International Journal of Pure and Applied Mathematics

Volume 119 No. 12 2018, 1417-1426

ISSN: 1314-3395 (on-line version)
url: http://www.ijpam.eu
Special Issue
ijpam.eu

COMPRESSION OF SAR RAW DATA USING BLOCK ADAPTIVE

QUANTIZATION

1 2 3
Ms.N.Kalaivani Reshma AR D.Devi
1. Assistant Professor
Sri Krishna College of Engineering and Technology
Kuniamuthur, Coimbatore,Tamilnadu,India
2. PG Scholar
Sri Krishna College of Engineering and Technology
Kuniamuthur, Coimbatore,Tamilnadu,India
3. Assistant Professor
Sri Krishna College of Engineering and Technology
Kuniamuthur, Coimbatore,Tamilnadu,India

Abstract

This paper focuses on synthetic aperture radar (SAR) raw data

compression using field-programmable gate array (FPGA) technology. In
particular, this paper centers around the locally available, ongoing
compression for spaceborne SAR frameworks.SAR is an essential instrument
for remote detecting of the World's surface, and creates huge amounts of
information that require compression before downlink by means of a
constrained limit channel. SAR framework configuration, including
information displaying is analysed from the point of view of data
compression, and algorithm usage prerequisites are determined. The
improvement of a straightforward, yet adaptable, FPGA implementation of a
block adaptive quantization (BAQ) compression algorithm is depicted, and
the analysis of its performance is done.

Keywords: - Synthetic Aperture Radar (SAR) ,Block Adaptive

Quantization (BAQ),Field Programmable Gate Array (FPGA), Signal to
Quantization Noise ratio (SQNR), Integrated Sidelobe Ratio (ISLR) Peak
Sidelobe Ratio (PSLR)

1. Introduction

Radar is a object identification syatem that utilizes radio waves to decide the
range,angle, or speed of items which can be utilized to recognize aero planes,

1417
International Journal of Pure and Applied Mathematics Special Issue

ships, shuttle, guided rockets, engine vehicles, climate conditions, and

territory. A radar system consists of a transmitter producing electromagnetic
waves in the radio or microwaves domain, a transmitting antenna, a
receiving antenna (often the same antenna is used for transmitting and
receiving) and a receiver and processor to determine properties of the
object(s). Radio waves (pulsed or continuous) from the transmitter reflect off
the object and return to the receiver, giving information.
Synthetic-aperture radar (SAR) is used to create 2D or 3D images of targets,
like topographies. SAR captures the radar antenna motion over a target
region to give better spatial resolution than traditional beam-scanning
radars. The statistics of a block of data is estimated by the block adaptive
quantizer (BAQ) algorithm and then scalar quantization is used to compress
that block of data. The block statistics are downlinked with the compressed
data to allow reconstruction of the signal. For spaceborne SAR, the raw data
input signal is assumed to fit the complex Gaussian model. In the literature,
the BAQ appears to be the de facto standard, against which all other
algorithms are compared.

2. Literature Review

The block adaptive quantization (BAQ) was created by Kwok and Johnson of
the NASA JPL (Kwok and Johnson, 1989) .As shown by its name, the BAQ
algorithm exploits the gradually changing echo power in range from pulse-to-
pulse to allow the block adaptation of the quantizer. This is proficient by the
assurance of the insights for each block of information samples, which is
utilized to modify the quantizer levels. Since the measurable appropriation of
the real(I) and quadrature (Q) information samples is assumed to be zero-
mean Gaussian with indistinguishable difference, the standard deviation is
the main parameter important to portray the distribution.The decision of
block size is a vital parameter in the BAQ algorithm. It is basically an trade
off between the quantity of samples important to guarantee that the block
will have a Gaussian distribution, and an endeavor to limit the block size to
keep legitimate the stationary signal assumption. The BAQ agorithm is a
non-uniform quantizer improved for a Gaussian probability distribution
where the threshold values are adjusted on a block by block basis, and are
obtained from the block variance.

BAQ Algorithm Compressed data

stream
Variance of
Calculation
data block
of variance

Block of SAR
raw data

Max- Lloyd Coded data

block

1418
International Journal of Pure and Applied Mathematics Special Issue

Fig 1: Block Diagram Of Standard BAQ Algorithm

The raw data is divided into small blocks. The block size is chosen so as to
guarantee Gaussian statistic distribution within a block and the the
maximum block size is constrained by signal power, which ought to be
constant all through the block.The standard deviation of each block is
evaluated by the following equation.

------------------(1)

Where I and Q are the real and imaginary part of the samples. For
the ideal working point, the threshold values are corresponding to the
standard deviation of every block of data. The standard deviation of every
block is transmitted and utilized for the decompression of the information
.For every block, the response levels of a non-uniform quantizer are scaled by
the standard deviation.
3. Research Objectives
The goal of adaptive quantization is to provide effective data compression of a
signal source with time-varying parameters. An adaptive quantizer estimates
the statistics of the source and attempts to match the quantizer to the

SPACECRAFT GROUND

ESTIMATING
SIGNAL STATISTICS

FASCICLE OF
SIGNAL ORIGINAL
RECONSTRUCTION SIGNAL

QUANTIZER
OPTIMIZED FOR
SIGNAL STATISTICS

observed in time-varying statistics.

1419
International Journal of Pure and Applied Mathematics Special Issue

Fig 2: Prospective BAQ block diagram

In the approach taken to design the adaptive quantization scheme, the

model shown in Fig 2 was used. The statistics of a block of incoming data
samples are estimated and a quantizer which is optimized for that source
model is selected for quantization of that block of data. An understanding of
the source statistics is therefore paramount in the design of an optimal
quantizer. The reconstruction of the data samples is also an issue in the
efficient mechanization of the quantization scheme. In a data rate limited
system, the selection of the number of levels of quantization is a trade-off
between the range resolution (or pulse bandwidth) and performance of the
quantizer (quantization noise). The range bandwidth determines the
sampling rate of the echo returns and therefore affects the data rate. In
addition, the performance of the quantizer affects the noise in the digital data
and therefore, the image quality.
4. COMPRESSION ALGORITHM
Lossless compression removes redundancy from a dataset in such a way
that the original information content can be recovered after compression. The
entropy of a dataset gives an indication of the potential for lossless
compression. The entropy H(X) of a discrete random variable X is defined by

------------------(2)

and is the number of bits on average required to describe X. Typical

SAR raw data has high entropy relative to the number of bits used to
quantize the analog signal, and thus the potential for lossless compression is
limited. In addition, SAR data exhibits low sample-to-sample correlation, a
lack of systematic patterns and a relatively high bandwidth compared to the
sampling rate. Therefore in order to achieve data rate reduction in SAR data,
lossy compression is necessary.
Lossy compression can achieve the required data rate reduction, but
does so at the expense of information content. However, due to the relatively
low SNR levels of satellite SAR data (between 5dB and 15dB), a moderate
amount of added noise due to lossy data reduction will not have a significant
impact on image quality.
A variety of lossy compression algorithms for compressing SAR data
can be found in literature, but most commonly implemented is the Block
Adaptive Quantization (BAQ), which uses scalar quantization to achieve
lossy compression. Scalar quantization is the mapping of each observed value
to the nearest approximating value in a predetermined finite set of allowed
values.Lossy compression is achieved by mapping M input levels to N output
levels, where M>N. Various ways to measure quantizer performance are
• RMS phase noise
• Signal to Quantization Noise ratio (SQNR)
• Integrated Sidelobe Ratio (ISLR)

1420
International Journal of Pure and Applied Mathematics Special Issue

• Peak Sidelobe Ratio (PSLR)

• Visual comparison of processed images
5. SAR DATA ACQUISITION
Radar echoes are down converted from Ku band to an IF of 1750 MHz in
analog receiver and ADC which is operating at 1GHz, samples the analog
signal to 12 bit wide samples.

There are two parallel chains each for vertical and horizontal
Fig 3 : SAR Data Acquisition

polarizations. ADC output is provided to a 1:4 DMUX for reducing

input data rate to the FPGA. Digital down conversion and filtering is
carried out in each of the four channels and decimated In phase and
quadrature phase signals with 12 bit wide samples are passed onto
block adaptive quantization modules to reduce the sample width to
4bits per sample. 32 range lines are considered as one batch since
block size considered for BAQ is 32x32. With a sampling frequency of
250MHz, transmit pulse width of 33uS and imaging swath of 5km,
250e6x (33e-6+33e-6) x32=528000 samples will be generated and when
it is further divided into blocks of 32x32 samples, 520 blocks will be
formed. On a raster scan basis, all the blocks of the batch will undergo
BAQ compression. Compression can be done on different blocks in
parallel. No. of parallel compression modules will be dictated by
hardware resource availability and data rate
requirements.Compression related information used for decompression
and other radar metadata are attached along with compressed IQ data
in the formatter and finally provided to satellite Baseband Data
Handler (BDH).

6. SIMULATION RESULTS

BAQ encoding is actualized by separating the SAR raw informational set into
information blocks of little size.For each block standard deviation esteem is
calculated.

1421
International Journal of Pure and Applied Mathematics Special Issue

a) Impulse response of range in stripmap mode 2 b) Impulse

response of azimuth in stripmap mode 2 c) image obtained

d) particular zoomed image of BAQ model

6.1 SIMULATION IN FPGA

FPGA technology is suitable for the implementation of signal processing in

space applications, especially given that most of the space-based systems are
highly specialized and custom-built. Radiation-hardened, high-reliability
versions of FPGAs are widely-used in space. In addition, FPGA features
include multiple clock domain support and a large number of 10 pins, thus
allowing various radar functions (e.g., pulse generation, timing and control,
filtering, data compression) to be implemented on the same chip.

1422
International Journal of Pure and Applied Mathematics Special Issue

FIG 4 : Simulink model

6.2 IMPLEMENTATION RESULTS

FPGA implementation is obtained by HDL coder which is used to generate

portable,synthesizable,verilog and vhdl codes from matlab functions,
simulink models and state flow charts. The generated hdl code can be used
for FPGA programming. It provides an traceblity between Simulink model
and generated vhdl and Verilog code enabling code verification for high
integrity applications

FIG 5. a) Impulse response of range in stripmap mode 2 b) Impulse response of

azimuth in stripmap mode 2 c) image obtained .

1423
International Journal of Pure and Applied Mathematics Special Issue

d) zoomed image obtained in FPGA implementation

7. DISCUSSIONS AND CONCLUSIONS

The paper described the simulation of BAQ process and its FPGA
implementation including the development of a practical test bench suitable
for further radar development. The combination of Matlab, Simulink and
HDL Coder provides an efficient method for developing, testing and
implementing digital designs.The standard BAQ algorithm utilizes a small
percentage of the available resources of modern FPGAs, and thus there is
significant scope for more advanced compression algorithms and/or more
radar functions on a single FPGA chip.
8. REFERENCES
[1] R. Kwok, and W.T.K. Johnson, “Block adaptive quantization of Magellan
SAR data”, IEEE Transactions on Geoscience and Remote Sensing, vol. 27,
no. 4, pp. 375-383, 1989.
[2] Y. M. Zhu, D. Y. Zhu, and Z. D. Zhu, “An improved algorithm in the
application of SAR raw data compression”, Airborne Radar, vol. 1, no. 145,
pp. 25-30, 2004.
[3] Z. H. Guan, D. Y. Zhu, and Z. D. Zhu, “Compression of SAR raw data by
block adaptive spherical vector quantization”, Acta Aeronoutica et
Astronautica Sinica, vol.27, no. 1, pp. 82-86, 2006.
[4] U. Benz, K. Strodl, and A. Moreia, “A comparison of several algorithms for
SAR raw data compression”, IEEE Transactions on Geoscience and Remote
Sensing, vol. 33, no.5, pp. 1266-1276, 1995.
[5] M.W. Marcellin, and T. R. Fischer, “Trellis coded quantization of
memoryless and Gauss-Markov sources”,IEEE Transactions on
Communications, vol. 38, no. 1, pp. 82-93, 1990.
[6] G. Poggi, A. R. P. Ragozini, and L. Verdoliva, “Compression of SAR data
through range focusing and
variable-rate vector quantization”, IEEE Transactions on Geoscience and
Remote Sensing, vol. 38, no. 3, pp. 1282- 1289, 2000.

1424
International Journal of Pure and Applied Mathematics Special Issue

[7] C. D’Elia, G. Poggi, and L. Verdoliva, “Compression of SAR data through

range focusing and variable-rate trelliscoded quantization”, IEEE
Transactions on Image Processing,vol. 10, no. 9, pp. 1278-1286, 2001.
[8] E. Magli, and G. Olmo, “Lossy predictive coding of SAR raw data”, IEEE
Transactions on Geoscience and Remote Sensing, vol. 41, no. 5, pp. 977-987,
2003.
[9] X. J. Wang, H. Sun, and B. Guan, “Evaluation for coherent speckle
suppression filters of SAR images”, Systems
[10] SAR RAW DATA COMPRESSION USING FPGA TECHNOLOGY
Robert C Huxtable BASc., Simon Fraser University, British Columbia,
Canada, 2005
[11] QUANTIZING FOR MINIMUM DISTORTION, JOEL MAX
[12] DESIGN AND IMPLEMENTATION OF SAR RAW DATA BAQ BASED
ON FPGA Sheng-Lei,Zheng- TaoYe,Zhang-XuJing,East China Research
Institute of Electronic Engineering, Hefei China
[13] BLOCK ADAPTIVE QUANTIZATION OF MAGELLAN SAR DATA
Ronald Kwok, Member IEEE,and William T.K Johnson

1425
1426