ION GNSS 2012
Session F1: Urban & Indoor Navigation: GNSS & Assisted-GNSS

Title: GNSS Shadow Matching: Improving Urban Positioning Accuracy Using a 3D City Model with Optimized Visibility Prediction Scoring
Author(s): L. Wang, P.D. Groves, M.K. Ziebart, University College London, UK
Date/Time: Wednesday, September 19, 2012, 8:35 a.m.
Room: Grand Ballroom Center (Renaissance)

Positioning using the Global Navigation Satellite System (GNSS) is unreliable in dense urban areas with tall buildings or narrow streets, known as "urban canyons". This is because the buildings block, reflect, or diffract the signals from many of the satellites. Combining GPS with other Global Navigation Satellite Systems (GNSS) significantly increases the availability of direct line-of-sight signals. However, even when these may be separated from the non-line-of-sight signals, the positioning accuracy is poor in the cross-street direction because the unobstructed satellite signals travel along the street, rather than across it, resulting in poor signal geometry. Due to these issues, the accuracy in the across-street direction can degrade to a few tens of meters in deep urban canyons.

A new solution to this cross-street positioning problem is to use 3D city models to predict which satellites are visible from different locations (Bradbury et al., 2007) and then compare this with the measured satellite visibility to determine position. This concept is known as shadow matching and enables non-visible satellites to contribute to the position solution (Groves, 2011). It is analogous to the "finger printing" technology deployed in WiFi positioning. preliminary implementation of shadow matching has been validated using real world experimental data from the City of London (Wang et al., 2011, Groves et al., 2012). This test demonstrated that shadow matching can reliably determine which side of the street the user is on under conditions when conventional stand-alone GNSS positioning cannot. It also showed that shadow matching has the potential for lane-level positioning but that further work is needed to achieve this reliably.

The improvements to shadow-matching presented here focus on the scheme for scoring candidate positioning selection based on the match between predicted and measured satellite visibility. The old scoring scheme simply awarded one point when the prediction and measurement matched and zero when they did not. The improved scoring scheme introduces two new factors. The first factor addresses the signal diffraction effects that occur when a satellite signal travels around the roofs or edges of buildings or other obstructions. This effect can significantly impact the accuracy of shadow matching, because the algorithm relies on the positions of the building boundaries to work. In the improved algorithm, the occurrence of signal diffraction is predicted using the 3D building model. The second factor makes better use of the observed data from the GNSS receiver by considering the signal-to-noise ratio (SNR) for each satellite. This is because a satellite signal received with low SNR is more likely to be a reflected-only or diffracted signal than a satellite signal with normal SNR, though direct line-of-sight signals may also be attenuated. Thus, a satellite matching result between prediction and observation will have a lower degree of confidence when the SNR is low, which should be reflected in the scoring scheme.

From this perspective, strategies for optimising the accuracy and reliability of shadow matching are investigated by developing and comparing different scoring schemes. Testing will be performed at different levels of urbanization. Real-world data sets will be collected in various urban canyons in London. Factors varied include the building height to street width ratio and the proximity of road junctions. A 3D city model of the Aldgate area of central London, which has a high level of detail and decimetre-level accuracy, is used.

Previous shadow-matching tests used both GPS and GLONASS signals, so the feasibility of performing shadow matching using GPS only will be assessed. Although most navigation devices used by vehicles and pedestrians only enabled GPS, the immerging new personal navigation devices (PNDs) and smartphones have deployed or are deploying GLONASS.

This improved shadow matching technique could enable new GNSS applications and improve performance of existing applications by improving the reliability of traffic lane identification for vehicles and side of street determination for pedestrians. Knowing which side of the street a pedestrian on is useful for visitor guidance and location based advertising, and critical for guiding the blind and visually impaired and for augmented-reality applications. Similarly, lane-level positioning is important for advanced intelligent transportation systems that can direct individual vehicles in order to maximize traffic flow and prioritize emergency vehicles. In practice, shadow matching would be implemented as part of an intelligent urban positioning system alongside conventional GNSS positioning and possibly additional sensors, such as odometers on cars and cell phone signals, WiFi and inertial sensors for pedestrian users.

REFERENCE

BRADBURY, J., ZIEBART, M., CROSS, P. A., BOULTON, P. & READ, A. (2007). Code Multipath Modelling in the Urban Environment Using Large Virtual Reality City Models: Determining the Local Environment. The Journal of Navigation, 60, 95-105. GROVES, P. D. (2011). Shadow Matching: A New GNSS Positioning Technique for Urban Canyons The Journal of Navigation, 64, pp417-430. GROVES, P. D., WANG, L. & ZIEBART, M. (2012). Shadow Matching Improved GNSS Accuracy in Urban Canyons. GPS World, 23, pp14-29, February 2012. WANG, L., GROVES, P. & ZIEBART, M. (2011). GNSS Shadow Matching Using A 3D Model of London. European Navigation Conference. Grange Tower Bridge, London.

Return to Session F1 Next Abstract