21 R
Bandwidth chirp signal entering RF system limited to Ms quadrature therefore
submeter resolution can be achieved. Chirp signal can be stated as equation below : y r
Where B is bandwidth of chirp signal, T is pulsewidth and r is pulse amplitude modulation signal.
Azimuth resolution depend on SAR antenna length, as can be stated : z
Where L is physical antenna length in azimuth direction.
.
2.4 Sample Window Time and PRF
Sampling data of SAR will be divided to be two dimension in range and azimuth, which fast time in range and slow time in azimuth direction. Fast time sample should be fast
enough to detect chirp signal, as nyquist criteria it has to be minimum twice of chirp bandwith. Sampling window time in one range will be depent on the maximum range
and pulse width of chirp signal as can be stated: t
rw
2 c
T While in azimuth direction, window sample timing depent on the speed of UAV and
antenna dB azimuth beamwidth as can be stated below: t
az az
v PRF selection should be considered in order to collect unambigue range and doppler
frequency which is represent by minimum and maximum chosen value criteria Mahafza,
. c
2 R
max
R
min
sin
2.5 Graphical User Interface
Graphical user interface was made to provide changing signal parameter, SAR mode operation and time duration to be implemented on SAR data acquisition system. Visual
c will be used to realize this GU, the flowchart can be shown in figure and .
22 Figure . GU Single Mode FlowChart
Figure . GU Observation Mode Flowchart From figure and operator can change some parameter of SAR mode, bandwidth,
pulsewidth, timing window, and ADC sampling frequency. Specific command will be sent to FPGA through UDP communication. After receiving data, in single mode they will
be displayed on chart directly while in observation mode they will be saved to file before some sample data displayed to chart.
23
3. Experiment
Experiment data acquisition CP‐SAR was tested using hardware in the loop methode which is described condition operation before real flight test.
3.1 Experimental Setup
n this hardware in the loop experiment we use JX platform as referenced, which antenna dimension . m length and . m width. The off nadir angle of antenna adjust
to . degree. JX fly on low altitude meter to obey UAV flight limitation rule in
ndonesia jonan, with cruise speed
kmh. 3.2
Experimental Result
This experiment setup condition lead . degree for Range direction dB beamwidth and .
degree azimuth direction dB beamwidth such as shown in figure .
Figure 6. Antenna 3dB Beamwidth
Combine with UAV flight condition and antenna on stripmap mode SAR provide timing result information on figure .
24
Figure 7. Stripmap Geometry SAR result
Baseband signal reference to be transmitted is shown in figure 8 which has Mz
bandwidth bilinear chirp signal inphase and quadrature.
Figure 8. Signal Transmitted Reference The backscattered signal raw data from fpgaPC through gigabit ethernet can be shown
in figure . t is delayed form of Tx signal due to target distance from UAV and added random noise function.
2
Figure 9. Signal Transmitted Reference
The matched filter signal is convolution signal reference and signal received as can be shown in figure .
Figure . Matched Filter Result t is clear that target can be detected using matched filter.
4. Result Analysis
Antenna as SAR sensor mount on JX‐ has length dimension = . meter and antenna width dimension . meter. These parameter will lead dB beamwidth in asimuth and
2 range as .
O
and .
O
respectively. n order to reach
O
to
O
incident angle of image the look angle of antenna should be adjust to .
O
inside JX fuselag. UAV JX will fly at
meter low altitude, with s Pulse repetition frequency
PRF . The PRF value is higher than PRF minimum limitation s due to cruise speed
of JX at kmh antenna length . meters. n addition chirp signal has very high
bandwith Ms in order to get sub meter range resolution.
This Single polarisation system C Band SAR will provide in‐phase and quadrature signal with bit resolution ADC and Gsps. At this low altitude SAR geometry the receiving
window time will be very short only µS. Raw data collected each pulse will be bit, then they will be delivered to laptop hardisk by gigabit ethernet UDP in order to be
saved to hardisk. The transfer time of raw data to laptop only µS or . from PR, which is very enough transfer time in one pulse. This low altitude SAR operation will be
collected data for seconds continuously for 8 meter flight distance, with PRF
it will collect Mbytes which is enough space for external RAM of FGPA to buffer all
measurement data. 5.
Conclusion
Network based data acquisition of C band CP‐SAR sensor on UAV was described. ardware in the loop test also was done for low altitude UAV operation for ensure that
the system working well. From the result and discussion it is ready for test using real embedded system in near future for natural disaster monitoring.
Acknowledge Thank you for reseach support from Riset‐Pro Ristek‐Dikti ndonesia and collaboration
research team in JMRSL Ceres Chiba University for this research project.
References
Bayuaji et al. . ALOS PALSAR D‐nSAR for Land Subsidence Mapping in Jakarta, ndonesia.
Canadian Journal of Remote Sensing. Vol. , No. , pp. ‐8. Fikar et al.
. A Decision Support System for Coordinated Disaster Relief Distribution. Expert System With Application
‐ .
Freeman and S. Saatchi. . On The Detection of Faraday Rotation in Linearly Polarized L‐
Band SAR Backscatter Signature. EEE Transactions On Geoscience And Remote Sensing, Vol. . No. 8.
Jonan. . Pengendalian Pengoperasian Pesawat Tanpa Awak di Ruang Udara Yang Dilayani
ndonesia. Peraturan Menteri Perhubungan Republik ndonesia. Nomor PM . Jakarta,
ndonesia. Li et al.
. Quick mage Processing Method of UAV Without Control Point Data in Earthquacke Disaster Area. Transaction of Nonferrous Metals Society Of China. S
‐S 8. Elsevier Science Press. China.
Mahafza . Matlab Simulations for Radar System Design. CRC Press. USA.
Joint Scientific Symposium
IJJSS 2016
Chiba, 20‐24 November 2016
27
Topic : Computer Science
CP ‐SAR Image Processing System with Kintex‐7 FPGA Board
Masaru Bunya
a
, Kazuteru Namba
a
, Josaphat Tetuko Sri Sumantyo
b
a
Graduate School of Advanced Integration Science, Chiba University, Chiba, Japan
b
Center for Environmental Remote Sensing, Chiba University, Chiba, Japan
Abstract
Synthetic Aperture Radar SAR is a class of multipurpose sensors, which can operate in all‐weather and day‐night time. n Chiba University, We have a plan a
verification experiment for Circularly Polarized SAR CP‐SAR using Unmanned Aerial Vehicle UAV . Since CP wave is not affected by an ionosphere, it’s expected to allow for
more precise observation. Raw image data received by CP‐SAR needs to be applied image processing, but it will not be processed on the ground. Considering the
communication time, only the image processed on UAV is sent. Our previous image processing system needs high power consumption and takes
up space because it uses Vertex‐ FPGA Board ML , Spartan‐ FPGA Board SP
and one PC which has ntel Atom processor and a Solid State Drive SSD . n this system,
Range Doppler Algorithm RDA which is major image processing algorithm for SAR is executed in ML
and the others is used to save the image to SSD. Our proposed system only uses Kintex‐ FPGA Board KC‐
with XM Connectivity Card. XM
makes it possible for KC‐ to use SATA; it can connect SSD without a PC. t’s expected
to reduce power consumption, space, time of development and modification. Our CP‐SAR image processing system is required semi‐real‐time observation.
KC has high performance than ML
so we expect it to satisfy this requirement and to reduce image processing time. The proposed system is applicable to SAR image
processing system on an UAV making semi‐real‐time observation.
Keywords
Keywords: CP‐SAR, SAR image processing, on‐board processing, FPGA, Kintex‐7
1. Introduction
To observe global environment, remote sensing technology is necessary. n particular, Synthetic Aperture Radar SAR which allows high‐resolution observation in
all‐weather and day‐night time is used in variety of fields [ ]. Raw image data received by SAR needs to be applied image processing. n order to apply image processing, DSP
had been often used. But recently, Field Programmable Gate Arrays FPGA has been used as its performance improved [ ]‐[ ].
28 We are developing a Circularly Polarized CP ‐SAR system on small satellite.
Since circularly polarized wave is not affected by earth’s ionosphere, CP‐SAR is expected to allow for more precise observation.
Now, we have a plan a verification experiment for CP‐SAR using Unmanned Aerial Vehicle UAV [ ]. To do image processing, our system apply Range Doppler
Algorithm RDA to raw image data. n generally, taking into time of communication with the ground, image processing is executed on platform plane and satellite . n this
verification experiment, CP‐SAR and image processing is executed on UAV with FPGA. After UAV landed on the ground, we analyze the data. To achieve semi‐real‐time
observation, our system is needed to make one image in second. The rest of the paper is organized as follow. Section introduces the
preliminaries of SAR and FPGA. Section shows our current system. Section presents the proposed system. Section evaluates and section draws the conclusion.
2. Preliminaries
2.1. FPGA
FPGA is well known devices concerning reconfigurable hardware. FPGAs consist of an array of programmable logic blocks surrounded by a programmable routing fabric
that allows blocks to be programmably interconnected. n high performance system, it has advantages in terms of cost and power consumption [ ] [ ]. n order to reduce the
development period, various P core has been provided by each vender. n this paper, we use P Core Generator provided by Xilinx. The FPGA configuration is generally
specified using a hardware description language. The features of flexibility, it is often used for control of the equipment with special purpose and aerospace equipment.
2.2. SAR
SAR, sends itself a microwave and measures the scattered waves, is active image radar enabling high azimuthal resolution in the resulting image despite a physically
small antenna. The larger the antenna, the resolution of the image is higher, but there is a limit of increase the size of the antenna on satellite and plane. To achieve a large
antenna virtually, small antenna observes many times in moving as shown in fig. . A radar on satellite and plane is called platform. Fig. is the instrument arrangement and
the name of the radar.Using a microwave characterized in that passing through the clouds, it can observe earth in all‐weather and day‐night time. Thus, it has been applied
to an earthquake prediction by the monitoring of small crustal movement, an observation in a disaster and a geological survey.
Figure 1. Synthetic Aperture Radar.
Figure 2. .
The instrument arrangement and the name of the radar
29 SAR needs to send the data to the ground. But the raw data received by SAR is too large
to communicate with the ground. t sends only the image processed on platform since it takes a long time and power.
SAR generally uses linearly polarized wave. Linearly polarized wave is received twice the Faraday effect at the time of transmission and reception as it passes through
the ionosphere. owever, circularly polarized is able to ignore the effect by having a uniform amplitude in all directions. Therefore CP‐SAR is expected to be observed in the
better than the SAR using linearly polarized wave [8]. 2.3.
Range Doppler Algorithm
Raw image data received by SAR needs to apply image processing in order to correct the gap which occurred due to the movement during observation. RDA is the
commonly used algorithm for processing continuously collected SAR data into an image. The main steps of RDA used in our system are shown in fig. . n collecting the SAR data,
a long‐duration linear FM pulse is transmitted. This allows the pulse energy to be transmitted with a lower peak power. The linear FM pulse has the property that, when
filtered with a matched filter, the result is a narrow pulse in which all the pulse energy has been collected to the peak value. This matched filtering of the received echo is called
range compression. Azimuth compression is a matched filtering of the azimuth signal,
performed efficiently using Fast Fourier Transforms [ ].Our system only make a Single Look Complex SLC image by using RDA.
Figure 3. Chart of Range Doppler Algorithm.
3. Current System
Our system consists of parts; signal processing of CP‐SAR part and image processing part, as shown in fig. . At first, signal processing part gets raw data by using
CP‐SAR and sends data including information of SAR to image processing part. After applying RDA at image processing part, the image data is saved in SSD.
Our image processing part in fig. has one PC which has ntel Atom processor and a Solid State Drive SSD and two FPGA Boards, Vertex‐ FPGA Board ML
for the target device xc vlx
t‐ ffg is used and Spartan‐ FPGA Board SP
. n ML ,
30 it receives raw data gained by signal processing part through Ethernet and apply RDA to
make a SLC image. SP which connected to ML
by FMC cable sends the image to PC to save. PCe is used between SP
and PC. These FPGAs is implemented in Verilog ardware Descriptive Language VDL . Since our systems has one PC and two FPGA
Board, it needs large power consumption, space and time.
Figure 4. Current System. Figure 5. Image Processing Part.
4. Proposed System
Our proposed system has only one FPGA Board Kintex‐ KC
for the target xc k t‐ ffg
c is used with XM Connectivity Card.
Since current system has low resources, it has to need PC for control and divide system into unit; image processing unit, data saving unit. When it saves the image to SSD,
SP had to use the PC because SP
does not have SATA port. owever, XM which has SATA port does not need PC and SP
.
Figure 6. Proposed System. Figure 7. Image Processing Part in Proposed System.
5. Evaluation
Our System is needed semi‐time‐observation, low space and low power consumption in order to observe on small satellite. Table shows specification ML
and KC [ ] [ ]. KC
has more logic cells than ML . Current systems used
8 resources of ML only image processing. KC
is expected to send or receive image data, to execute image processing, to save the image to SSD on only one board.
Equipment on UAV has to be a space‐saving, power‐saving, lightweight. Fig.8, and table show comparisons of current system and proposed system. Proposed system can
reduce . kg than current system.
Table . Specification ML and KC