DefenseElectronics

High-resolution radar system modeling

with MATLAB/SIMULINK
Modeling novel types of imaging radar systems is a robust and inexpensive way to gauge the
system’s expected performance and create a proof-of-concept design guide. The article illustrates
this with the example of a simulation study of UWB SAR based on a multifrequency signal.

By Dmitriy Garmatyuk

R adar systems have been used extensively

over the past decade for a variety of ap-
plications and in a multitude of configurations.
as discovery of buried mines and unexploded
ordnance [2-3], or assessment of polar ice cap
dynamics[4], or as a surveillance and target
resolution that corresponds to that size.
In recent years, the interest in submeter
resolution SAR has grown, in part thanks
One of the relatively more modern implemen- tracking tool in reconnaissance operations[5]. to continued research into automatic target
tations—although its origins are traced back In all scenarios, however, radar engineers recognition (ATR) algorithms. When it is
to the end of the 1950s[1]—is imaging radar, wishing to improve the resolution of resultant desired to not only detect, but also to identify
in particular, synthetic aperture radar (SAR). imagery cannot evade the underlying principle the target as belonging to a certain category
Imaging radars are used to obtain visual in- of inverse relationship between a radar signal’s (e.g., make and model of a tank), high resolu-
formation about the environment of interest, bandwidth and a minimum imaged feature di- tion of obtained radar images is of much im-
often with the goal of discerning particular mension in range coordinate. Thus, to properly portance[6 -7]. Thus, among the relatively new
objects concealed either intentionally or un- distinguish between target components of par- R&D directions in the field of radar systems
intentionally in the background. These radars ticular size, one needs to select the bandwidth is design and analysis of ultrawideband, or
can be geared toward certain scenarios, such of imaging radar signal accommodating range UWB (see Author’s Note) waveforms and

�
����������� �
�����
������ � ���� ��������
�����
������� ����������
������ ���� ������ ��� �
�������������� �
�����������
����������

��������������
������������ ���������������������������
��������� ������� � �
������ �������� �������
���������������� ��������� � �
���������
������ �
�����������
����������� ������
�����������
��������� �����������

�
� � � �
�������������� � � ������
� ��
��������� ����������� �������
�������

��
��
�
��������� � �
� ���������� �
����������� �������
���������� ��
�� � � �����

� ��������� ��������
������������ �������

���������� ��
��
�� ���������
������������
���������� Figure 1. Transmitter model.
��
��
���������
������������
���������
��������� ������
�

12 www.rfdesign.com August 2006

 Data Type Conversion
1 In1 Out1 In Out
In1 Free Space
Path Loss In Out
Propagation Data Type Conversion Transport
2 In2 Out2 Delay
Path
In2 Free Space
Path Loss In Out

Propagation Data Type Conversion Transport

3 In3 Out3 Delay
Path
In3
Free Space
Path Loss In Out
Propagation Data Type Conversion Transport
4 In4 Out4
Path Delay
In4
Free Space
Path Loss In Out

Propagation Data Type Conversion Transport

5 In5 Out5
Path Delay
In5
Free Space
Path Loss In Out

6 In6 Out6 Propagation

Path
Data Type Conversion Transport
Delay Sum all point scatterer returns
In6
Free Space
Path Loss In Out
Propagation Data Type Conversion Transport
7 In7 Out7
Path Delay
In7
Free Space
Path Loss In Out 1
Data Type Conversion Transport Multirange Sum (noisy)
In8 Out8 Propagation
8 Delay
In8 Path
Free Space
2
Path Loss In Out Multirange Sum (no noise)
Propagation Data Type Conversion Transport
9 In9 Out9
Path Delay
In9
Free Space
Path Loss In Out
Propagation Data Type Conversion Transport
Sum result with noise
10 In10 Out10
Path Delay
In10
Free Space
Path Loss In Out

11 In11 Out11 Propagation Data Type Conversion Transport

In11 Path Delay
Free Space
Path Loss In Out
12 In12 Out12
Propagation Data Type Conversion Transport
In12 Path Delay
Free Space
Path Loss In Out
13 In13 Out13
Transport
Propagation Data Type Conversion
In13 Delay
Path

Free Space
Path Loss In Out
14 In14 Out14
In14 Propagation Data Type Conversion Transport
Path Delay
Free Space
15 In15 Out15 Path Loss In Out
In15 Propagation Transport
Data Type Conversion
Path Delay
Free Space
16 In16 Out16 Path Loss In Out
In16
Propagation Data Type Conversion Transport
Target Function Delay
17 Path

Path loss and time delays

AWGN input Splitting the signal into 16 paths and multiplying each copy by appropriate reflection coefficient

Figure 2. Target model.

architectures[8]. For example, in the somewhat are currently being explored for wideband sufficiently large dynamic range to accom-
extreme case of using SAR to monitor single- applications, which may include dual-use modate non-constant-envelope random noise
person targets one cannot avoid implementing (radar and communication system) and anti- signals. Furthermore, the nature of the noise
UWB waveforms[9]. jamming characteristics[13-15]. More flexibility signal allows little control of its parameters by
Because designing high-resolution radar in spectrum allocation and in handling the an engineer or operator.
often amounted to building UWB radar, instantaneous frequency and phase of radar Another promising area is multicarrier
several approaches to implementing UWB signal translates into better opportunities radar systems, which employ signals that
signals were used. Conventionally, UWB for combating ECM—e.g., via pulse diver- represent a collection of sinusoids at various
radars were based on generation and coherent sity—and expanding radar functionalities. center frequencies. Using these independent
reception—e.g., via matched filtering—of The main disadvantage of those implemen- carriers is advantageous in terms of pulse di-
ultrashort pulses, e.g., Gaussian pulses[10-11]. tations is increased complexity of resultant versity and also in terms of exploiting frequen-
This approach does, indeed, provide for systems if UWB waveform generation and cy-dependent characteristics of the channel
high-range resolution if pulse duration is reception are desired. and signal returns[20-21]. Due to recent advances
small enough. However, it also has certain A class of UWB true-noise and pseudo- in digital signal processing—particularly in
disadvantages, such as low spectral efficiency noise radars, which can be pulse-based or sampling technology—it is becoming possible
and ease of signal repeatability (once the signal continuous-wave (CW), is being investigated to extend multicarrier signal bandwidth to
has been intercepted by an adversary), which by several researchers, who point out inher- make them truly UWB waveforms.
can make these imaging systems susceptible ent ECM resistance of random waveforms as One uniting characteristic of many UWB
to certain types of electronic counter-measures a natural advantage[16-19]. While these radars imaging radars is the complexity of their
(ECM). exhibit good low probability of intercept and design and data analysis. Because of the multi-
Other existing alternatives to pulse-based detect (LPI/LPD) characteristics, they are also tude of different scenarios in which particular
radars include linear-frequency modulation more difficult to implement for medium- and types of imaging radars will operate, there is
(LFM) and stepped-frequency radars[12] . long-range applications due to the complex- no one single model that describes radar signal
These types of radar, in particular LFM-based, ity of building a high-power transmitter with propagation and reflection characteristics. The

14 www.rfdesign.com August 2006

 (a.)

Normalized
Data Out

Data Out
Raw
2

1
Frame Conversion1

Carrier Reference
Frame Conversion

signals from Tx
Frame

Frame
To

To

Out_Decision In_Signal Real_N

DecisionMaker I I_sampl Out1

Product

x Carrier In (cos)

From Target Model

In1 I
Out_Decision In_Signal Imag_N
Function

DecisionMaker Q Signal In 1
Math
1
–
u

rxi1.mat Real_R Rx ANT

MinMax

In2 Q
max

To File Carrier In (sin)

Q_sampl Out2
Abs

Imag_R
(u)

rxq1.mat IQ Detector
To File1
FFT
FFT

FFT Delayed Sampler

In
1

Carrier In (cos)
1
buffer
x 2 1
I
Mixer1 Gain1
2 Analog
1 Signal In Filter Design
Real_N
buffer
Unbuffer Normalized 3
x 2 2
Re(u) Data Carrier In
(sin) Q
Im(u) Mixer Gain
Analog
Complex to Filter Design
2 Real-Imag2 Real-Imag to
Imag_N Complex 1
Re I_sampl Integer Delay1
Unbuffer1
FFT
Im -15
1 Z 1
Buffer 2
3
Q_sampl
In1 Out1
Real_R 1Gs/s1
Unbuffer2
Re(u) Raw Data
Im(u) 2 -15 2
Z
Complex to In2 Out2
1Gs/s4 Integer Delay4
4 Real-Imag1
Imag_R
Unbuffer3

Math
(b.)
Function

1 |u|2 1
–
Signal In s
Integrator2
dB
3 1
Reset
|u|2 1 1
Product dB Conversion 1Gs/s3 Sum of SNR_dB
– – 4-Element Elements Product1
s u Buffer
Math
2 Function1 Integrator1 Math
Noise In Function2 1/4

Constant

Figure 3. (a) Receiver model and (b) SNR-computing block.

indoor environment of a building based on radar architectures; however, they are difficult learning curve. As a starting point, users may
steel beam construction, for example, will be to implement due to a considerable number refer to the demonstration test bench “RF
vastly different from the indoor environment of variables and conditions to be taken into Satellite Link” built into Matlab Help (Help
of an all-brick house from the perspective account. With new advances in modeling  Demos (tab)  Blocksets  Communica-
of interior signal propagation. Adding ECM software and computer speeds such simulation tions  Channel Models and Impairments 
effects will further complicate the picture. analyses are becoming more attractive as they RF Satellite Link). This demo contains all the
Therefore, a comprehensive radar simulation also allow identifying potential bottlenecks in major components of an RF communication
test bench is often a requirement for any novel the entire system, from the overall radar archi- system—modulator, high-power transmitter
implementation. tecture to particular channel models in various amplifier model, free space propagation chan-
conditions. Different options with respect to nel model, thermal noise and phase offset in
UWB imaging radar simulation choosing a particular simulator exist; for this the receiver, automatic gain control model,
Modeling radar systems is a low-cost way example we have chosen Matlab Simulink demodulator and results displays.
to create “proof-of-concept” results for novel platform due to relative simplicity and fast The test bench discussed in this article is

RF Design www.rfdesign.com 15
 ous noise conditions and target
configurations, along with
system-level observations in-
tended to assist in subsequent
hardware design and test of
the prototype. The partition-
ing between using Simulink
and Matlab appropriate for
imaging radar simulations is
done the following way: sig-
nal generation, transmission,
reception of radar returns,
sampling and recording them
into digital arrays are per-
formed in Simulink, while
image generation and process-
ing is performed using Matlab
scripts. Below is a description
of the Simulink-based part of
the test bench.
Transmitter: Consists
mainly of digital blocks—
QPSK encoder, IFFT block
(to translate subcarriers from
frequency domain repre-
sentation into time-domain
I/Q samples) and auxilia-
ry blocks. All sub-blocks
Figure 4. Simulink test bench scope capture.
shown in inserts in Fig-
aimed at “proof-of-concept” simulation analy- bench incorporates full transceiver design on a ure 1 are standard Simulink components.
sis for the synthetic aperture radar implemen- block diagram level, with the potential of add- A Bernoulli binary generator provides ran-
tation based on orthogonal multicarrier UWB ing RF effects and an upgradable channel and dom numbers from which the radar signal
pulses. It assumes the general transceiver target scene model, with free space loss and is formed.
architecture similar to the one used for OFDM- multipath delays. The resultant product of the Free space loss: Simulink has a pre-made
based communication systems[22-23]. The test test bench is simulated radar imagery in vari- block that implements the general formula for
a single-frequency signal loss for one-way
propagation. For radars, we need to imple-
ment the following formula for roundtrip
propagation:

 4π ⋅ 2r min 
Lr min = 20 ⋅ log10 
 λ min 
where rmin is the minimum target range
from the radar and λmin is the wavelength of
the minimum frequency in the signal’s spec-
trum, which is achieved by simply doubling
the path distance in Simulink free space path
loss block parameters.
Target model: The simple way to imple-
ment a target model is to represent it as a
collection of point scatterers. Each scatterer,
therefore, will be characterized by a distance
from radar (and, thus, a path loss associated
with that distance) and by the strength of
reflection. In our model we designated 16-
point scatterers within a 24-meter range
swath, of which five scatterers were defined
as “strong” and the rest were defined as
“clutter” with reflection coefficients around
20 dB below those of the “strong” scatterers.
The target model is shown in Figure 2.
 Receiver and signal-to-noise ratio
Figure 5. Simulated point target scene reconstruction using Simulink radar test bench data (SNR)-computation models: As shown in
(true target locations are circled). Figure 3a, the receiver consists of an I/Q

16 www.rfdesign.com August 2006

detector (it is possible to introduce carrier offset in the target model to sim-
ulate non-ideal mixing processes; it is also possible to introduce I/Q imbal-
ance by randomly multiplying the output channels by non-unity values),
delayed sampler (the delay is introduced to align the receive signal with
the transmit signal for better visual representation), and FFT block (it
performs demodulation process for the digitally constructed transmitter
multicarrier signal). The decision blocks map processed-signal samples
to digital values for further QPSK demodulation.
The SNR-computing block shown in Figure 3b implements a
conventional expression for calculating a ratio of the receive signal’s
energy to noise energy, averaged over four sample periods.
These components fully describe the functionality of a multicar-
rier UWB radar, yielding radar return samples in digital format, for a
given target function and SNR. Standard Simulink blocks that can be
added to the test bench to enhance the simulation process are: realistic
power amplifier models; phase noise in the receiver; frequency offset
to simulate target’s Doppler; receiver noise temperature; complex filter
models to simulate transmitter and receiver interconnect (e.g., cables
connecting RF front-end to the antenna); non-Gaussian noise, etc.

Simulation results
A sample scope capture from Simulink test bench is shown in
Figure 4. Transmit and receive signals’ spectra are shown in the inset.
A transmit signal of the 16 ns duration is shown in the uppermost graph,
the receive signals at different stages (at the antenna; after I/Q detector
with carrier removed; after 1 Gsps sample-and-delay block) are shown
below. The SNR-computing block showed approximately –5.5 dB. Note
that Matlab and Simulink operate with discrete numbers only, thus even
“analog” signals in Simulink have to be represented by digital samples.
It is, therefore, critical to choose the proper simulation step, which
is different from sampling interval used in the test bench to model
ADC/DAC performance. Maximum time step size was chosen to be
20 ps for this test bench, while the sampling rate in the transceiver
itself was 1 Gsps.
Then, the resultant samples were recorded by the test bench to data
files in the format ready for processing using Matlab scripts. One com-
mon method of target scene reconstruction for UWB signals is matched
filtering in range and cross-range coordinates. Due to the nature of the
test bench, we can generate a range scan in one simulation. To obtain
a range profile, then, we will simply need to perform matched filtering
of the recorded return signal, using a stored copy of the transmitted
pulse[24]. Cross-range profile reconstruction is performed in a similar
manner. However, one simulation test bench run will only produce
one cross-range data point for a given range coordinate. The total
number of cross-range signal samples required to produce alias-free
image depends on synthetic aperture length (total path described by
a radar in the direction perpendicular to range axis) and the signal’s
wavelength. For our scenario we found that the minimum number
of samples required was about 500. While it may be possible to run
500 simulations for each radar position in the cross-range coordinate,
we chose to run the simulation only once and simulate the rest of the
499 range profiles, based on the data obtained, by assuming a certain
distribution of targets in cross-range and adjusting the phase of each
return depending on the current radar position. To add realism to this
method, we should also vary the noise added to the signal for each range
profile, which is addressed in a Matlab image-processing script.
The back projection method of image construction was used[24]
to obtain the 2-D array representing the final image. Shown in
Figure 5 is a sample image of 15 strong point targets (with ellipses
marking the true target locations as defined in the model) amid clutter,
with a SNR≈ –5.5 dB.
Simulating the performance of a high-resolution imaging radar
system is achievable using Simulink/Matlab. Simulink contains a
large number of pre-made blocks (provided all appropriate toolboxes
and blocksets are included—the author used communication, RF and

RF Design www.rfdesign.com 17
 signal processing blocksets in the test bench bandwidth, RF impairments and secondary on Military Electronics, vol. 5, pp. 127-131,
described) which can be used to model various methods of image enhancement. Due to the April 1961.
types of communication and radar systems. relatively low time and material expenses 2. L. Carin, N. Geng, M. McClure, J. Sichi-
Performed on a 3.4 GHz Pentium 4, one range involved, this type of radar simulation analysis na, and L. Nguyen, “Ultrawideband Synthetic-
profile run computing approximately 10,000 is deemed promising, particularly for UWB Aperture Radar for Mine-field Detection,”
pseudo-analog signal samples per block and radars with complex architectures that are IEEE Trans. Antennas and Propagation, vol.
yielding 160 radar data points takes about two heavily reliant on digital technology, such as 41, No. 1, pp. 18-33, February 1999.
minutes. Subsequent signal phase processing multicarrier OFDM systems. 3. A. J. Dumanian and C. M. Rappaport,
for cross-range profile simulation and image “Enhanced Detection and Classification of
reconstruction are completed in Matlab within References Buried Mines with an UWB Multistatic GPR,”
one minute. Overall results can be checked 1. L. J. Cutrona, W. E. Vivian, E. N. Leith, Proc. of 2005 IEEE Antennas and Propagation
vs. different parameters, such as system and G. O. Hall, “A High-resolution Radar Soc. Int. Symp., Washington, D.C., vol. 3B,
noise, target’s reflectivity function, signal’s Combat-Surveillance System,” IRE Trans. pp. 88-91, July 2005.
4. W. Dierking and T. Busche, “Sea Ice
Monitoring by L-band SAR: An Assessment
Based on Literature and Comparisons of
JERS-1 and ERS-1 imagery,” IEEE Trans.
Geoscience and Remote Sensing, vol. 44, No.
2, pp. 957-970, April 2006.
5. M. Soumekh, “Reconnaissance with
Ultrawideband UHF Synthetic Aperture Ra-
dar,” IEEE Signal Processing Magazine, vol.
12, No. 4, pp. 21-40, July 1995.
6. S. P. Jacobs and J. A. O’Sullivan, “Auto-
matic Target Recognition Using Sequences of
High Resolution Radar Range-Profiles,” IEEE
Trans. Aerospace and Electronic Systems, vol.
36, No. 2, pp. 364-381, April 2000.
7. P. Bajcsy and R. Chaudhuri, “Benefits
of High-resolution SAR for ATR of Targets in
Proximity,” Proc. of 2002 IEEE Radar Conf.,
Long Beach, CA, pp. 29-34, April 2002.
8. P. Withington, H. Fluhler, and S. Nag,
“Enhancing Homeland Security with Advanced
UWB sensors,” IEEE Microwave Magazine,
vol. 4, No. 3, pp. 51-58, September 2003.
9. SoldierVision, Product brochure, (http://
www.uwb.org/files/pdf/cutsheet/SV_cut-
sheet_Aug04.pdf)
10. S. Vitebskiy, L. Carin, M. A. Ressler,
and F. H. Le, “Ultrawideband, Short-pulse
Ground-penetrating Radar: Simulation and
Measurement,” IEEE Trans. Geoscience and
Remote Sensing, vol. 35, No. 3, pp. 762-772,
May 1997.
11. Y. Yang and A. E. Fathi, “See-through-
wall Imaging Using Ultrawideband Short-
pulse Radar System,” Proc. of 2005 IEEE
Antennas and Propagation Society Int. Symp.,
Washington, D.C., vol. 3B, pp. 334-337, July
2005.
12. D. R. Wehner, High Resolution Radar,
Artech House, Norwood, MA, 1987.
13. R. Appelman, Z. Zalevsky, D. Mend-
lovic, and G. Shabtay, “Hybrid Optical-RF
System for Generating an Improved Linear
Frequency Modulated Pulses for Radar Ap-
plications,” The Record of 2000 IEEE Int.
Radar Conf., Alexandria, VA, pp. 775-780,
May 2000.
14. M. Roberton and E. R. Brown, “In-
tegrated Radar and Communications Based
On Chirped Spread-spectrum Techniques,”
Proc. of 2003 IEEE MTT-S Int. Microwave
Symp., Philadelphia, PA, vol. 1, pp. 611-614,

18 www.rfdesign.com August 2006

June 2003. of 2005 Institute of Navigation (ION) National multiband OFDM system for realistic UWB
15. M. Soumekh, “SAR-ECCM Using Tech. Meeting, San Diego, Calif, January 2005 channel environments,” IEEE Trans. Micro-
Phase-perturbed LFM Chirp Signals and (http://www.ece.wpi.edu/Research/PPL/Pub- wave Theory and Techniques, vol. 52, No. 9,
DRFM Repeat Jammer Penalization,” Proc. lications/ionntm2005.pdf) Part 1, pp. 2123-2138, September 2004.
of 2005 IEEE Int. Radar Conf., Arlington, 22. A. Batra, J. Balakrishnan, A. Da- 24. M. Soumekh, Synthetic Aperture Radar
VA, pp. 507-512, May 2005. bak, et al., “TI physical layer proposal: Signal Processing with MATLAB Algorithms,
16. X. Xu and R. M. Narayanan, “FOPEN Time-frequency interleaved OFDM,” IEEE John Wiley & Sons, New York, 1999.
SAR Imaging Using UWB Step-frequency 802.15 Working Group Document Archive,
and Random Noise Waveforms,” IEEE Trans. 2003 (http://grouper.ieee.org/groups/802/15/ Author’s Note
Aerospace and Electronic Systems, vol. 37, pub/2003/Mar03/0314r0P802-15_TG3a-TI- The most common definition of ultrawide-
No. 4, pp. 1287-1300, October 2001. CFP-Presentation.ppt). band signal is the type of signal which occupies
17. D. S. Garmatyuk and R. M. Naray- 23. A. Batra, J. Balakrishnan, G. R. Aiello, a fractional bandwidth of 20% or greater, or
anan, “ECCM Capabilities of an Ultrawide- J. R. Foerster, and A. Dabak, “Design of a whose overall bandwidth is 500 MHz or more.
band Bandlimited Random Noise Imaging
Radar,” IEEE Trans. Aerospace and Elec-
tronic Systems, vol. 38, No. 4, pp. 1243-1255,
October 2002.
18. G.-S. Liu, H. Gu, W.-M. Su, H.-B. Sun,
and J.-H. Zhang, “Random Signal Radar—A
Winner in Both the Military and Civilian Op-
erating environments,” IEEE Trans. Aerospace
and Electronic Systems, vol. 39, No. 2, pp.
489-498, April 2003.
19. S. R. J. Axelsson, “Noise Radar Using
Random Phase and Frequency Modulation,”
IEEE Trans. Geoscience and Remote Sensing,
vol. 42, No. 11, pp. 2370-2384, November
2004.
20. E. Mozeson and N. Levanon, “Multi-
carrier Radar Signals with Low Peak-to-mean
Envelope Power Ratio,” IEE Proc. Radar,
Sonar and Navigation, vol. 150, No. 2, pp.
71-77, April 2003.
21. R. J. Duckworth, H. K. Parikh, and
W. R. Michalson, “Radio Design and Perfor-
mance Analysis of Multicarrier-Ultrawide-
band (MC-UWB) Positioning System,” Proc.

ABOUT THE AUTHOR

Dmitriy Garmatyuk received a Dipl. Eng.
in Radar/Communication Systems in
1996 from Taganrog State University of
Radioengineering, Taganrog, Russia and
a Ph.D. in Electrical Engineering from the
University of Nebraska-Lincoln (UNL)
in 2001. From 2001 until 2005 he held a
position of senior analog design engineer
at Chipset Design Group of Intel Corpora-
tion in Folsom, Calif. working on gigabit
architectures, circuits, MB-OFDM UWB
feasibility analysis and high-speed inter-
connect signal integrity.
In 2005 he joined Miami University
(Oxford, Ohio) as a Department of Elec-
trical & Computer Engineering faculty
member. His research interests include
ultrawideband radar for imaging and
surveillance, imaging sensor networks,
signal integrity in gigabit links and sys-
tem-level modeling/simulation of high-
speed data transfer architectures.

RF Design www.rfdesign.com 19