|Abstract:||Satellite navigation signals transmitted from altitudes of more than 20,000km generate tens of meter error due to atmosphere effect, orbits and clock estimating error, reflected signal error. Since most of the error sources such as ephemeris, clock, and atmospheric errors are correlated temporally and spatially, they can be eliminated by applying Differential GNSS(DGNSS) and Real Time Kinematics (RTK). User-specific errors such as multipath error and receiver noise, however, cannot be eliminated by the differential technique. Their error types and effects on user are so sensitively dependent on the signal reception environment that it is very difficult to mitigate or model them. Especially, non-line of sight (NLOS) type multipath error could be dominant error source in urban canyon area, thus accurate positioning in urban areas has been a major challenge in the GNSS field for a long time. Recently various studies such as multi-constellation GNSS, NLOS measurement rejection, Ray tracing, and Shadow matching have been introduced to reduce the multipath effect. Multipath error estimation algorithm using previously computed position and code-minus-carrier (CMC) variation has been suggested in 2019 (Lee, 2019). By including inter system bias between other constellation and GPS into the initial multipath value, multi-constellation positioning was possible with only four satellites, and its 95% error was reduced from 18m to 5m in a deep urban area of Teheran-ro, Seoul, S.Korea. However, this algorithm should be used under very limited constraint that the user should start from a known coordinate for estimating initial multipath for each satellite. Besides, multipath estimation uncertainty gradually accumulates, which could lead to increase error with time. To solve the inherent problem of the previous algorithm, this paper suggests an automatic switching technique between general DGNSS and CMC-based multipath mitigation. The determination of which positioning mode should be selected is made by a validation test using the DGNSS residuals. Under an open sky, most bias error should be eliminated by the DGNSS technique, and the residuals are close to white noises. Accordingly, we can consider that the GNSS signals has been received in suburban area, if sum of the normalized residuals passes the chi-square test whose threshold is determined by the CFAR (Constants False Alarm Rate) method. Once the algorithm determine current position is suburban area based on the residual test results, the positioning mode is switched to DGNSS, and all the multipath estimates are initialized based on the DGNSS position. On the other hand, the sum of normalized residual cannot pass the chi-square distribution due to multipath error that remained in any of the residuals, the DGNSS position is considered as no longer valid, and the positioning should be computed by CMC-based multipath mitigation mode till the DGNSS residuals pass the chi-square test. To assess performance of the suggested algorithm in an urban canyon, a dynamic test was conducted in Seoul, Korea. We drove a vehicle for this test from Sports Complex Station to Gangnam Station along the Teheran-ro. The start and end spots were relatively sub-urban area, but most of the road was deep urban area. A dual frequency GNSS receiver, Novatel flexpak-6 was used for the positioning and Novatel SPAN provided a reference trajectory. According to our DGNSS validation test results during the half hour driving, DGNSS positions were acceptable for 6% of the test sections. In addition to the start and end spots, the DGNSS-valid places also include several points where relatively good satellite visibilities are temporarily provided in urban such as intersections. These points in the deep urban areas greatly contribute to preventing the multipath estimates error from accumulating severly by initializing them. While the DGNSS technique of the receiver could calculate its positions only for 64.7% of the test section, the suggested method provided the rover’s real time results through all the period. In addition to this improvement of position available time to 100%, the position error is significantly reduced compared to conventional methods. Root Mean Square (RMS) horizontal error of the flexpak-6 in DGNSS mode has been reduced from 11.1m to 1.2m by applying the suggested algorithm to the receiver. Most impressive thing is that the position error of the algorithm is bounded to 3.0m, while the maximum error of the receiver was 101.0m. From these results, we confirmed that the mode switching between DGNSS and CMC-based multipath mitigation technique was done successfully in the deep urban area by residual test. We expect this method to produce similar performance in other cities than Seoul since rovers can apply this algorithm to acquire their position without any prior information for the operation such as the layout and 3D information of buildings or known positions at specific points. REFERNECE  Lee, Yongjun, Park, Byungwoon, Hwang, Yoola, Lee, Byoung-Sun, Ahn, Jaeyoung, "Direct Estimation of Multipath in a Deep Urban Area using Multi-GNSS Carrier Phase Variation and Previous Position," Proceedings of the ION 2019 Pacific PNT Meeting, Honolulu, Hawaii, April 2019, pp. 728-736. https://doi.org/10.33012/2019.16835|
Proceedings of the 33rd International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2020)
September 21 - 25, 2020
|Pages:||1690 - 1719|
|Cite this article:||
Lee, YongJun, Park, Byungwoon, "Seamless Accurate Positioning in Deep Urban Area using DGNSS and CMC-based Multipath Mitigation," Proceedings of the 33rd International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2020), September 2020, pp. 1690-1719.
ION Members/Non-Members: 1 Download Credit