BROWSE
Detail
Conference Paper
Enhancing GNSS Performance in Urban Canyon using GNSS Visibility Map and Recurrence Vector
Cited 1 time in 
Share 
Share
- Authors
- Yongjun Lee, Byungwoon Park, Jungyo Park, Hyunseung Jei, Youngsun Yun, Haesung Park
- Issue Date
- 2025-09
- Citation
- International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+) 2025, pp.1907-1913
- Language
- English
- Type
- Conference Paper
- Abstract
- As Global Navigation Satellite System (GNSS) technology becomes increasingly prevalent in applications such as smartphones, autonomous vehicles, and Unmanned Air Mobility (UAM), the demand for enhanced GNSS performance in dense urban environments has grown significantly. However, urban areas present considerable challenges for GNSS positioning due to signal degradation caused by high-rise buildings and vehicular obstructions. In such environments, GNSS signals are frequently subject to reflection, leading to multipath and Non-Line-of-Sight (NLOS) errors. These errors exhibit significant variability depending on the reception environment, making them difficult to distinguish and model accurately. To improve GNSS performance in urban environments, various positioning techniques integrating GNSS with 3D building models, such as Shadow Matching and Ray Tracing, have been studied. However, a main challenge of these approaches is their reliance on accurate user position information to effectively integrate 3D building models with GNSS. Given that GNSS alone struggles to provide accurate positioning in dense urban areas, these methods typically generate multiple candidate positions around an initial GNSS position estimate and perform simulations using 3D building models at each candidate position. In urban environments, GNSS positioning errors caused by multipath and NLOS effects can reach several hundred meters, necessitating the generation of numerous candidate positions when applying techniques integrating GNSS with 3D building models. This study proposes a novel methodology for estimating GNSS measurement ranging errors by projecting a recurrence vector onto the Line-of-Sight (LOS) directions of the satellites utilized in GNSS positioning. The recurrence vector represents the discrepancy between the initial GNSS position and each candidate position, under the assumption that one of the candidates corresponds to the actual user position. To directly project the recurrence vector onto LOS directions, subsets of four satellites are formed. Once these subset positions are computed initially, recurrence vectors for all candidate positions can be derived through differencing, enabling batch computation of ranging errors across all candidates. The computed ranging errors at each candidate position are subsequently compared with a visibility map derived from 3D building models to identify the candidate most likely to be nearest to the true user position. Received satellite signals are classified as either LOS or NLOS at each candidate position based on the visibility map. The probability of correct signal classification is then determined using probabilistic models of signal types in conjunction with recurrence vector-derived ranging errors. Fianlly, the candidate with the highest cumulative probability of correct signal classification is selected as the final candidate. The proposed methodology was validated using data collected from a moving vehicle in one of the most densely populated urban canyons of Seoul.
- KSP Keywords
- 3D building models, Accurate positioning, Autonomous vehicle, Batch computation, Cumulative probability, Data collected, Dense urban, GNSS positioning, GNSS signal, Line of sight(LOS), Moving Vehicle