Modeling multifrequency GPS multipath fading in land vehicle environments

Abstract

Two of the most used models to characterize GPS signal fading are the Nakagami-m and Rice, but in this work, we present evidence that supports the κ–μ distribution as the best fit to deal with multifrequency GPS multipath channels inside urban, rural, and forest areas. Experimental measurements are presented to confirm the κ–μ distribution as the best distribution to characterize different situations on the available three GPS frequencies. We also present typical values of fading coefficients in L1, L2C, and L5 signals, for cases involving urban canyons, regular urban, rural, and dense vegetation areas. These coefficients can also be used to evaluate the receiver performance under similar cases or may be applied in weights measurement methods for positioning computation improvement.

Full text

Vol.:(0123456789)
1 3
GPS Solutions (2021) 25:3
https://doi.org/10.1007/s10291-020-01040-8
ORIGINAL ARTICLE
Modeling multifrequency GPS multipath fading inland vehicle
environments
VicenteCarvalhoLimaFilho1 · AlisonMoraes2
Received: 18 October 2019 / Accepted: 28 September 2020 / Published online: 9 October 2020
© Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract
The reliability and performance of GPS receivers depend on the quality of the signal received, which can be largely affected
by the interference caused by buildings, trees, and other obstacles. Since obstacles are always present in practical applications,
several statistical representations have been developed along the years to measure, predict, and compensate errors induced
by interferences. Two of the most used models to characterize GPS signal fading are the Nakagami-m and Rice, but in this
work, we present evidence that supports the κ–μ distribution as the best fit to deal with multifrequency GPS multipath chan-
nels inside urban, rural, and forest areas. A synthetic signal simulator was developed to create propagation cases involving
scattering clusters and specular reflections. Additionally, experimental measurements are presented to confirm the κ–μ dis-
tribution as the best distribution to characterize different situations on the available three GPS frequencies. We then present
typical values of fading coefficients in L1, L2C, and L5 signals, for cases involving urban canyons, regular urban, rural, and
dense vegetation areas. These coefficients can also be used to evaluate the receiver performance under similar cases or may
be applied in weights measurement methods for positioning computation improvement.
Keywords Fading distribution· Multipath· Urban environments
Introduction
In the near future, systems such as smart cities, autono-
mous cars, vehicle ad-hoc networks, and drones will hugely
increase the demand for radio communication services.
For mobile applications in land vehicles such as vehicular
ad-hoc network (VANET) and autonomous cars, real-time
positioning is needed with high availability and accuracy Li
and Wang (2007), indicating the relevance of global naviga-
tion satellite system (GNSS) channel modeling inside urban
areas. Mobile multipath modeling in the urban environment
is a relevant subject such as Lehner and Steingass (2005),
where a model that takes into account changes in elevation,
azimuth, speed, and number of reflectors was developed.
In crowded cities, scattering and specular reflections of
the direct signal pollutes the received signal by the receptor.
In Strode and Groves (2016), for example, signal-to-noise
measurements on three different GNSS frequencies are
compared to detect multipath signals. In Håkansson (2019),
GNSS observations made using a tablet showed that mul-
tipath impacts the expected accuracy of calculated positions
because of induced measurement errors and also because
of loss of lock of GNSS signals. Zhang etal. (2018) also
reported multipath effect is a challenge to achieve submeter
level in smartphone positioning. To process those polluted
signals, it is necessary to deploy signal processing tech-
niques that use some specific distribution of the received
signal for channel modeling. In addition, it is important to
decide which distribution must be used, including the coef-
ficients that best describe the channel effect on the signal.
Traditionally, multipath is modeled using Nakagami-
m and Rice models, for example, Gaertner and Nuallain
(2007) used these models to characterize fading events in
microcell urban environments. Nakagami-m is an interest-
ing distribution for the scattering effect, especially when the
scattering cluster is not homogeneously disposed or there is
more than one cluster. The Rice distribution is adequate for
environments where specular reflection is present. Yacoub
(2007) proposed a more general model, called κ–μ, of which
* Alison Moraes
aom@ita.br
1 Instituto Tecnológico de Aeronáutica (ITA),
SãoJosédosCampos, SP, Brazil
2 Instituto de Aeronáutica e Espaço (IAE),
SãoJosédosCampos, SP, Brazil
Content courtesy of Springer Nature, terms of use apply. Rights reserved.