scispace - formally typeset
Search or ask a question
Proceedings ArticleDOI

Classification of GNSS SNR data for different environments and satellite orbital information

19 Mar 2017-pp 2088-2092
TL;DR: Results show good correlation ofSNR's between same sub environments for different satellite elevation ranges which offer useful insight to regenerate a generalized set of SNR parameters in the laboratory environment for the development of 3D GNSS channel model.
Abstract: In this paper, a data classification method for analyzing the aspects of Signal-to-Noise Ratio (SNR) for Global Navigation Satellite System (GNSS) in real conditions is introduced Different parts of measured environments and the orbital information of satellites are used as criteria for data classification It consists of: 1) taking fish eye images of measured routes; 2) dividing measured environments into four potential sub environments (open area, forest area, single building blockage, and street canyon); 3) classifying satellites into nine different groups as function of elevation angles; and 4) creating a table containing the information of mean and standard deviation of SNR for different environments and satellite elevation angles Results show good correlation of SNR's between same sub environments for different satellite elevation ranges which offer useful insight to regenerate a generalized set of SNR parameters in the laboratory environment for the development of 3D GNSS channel model

Summary (3 min read)

Introduction

  • In order to guarantee an acceptable user experience, GNSS receiver’s location accuracy needs to be proven cautiously on the grounds of interference, sensitivity, and use cases in various surroundings and in different atmospheric conditions.
  • GNSS laboratory measurement environments are also marketed by companies, e.g. ETS-Lindgren [4] and Keysight Technologies [5].
  • As the result of measurements, characteristics of polarization based reflections, position error, coverage efficiency (i.e. mean number of tracked satellites), and the impact of satellite elevation angle on received Signalto-Noise Ratio (SNR) for a typical multipath environment are investigated in [7] to create a dynamic channel model for the laboratory environment.
  • Third, the data classification approach and results are presented in section IV.

II. MEASUREMENT SETUP

  • The measurement system consists of two GNSS receivers, capable of recording signals that are transmitted by various constellations of satellites (GPS, GLONASS, and Beidou).
  • In addition to video recording continuous images with fish-eye lens were taken to track the changing environment.
  • The camera was held head-high pointing towards zenith.
  • During the test, both receivers collected GPS satellite data at a rate of 1Hz including the Coordinated Universal Time (UTC), received SNR, number of satellites tracked and used in position fix, and the orbital information of each visible satellite (i.e. satellite elevation, and azimuth).
  • A detailed description of the measurement setup along with antenna orientations is presented earlier [6].

III. ENVIRONMENTS

  • The mentioned GPS data were collected for three different environments; consisting of, university campus, suburban residential, and urban downtown areas.
  • Starting position of all three measured routes was nearly an open-sky area to obtain complete GPS ephemeris and good position fix.
  • Measurements were made in Oulu, Finland during the months, when the trees were with and without leaves, respectively.
  • Measurements for university campus area represent the case in which the trees were with leaves and measurements for suburban residential, and urban downtown areas represent the case when deciduous trees were almost without leaves.
  • Fig. 1 shows the aerial view of three environments.

B. Suburban Residential Area

  • The measured route consist of a typical residential area occupied primarily by private residences.
  • The measured route included trees surrounding the roads, two storey houses with wooden structures, narrow streets, and a three storey office building located nearly 30 meters away from the measured route.
  • The test vehicle took about 10 min to finish the trajectory and about 720 data points are recorded by GNSS receivers.

C. Urban Downtown Area

  • A heavily dense area with many businesses, recreations, narrow streets, underpass, and tall buildings close together.
  • The subway underneath the roads and railway lines is roughly 80 meters long.
  • The test vehicle took about 13 min to finish the trajectory and about 850 data points are recorded by GNSS receivers.

IV. DATA CLASSIFICATION

  • Based on the measurements from various environments shown in Fig. 1, the different parts of each environment and the orbital information of satellites are used as criterion for data classification.
  • Each measured environment is further divided into four different environments: (1) open area; (2) forest area; (3) single building blockage; and (4) street canyon.
  • Classification of satellites is based on their elevation, however in case of single building blockage azimuth information of satellites is also taken into account.
  • An elevation mask of 10◦ is used for the forest area due to insufficient data and visibility of satellites below 10◦, and the tall building structures of street canyon prevented visibility of satellites for all elevations below 30◦.

A. Notations

  • Elevation and azimuth angles of satellite are represented by θS and φS , respectively and, similarly, the elevation tilt angles of directional receiver antennas are presented with θRHCP and θLHCP .
  • Environments consisting of university campus, and suburban residential areas are presented with UCA, and SRA, respectively.
  • Merged SNR data from different sub environments are denoted by Md.
  • In single building case, Blocked side of building is represented by B, whereas, the other (open) side is presented with O. Finally, the mean and standard deviation of SNR are represented by µSNR, and σSNR.

B. Approach

  • Fig. 2 represents the flow chart of the data classification approach.
  • After the route selection and measurement setup, the selected route is measured with GNSS receivers in parallel with a video camera to track the continuously changing environment.
  • Then, satellites are classified in nine different groups based on their elevation and the data form all the satellites that fall in same group are merged together.
  • The same process is repeated for each environment and data from each sub environment are merged together.
  • Finally, a table is created containing the information of mean and standard deviation of SNR for different environments and elevation ranges.

C. Results

  • In order to get the full coverage for both upper and lower hemisphere, measurements were performed at different antenna orientations (i.e. θRHCP and θLHCP ).
  • It should be noted that there was no area in measured environments open enough to be considered as an open area.
  • Bar charts of the µSNR and σSNR from the forest area (as described under Data Classification) for RHCP antenna are shown in Figs. 4 and 5.
  • Therefore, the rest of results are based on merged data from each sub environment instead of data from individual environments.
  • At higher θS , RH has LOS situation and lower levels of LH pointing towards weaker reflections from ground below.

V. CONCLUSION

  • In this article the authors have presented a data classification method based on the measurements from various environments.
  • Different parts of each environment and the orbital information of satellites are used as criterion for data classification.
  • First, the method involves taking fish eye images of measured routes and analyzing the images to divide each environment into four potential sub environments: (1) open area; (2) forest area; (3) single building blockage; and (4) street canyon.
  • The results obtained with significant amount of data from three different environments showed good correlation of SNR between same sub environments.

Did you find this useful? Give us your feedback

Content maybe subject to copyright    Report

Abstract—In this paper, a data classification method for
analyzing the aspects of Signal-to-Noise Ratio (SNR) for Global
Navigation Satellite System (GNSS) in real conditions is in-
troduced. Different parts of measured environments and the
orbital information of satellites are used as criteria for data
classification. It consists of: 1) taking fish eye images of measured
routes; 2) dividing measured environments into four potential sub
environments (open area, forest area, single building blockage, and
street canyon); 3) classifying satellites into nine different groups
as function of elevation angles; and 4) creating a table containing
the information of mean and standard deviation of SNR for
different environments and satellite elevation angles. Results show
good correlation of SNR’s between same sub environments for
different satellite elevation ranges which offer useful insight to
regenerate a generalized set of SNR parameters in the laboratory
environment for the development of 3D GNSS channel model.
Index Terms—GNSS, measurements, data classification, satel-
lite communication.
I. INTRODUCTION
Lately, Global Navigation Satellite System (GNSS) or more
commonly Global Positioning System (GPS) receivers have
become a must have feature in mobile phones and multi-
purpose wearable device. They are likewise being used in
material logistic systems, automobile (burglary assurance), and
transport tracking systems. GPS can measure the geographical
location of a user’s receiver anywhere on earth with the
accuracy up to 1 meter. However, the presence of multipath
signals especially in urban areas, and the position of satellites
can have serious impacts on the positioning accuracy of a GPS
device.
In order to guarantee an acceptable user experience, GNSS
receiver’s location accuracy needs to be proven cautiously
on the grounds of interference, sensitivity, and use cases in
various surroundings and in different atmospheric conditions.
Therefore, the performance evaluation of GNSS devices in 3D
laboratory measurement environment is gaining increasing in-
terest among scientific community and device manufacturers.
Additionally, bringing the time consuming, costly, and uncon-
trollable field tests in to a controllable laboratory environment
gives cost savings and speeds up the product development.
Some of previous investigations have used 3D city models
to improve the accuracy of GNSS receiver by detecting non-
This research is supported by the Finnish Funding Agency for Innova-
tion (Tekes) through the Hilla research program and Centre for Wireless
Communications.
line-of-sight (NLOS) signals for Land Mobile Satellite (LMS)
channel in urban canyons [1]–[3]. However, 3D maps of cities
are hard to come by and also, the polarization of multi-
path components is not taken into account. GNSS laboratory
measurement environments are also marketed by companies,
e.g. ETS-Lindgren [4] and Keysight Technologies [5]. Despite
the provided reference setups for GNSS over-the-air (OTA)
testing, publicly available measurement results and channel
models are not available from these companies.
This paper is based on the recent measurement campaign
performed with polarization based measurement system for
the development of 3D channel model [6]. Both Right Hand
Circularly Polarized (RHCP) and Left Hand Circularly Polar-
ized (LHCP) antennas are employed to obtain direct and re-
flected signals simultaneously. As the result of measurements,
characteristics of polarization based reflections, position error,
coverage efficiency (i.e. mean number of tracked satellites),
and the impact of satellite elevation angle on received Signal-
to-Noise Ratio (SNR) for a typical multipath environment are
investigated in [7] to create a dynamic channel model for the
laboratory environment.
In this paper, we present data classification approach based
on classification of measured environments into four sub envi-
ronments, orbital information of satellites, and video recording
of environments with fish-eye lens. The method of using
fish-eye lens to track the environment first time appeared in
one of the earlier papers on GNSS in mid 1990s [8]. The
data classification method, with inputs of video recording
and GPS signal measures, is accordingly designed to output
SNR distributions and look up table containing the mean and
standard deviation of SNR for sub environments at different
satellite elevation angles. By applying the proposed method to
data collected in different environments, an analysis of SNR in
real conditions is provided, and SNR parameters between sub
environments are compared as function of satellite elevation
angle, which will provide useful insight for the development
of 3D GNSS channel model.
The rest of paper is organized as follows. We present brief
description of measurement setup in section II. Next, the
measured environments are introduced in section III. Third,
the data classification approach and results are presented in
section IV. Finally, relevant conclusions are drawn based on
the results in section V.
Classification o f G NSS S NR D ata f or Different
Environments and Satellite Orbital Information
Rameez UR Rahman Lighari, Markus Berg, Erkki T. Salonen, Aarno Parssinen
Centre for Wireless Communications Radio Technology Research Unit, University of Oulu, Finland
rameez.lighari@oulu.fi

(a) University Campus Area (b) Suburban Residential Area (c) Urban Downtown Area
Fig. 1: Aerial view of measured environments in Oulu, Finland.
II. MEASUREMENT SETUP
The measurement system consists of two GNSS receivers,
capable of recording signals that are transmitted by various
constellations of satellites (GPS, GLONASS, and Beidou).
For this paper, only GPS signals were recorded using RHCP
(sensitive for line-of-sight component and a double reflected
component) and LHCP (sensitive for first order reflections)
antennas. In addition to video recording continuous images
with fish-eye lens were taken to track the changing environ-
ment. The camera was held head-high pointing towards zenith.
During the test, both receivers collected GPS satellite data
at a rate of 1Hz including the Coordinated Universal Time
(UTC), received SNR, number of satellites tracked and used in
position fix, and the orbital information of each visible satellite
(i.e. satellite elevation, and azimuth). A detailed description
of the measurement setup along with antenna orientations is
presented earlier [6].
III. ENVIRONMENTS
The mentioned GPS data were collected for three different
environments; consisting of, university campus, suburban res-
idential, and urban downtown areas. Starting position of all
three measured routes was nearly an open-sky area to obtain
complete GPS ephemeris and good position fix. Measurements
were made in Oulu, Finland during the months, when the trees
were with and without leaves, respectively. Measurements for
university campus area represent the case in which the trees
were with leaves and measurements for suburban residential,
and urban downtown areas represent the case when deciduous
trees were almost without leaves. Fig. 1 shows the aerial view
of three environments.
A. University Campus Area
The measurement route included a typical campus area envi-
ronment with built-up area, narrow street canyons, walkways,
canopy of metal over the walkway, and a small forest area.
The total length of the route is 1 km. The test vehicle took
about 10 min to finish the trajectory and about 650 data points
are recorded by GNSS receivers.
B. Suburban Residential Area
The measured route consist of a typical residential area
occupied primarily by private residences. The measured route
included trees surrounding the roads, two storey houses with
wooden structures, narrow streets, and a three storey office
building located nearly 30 meters away from the measured
route. The total length of the route is 1.1 km. The test vehicle
took about 10 min to finish the trajectory and about 720 data
points are recorded by GNSS receivers.
C. Urban Downtown Area
A heavily dense area with many businesses, recreations,
narrow streets, underpass, and tall buildings close together.
The subway (underpass) underneath the roads and railway
lines is roughly 80 meters long. The total length of the route
is 1.4 km. The test vehicle took about 13 min to finish the
trajectory and about 850 data points are recorded by GNSS
receivers.
IV. DATA CLASSIFICATION
Based on the measurements from various environments
shown in Fig. 1, the different parts of each environment
and the orbital information of satellites are used as criterion
for data classification. Each measured environment is further
divided into four different environments: (1) open area; (2)
forest area; (3) single building blockage; and (4) street canyon.
Classification of satellites is based on their elevation, however
in case of single building blockage azimuth information of
satellites is also taken into account. An elevation mask of 10
is used for the forest area due to insufficient data and visibility
of satellites below 10
, and the tall building structures of street
canyon prevented visibility of satellites for all elevations below
30
.
A. Notations
Elevation and azimuth angles of satellite are represented
by θ
S
and φ
S
, respectively and, similarly, the elevation tilt
angles of directional receiver antennas are presented with
θ
RHCP
and θ
LHCP
. Environments consisting of university
campus, and suburban residential areas are presented with
U
CA
, and S
RA
, respectively. Merged SNR data from different

sub environments are denoted by M
d
. In single building case,
Blocked side of building is represented by B, whereas, the
other (open) side is presented with O. Finally, the mean and
standard deviation of SNR are represented by µ
SNR
, and
σ
SNR
.
B. Approach
Fig. 2 represents the flow chart of the data classification
approach. After the route selection and measurement setup,
the selected route is measured with GNSS receivers in parallel
with a video camera to track the continuously changing
environment.
The first and the foremost important step in data classifi-
cation is to synchronize the clocks between National Marine
Electronics Association (NMEA) data and video recording.
Now, by the means of video recording each environment is
classified into four different environments as mentioned above
in section IV.
Then, satellites are classified in nine different groups based
on their elevation and the data form all the satellites that fall
in same group are merged together. In the next step, elevation
based data are sorted for each sub environment.
The same process is repeated for each environment and
data from each sub environment are merged together. Finally,
a table is created containing the information of mean and
standard deviation of SNR for different environments and
elevation ranges.
However, in the case of single building blockage satellites
are classified into two different classes (1) blocked by building
(180
in φ
S
); and (2) other side which we refer as line-of-sight
(LOS) situation (180
in φ
S
) as illustrated in Fig 3.
C. Results
In order to get the full coverage for both upper and
lower hemisphere, measurements were performed at different
antenna orientations (i.e. θ
RHCP
and θ
LHCP
). However, in
this paper the results are presented for θ
RHCP
= θ
LHCP
= 45
.
More details about different antenna orientations can be found
in [6], [7]. Furthermore, It should be noted that there was no
area in measured environments open enough to be considered
as an open area. Also, during the measurement campaign
no satellites between the elevation angles of 81
90
were
received because of the northern location of the measurement
routes. For the deeper investigation two different operation
environments were selected where mean and variance of the
Left Hand (LH) and Right Hand (RH) SNR values are plotted
as a function of elevation angles.
Bar charts of the µ
SNR
and σ
SNR
from the forest area
(as described under Data Classification) for RHCP antenna
are shown in Figs. 4 and 5. The small difference between the
µ
SNR
and σ
SNR
values from two environments might be due
to lack of leaves in suburban residential area. In general, the
merged data from two environments gives a better prospect
of tree shadowing effect. Similar kind of behavior is observed
for different sub environments (e.g. single building blockage
and street canyon). Therefore, the rest of results are based on
Fig. 2: Flow chart of data classification approach.
Fig. 3: Case of single building blockage.
merged data from each sub environment instead of data from
individual environments.

Fig. 4: Comparison of µ
SNR
between environments for RHCP.
Fig. 6: Cumulative probability distributions of SNR for RHCP.
Shown in Figs. 6 and 7 are the cumulative probability
distributions of SNR for single building blockage. For RH
(Fig. 6) most satellites are shadowed by building and satellites
on opposite to buildings have higher values. RH levels of
satellites on both sides of building tends to increase with
increase in θ
S
and RH values are closely matched at higher θ
S
.
Higher LH values (Fig. 7) of other satellites suggest stronger
reflection from the building and lower values are due weaker
reflections from shadowed satellites.
The results of cumulative probability distributions of SNR
(RH and LH) for street canyon are presented in Fig 8.
At lower θ
S
, RH and LH levels are closely matched and
pointing towards stronger reflections and weak LOS condition.
However, at higher θ
S
, RH has LOS situation and lower levels
of LH pointing towards weaker reflections from ground below.
An increasing trend in mean SNR values can be observed
for all sub environments for RHCP antenna as presented in
Table I. Furthermore, at higher θ
S
mean values are closely
matched pointing towards LOS situation. On the other hand,
at lower θ
S
, LH–SNR’s are either closely matched or higher
Fig. 5: Comparison of σ
SNR
between environments for RHCP.
Fig. 7: Cumulative probability distributions of SNR for LHCP.
Fig. 8: Comparison of SNR distribution between RHCP and
LHCP.
than RH–SNR’s as reported in Table II.
Overall decreasing pattern can be seen in the values of

TABLE I: MEAN AND STANDARD DEVIATION OF RHCP SNR
θ
S
Forest Area Single Building (Blocked) Single Building (Others) Street Canyon
µ
SNR
σ
SNR
µ
SNR
σ
SNR
µ
SNR
σ
SNR
µ
SNR
σ
SNR
01
10
11
20
25.06 5.01 15.06 4.70 25.59 4.32
21
30
28.25 5.41 24.34 7.42 28.67 5.48
31
40
31.48 4.85 28.01 5.75 30.08 4.28 25.90 7.59
41
50
32.87 4.02 33.26 3.11 32.83 3.75 31.79 4.71
51
60
32.30 3.73 31.82 3.75 33.94 2.92 28.92 6.75
61
70
33.46 3.92 33.13 1.83 33.70 2.92 33.64 2.92
71
80
35.02 2.72 34.611 1.69 35.52 2.35 34.33 2.79
81
90
TABLE II: MEAN AND STANDARD DEVIATION OF LHCP SNR
θ
S
Forest Area Single Building (Blocked) Single Building (Others) Street Canyon
µ
SNR
σ
SNR
µ
SNR
σ
SNR
µ
SNR
σ
SNR
µ
SNR
σ
SNR
01
10
11
20
22.97 4.12 19.49 3.68 28.63 4.88
21
30
23.69 4.27 21.78 4.24 28.50 4.30
31
40
25.13 3.52 24.76 4.14 32.07 5.27 30.13 6.09
41
50
25.48 3.53 25.47 4.02 29.46 5.04 29.16 6.55
51
60
23.90 4.40 27.82 3.91 29.61 4.76 29.65 6.09
61
70
23.13 3.98 24.75 4.35 28.90 5.11 28.84 4.53
71
80
24.79 3.15 26.71 2.59 31.91 3.06 27.06 3.46
81
90
variance for both RH and LH in Table I and II. Variance
values are widely spread at lower θ
S
and at higher θ
S
, values
are closely distributed for both RH and LH.
V. CONCLUSION
In this article we have presented a data classification method
based on the measurements from various environments. Dif-
ferent parts of each environment and the orbital information
of satellites are used as criterion for data classification. First,
the method involves taking fish eye images of measured routes
and analyzing the images to divide each environment into four
potential sub environments: (1) open area; (2) forest area; (3)
single building blockage; and (4) street canyon. Then, satellites
are classified in nine different groups based on their elevation
and the data form all the satellites that fall in same group
are merged together. Finally, a table is created containing
the information of mean and standard deviation of SNR for
different environments and elevation ranges.
The main drawback of this approach is the tedious and
precise work required to classify data into sub environments
if the area to be covered is extended: the process is only
automated to limited extend. However, in the future more
automatic pattern recognition of the still images could offer a
faster way for SNR evaluation in different environments.
The results obtained with significant amount of data from
three different environments showed good correlation of SNR
between same sub environments. This information will be
used to regenerate a generalized set of SNR parameters in
the laboratory environment for 3D GNSS channel model.
Future efforts will include taking second round of measure-
ments with 2nd generation measurement system equipped with
dual polarized GPS antennas [9], and 360 degree images of
environments. In addition to three environments described in
this paper open and offshore areas will be measured separately.
REFERENCES
[1] A. Bourdeau, M. Sahmoudi, and J.-Y. Tourneret, “Tight Integration of
GNSS and a 3D City Model for Robust Positioning in Urban Canyons,
in 2012, ION GNSS, USA, September 2012.
[2] E. Takeuchi, M. Yamazaki, K. Ohno, and S. Tadokoro, “GPS Measure-
ment Model with Satellite Visibility using 3D map for Particle Filter, in
2011, International Conference on Robotics and Biomimetics (ROBIO),
Phuket, Thailand, December 2011.
[3] K. A. B. Ahmad, M. Sahmoudi, C. Macabiau, A. Bourdeau, and G.
Moura, “Reliable GNSS Positioning in Mixed LOS/NLOS Environments
Using a 3D Model, in 2013, European Navigation Conference (ENC),
2013.
[4] M. D. Foegelle, and R. Borsato, A-GPS Over-The-Air Test
Method: Business and Technology Implications, 2009, White Paper,
[online], Available: http://www.ets-lindgren.com/WhitePaper-ETSL0709
(Accessed: September 20, 2016).
[5] Keysight Technologies, A-GPS OTA Measurements, 2015, [online],
Available: http://literature.cdn.keysight.com/litweb/pdf/5990-6043EN.pdf
(Accessed: September 20, 2016).
[6] M. Berg, R. Lighari, J. Kallankari, V. Majava, A. Parssinen, and E. T.
Salonen, “Polarization Based Measurement System for Analysis of GNSS
Multipath Signals, in 2016, 10th European Conference on Antennas and
Propagation (EuCAP), Davos, Switzerland, April 2016.
[7] R. Lighari, M. Berg, J. Kallankari, A. Parssinen, and E. T. Salonen,
Analysis of the Measured RHCP and LHCP GNSS Signals in Multipath
Environment, in 2016, 6th International Conference on Localization and
GNSS (ICL-GNSS), Barcelona, Spain, June 2016.
[8] R. Akturan, and W. J. Vogel, “Photogrammetric Mobile Satellite Service
Prediction, Electron. Lett., vol. 31, no. 3, pp. 165–166, Feb. 1995.
[9] M. Berg, R. Lighari, T. Tuovinen, and E. T. Salonen, “Circularly Polarized
GPS Antenna for Simultaneous LHCP and RHCP Reception with High
Isolation, in press 2016, Loughborough Antennas and Propagation
Conference (LAPC), Loughborough, UK, November 2016.
Citations
More filters
Journal ArticleDOI
TL;DR: In this paper, a non-destructive approach based on Global Positioning System (GPS) signals was developed to derive SWE, snow height (HS), and snow liquid water content (LWC) simultaneously using one sensor setup only.
Abstract: For numerous hydrological applications, information on snow water equivalent (SWE) and snow liquid water content (LWC) are fundamental. In situ data are much needed for the validation of model and remote sensing products; however, they are often scarce, invasive, expensive, or labor‐intense. We developed a novel nondestructive approach based on Global Positioning System (GPS) signals to derive SWE, snow height (HS), and LWC simultaneously using one sensor setup only. We installed two low‐cost GPS sensors at the high‐alpine site Weissfluhjoch (Switzerland) and processed data for three entire winter seasons between October 2015 and July 2018. One antenna was mounted on a pole, being permanently snow‐free; the other one was placed on the ground and hence seasonally covered by snow. While SWE can be derived by exploiting GPS carrier phases for dry‐snow conditions, the GPS signals are increasingly delayed and attenuated under wet snow. Therefore, we combined carrier phase and signal strength information, dielectric models, and simple snow densification approaches to jointly derive SWE, HS, and LWC. The agreement with the validationmeasurements was very good, even for large values of SWE (>1,000 mm) and HS (> 3 m). Regarding SWE, the agreement (root‐mean‐square error (RMSE); coefficient of determination (R)) for dry snow (41 mm; 0.99) was very high and slightly better than for wet snow (73 mm; 0.93). Regarding HS, the agreement was even better and almost equally good for dry (0.13 m; 0.98) and wet snow (0.14 m; 0.95). The approach presented is suited to establish sensor networks that may improve the spatial and temporal resolution of snow data in remote areas.

31 citations


Cites background from "Classification of GNSS SNR data for..."

  • ...The recorded signal strengths depend mainly on the elevation angle of each GPS satellite passing by, as presented, for example, in Lighari et al. (2017)....

    [...]

Journal ArticleDOI
TL;DR: Zhang et al. as discussed by the authors proposed an environment scenario identification model, laying a foundation for designing a scenario-based navigation strategy, which can identify all defined scenarios, with an overall accuracy of 97.08%.

4 citations

Book ChapterDOI
08 May 2020
TL;DR: In this paper, an attention-based recurrent neural network (RNN) is constructed; the historical characteristics extracted above are fed into the RNN, and high dimensional features are then constructed by soft attention and are used as the input of a fully connected network for classification.
Abstract: In this paper, an environment classification method for Global Navigation Satellite System (GNSS) is presented. The goal of the study is to characterize the statistical properties of the historical GNSS data in certain typical environments, so that appropriate localization or navigation algorithms can be chosen to achieve better performances once any environments are recognized in real practice. We extract Dilute of Precision (DOP) value, Carrier-to-Noise Ratio (C/N) and Number of Satellite in View from NMEA-0183 data collected in three real typical environments to characterize the environments. Further, an attention-based Recurrent Neural Network (RNN) is constructed; the historical characteristics extracted above are fed into the RNN. Attention values are then calculated using real-time characteristics and the RNN output in each time steps. High dimensional features are then constructed by soft attention and are used as the input of a fully connected network for classification. The performance of proposed method on the classification task of three typical environments has significantly improvement compared to recurrent neural networks without attention mechanism, and achieves an average accuracy of 94% on the testing set.

2 citations

Proceedings ArticleDOI
03 Oct 2021
TL;DR: In this article, a machine learning based performance prediction algorithm is suggested considering environment recognition, and valid environmental parameters that support recognition and prediction stages are introduced, and K-Nearest Neighbour, Support Vector Regression and Random Forest algorithms are tested based on their prediction performance with using these parameters.
Abstract: As the primary navigation source, GNSS performance monitoring and prediction have critical importance for the success of mission-critical urban air mobility and cargo applications. In this paper, a novel machine learning based performance prediction algorithm is suggested considering environment recognition. Valid environmental parameters that support recognition and prediction stages are introduced, and K-Nearest Neighbour, Support Vector Regression and Random Forest algorithms are tested based on their prediction performance with using these environmental parameters. Performance prediction results and parameter importances are analyzed based on three types of urban environments (suburban, urban and urban-canyon) with the synthetic data generated by a high quality GNSS simulator.

1 citations

Proceedings ArticleDOI
20 Sep 2022
TL;DR: In this paper , a GNSS-based environmental context detector is proposed to detect the environment surrounding a vehicle in four classes: canyon, open-sky, trees and urban, and a support vector machine classifier is trained on a dataset collected around Toulouse.
Abstract: Context-adaptive navigation is currently considered as one of the potential solutions to achieve a more precise and robust positioning. The goal would be to adapt the sensor parameters and the navigation filter structure so that it takes into account the context-dependant sensor performance, notably GNSS signal degradations. For that, a reliable context detection is essential. This paper proposes a GNSS-based environmental context detector which classifies the environment surrounding a vehicle into four classes: canyon, open-sky, trees and urban. A support-vector machine classifier is trained on our database collected around Toulouse. We first show the classification results of a model based on GNSS data only, revealing its limitation to distinguish trees and urban contexts. For addressing this issue, this paper proposes the vision-enhanced model by adding satellite visibility information from sky segmentation on fisheye camera images. Compared to the GNSS-only model, the proposed vision-enhanced model significantly improved the classification performance and raised an average F1-score from 78% to 86%.
References
More filters
Proceedings ArticleDOI
10 Apr 2016
TL;DR: It is concluded that the polarization based measurement system is able to separate LOS and NLOS signal components thus enabling first steps toward the three-dimensional GNSS channel model development.
Abstract: In this paper, the principle of a polarization-based measurement system intended for the data recording of the Global Navigation Satellite System (GNSS) signals is described. The measurement arrangement includes left- and right hand circular polarized antennas with separate satellite RF recorders for both polarizations. Based on the recorded data for different antenna polarizations the Line-Of-Sight (LOS) and the reflected Non-Line-Of-Sight (NLOS) signal components can be analyzed in order to generate a three-dimensional channel model. Initial field measurements are performed for a route of one kilometer in a university campus area. The initial recorded data include digitized RF signals for all visible Global Positioning System (GPS) satellites, the National Marine Electronics Association (NMEA) data, video recording of the environment and reference velocity from a separate speed sensor. In this paper, the preliminary analysis of the recorded data is presented. Based on the recorded data it is concluded that the polarization based measurement system is able to separate LOS and NLOS signal components thus enabling first steps toward the three-dimensional GNSS channel model development.

9 citations

Proceedings ArticleDOI
08 Aug 2016
TL;DR: Results show that satellite elevation angle, and multipath propagation affect both the position precision measured by the receiver and SNR, and will serve as basis for the development of 3D GNSS channel model to work for both static and dynamic environments.
Abstract: The performance evaluation of Global Navigation Satellite System (GNSS) device in 3D laboratory measurement environment is gaining increasing importance. Even though GNSS is a mature technology the 3D channel model to be implemented in laboratory environment does not exist due to the challenges encountered in creating controllable and repeatable multipath conditions. This research work is a first step toward the one solution of these problems. In this paper, the GNSS data set recorded with the polarization based measurement system is analyzed. Both Right Hand Circularly Polarized (RHCP) and Left Hand Circularly Polarized (LHCP) antennas are employed so that direct and reflected signals can be acquired simultaneously. The goal of the study is to investigate the characteristics of polarization based reflections, path length of delayed multipath signal, position error, coverage efficiency (mean number of tracked satellites), and the impact of satellite elevation angle on received Signal-to-Noise Ratio (SNR) for a typical multipath environment. Results show that satellite elevation angle, and multipath propagation affect both the position precision measured by the receiver and SNR. Additionally, presented results will serve as basis for the development of 3D GNSS channel model to work for both static and dynamic environments.

7 citations

Frequently Asked Questions (10)
Q1. What are the contributions in this paper?

In this paper, a data classification method for analyzing the aspects of Signal-to-Noise Ratio ( SNR ) for Global Navigation Satellite System ( GNSS ) in real conditions is introduced. It consists of: 1 ) taking fish eye images of measured routes ; 2 ) dividing measured environments into four potential sub environments ( open area, forest area, single building blockage, and street canyon ) ; 3 ) classifying satellites into nine different groups as function of elevation angles ; and 4 ) creating a table containing the information of mean and standard deviation of SNR for different environments and satellite elevation angles. 

The main drawback of this approach is the tedious and precise work required to classify data into sub environments if the area to be covered is extended: the process is only automated to limited extend. However, in the future more automatic pattern recognition of the still images could offer a faster way for SNR evaluation in different environments. This information will be used to regenerate a generalized set of SNR parameters in the laboratory environment for 3D GNSS channel model. 

An elevation mask of 10◦ is used for the forest area due to insufficient data and visibility of satellites below 10◦, and the tall building structures of street canyon prevented visibility of satellites for all elevations below 30◦. 

The measured route included trees surrounding the roads, two storey houses with wooden structures, narrow streets, and a three storey office building located nearly 30 meters away from the measured route. 

during the measurement campaign no satellites between the elevation angles of 81◦– 90◦ were received because of the northern location of the measurement routes. 

Future efforts will include taking second round of measurements with 2nd generation measurement system equipped withdual polarized GPS antennas [9], and 360 degree images of environments. 

Elevation and azimuth angles of satellite are represented by θS and φS , respectively and, similarly, the elevation tilt angles of directional receiver antennas are presented with θRHCP and θLHCP . 

The main drawback of this approach is the tedious and precise work required to classify data into sub environments if the area to be covered is extended: the process is only automated to limited extend. 

This information will be used to regenerate a generalized set of SNR parameters in the laboratory environment for 3D GNSS channel model. 

The mentioned GPS data were collected for three different environments; consisting of, university campus, suburban residential, and urban downtown areas.