02 and 4.10) than GPS (1.97 and 3.16) for both cases. This is the main reason that a low-cost receiver with only GPS signals provides lower RMSE values than with BDS in kinematic experiments.

#### **4. Discussion**

As the low-cost single-frequency GNSS receivers dominate most of the GNSS market [40], there is a strong interest in enhancing their accuracy. Low-cost DBA altitude plays a significant constraining role in improving the DGNSS positioning accuracy.

In the DBA altitude accuracy evaluation experiment, BMP280 barometers can achieve better than 2 m altitude accuracy within 10 km baseline lengths in static environments. The DBA altitude consistent with GNSS reference altitude in Figures 14 and 18 implies that it is also reliable and stable in complex environments. Low-cost single-frequency GNSS receivers with a patch antenna have become increasingly popular due to their lower and lower price. The DGNSS positioning accuracy of single-frequency low-cost GNSS receivers can still meet the submeter positioning accuracy needed by the general public in GNSS-friendly environments. However, the RMSE in the N/E/U directions are all at the meter-level in complex urban environments since low-cost GNSS receivers have poor observation quality, and the positioning accuracy of GPS+BDS dual system is significantly improved compared to single system. There is only a single-epoch resolution algorithm rather than a filtering algorithm is used in this study. In the future, with more and more satellites available for low-cost GNSS receivers and the use of multiple filtering algorithms, DGNSS positioning accuracy is expected to be further improved.

The DGNSS/DBA combined positioning can effectively improve the DGNSS positioning accuracy and meet the demand for real-time positioning applications. The Earth ellipsoid constraint equation constructed by the DBA altitude is equivalent to adding a virtual satellite located at the center of the Earth, effectively improving the spatial geometry structure of the observation satellite. The DGNSS/DBA combined positioning improves the positioning accuracy in the U direction by 30% to 80% compared with the DGNSS positioning, while the positioning accuracy in N and E directions also has a certain improvement effect.

Nowadays, most smartphones integrate both an inexpensive GNSS chip and a barometric pressure sensor. WADGNSS services [10] and a large number of meteorological stations [26] can provide correction information to users. We can achieve higher positioning accuracy without increasing hardware costs. The applications of low-cost DGNSS/DBA, such as indoor and outdoor seamless switching positioning, car navigation, emergency mapping, LBS, and rescue, etc. are likely to increase dramatically. Subsequently, based on the combined positioning of low-cost DGNSS/DBA, the positioning performance research by integrating other sensors, such as MEMS IMU and geomagnetic, etc., will be worth further investigation.

#### **5. Conclusions**

In this study, low-cost single-frequency DGNSS/DBA combined positioning research and performance evaluation was carried out. First, a DGNSS/DBA combined positioning model is proposed. The Earth ellipsoid constraint equation act as a virtual satellite observation at the center of the Earth, effectively improving the spatial geometry structure and PDOP value of the observation satellite. The low-cost BMP280 barometer DBA altitude accuracy is evaluated by different baseline lengths, which is better than the submeter level within 2 km and better than 2 m within 10 km baseline length. In both open and complex urban environment kinematic vehicle experiments, the DBA altitude accuracy is better than 2.20 m, which indicates that the DBA system has highly reliable and stability in different environments in local area.

The low-cost single-frequency NEO-M8T receiver with a patch antenna can achieve submeter level positioning accuracy for DGNSS positioning in the N/E directions and better than 1.5 m in the U direction in a short baseline static environment; as the baseline length increases, the DGNSS positioning accuracy gradually decreases. The positioning accuracy in a kinematic vehicle environment is significantly lower than in the static environment, and the RMSE in the N/E/U directions are all at the meter level in the complex urban environment, and the positioning accuracy of both GPS+BDS dual system is significantly improved compared to single system. The DGNSS/DBA combined positioning for low-cost NEO-M8T receiver and BMP280 barometer improves the positioning accuracy in the U direction by 30% to 80% compared with the DGNSS positioning, while the positioning accuracy in the N and E directions also has a certain improvement effect.

**Author Contributions:** Conceptualization, S.W.; methodology, S.W.; software, S.W., M.G., W.Z., D.L. and S.C.; validation, M.G., W.Z. and D.L.; formal analysis, M.G.; investigation, S.W., M.G. and S.C.; resources, G.L.; data curation, S.W.; writing—original draft preparation, S.W.; writing—review and editing, X.D., G.L., M.G., W.Z., D.L. and S.C.; visualization, S.W. and M.G.; supervision, X.D. and G.L.; project administration, X.D. and G.L.; funding acquisition, S.W., X.D. and G.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was jointly supported by the National Key Research Program of China "Collaborative Precision Positioning Project" (No. 2016YFB0501900) and the National Natural Science Foundation of China (Grant No. 41774017).

**Acknowledgments:** The authors would like to thank Gongwei Xiao and Aizhi Guo for their help in setting up the multi-sensor platform used in the kinematic vehicle experiments. Meanwhile, the authors are grateful to Chengfeng Zhang for his valuable advice in writing.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

