Content uploaded by Amir Allahvirdi-Zadeh
Author content
All content in this area was uploaded by Amir Allahvirdi-Zadeh on Feb 12, 2024
Content may be subject to copyright.
Ginan Supporting Future LEO-PNT
1
Amir Allahvirdi-Zadeh, Ahmed El-Mowafy, Simon
McClusky, Sebastien Allgeyer, Aaron Hammond
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
GNSS-SPAN GROUP, SCHOOL OF EARTH AND PLANETARY SCIENCES, CURTIN UNIVERSITY
GEOSCIENCE AUSTRALIA
2
Outline
•Why do we need LEO-PNT?
•Precise orbit determination of LEO satellites (LEO POD)
•LEO POD modules in Ginan
•LEO-PNT simulation
•Processing LEO-PNT observations by Ginan
•Summary and conclusion
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
3
Why do we need LEO-PNT?
Signals of opportunity (SoP) Navigation signals from LEO-PNT constellation
•Challenges in using GNSS in critical environments, such as urban area and indoor spaces, etc.
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
LEO-PNT systems
4
Satellites in Low Earth Orbit (LEO) are used for:
-Satellite Gravimetry (SWARM, GRACE-FO)
-GNSS radio occultation (COSMIC-1, -2)
-Satellite altimetry (Jason)
-InSAR (Sentinel 1-6)
-Future LEO-PNT system
Precise Orbits
are required
for these
missions
(Cm- to dm-level of accuracy)
Precise orbit determination of LEO satellites (LEO POD)
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
Project Title: Developing and incorporating Low Earth
Orbiter (LEO) GNSS data analysis capability into Ginan
5
Reduced-Dynamic POD
•Based on solving the equation of motion
•Integrating with the GNSS observations
•Estimating stochastic accelerations to compensate
for dynamic model deficiencies
•Continuous and more accurate orbit
•Cumbersome processing
Kinematic POD:
•Based on kinematic Precise Point Positioning (PPP)
•Sensitive to outliers
•No observation → No orbit
•Bad observation → Low accuracy
Precise orbit determination of LEO satellites (LEO POD) (Cont.)
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
6
Kinematic POD in Ginan
()=+
=[ ]
=,1
,…,,
=
,1
, … ,
,
All observations collected by
LEO satellites at epoch
,=,
1,…,,
,=,
1, … , ,
Observation model
=,,,
,,
Unknown State vector
Position
Velocity Receiver
clock offset
error and
its rate
Ambiguities
Ionospheric
slant delay
•Ginan is the Australian open-source comprehensive software developed by Geoscience
Australia and its partners for processing GNSS observations (for classical PPP, PPP-AR,
PPP-RTK and POD).
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
7
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
Kinematic POD in Ginan
POD in Ginan is based on Extended Kalman Filtering (EKF)
including Time update and Measurement update steps
State transition
matrix (STM)
=
=1
0 1
Identity matrix
for all states
Except for states
with rate terms
Time difference
between two
consecutive epochs
POD at Ginan is
Based on Extended
Kalman Filter
8
Reduced-dynamic POD in Ginan
(
)=
+
Observation model
Unknown state vector
orbit
Velocity
Receiver clock
offset error Ambiguities
Ionospheric
slant delay
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
Empirical accelerations in
radial (R), along-track (T), and
cross-track (N) directions
Equation of motion:
=,,,
=
Satellite state vector: = Initial
condition
accelerations
Numerical
integration
State
vector
9
State Standard deviation ()STD of Process noise
Position 30 m 0 m
Position rate (velocity) 5000 m/s 1000 m/s
Clock 500 m 500 m
Clock rate 500 m/s 0.0001 m/s
Ambiguity 6000 m 0 m
Ionosphere slant TEC 1000 m 8000 m
Empirical acceleration (RTN) 50 m/s0.2 m/s
Initial values for the state vector in the EKF in Ginan
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
POD in Ginan – Models and initial values
Item Description
Dynamic
models
Gravity field:Earth gravitational model (EGM 2008); Tidal corrections:Finite element solution tidal model –
FES2014b; General Relativity:IERS 2010; Planets ephemeris:JPL DE436.1950.2050; Empirical acceleration:
in RTN directions
Observation
models
Observation:Ionospheric-free of code and phase;Attitude:quaternions in ORBEX files;
Antenna corrections: PCO, PC V, antenna sensor offsets
10
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
POD validation in Ginan (Tested satellite GRACE-FO C)
Date RMSE X (cm) RMSE Y (cm) RMSE Z (cm) 3D RMSE (cm)
2019-02-14 3.2 1.2 1.9 3.9
2019-02-15 1.5 1.1 1 2.1
2019-02-16 1.6 1.4 0.9 2.3
2019-02-17 1.6 1 1 2.1
2019-02-18 1.3 1.2 0.9 1.9
2019-02-19 1.5 1.2 1.3 2.3
2019-02-20 2.4 2.5 2.3 4.1
Date RMSE X (cm) RMSE Y (cm) RMSE Z (cm) 3D RMSE (cm)
2019-02-14 5.2 5.3 6.6 9.9
2019-02-15 7.2 7.9 7.6 13.1
2019-02-16 5.1 5.6 6.2 9.7
2019-02-17 4.6 4.6 6.8 9.4
2019-02-18 5.1 5.1 6.9 9.9
2019-02-19 4.8 4.5 6.8 9.4
2019-02-20 4.3 3.4 5.3 7.6
RMSE of
Kinematic
POD
RMSE of
Reduced-
dynamic
POD
External validation: comparing to the JPL reference orbit
11
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
POD validation in Ginan (Tested satellite GRACE-FO C)
Residuals of phase
observations
RMS = 2mm
Internal validation: Residuals of ionosphere-free phase observations
(Each satellite’s phase
residual is represented by a
distinct color)
12
Positioning using Ginan from LEO-PNT systems
Use future LEO-PNT
systems for the positioning
of ground users:
I- independent from GNSS,
as a backup.
II- combined with GNSS
Why Ginan?
EKF suitable for real-
time applications, low-
cost receivers, low-budget
CPUs, etc.
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
LEO-PNT constellation
Ground user
13
Simulating LEO-PNT Constellations
Walker Delta Pattern:
85.64°: 240/12/6
inclination
total
number of
satellites ()
number of
equally spaced
orbital planes ()
interplane
phase
increment ()
=2
( 1)
Right ascension of
ascending node
=2
( 1)+2
( 1)
Mean anomaly
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
240 satellite constellation
with 1000 km altitude
14
Simulating LEO-PNT Observations
Target: Simulating 1 Hz
observations from LEO-PNT
system to CUT0 CORS on
Curtin campus
Ground truth of CUT0
(derived from AUSPOS)
1
Orbits of the LEO-PNT
constellation
2
Considering the observation errors
The simulated code and phase observations are:
3
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
,
=
+
+,
+
,
=
,
+
+,
+
Ionospheric delay
,
=82.1 ×
2sin2+ 0.076 +sin
Ambiguities
Troposphere
15
Coverage of the simulated LEO-PNT constellations for CUT0
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
(Each satellite’s access is
represented by a distinct color)
Satellites coverage
16
Number of the satellites in the simulated LEO-PNT Observations
7 ≤ Number of available LEO satellites ≤ 13 15 ≤ Number of available GPS+LEO satellites ≤ 24
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
17
Processed LEO-PNT observations by Ginan
Ginan has been compiled on a Raspberry Pi 4 equipped with
ARM CPU v8 for precise point positioning (PPP) of CUT0
Less than 1 second for processing each epoch
Showing the
capability of Ginan
for real-time LEO-
PNT applications
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
Processing case RMSE X
(m)
RMSE Y
(m)
RMSE Z
(m)
LEO-PNT 0.042 0.059 0.070
GPS only 0.023 0.018 0.031
GPS + LEO-PNT 0.020 0.017 0.026
Comparing the positioning
results with AUPOS output
18
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
Ginan has the capability to be used for LEO-PNT
applications:
•The LEO POD module can perfume both reduced-
dynamic and kinematic POD of LEOs.
•With 85.64°: 240/12/6 constellation at 1000 km
altitude, 7 to 13 LEO satellites were available for a
tested CORS: CUT0 at the Curtin campus
•Processing of simulated LEO-PNT observations
were provide PPP accuracy of few centimeters
compared to the AUSPOS output
•Adding LEO to the GPS-only case brings more
observations and improves the PPP accuracy.
Using Ginan is promising for the LEO-PNT
applications, it can be used in both LEO-POD and
LEO-PNT parts.
Summary and conclusion
19
Thanks
http://gnss.curtin.edu.au/
IGNSS 2024, GINAN SUPPORTING FUTURE LEO-PNTSYDNEY, NSW, FEB 7 - 9,
ACKNOWLEDGMENTS
We would like to acknowledge Spirent for providing access to the
SimGEN software, and LEAP Australia for providing access to the
Ansys Satellite Tool Kit (Ansys STK) for the LEO-PNT simulation.