A Matched-filter based method in the Synthetic

Aperture Radar Images Using FMCW radar

1st Moein Movafagh 2nd Avik Santra 3rd Daniel Oloumi
Dept.of electrical engineering
K. N. Toosi University of Technology Infineon Technologies AG Infineon Technologies AG
Tehran, Iran Neubiberg, Germany Villach, Austria
m.movafagh@kntu.ac.ir avik.santra@infineon.com Daniel.oloumi@infineon.com
arXiv:2111.00613v1 [eess.SP] 31 Oct 2021

Abstract—The stretch processing architecture is commonly beat frequency signal. It presents high-resolution imaging with
used for frequency modulated continuous wave (FMCW) radar a low sampling rate which is very beneficial and cost-effective
due to its inexpensive hardware, low sampling rate, and simple for various applications. In this architecture, to identify targets’
architecture. However, the stretch processing architecture is not
able to achieve optimal Signal to Noise ratio (SNR) in comparison location or ranges profile, the Fast Fourier Transform (FFT) is
to the matched-filter architecture. In this paper, we aim to applied to the sampled signal. In addition to stretch processing,
propose a method whereby stretch processing can achieve optimal the matched filter architecture is also employed for FMCW
SNR. Hence, we develop a novel processing method to enable radars. In this architecture, the received signal, after being
applying a matched filter to the output of the stretch processing. down-converted, is sampled. Then, the matched filter is used
The proposed architecture achieves optimal SNR while it can
operate on a low sampling rate. In addition, the combination of to identify the targets’ location or range profiles. Therefore, it
the proposed radar architecture and SAR technique can generate would be able to achieve optimal Signal to Noise Ratio (SNR)
high-quality images. To evaluate the performance of the proposed because of using the matched filter. However, this architecture
architecture, four scenarios are considered. Simulation is carried requires a high sampling rate analog to digital converter
out based on these scenarios. The simulation results show that (ADC). This is a major disadvantage of this architecture and
the proposed radar demonstrates the ability to generate an image
with higher quality over stretch processing. This proposed radar leads to an increase in the cost.
can also bring a bigger gain compression. In this paper, a novel signal processing method is proposed.
Index Terms—FMCW radar, SAR, Matched-filter architecture This method enables the use of a matched filter in the
,stretch processing architecture output of the stretch processing architecture. So, this new
architecture can operate on a low sampling rate for sake of
I. I NTRODUCTION using stretch processing architecture. On the other hand, it

R ECENTLY, the frequency modulated continuous wave

(FMCW) radars draw attention due to their advantages
such as low sampling rate and the minimum target range
exploits a matched filter in the output of the stretch processing,
so it can achieve optimal SNR.
This paper is organized as follows: the FMCW radar is
[1]. The FMCW radar can be a reasonable alternative for the elaborated in section II to prepare the foundation for the next
pulsed radar since it requires a lower transmitting power [2]. section. This section also presents the mathematical analysis
Besides, simplicity, low cost, and miniaturized system design for matched filter and the stretch processing architecture. In
are potential advantages that stimulate demands for FMCW section III, the proposed radar architecture which benefits
radar in many applications [3]. Therefore, the FMCW radar is from a novel signal processing method is completely studied.
an ideal choice for both industrial and academic purposes. Section III also gives a mathematical analysis related to the
The combination of FMCW radar and synthetic aperture proposed architecture. Section VI evaluates the simulation
radar (SAR) principle presents a high-resolution imaging [1]. results. Finally, section VI includes the conclusion.
The FMCW radar provides a high-range resolution. So, it
II. FMCW R ADAR
enables the creation of high-resolution images using SAR. The
FMCW-SAR is a cheaper alternative for pulsed SAR since A. Stretch Processing
it requires lower transmitted power. Hence, these advantages The stretch processing architecture is widely employed in
make the FMCW-SAR a possible choice for various appli- FMCW radars for sake of It’s great benefits of low sampling
cation [4]. In addition, the FMCW-SAR transceiver provides rate and inexpensive hardware [1]. It also provides a fine
a major benefit on cost-effectiveness, power consumption, resolution for linear frequency modulated (LFM) radars [10].
volume, and weight [5]. The stretch processing architecture is illustrated by Fig.1. As
The stretch processing [6], [7], [8] and the matched-filter [9] it can be seen, the stretch processing architecture is based
are two common architectures which are widely employed in on mixing the reflected signal off a target with a local
the FMCW radars. The stretch processing architecture is based FMCW signal. Therefore, a beat frequency signal is generated
on mixing the received and transmitted signal to generate a by the mixer. The frequency of the beat frequency signal
directly depends on the target range. According to Fig.1, the B. Matched filter
transmitted signal is defined as: The matched filter architecture is an alternative to stretch

t − T /2 processing architecture in FMCW radars. This architecture is
St (t) = Aexp φ(t) .rect( ) (1)
T widely employed in many SAR imaging algorithms such as
φ(t) = πkt2 (2) chirp scaling, Range-Doppler, frequency scaling, etc. [11]. The
matched-filter architecture is shown in Fig.2. The transmitted
T is the chirp time and k is the chirp rate. The transmitted signal is defined as:
signal is reflected off a target at a range of R and the received
signal can be represented as:
t − T /2
St (t) = Aexp φ(t) .rect( ) (6)

t − τ − T /2 T
Sr (t) = Aexp φ(t − τ ) .rect( ) (3)
T φ(t) = πkt2 (7)
τ = 2R/C is the time delay and C is the speed of light.
Then, the received signal is mixed with a local FMCW signal T is the chirp time duration and k is the chirp rate. Likewise
in the mixer. The beat frequency signal is generated as follow: the stretch processing architecture, the received signal can be
represented as:
Sif (t) = ASt∗ (t).Sr (t) = exp φ(t − τ ) − φ(t)

t − τ − T /2 t − T /2 (4)
t − τ − T /2
rect( ).rect( ) Sr (t) = Aexp φ(t − τ ) .rect( ) (8)
T T T
the Sif consists of a beat frequency signal whose frequency is In this architecture, the received signal is down-converted to
directly proportional to the target range. Hence, the Sif signal baseband and then is sampled by ADC. The ADC sampling
can be represented as: rate must be twice the bandwidth of the received signal based
on the Nyquist theorem. Hence, the high-resolution application
Sif (t) = Aexp(πktd t − πktd ) (5) necessitates a higher sampling rate ADC which leads to an
Since the frequency of Sif directly depends on the target increase in the cost of radar devices. Just after the signal
range, the Fast Fourier Transform (FFT) can precisely identify is sampled by ADC, the matched filter is applied to the
the target location. In near-range applications, the frequency signal to identify the target location. Thus, the matched filter
of Sif is often low. So, it requires a low sampling rate ADC architecture can achieve the optimal SNR owing to the use
which is a great benefit of a stretch processing architecture. of the matched filter. This radar architecture is capable to
operate in a noisy environment. Both stretch processing and
matched filter architecture are used for SAR imaging. It should
be noted that the matched filter architecture generates images
with higher quality.

III. NEW MATCHED - FILTER BASED METHOD

ARCHITECTURE

In this section, a novel radar architecture based on the

matched filter is proposed. This new architecture aims to
operate on a low sampling rate and achieve optimal SNR.
To this end, stretch processing architecture (refer to Fig.1) is
used to keep the sampling rate low. A matched filter is also
Fig. 1: Simplified block diagram of a stretch processing employed in the output of the stretch processing to achieve
architecture optimal SNR. Fig.4 illustrates the proposed architecture. In
this architecture, a novel processing method is exploited (refer
to Fig.3). This method enables employing a matched filter in
the output of the stretch processing. Thus, proposed radar can
easily achieve optimal SNR.

A. mathematical analysis
This section aims to give a mathematical analysis of the pro-
posed radar architecture. As mentioned earlier, the proposed
radar uses stretch processing in the hardware section due to
lower the sampling rate. According to Fig.4, the signal which
is sampled by ADC is expressed as:
Fig. 2: Simplified block diagram of a matched-filter architec-
ture Sif (t) = ASr (t).St∗ (t) (9)
According to Fig.3, the sampled signal is up-sampled and then W and T are FMCW bandwidth and chirp time duration
is multiples by exp(j2πkt2 ). So, St∗ (t) term will be eliminated respectively. According to (11), the compression gain of a
from Sif (t) and the received signal is reconstructed as follow: matched-filter which is applied to an FMCW signal (W =
1GHz, T= 10µs) is about 40 dB. The compression gain value
Srec (t) = ASr (t) (10) for the proposed method is higher than the analytical value
(refer to Table.I). This additional gain (approximately 8 dB)
So, the target location can be readily identified by a matched is achieved due to the SAR algorithm. The SAR algorithm
filter. Using the proposed processing method (Fig.3) enables utilizes a kind of averaging method [14]. In this algorithm,
employing matched filter in the output of the stretch process- a radar moves along a direction and acquires range profiles.
ing. Thus, proposed radar can achieve optimal SNR while Ultimately, these range profiles are combined to generate
benefits from a low sampling rate. an image. Therefore, the SAR algorithm benefits from an
averaging method indirectly.
IV. SIMULATION RESULT
TABLE I: Compression Gain
In this section, the performance of traditional FMCW radar
and proposed radar architecture in improving SNR is fully an- SN Rin (dB) Matched-Filter based method (dB) Stretch Processing (dB)
alyzed. In this paper, the circular global back projection (GBP) 0 48 37
algorithm [12] is used to generate an image. We consider four -10 48.2 34
scenarios in which a target is located at (x, y) = (2, −2). An -20 48.6 33.8
-30 49 33.4
FMCW radar with 1GHz bandwidth and 10µs chirp duration
time is also utilized. Fig.5 and Fig.6 show the images gen-
erated by two radar architectures with different SNR. As can
be seen from these images, the proposed radar can generate V. CONCLUSION
higher quality images in comparison to the traditional stretch In this paper, a novel signal processing method is pre-
processing architecture. Table.I also shows the compression sented. This method enables the exploit of matched-filter in
gain of these two radars architecture in different SNR. As can the output of the stretch processing. Thus, this new radar
be seen, the proposed radar brings a better gain over stretch architecture can achieve optimal SNR while operates on a low
processing. It is worthy to note that the compression gain of sampling rate. The analytical analysis of the proposed radar
the matched filter is defined as [13]: architecture is thoroughly studied. The impact of the proposed
GP = W T (11) radar architecture on improving SNR is validated. The GPB
algorithm is also employed to generate images based on the
SAR principle in both stretch processing and the proposed
architecture. The simulation results show that the proposed
architecture demonstrates a potential capability for generating
an image with higher quality compared to stretch processing.

Fig. 3: Proposed Pre-processing section which enable stretch

processing architecture to exploit matched-filter to archive R EFERENCES
optimal SNR
[1] J.-W. Ting, D. Oloumi, and K. Rambabu, “Fmcw sar
system for near-distance imaging applications—practical
considerations and calibrations,” IEEE Transactions on
Microwave Theory and Techniques, vol. 66, no. 1, pp.
450–461, 2017.
[2] A. Meta, P. Hoogeboom, and L. P. Ligthart, “Signal pro-
cessing for fmcw sar,” IEEE Transactions on Geoscience
and Remote Sensing, vol. 45, no. 11, pp. 3519–3532,
2007.
[3] G. L. Charvat and L. C. Kempel, “Low-cost, high resolu-
tion x-band laboratory radar system for synthetic aperture
radar applications,” pp. 529–531, 2006.
[4] S. Navneet, A. Roy, and C. Bhattacharya, “High-
resolution sar image generation by subaperture process-
ing of fmcw radar signal,” IEEE Geoscience and Remote
Sensing Letters, vol. 11, no. 11, pp. 1866–1870, 2014.
Fig. 4: the Proposed matched-filter based stretch processing [5] Y. Wang, L. Lou, B. Chen, Y. Zhang, K. Tang, L. Qiu,
architecture S. Liu, and Y. Zheng, “A 260-mw ku-band fmcw
 (a) SN Rin = 0 (a) SN Rin = 0

(b) SN Rin = −10 (b) SN Rin = −10

(c) SN Rin = −20 (c) SN Rin = −20

(d) SN Rin = −30 (d) SN Rin = −30

Fig. 5: The generated Images using the traditional Stretch Fig. 6: The generated Images using the proposed radar archi-
processing architecture tecture
 transceiver for synthetic aperture radar sensor with 1.48-
ghz bandwidth in 65-nm cmos technology,” IEEE Trans-
actions on Microwave Theory and Techniques, vol. 65,
no. 11, pp. 4385–4399, 2017.
[6] J. A. Torres, R. M. Davis, J. D. R. Kramer, and R. L.
Fante, “Efficient wideband jammer nulling when using
stretch processing,” IEEE Transactions on Aerospace and
Electronic Systems, vol. 36, no. 4, pp. 1167–1178, 2000.
[7] W. J. Caputi, “Stretch: A time-transformation technique,”
IEEE Transactions on Aerospace and Electronic Systems,
no. 2, pp. 269–278, 1971.
[8] R. Middleton, “Dechirp-on-receive linearly frequency
modulated radar as a matched-filter detector,” IEEE
Transactions on Aerospace and Electronic Systems,
vol. 48, no. 3, pp. 2716–2718, 2012.
[9] J. Wang, D. Cai, and Y. Wen, “Comparison of matched
filter and dechirp processing used in linear frequency
modulation,” in 2011 IEEE 2nd International Confer-
ence on Computing, Control and Industrial Engineering,
vol. 2. IEEE, 2011, pp. 70–73.
[10] H. S. Mir and U. K. T. Wong, “Low-rate sampling tech-
nique for range-windowed radar/sonar using nonlinear
frequency modulation,” IEEE Transactions on Aerospace
and Electronic Systems, vol. 51, no. 3, pp. 1972–1979,
2015.
[11] R. Yang, H. Li, S. Li, P. Zhang, L. Tan, X. Gao,
and X. Kang, High-Resolution Microwave Imaging.
Springer, 2018.
[12] D. Oloumi, “Oil well monitoring by ultra-wideband
ground penetrating synthetic aperture radar,” 2012.
[13] A. Santra, R. Srinivasan, K. Jadia, and G. Alleon, “Ambi-
guity functions, processing gains, and cramer-rao bounds
for matched illumination radar signals,” IEEE Transac-
tions on Aerospace and Electronic Systems, vol. 51, no. 3,
pp. 2225–2235, 2015.
[14] D. Oloumi, J.-W. Ting, and K. Rambabu, “Design of
pulse characteristics for near-field uwb-sar imaging,”
IEEE Transactions on Microwave Theory and Tech-
niques, vol. 64, no. 8, pp. 2684–2693, 2016.