GPS Multipath Detection and Mitigation in Urban EnvironmentsShiwen Zhang, Sherman Lo, Yu-Hsuan Chen, J. David Powell

GPS Laboratory, Department of Aeronautics & Astronautics, Stanford University

M. Obst, S. Bauer and G. Wanielik, "Urban multipath detection and mitigation with dynamic 3D

maps for reliable land vehicle localization," Proceedings of the 2012 IEEE/ION Position,

Location and Navigation Symposium, Myrtle Beach, SC, 2012, pp. 685-691.

GPS multipath refers to the phenomenon when a satellite signal is

received off a reflecting surface (such as buildings) rather than

unimpeded from the satellite. Such reflections can cause significant

errors in the user navigation solutions. Multipath is particularly significant

and common in urban environment. Identifying and reducing the effect of

multipath would enable GPS and the other satellite navigation

constellations to contribute to high integrity railway control and

autonomous vehicle operating in urban environments.

Introduction

The purpose of this study is to detect and mitigate GPS multipath error

to improve position accuracy and trustworthiness in urban environment.

Purpose

This study examines a multipath detection algorithm followed by satellite

exclusion. The proposed algorithm uses a ray-tracing algorithm on a 3D

building model to predict the presence of both LOS and reflected signals

given a satellite-user geometry. A statistical analysis was then performed

to help quantify the confidence level of the building model’s multipath

prediction under modeling uncertainty. Experimental data were collected

to evaluate the performance of the algorithm.

Methods

GPS data were collected at 13 ground locations on the Engineering

Quad using a NovAtel receiver. Four different algorithms were applied at

each location. The no exclusion algorithm uses all received signals to

calculate position solution. The residual checking algorithm excludes

outlier signals one by one until the residuals are consistent with each

other. The hard exclusion algorithm excludes signals based on the

building model’s predictions of LOS and reflection. The soft exclusion

algorithm performs satellite exclusion based on the confidence level of

the model’s prediction . The hard and soft exclusion may be explained

more clearly – you may do that in your talk rather than in the text

Results

This study examined a multipath detection method that uses 3D building

model. The soft exclusion algorithm reduced multipath effects on

position error from as large as 17 meters to 1.2 meters and achieved an

average position error of 1.6 meters. Experimental results show the

need for a reasonable initial user position estimate.

Conclusions

Bibliography

The authors would like to thank the Center for Automotive Research at

Stanford (CARS) and the Stanford Center for Position Navigation and

Time (SCPNT) for their support in this research.

Acknowledgments

The author can be reached at shiwenzh@stanford.edu

A PDF version of this poster can be found at

https://stanford.box.com/s/utl800pop1hllfvzrjlytiubgicplyn0

Further Information

More LogosAbstract #

20980123

Line of sight coverage Multipath range errorNumber of reflection

30 m

10 m

15 m

60-deg

elevation

LOS

Reflection

A 3D building model was developed for ray-tracing simulation and multipath prediction. The site

chosen for simulation and testing is the Engineering Quad at Stanford University. The building

model was constructed using building corner coordinates and building heights estimated from

Google Earth. Detail structure of the building walls and roofs were not captured in the model to

simplify simulation. The ray-tracing algorithm simulates the LOS signal path as well as all the

building-reflected signal paths from the satellite to the user receiver.

Modeling and Simulation