Prototype Software-based Receiver
for Remote Sensing using
Reflected GPS Signals
Dinesh Manandhar
The University of Tokyo
[email protected]
1
Contents
•
•
•
•
•
•
Background
Remote Sensing Capability
System Architecture
Data Observation
Algorithm (Signal Processing)
Results and Discussions
2
GPS Signal Characteristic
• GPS signal is RHCP (Right Hand Circular Polarization) type
• The polarization of GPS signal may change when reflected from RHCP to
LHCP and vice versa
– Based on reflecting material type and signal incidence angle
• The amplitude reduces for every reflection, because:
–
–
–
–
The reflection coefficient is less than one
Some of the signal is absorbed
Reflection may not be perfectly specular
The phase of the signal changes that might be neither RHCP nor LHCP
(elliptical polarization)
• The chip delay increases for every reflection
– Since it needs to travel extra distance
• Thus, the analysis of relative amplitude between the direct signal and
reflected signal provides information about reflecting material type
• The chip delay corresponds to the path delay length or multipath amount
3
Remote Sensing Capability using
GPS/GNSS Signal
• GPS / GNSS signals around 1.5Ghz is good for soil moisture
analysis
– The dielectric value difference is about 10 times between dry soil and
wet soil
• This value is about 30times more for water than dry soil
• Hence it provides very strong sensitivity to soil moisture at L-band
– Beside soil moisture, there are many other applications where GPS
signals can be used for remote sensing applications
• Wind velocity over ocean, ocean observation, ice monitoring etc
– GPS / GNSS signals is available to all, at any time and at free of cost,
• An active radar can be tracked or detected while it’s
observing, but not the passive one like GPS because it does
not transmit any signal for observation
– The signals are continuously transmitted only for navigation purpose
– You can observe the things without being detected
• Important aspect in military applications
4
Reflection Coefficient of materials for
Horizontal and Verticle Polarization
Linear Reflection Coef ficient, Horizontal
Linear Reflection Coeff.
1
Fresh W ater
0.8
W et Ground
Medium Dry Ground
Dry Ground
0.6
0.4
0.2
Sea W ater
Concrete
0
10
20
30
40
50
60
70
80
90
Linear Reflection Coeff icient, V erticle
Linear Reflection Coeff.
1
Fresh W ater
0.8
Sea W ater
W et Ground
0.6
Medium Dry Ground
0.4
Dry Ground
0.2
0
Concrete
0
10
20
30
40
50
Propagation A ngle
60
70
80
90
This slide shows coefficient of reflection at 1.5Ghz fro different incident angles.
Circular polarization can be divided into two components, which are horizontal and
verticle components. As we can see in the graphs that, the reflection coefficient
values smoothly decreases with respect to incident angle for horizontal component.
Where as for verticle component, it decreases fast upto certain angle called
Brewster's angle then it raises smoothly after that angle. Hence, the characteristics
of horizontal and verticle components are quite different. A circular polarization can
be modeled by the vector sum of horizontal land vertical components to estimate
the overall effect.
5
Change in Phase Angle for Horizontal and
Verticle Polarization
Phase Angle of Reflection, Horizontal
200
Phase Angle
150
Concrete
Dry Ground
100
Medium Dry Ground
Wet Ground
50
Fresh Water
Sea Water
0
0
10
20
30
40
50
Propagation Angle
60
70
80
90
Phase Angle of Reflection, Verticle
200
Phase Angle
Concrete
150
Dry Ground
100
Medium Dry Ground
Wet Ground
Fresh Water
Sea Water
50
0
0
10
20
30
40
50
Propagation Angle
60
70
80
90
This slide shows the effect of reflection on phase for horizontal and vertical
components. When a circularly polarized signal is reflected, it’s horizontal
component will always have 180 degree phase reversal. Where as the verticle
component will have phase reversal only if the propagation angle is below
Brewster’s angle. If the phase changes for both horizontal and vertical components,
then we will have left hand circular polarization since GPS signal is right hand
circular polarization. However, if the propagation angle is greater than the
Brewster’s angle, then we will have elliptical polarization which is neither perfectly
left hand nor right hand. This is the case in most of the cases for reflected signals.
6
System Architecture
RH
Master Antenna (RHCP)
RF
Front-End
GPS signal
Processing pc
Hard Disk 1
LH
Slave Antennas (LHCP)
Antenna Orientation
RHCP : Zenith (Sky)
LHCP : Nadir (Ground)
The system consists of two antennas, a front-end device, a PC and an external high
speed hard disk.
The front-end down-converts the analog signal from the antenna at 1.5Ghz to
16Mhz with a centre frequency of 4Mhz at 4bit resolution. It is necessary to do so
since we need a sampled digital data of the original analog data at a level data a
standard PC can process.
We use RHCP (right hand circular polarization) antenna for GPS since GPS signal
transmitted from the satellite is RHCP. However, in our case we also use reversed
polarized antenna, which is LHCP (left hand circular polarization). The reason for
using LHCP antenna is that the reflected signal in most of the cases is no more
RHCP unless it reflected twice where the signal changes polarization from RHCP to
LHCP and then back to RHCP again. But, multiple reflection signals are very weak
compared to single reflection. As, we have seen in the previous slides, normally
reflected signals are either LHCP or elliptical. Hence the use of LHCP antenna
provides us better SNR (signal to noise ratio). The RHCP antenna is oriented
towards the sky to receive direct signal, without multipath as far as possible. The
LHCP antenna is oriented downwards to receive reflected signal. The orientation
angle of the LHCP antenna has big impact on received signal based on satellite
geometry in view. As we discussed in the previous slide, the reflected signal
coefficient and phase change with respect to incident or propagation angle. Hence,
it is very important to understand the impact of orientation angle with respect to
satellite geometry if the application is intended for remote sensing purpose.
7
Data Observation, Rooftop of University Bldg
• Data observation height :
about 20m
• Antenna Used
:
LPA Passive
• Data logged on different days
8
Data Observation, Tower
• Data Obs. Ht: 87m
• Antenna Used: LPA
Passive
• Data logged for a
single day only
Location of Antenna
9
Power Spectrum and Histogram
of Raw Data
This slide shows the PSD (Power Spectrum Density) of raw GPS (IF Data) signals
(TOP) and histograms of the data (bottom) .
TOP: Blue color shows data from Reflected Signal or LHCP antenna
Red Color shows data from Direct Signal or RHCP antenna
BOTTOM Right: Histogram for Direct or RHCP antenna data
Bottom Left : Histogram for Reflect or LHCP data
10
Algorithm, Signal Processing
RHCP
Data
Tracking
∑ (I
PRN
only
PRN for LHCP
PRN &
Carrier
Code and Carrier Phase from RHCP tracking
I-Channel
0.125 chip
Q-Channel
0.125 chip
I-Channel
nth chip
X, Y, Z
Vx, Vy, Vz
Position
Navigation
2
+ Q2)
RHCP Power
∑ (I
2
+ Q2)
Q-Channel
nth chip
2
+ Q2)
∑ (I
2
+ Q2)
Corresponds
to Path Delay
∑ (I
LHCP Power
LHCP
Data
Corresponds
to Reflecting
Material
Character
Acquisition
The basic algorithm concept:
Perform acquisition on RHCP data and compute code phase and doppler frequency
Assume that the doppler frequency is the same since both the antennas are at the
same platform and they move together.
Use the code phase and doppler frequency computed from RHCP antenna to
compute the relative delay in LHCP data with respect to RHCP data.
Doppler frequency computed from RHCP is used for acquisition in LHCP data. Only
code phase is computed at various chip delays beginning from the chip delay
computed for RHCP data.
This is also called Master-Slave processing. In our case, Master is RHCP and slave
is LHCP.
11
Basic Concept
Difference in Amplitude
Provides Information
about Reflecting Material
Chip Delay corresponds to
Extra Path (Multipath?)
Basically, we can observe two parameters:
(1) Relative difference in Chip Delays
(2) Relative difference in Amplitude
If there is a delay in incoming signal in LHCP antenna, there will be chip delay
difference between the RHCP and LHCP peaks.
By observing the relative chip delay between the two antennas, we can estimate the
amount of path delay.
Another point of observation is the relative difference between the amplitude of the
two signals.
The difference in amplitude is due to the difference in electrical property of the
reflecting material.
Hence, by observing the relative amplitude difference, it might be possible to model
the reflecting object type.
This is a sort of remote sensing using GNSS signal.
12
Results, Tower Data
• PRN ID : 30
• Antenna Orientation
– RHCP – UP
– LHCP – Down
• Antenna Height
– 87mtr.
This slide shows results of data taken from tower at a height of about 87m from the ground. The satellite visibility chart shows
the visible satellites and their location at the time of survey. In this case, we have taken satellite 30 for analysis. Satellite 30 is
visible at an elevation of about 58 degrees and azimuth of about 40degrees. The LHCP antenna was oriented towards the
ground at an “down looking angle” of about 55degrees from nadir. The azimuth direction is similar to satellite 30.
The figure on the bottom left shows C/No (carrier to noise) with respect to satellite azimuth and elevation angle for RHCP and
LHCP antenna data. As expected, we can see stronger C/No for RHCP data since the signal is coming directly into the
antenna. The LHCP signal is weaker compared to RHCP signal but it is not so weak. We can also see the fading patterns in
LHCP signal which is due to reflection of the signal. We have data just for a few minutes.
The top figure at the right side of the slide shows correlation output power for RHCP (red) and LHCP (blue) signal. We can
see very clear difference between the two. Direct signal is stronger and reflected signal is weaker. Also, the reflected signal is
offset from the centre (prompt value) by half a chip (one chip is one bit of PRN code which is about 300m long in terms of
distance). This is due to extra distance the signal has to travel because of reflection.
The middle figure on the right side shows I and Q channel (cosine and sine components) powers. A normal signal will exhibit
only power in I channel and noise in Q channel as we see in the figure. The I and Q plot also represents the navigation bits.
We can clearly see the sequence of navigation bits with phase reversal at multiples of 20msec, since a navigation bit is
20msec long.
However, in the bottom figure, we can see power on both I and Q channels. This is probably due to the fact that the reflected
signal is neither LHCP nor RHCP. It is a sort of elliptical polarization and hence the power is divided into both channels. As
time lapses, the power level in Q channel will be changing. It may also change in I channel.
The key observation here is power level difference between the two antenna data and their relative chip delay. The difference
in power level is due to different reflection characteristic of the reflecting material. Dry ground will have very low reflection
coefficient and hence the power level will be very low compared to wet ground or water.
13
Antenna Height Estimation
Extra Path Delay:
δ R = 2h sin (θ )
h=
θ
P
δR
θ
2 sin (θ )
h
θ
θ
-h
• Multipath Model
δR
– Forward Scattering as shown above
P’
• The extra path delay is about 0.5 chip delay which is about 150m
• The antenna elevation angle is about 55degree from the nadir
• Thus, height of the antenna, h = 150/2*sin(55*pi/180) = 92m
– This is in very close proximity of actual antenna height from the ground
14
Results, Tower Data
• PRN ID : 1
• Antenna Orientation
– RHCP – UP
– LHCP – Down
• Antenna Height
– 87mtr
15
Results, Tower Data
• PRN ID : 5
• Antenna Orientation
– RHCP – UP
– LHCP – Down
• Antenna Height
– 87mtr
16
Results, Tower Data
• PRN ID : 25
• Antenna Orientation
– RHCP – UP
– LHCP – Down
• Antenna Height
– 87mtr.
17
Results, Roof Data
• PRN ID : 14
• Antenna Orientation
– RHCP – UP
– LHCP – Down
• Antenna Height
– 20mtr.
18
Results, Roof Data
• PRN ID : 6
• Antenna Orientation
– RHCP – UP
– LHCP – Down
• Antenna Height
– 20mtr.
19
Summary
• LHCP Antenna Orientation and Gain Pattern
– The orientation of the antenna with respect to the satellite geometry
(azimuth and elevation) has effect on power level
– Hence, the total power computation shall be done by considering the
satellite geometry and antenna orientation considering the gain pattern.
The power value shall be normalized for easy comparison
– A narrow beam antenna will provide good footprint resolution, but this
also limits the satellite visibility duration
• Probably, an array of antennas is needed for wide coverage with better
resolution
• Difficult to automate the selection of satellite that has good reflected
signal
– Both amplitude and delay
– It seems that similar elevation angle of the satellite and LHCP antenna
provides best result
• For example, if the satellite is at 25degrees above horizon and the LHCP
antenna is oriented at 60degrees down, probably there will be no reflected
signal
– It seems that an angle of 40-60 degree provide good data provided
there are visible satellites around that elevation as well
• This needs further experiments and also depends on the antenna gain
pattern
20
Summary
• Correlator Output and spacing
– Very small chip delays (less that two sampled chips) are not
clearly identifiable
• But, we can see reflected signal with both the direct and reflected
peak almost at the same correlator location
• Probably, we need wider bandwidth front-end in order to detect the
peaks at very narrow correlator values, e.g. 0.0625 or less in our
case
• Integration of Samples
– Long integration (averaging of I and Q power) period provides
reduced noise
– However, we may loose the important observation point,
especially in dynamic platform
• We have not modeled the amplitude value with respect
to ground condition due to lack of enough samples at
different time intervals
– This needs observation together with validation (reference) data
– Modeling for remote sensing applications need data from both
static and dynamic platforms
21
Future Plans
• Conduct Survey using UAV (remote
controlled helicopter)
– Test Flight Scheduled in Oct
• Conduct Survey at Experimental Ground
of the University where a ground based
Microwave Radar is stationed
– This will provide good reference data,
especially for soil moisture
22
Thanks a lot !
23