|The release of GNSS data collection APIs along with Android 7.0 (Nougat) in 2016 makes it possible for common users to access GNSS raw measurements originating in portable smart devices. With data recording APPs (e.g., Google GNSS logger and Geo ++ RINEX Logger), developers targeting Android platforms can easily obtain pseudorange, Doppler and carrier-phase measurements which can in turn be exploited to achieve better positioning performance than those reported by smart devices themselves. Since usual GNSS position outputs from smart devices is at meter-level accuracy, a sub-meter or even decimeter level accuracy will greatly expand the application fields of portable devices in location based services for mass-market users and even professionals. However, it is challenging to achieve such high position precisions based on the carrier-phase and pseudorange data from smart devices. The difficulties consist primarily in their low signal strength, poor multipath suppression, and inability to provide high reliability and continuous measurements, as these devices use linearly polarized antennas and low-cost single frequency GNSS chipsets. Previous studies have shown that the primary impediment to the implementation of high-precision GNSS on smart devices comes from the antenna, which is usually a passive antenna with linearly polarized rather than right-hand circular polarized patterns used by typical GNSS antennas, and thus the antenna gain pattern is irregular and of low gain. In addition, the antenna can introduce significant multipath errors, not only owing to the antenna’s poor multipath suppression, but also because of the antenna pattern which is omni-directional rather than hemispherical, consequently increasing the contamination of multipath effects. This also complicates the quality analysis of GNSS raw measurements from smart devices, as the observation quality from smart device internal oscillators and GNSS chips cannot be accurately evaluated. In this paper, an external antenna was designed to replace the GNSS antenna of smart devices to enhance the signal strength and reduce multipath errors, which had the same signal-to-noise ratio (C/N0) and multipath suppression capability which are both comparable to a survey-grade GNSS antenna. In addition, zero-baseline tests were carried out between smart devices and survey-grade receivers, smart devices and low-cost receivers, and smart devices and smart devices, respectively. With these tests, we could eliminate the effects of atmospheric delays and other errors; analyze the double-difference observation noise and the double-difference residuals to explore the characteristics of GNSS raw measurements from smart devices and the noise characteristics that may affect the double difference ambiguity fixing . The real raw multi-GNSS measurement quality of single-frequency GNSS chipsets for Nexus 9, Honor V8 and Samsung S8 were analyzed and compared with the low-cost receiver ?-blox and the survey-grade receiver Trimble NetR9. The results show that the raw GNSS measurements produced by the internal oscillator and the GNSS chip of the smart devices are usable, but the quality is inferior to that of geodetic receivers and low-cost receiver. Their pseudorange noise is 10-fold larger than the noise of Trimble R9 and ?-blox. Specifically, the pseudorange noise of Nexus 9 is lower than that of Honor V8 and Samsung S8, possibly because they used different chipset. GLONASS pseudorange noise in all test smart devices surpasses those of GPS and BDS pseudorange. In addition, the amplitude of BDS pseudorange noise is clearly smaller than those of GPS and GLONASS pseudorange noise in Honor V8 and Samsung S8. Therefore, it is necessary to assign different weights when performing integrated positioning of those observations from different satellite constellations. Their phase noise is less than 0.2 cycles (duty cycle off), only 3-5 times larger than that of the Trimble R9, and even close to that of ?-blox. This phenomenon shows that the GNSS chip of the smart devices can provide high precision carrier-phase measurements, have high precision navigation potential, and may even provide better positioning accuracy than some low-cost GNSS receivers. However, for the zero-baseline tests, the double-difference phase residuals (duty cycle off) exhibit an anomalous "jagged" distribution (i.e., aperiodic jumps). This phenomenon exists in the zero-baseline tests between different receivers, such as smart devices and survey-grade receivers, smart devices and low-cost receivers, and smart devices and smart devices. In addition, the double-difference phase residuals are not reduced or disappeared by using identical smart devices. Therefore, the errors contained in the double-difference phase residuals may not stem from any systematic errors of the smart-device GNSS chip. Also, it may be that the same type of smart device has different phase errors. There are also large differences in the double-difference phase residuals from different GNSS. For GPS, the standard deviation of phase residuals of the all zero-baselines based on Nexus 9 is less than 0.1 cycles, while about 10% of the residuals exceed 0.1 cycles. For Trimble R9 and ?-blox, in comparison, they are less than 0.01 cycles, and there are no residuals exceeding 0.1 cycles. For GLONASS, in contrast, the performance of double-difference phase residuals for smart devices is even worse. The standard deviation is more than 1 cycles, 10 -100 times larger than those of survey-grade receivers. We propose a multi-GNSS single point positioning (SPP) strategy based on weighted least squares adjustment according to the pseudorange noise of various GNSS. Multi-GNSS can provide better positioning performance than single GPS, especially in terms of reliability. In this open-sky test environment, the availability of multi-GNSS is about 10% higher than that of single GPS, which is expected to improve the positioning performance in case of GNSS-adverse environments. In addition, the accuracy of the multi-GNSS SPP in this study is better than that of multi-GNSS SPP based on traditional weighting methods for smart device. Because the traditional method is based on the error models obtained by the survey-grade receiver for each system, this is not suitable for smart devices, so its horizontal error is over 10m and the vertical error is close to 15m. The multi-GNSS SPP in this study, in contrast, can provide 10m of horizontal and vertical positioning accuracy in 95%. Similarly, we optimized the carrier-phase differential multi-GNSS positioning for smart devices according to the characteristics of the double-difference phase residuals, which realizes the centimeter-accurate positioning (float solution). For this study, a baseline of approximately 400 meters taken between the Nexus 9 and IGS station WUHN (Trimble NetR9) was used to verify the accuracy of the carrier phase differential positioning and the benefits of multi-GNSS positioning. For GPS, its availability (the ratio of the total number of epochs that can be solved) is 89.86%, the convergence time (the first time when the North, East and Up error are less than 10cm) is 2935s, and the RMS in horizontal errors are less than 15cm. For multi-GNSS (GPS/GLONASS), in contrast, its availability was increased to 91.98%, the convergence time was reduced to 906s, but the RMS in horizontal errors remains. Although the final positioning accuracy compared to single GPS is not improved significantly, the convergence time and position availability has been improved appreciably. This is more meaningful because smart devices are often used in GNSS-adverse environments where some of the signals are obstructed, such as within urban environments. Finally, we also try single-frequency precise point positioning (PPP) by introducing external ionosphere corrections to investigate carrier-phase based positioning performance. Finally, we demonstrate that this study is to investigate the potential of precise positioning with raw multi-GNSS measurements from mainstream portable smart devices. The analysis on the noise characteristics of carrier-phase and pseudorange from different instrument and different GNSS is especially useful to the establishment of precise error mitigation models for achieving high-precision positioning with smart devices.
Proceedings of the 31st International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2018)
September 24 - 28, 2018
Hyatt Regency Miami
|392 - 412
|Cite this article:
Geng, Jianghui, Li, Guangcai, Zeng, Ran, Wen, Qiang, Jiang, Enming, "A Comprehensive Assessment of Raw Multi-GNSS Measurements from Mainstream Portable Smart Devices," Proceedings of the 31st International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2018), Miami, Florida, September 2018, pp. 392-412.
ION Members/Non-Members: 1 Download Credit