ION GNSS 2012
Session B3: Robust Navigation in GNSS-Challenged Environments

Title: Tight Integration of GNSS and a 3D City Model for Robust Positioning in Urban Canyons
Author(s): A. Bourdeau, M. Sahmoudi, ISAE/TeSA, University of Toulouse, France; J-Y. Tourneret, IRIT/ENSEEIHT/TeSA, University of Toulouse, France
Date/Time: Thursday, September 20, 2012, 11:48 a.m.
Room: Grand Ballroom Center (Renaissance)

The number of global navigation satellite systems (GNSS) applications has steadily increased over the last decade, in particular for personal mobility. However, the urban environment presents significant challenges for satellite positioning. On the one hand, the user is expecting for a positioning accuracy greater than that obtained in open sky areas, because of the proximity of the various points of interest and intersections. On the other hand, the urban environment creates difficulties in the GNSS signal reception, particularly because of satellite masking and multipath phenomena. As a consequence, the receiver delivers a position that can be biased by an error of several tens of meters, when it is not totally impossible to calculate a position. This is particularly true in the context of urban canyons, i.e., when the streets are very narrow and/or the buildings are very high. The main undesirable phenomena encountered in urban areas are attenuations, shadowing effects and multipath. Multipath signals can be very strong and have small relative delays which makes them difficult to be distinguished from the desired path signal. The most common approach to deal with the multipath problem is through mitigation methods during the signal processing step, such as the narrow correlator. However, in urban canyon environments, the number of line-of-sight (LOS) satellites is very low and the position dilution of precision (PDOP) of these satellites is usually unsatisfactory. In this work, we suggest to investigate the constructive use of multipath signals instead of simply mitigate those reflections as in most current GNSS receivers. The difficulty to use non-line-of-sight (NLOS) multipath is in the capability to model the lengths of these indirect paths in order to correct the distance error introduced. To resolve the problem of GNSS reduced availability in the urban context, some solutions have been proposed in the literature, as shadow-matching [1], geometric path modeling thanks to parameter estimation by a nonlinear filter [2] or path identification by laser scanning of the environment [3]. In this paper, we propose a new navigation strategy based on the augmentation of GNSS measurements by a 3D model of the environment. We introduce a new approach of position computation with indirect path measurements by tightly integrating the 3D model information in an extended Kalman filter (EKF). The SE-NAV software [4] is used to predict the signal reception of systems such as GPS and GALILEO into 3D virtual scenes of known urban areas. This software is based on a ray-tracing algorithm that computes the shadowing effects and the multipaths generated by the objects of a given environment. In this method, the measurement model, traditionally based on the trilateration equations, is constructed from the geometric information estimated by SE-NAV for each predicted state by the EKF. The point of reflection of the signal is added in the measurement model. Since the point of reflection is depending of the receiver position and time, it is expressed as a function of these two parameters. The geometric Jacobian matrix of the resulting measurement model is then calculated through the knowledge of the wall on which the received signal has been reflected. To use even less reliable measurements, a robust version of the proposed filter is also introduced. When SE-NAV is used to construct the measurement model, a new problem appears. If some signals received by the receiver are not estimated to be receivable according to the SE-NAV model at the position provided, the information they provide cannot be used. To use the information obtained for these signals, one solution is to use temporarily a less accurate but available model defined by the trilateration equations. However, the use of these equations should be done carefully. If the received signal is actually a multipath, the filter is biased because of the modeling error resulting from the trilateration, and a more accurate solution computed by our proposed filter may be corrupted. To avoid such cases, we propose to use a robust version of the EKF that mitigates the modeling errors due to trilateration. The robust approach proposed is adapted to the special context of urban canyons and to the tight integration of GNSS and the 3D city model. The paper will be organized as follows. First a state of the art about constructive use of multipath signals and 3D city models for GNSS positioning will be presented. The proposed tight integration of GNSS with the SE-NAV 3D city model will be introduced in a second step. The robust version of the proposed filter will be also described. Finally, the proposed filter and its robust version will be compared with the classical navigation filter via realistic simulations of trajectories in Toulouse, France. The proposed approach has been developed with a simplified 3D city model and is currently experimented with the high realistic 3D city model SE-NAV provided by the company OKTAL-SE. The prior results of this analysis show that integration of the 3D city model information into the GNSS navigator offers improvements over the stand alone GNSS filter, in particular in an urban canyon context.
To the authors´ best knowledge, the proposed navigation strategy based on a 3D city model has not been previously investigated in the context of GNSS applications. The findings and conclusions of this study will be described in the final paper including the detailed analysis in terms of positioning accuracy, robustness and integrity.

References [1] P. D. Groves, "Shadow Matching: A New GNSS positioning technique for urban canyons", The Journal of Navigation: The Royal Institut of Navigation, vol. 64, pp. 417-430, 2011. [2] D. E. Gustafson, J. M. Elwell and J. A. Soltz, "Innovative Indoor Geolocation Using RF Multipath Diversity", Position, Location, And Navigation Symposium, 2006 IEEE/ION, pp. 904-912, Apr. 2006. [3] A. Soloviev and F. van Graas, "Use of Deeply Integrated GPS/INS Architecture and Laser Scanners for the Identi?cation of Multipath Re?ections in Urban Environments", Selected Topics in Signal Processing, IEEE Journal of, vol. 3, no. 5, pp. 786-797, Oct. 2009. [4] http://www.oktal-se.fr/

Previous Abstract Return to Session B3 Next Abstract