http://dx.doi.org/10.5755/j01.eie.24.6.22292 ELEKTRONIKA IR ELEKTROTECHNIKA, ISSN 1392-1215, VOL. 24, NO.

6, 2018

Design and Implementation of FMCW

Surveillance Radar Based on Dual Chirps
Youngseok Jin, Bongseok Kim, Sangdong Kim, Jonghun Lee
ART Lab., Daegu Gyeongbuk Institute of Science & Technologies,
Techno-Jungangdaero 333, Daegu, South Korea
jhlee@dgist.ac.kr

1 Abstract—We propose and implement a low-complexity architecture for acquisition and a FPGA-based processing
dual-chirp FMCW radar system for surveillance applications. platform of a radar sensor with a single transmitter and
The FMCW radar is suitable for the detection of various multiple receivers. However, these algorithms usually
positions of multiple targets for the monitoring of concealed
perform full-dimensional FFT when detecting in order to
humans. For a surveillance FMCW radar system, it is necessary
to distinguish between stationary and moving targets while distinguish between stationary and moving targets and thus
maintaining a low level of complexity. However, conventional require high complexity levels. The most important issue in
FMCW radar systems are very complex with many chirps to relation to surveillance radar is a quick identification of the
distinguish between stationary and moving targets. Even in presence of a moving target. Therefore, these algorithms are
hardware with low complexity levels, in order to distinguish not suitable for surveillance radar systems due to their high
between a stationary and a moving target, the proposed
complexity.
algorithm employs only dual chirps. Experimental results show
that the proposed algorithm can effectively distinguish between In this paper, we propose and implement a low-cost
moving and stationary targets despite its low complexity and FMCW radar algorithm with very low complexity for
low-level hardware. surveillance applications. The proposed radar algorithm
recognizes a moving target with an inexpensive chip and a
Index Terms—FMCW; Radar; Surveillance; Low simple RF. In order to distinguish between stationary and
complexity. moving targets, we initially employ only dual chirps. Due to
the characteristics of FMCW radar signals, when the target
I. INTRODUCTION moves, the phase according to the chirp changes due to the
Recently, there have been several studies of radar sensors Doppler effect. That is, the proposed algorithm recognizes
for surveillance applications [1]–[10]. These areas are the existence of the target by observing the phase change
popular because radar sensors have many advantages between the dual chirp signals. Furthermore, this system is
compared to other sensors. For example, radar sensors can implemented as a 24 GHz module with low complexity in
detect at longer distances than an ultrasonic sensor with a order to verify that the proposed algorithm is suitable for
short detection distance and can detect targets even in surveillance applications.
environments in which a camera sensor cannot operate easily, The structure of the paper is as follows. In Section II, we
such as at night or in fog [1]–[3]. introduce and define the system model and the main notations
Meanwhile, in the various radar systems used for used in order to ensure a proper understanding of the FMCW
surveillance applications, the frequency-modulated algorithm, such as the TX and RX signals used in FMCW
continuous wave (FMCW) radar system represents a radar systems. Section III shows the conventional FMCW
promising technology due to advantages such as lower radar detection algorithm using the 3D FFT. In Section IV,
complexity and cost levels [6]–[15]. Compared to pulse radar we present the structure of the proposed algorithm. In
systems, FMCW radar systems have relatively low Sections V and VI, simulation and experimental results for
transmitted peak power levels and can meet certain range and various cases are provided. Finally, Section VII concludes
velocity resolution requirements with relatively simple this paper.
hardware and architectures [6].
In earlier work [10]–[12], FMCW radar algorithms with II. SYSTEM MODELS
multiple channels for surveillance were designed and This section discusses the different models of FMCW
implemented. In one study [10], the authors investigated the radar systems and the notations used in them. We consider
implementation of 24 GHz FMCW radar for surveillance and reflected signals of FMCW radar in order to detect the
detected the range, velocity and angle of targets. Meanwhile, ranges, the velocities, and the azimuth distributions of targets
in other words [11]–[13], researchers presented a scalable [6]. To facilitate an easier understanding of the concepts, we
initially consider the case of a single target, after which we
Manuscript received 14 December, 2017; accepted 10 August, 2018. extend this to the case of multiple targets.
This work was supported in DGIST was supported in part by the DGIST Consider a uniform linear array (ULA) of K receiving
R&D Program of the Ministry of Science, ICT and Future Planning, Korea
(18-IT-01).
(RX) antenna arrays with distance d between adjacent arrays

60
 ELEKTRONIKA IR ELEKTROTECHNIKA, ISSN 1392-1215, VOL. 24, NO. 6, 2018

and a single transmitted (TX) antenna. The TX FMCW signal xl ,k (t )  a0 x0  t    exp  j  f D lT  

denoted by x0 (t ) is shown in Fig. 1(a) and is expressed as
 exp  j 2 dk sin   /    w l ,k (t ), (2)

 
x0 (t )  exp j 2 f 0t   t 2 / 2 ,  (1)
where a0 is the complex amplitude, w l ,k is the additive white
where 0  t  T , f0 is the carrier frequency, T is the Gaussian noise (AWGN) signal, fD is the Doppler frequency
frequency modulation period during which the TX frequency due to the moving of the target and  is the wavelength. By
sweeps over the entire bandwidth B, and  is the chirp rate, denoting the velocity term as vl  f D   exp  j 2 lf DT  and
i.e.,   B / T . All L chirp signals are transmitted during a
the DOA term as  k    exp  j 2 dk sin   /   , xl ,k (t )
frame; i.e., TF = TL. The signal reflected from the target is
received with delay  and direction of arrival (DOA)  at can simply be represented as follows
the k-th array, as shown in Fig. 2. Here, the l-th chirp slot is
denoted by xl,k(t) and is expressed as xl ,k (t )  a0 x0  t    vl  f D  k    w l ,k (t ). (3)

Fig. 1. Waveforms of FMCW radar.

After an analogue-to-digital conversion (ADC) of yl ,k (t ) ,

the discrete time model of (4) with the sampling frequency
fs = 1/Ts is denoted by yl , k [n] ; i.e., yl ,k [n]  yl ,k ( nTs ) for
n = 0, 1, …, Ns ˗ 1 where ts is the sampling time interval and
Ns = T/Ts and yl ,k [n] can be represented as follows

yl ,k  n   axb  nTs  l k  wl ,k  n  . (7)

Fig. 2. DOA modeling. III. CONVENTIONAL 3D -FFT DETECTION ALGORITHM FOR

SURVEILLANCE APPLICATIONS
The beat signal yl,k(t) is expressed by omitting  and fD for
This section introduces the conventional detection
simplicity, as follows
algorithm using the 3D-FFT. In order to estimate the range,
the velocity and the angle of a target, a data cube composed
yl ,k (t )  xl ,k (t )  x0 (t )  of the 3D beat signal in (7) is employed. The conventional
 a exp   j 2 fb t  l k  wl ,k (t )  algorithm performs 3D-FFT operations on the data cube in
the time (range) domain, the ramp (chirp) domain, and the
 axb (t ) l k  wl ,k (t ), (4) array domain.
First, as shown in Fig. 3, a NRFFT-point FFT operation on
where fb   is the beat frequency and xb (t ) is the the data cube in the time (range) domain is performed, where
sinusoid term in the beat signal. In addition, a and wl ,k (t ) N RFFT denotes the number of points of the FFT in the range
are expressed as follows: domain
Ns 1

 
a  a0 exp  j 2 f 0   / 2 , 2
 (5)
Yl ,k u    yl ,k [n]WNqn ,
n 0
RFFT
(8)

wl ,k (t )  w l ,k (t ) x0* (t ). (6) where 0  u  N RFFT  1 , WN is the N-point DFT

61
 ELEKTRONIKA IR ELEKTROTECHNIKA, ISSN 1392-1215, VOL. 24, NO. 6, 2018

operation; i.e., W  exp   j 2 / N  . That is, the range the range and velocity map with N RFFT  N CFFT . Finally,
domain NRFFT-point FFT output vector at the l-th chirp and with an NAFFT-point FFT operation on the range and velocity
the k-th array Yl ,k  Yl ,k [1], Yl ,k [2],..., Yl ,k [ N RFFT ] is map with N RFFT  N CFFT in the array domain, the 3D map
which reflects the range, velocity and DOA is obtained.
obtained. Subsequently, the N CFFT point FFT operation on
However, in order to perform a 3D FFT, very high
Yl , k in the chirp domain is conducted where N CFFT is the complexity is required, as is well known. This causes the
number of points for the FFT in the chirp domain. This gives processing time and cost to increase.

Fig. 3. The conventional 3D FFT FMCW radar detection algorithm.

Meanwhile, in order to reduce the computational two beat signals among all L beat signals according to the
complexity of multi-dimensional FFT operations, FFT velocity of interesting target; that is, the p-th and the q-th beat
algorithms with reduced dimensions have been proposed [16]. signals, yp,k[n] and yq,k[n], are selected. We then subtract the
In these algorithms, the range of the FFT output is initially two signals, with this action denoted by d pq,k [n] , as follows
determined, after which the peak of the magnitude of the FFT
output is detected. The authors use the characteristic by
d pq,k [n]  y p ,k [n]  yq ,k [n] 
which the targets exist only in the range bin with the peak of
the magnitude of the FFT output. They conduct a
chirp-domain FFT operation on only the range bins in which
 
 axb [nts ] k  p  q  w p,k [n]  wq,k [n]. (9)

the target exists. By not performing the chirp-domain FFT on

Generally, clutter or stationary targets do not move.
the range bins without a target, they reduce the computational
Accordingly, the Doppler effect does not occur. On the other
complexity of the multi-dimensional FFT operation.
hand, when the target moves, the Doppler effect arises, and
However, these algorithms require additional algorithms
the phase changes as the index of the chirp increases. If there
such as clutter mitigation or moving target indicator
is no moving target, the Doppler frequency equals zero; i.e.,
algorithms because they do not distinguish between clutter
fD = 0. Therefore, only the noise terms remains in (9) because
and the target.
v p  0   v q  0  . This implies that only the noise term is input
IV. PROPOSED FMCW RADAR ALGORITHM FOR as the NRFFT-point FFT input. On the other hand, if there is a
SURVEILLANCE APPLICATIONS moving target, vp and vq are not equal except when p = q and
In the surveillance radar systems, the first important issue q = p + iT, where i is the integer number and T is the chirp
is a quick determination of the presence of the targets of duration. Therefore, for simplicity, by omitting the noise
interest. In cases where there are no targets of interest, we do terms, (8) is simply expressed as
not need to use resources unnecessarily. That is, the detection
algorithm should be run only if there is a target of interest. To d pq,k [n]   xb [nts ] k , (10)
this end, the proposed method initially determines whether or
not a target exists. Parameter detection operations are where   a (v p  v q ) . From (10), it can be seen that the beat
performed after it is determined that a target exists.
signal xb [nts ] including the range information and the DOA
Figure 4 illustrates the structure of the proposed algorithm.
As shown in Fig. 4, the proposed algorithm properly selects term  k still exists.

62
 ELEKTRONIKA IR ELEKTROTECHNIKA, ISSN 1392-1215, VOL. 24, NO. 6, 2018

Here, we properly select two chirps, i.e., p and q. This is In order to estimate the range of the target, we perform
done because we want to avoid a case in which the proposed FFT on d pq ,k [n] with respect to the sample domain n, as in
algorithm does not work properly. For example, if the phase
(8); thus, the NRFFT-FFT output of d pq,k [n] denoted by
changes at the two chirps are symmetrical, the output after
subtraction may be zero, as if there is no moving target. D pq ,k [u ] is obtained. We then obtain the index ipeak
However, the duration of one frame is relatively short corresponding to the beat frequency of d pq ,k [n] through the
compared to the moving time of the target. By properly
selecting two chirps, we can avoid cases in which the phase peak detection of | D pq,k [u ] | for 0  u  N RFFT  1 . Figure
changes of two chirps are symmetrical. 6 shows an example of the result of the peak detection of
| D pq ,k [u ] | when ipeak  35 . From the obtained ipeak , the
estimated range R̂ is calculated as [16]

c  ˆ c  fˆb c  ipeak  f s
Rˆ    , (11)
2 2 2  N RFFT

where c is the speed of light, ˆ is the estimated delay, and

fˆ is the estimated beat frequency.
b

0.8

0.6
|Dpq,k[u ]|

0.4
ipeak =35
0.2

0
Fig. 4. The structure of range detection of the proposed algorithm. 15 20 30 2535 40 45 50 55
sample index, n
Figure 5 shows the waveforms of beat signals and their Fig. 6. Example of the result of peak detection of | D pq,k [u ] | .
difference according to a sample index for stationary and
moving targets, respectively, when l = 1 and 3. As shown in Furthermore, we use only two RX antennas in order to
Fig. 5(a), in the case of the stationary target, it can be seen ensure a low cost and low complexity. The proposed
that the difference between the two beat signals becomes algorithm does not perform array-domain FFT in order to
almost zero. In the other hand, in the case of the moving detect the DOA information. In the proposed algorithm, the
target, the difference between the two beat signals follows a DOA is estimated using the phase change between two arrays
sinusoidal waveform with the beat frequency as shown in instead of an array-domain FFT, as follows [17]
Fig. 5(b).
 
ˆ  sin 1  Y p,1[ipeak ]Y p,2 [ipeak ]  

 
1 y1,1 [n ]
 sin 1    exp  j 2 d sin   /    , (12)
Amplitude

y3,1 [n ]
 
0
d 13,1 [n ] where  denotes the term apart from except DOA term in
-1
(3). Of course, this simple DOA estimation is only possible in
20 40 60 80 100 120 140 160 180 200 the condition with a single target. However, even with
sample index, n
(a)
multiple targets, if the ranges or velocities of the targets are
not identical, the outcome will be similar to that of a
1 single-target condition in a FMCW radar system [10].
Amplitude

y1,1 [n ]
0
y3,1 [n ] V. SIMULATION RESULTS
-1 d 13,1 [n ]
This section addresses the simulation results in order to
20 40 60 80 100 120 140 160 180 200
observe the feasibility of the proposed algorithm. The
sample index, n parameters for the simulation of FMCW radar system are
(b)
shown in Table I. In order to consider the multiple targets
Fig. 5. Difference between beat signals with l = 1 and 3 according to sample
index in the cases of stationary and moving targets: (a) stationary target; (b) condition, the number of targets is set to four and the 4th
moving target. target is set to stationary.

63
 ELEKTRONIKA IR ELEKTROTECHNIKA, ISSN 1392-1215, VOL. 24, NO. 6, 2018

TABLE I. PARAMETERS FOR SIMULATION. VI. EXPERIMENT

Parameter Value
In order to confirm the performance of the proposed
Center frequency (GHz) 24
algorithm using two chirps in an actual environment, we
Bandwidth (MHz) 200 performed a variety of experiments inside an advanced radar
Symbol duration (μs) 80 technology (ART) laboratory and an anechoic chamber
Sampling rate (MHz) 5 located at DGIST in South Korea [17]. By using the chamber
Number of chirps 40 environment, we can find the exact positions of the targets to
Distance of targets (m) [10, 15, 20, 25] assess the performance of the algorithm by avoiding
Speed of targets (m/s) [10, 10, 15, 0] undesired echo terms. Then, we show the results of actual
human movement by the experiment in the ART laboratory.
Figure 7 shows the snapshot of the simulation result of A. Front-end and Back-end Modules and Radar Signal
range estimation by the proposed scheme. Even though the Processing
number of targets is four, there are only three peaks. This is
From Fig. 9 to Fig. 11 show a block diagram of the
because the 4th target is stationary and thus only three moving
front-end module (FEM), antenna pattern and photo of back
targets are detected. This result implies that the proposed
end module (BEM) of our implemented FMCW radar system
scheme distinguishes between stationary and moving targets
[18]. The implemented FEM utilizes outputs signals of
by using only two chirps.
200 MHz in the range of 24.05 GHz~24.25 GHz at 10 dBm
1 of power in the TX part of a FMCW radar system [18]. As
Target #1
shown in Fig. 10, the antenna is composed of one TX and
Target #2
0.75
three RX antennas. The TX antenna is composed of five
antenna element arrays. The BEM is based on the TMS320
DSP processor with maximum 150 MHz [18].
PSD

0.5 Target #3

0.25

0
0 10 20 30 40 50
Range (m)
Fig. 7. Snapshot of the result of detection of the proposed scheme.

Figure 8 shows the root mean square error (RMSE) of

range estimation of the proposed scheme. In order to observe
the effect by complex amplitude of beat signal, the simulation
of two cases is performed. Fist, in Case I, the magnitude and
phase terms of the complex amplitude a in (7) are set to be
uniformly and randomly distributed between 0 and 1 and 0 Fig. 9. Functional block of front end modulation (FEM) [18].
and 2π, respectively, i.e., | a |  0, 1 and a   0, 2  . On
the other hand, in Case II, the magnitude term | a | is fixed to
unit amplitude, i.e., | a | 1 but the phase term a is set to
be uniformly and randomly distributed between
a   0, 2  . In the case of Case I, the performance is
degraded in the low SNR region because of fluctuation of
magnitude. However, both of two cases achieve the almost
same RMSE in the region of SNR > 10 dB.
2.5
Case I
Case II
2 Fig. 10. Photo of antenna pattern.

Figure 12 and Fig. 13 show the experiment results of

RMSE (m)

1.5
estimated range and DOA of the proposed algorithm which is
1 experimented in the anechoic chamber. For this experiment,
out of three targets, two are stationary and only one target is
0.5 moving. The two stationary targets are located at 3.3 m and
6.2 m, respectively. The moving target moves between the
0 two stationary targets. As shown in Fig. 12, the power
0 5 10 15 20
SNR [dB] spectral density (PSD) of two stationary targets is
Fig. 8. RMSE results according to SNR. insignificant. On the other hand, for the moving target, the

64
 ELEKTRONIKA IR ELEKTROTECHNIKA, ISSN 1392-1215, VOL. 24, NO. 6, 2018

peak of the PSD is located at approximately 5 m. From this slow mode. These results imply that the proposed algorithm
result, the proposed algorithm can detect only the moving can detect a moving target well in condition which contain
target. Fig. 13 shows that the proposed algorithm accurately clutter and stationary targets despite its low complexity.
detects the DOA of moving target.

Fig. 14. Photo of surveillance radar experiment for actual moving with
single target.

10
Fig. 11. Photo of back end modulation (BEM) [18].

range (m)
1
5
Moving target (5.2 m)
0.75 Stationary target (3.3 m)
Stationary target (6.2 m)
0
0 10 20 30 40 50 60
frame index
PSD

0.5
(a)
10
0.25
range (m)

5
0
0 5 10 15 20
Range (m)
Fig. 12. Experiment result of range detection of the proposed algorithm. 0
0 10 20 30 40 50 60
frame index
1 (b)
Fig. 15. Experiment results of range according to frame index (time) by our
0.8 developed surveillance radar: (a) slow mode; (b) fast mode.

0.6 Moreover, in order to verify the detection performance in

PSD

the multiple targets condition, the experiment is performed in

0.4 the condition of two persons. As shown in Fig. 16, one person
(target 1) walked toward the radar and the other person
0.2 estimated DOA using (12)
actual DOA
(target 2) walked away from the radar. Then, target 1 walked
away from the radar and target 2 walked toward the radar.
0
-60 -40 -20 0 20 40 60 80
Degree ()
Fig. 13. Experiment result of DOA detection of the proposed algorithm.

We performed experiments in the ART laboratory when

people were actually moving. As shown in Fig. 14(a) person
moves from the radar to the wall at a range of 10 m. In order
to confirm the performance of the proposed algorithm under
several speed conditions, the experiments utilize two speed
conditions, i.e., slow mode (walking) and fast mode
(jogging).
Figure 15 shows the result of used estimated range
according to used frame index TF ,i  iTL with the developed Fig. 16. Photo of surveillance radar experiment for actual moving with
multiple targets.
surveillance radar system, as performed in Fig. 15. As shown
in Fig. 15, the proposed algorithm can detect ranges of Figure 17 shows the snapshot of detection result (FFT
several steps, i.e., several speeds condition. In the slow mode, output) of the proposed algorithm. Despite many clutter, the
the range of the target changes slowly as the frame index proposed algorithm detects two moving targets.
increases. On the other hand, in the fast mode, the change in Figure 18 shows the result of the estimated range
the range of the target is significant compared to that in the according to the frame index for the two persons. In Fig. 18,

65
 ELEKTRONIKA IR ELEKTROTECHNIKA, ISSN 1392-1215, VOL. 24, NO. 6, 2018

the dashed and solid arrows show the movement of the target REFERENCES
1 and target 2, respectively. Figure 18 shows that according [1] M. Richards, M. A. Richards, J. A. Scheer, W. A. Holm, Principles of
to the elapsed time the movement of two targets is detected Modern Radar: Basic Principles. Institution of Engineering and
Technology. 2010. DOI: 10.1049/sbra021e.
well. It implies that the proposed algorithm can detect not
[2] J. Li, P. Stoica, “MIMO radar with collocated antennas”, IEEE Signal
only a moving target well but also even in multiple targets Process. Mag., vol. 24, no. 5, pp. 106–114, 2007. DOI:
under many clutter and stationary obstacle conditions. 10.1109/msp.2007.904812.
[3] H. H. Meinel, “Evolving automotive radar: From the very beginnings
1 into the future”, in Proc. European Conf. Antennas and Propagation,
Hague, Nethelands, 2014, pp. 3107–3114. DOI: 10.1109/eucap.2014.
Target 1 6902486.
0.75 [4] G. Reina, D. Johnson, J. Underwood, “Radar sensing for intelligent
vehicles in urban environments”, Sensors, vol. 15, no. 6,
pp. 14661–14678, 2015. DOI: 10.3390/s150614661.
[5] A. Natarajan, A. Komijani, X. Guan, A. Babakhani, A. Hajimiri, “A
PSD

0.5
77-GHz phased-array transceiver with on-chip antennas in silicon: [3]
Target 2 Transmitter and local LO-path phase shifting”, IEEE J. Solid-state
circuits, vol. 41, no. 12, pp. 2807–2819, 2006. DOI: 10.1109/jssc.
0.25 2006.884817.
[6] C. Hu, Y. Liu, H. Meng, “Randomized switched antenna array FMCW
radar for automotive applications”, IEEE Trans. on Vehi. Tech., vol. 63,
0 no. 8, 2014. DOI: 10.1109/tvt.2014.2308895.
0 5 10 15 20 25 30
[7] A. Tessmann, S. Kudszus, T. Feltgen, M. Riessle, C. Sklarczyk,
range (m) W. H. Haydl, “Compact single-chip W-band FMCW radar modules for
Fig. 17. Photo of surveillance radar experiment for actual moving with commercial high-resolution sensor applications”, IEEE Trans. on
multiple targets. Microwave Theory and Tech., vol. 50, no. 12, pp. 2995–2997, 2002.
DOI: 10.1109/tmtt.2002.805162.
10 [8] D. C. W. Low, K. W. Chang, R. Lin, E. W. Lin, H. Wang,
Target 1 Target 2 M. Biedenbender, G. S. Dow, B. R. Allen, “A single-chip W-band
8 transceiver with front-end switching receiver for FMCW radar
applications”, IEEE Microwave Millimeter-Wave Monolithic Circuits
Symp. Dig., pp. 225–228, 1995. DOI: 10.1109/mcs.1995.470952.
range (m)

6 [9] A. G. Stove, “Linear FMCW radar techniques”, in Proc. Inst. Elect.

Eng. F—Radar Signal Process., 1992, vol. 139, no. 5, pp. 343–350.
4 DOI: 10.1049/ip-f-2.1992.0048.
Target 1 [10] E. Hyun, Y. Jin, J. Lee, “Design and implementation of 24 GHz
Target 2
multichannel FMCW surveillance radar with a software-reconfigurable
2
baseband”, Journal of Sensors, 2017. DOI: 10.1155/2017/3148237.
[11] S. Saponara, B. Neri, “Radar sensor signal acquisition and
0 25 75 50100 125 150 multidimensional FFT processing for surveillance applications in
frame index transport systems”, IEEE Trans. on Instrumentation and Measurement,
Fig. 18. Experiment results of multiple target detection according to frame vol. 66, no. 4, 2017. DOI: 10.1109/tim.2016.2640518.
index (time). [12] S. Saponara, B. Neri, “Radar sensor signal acquisition and 3D FFT
processing for smart mobility surveillance systems”, in Proc. IEEE
Sens. Appl. Symp., 2016, pp. 1–6. DOI: 10.1109/sas.2016.7479883.
VII. CONCLUSIONS [13] S. Saponara, S. Lischi, R., Massini, L. Musetti, D. Stagliano, F. Berizzi,
B. Neri, “Low cost FMCW radar design and implementation for
We proposed and implemented 24 GHz FMCW radar Harbour surveillance applications”, Applications in Electronics
algorithm using dual chirps for surveillance applications. In Pervading Industry, Environment Society, pp 139–144, 2015. DOI:
order to reduce the complexity of surveillance FMCW radar 10.1007/978-3-319-20227-3_18.
algorithm, we performed subtracting between two range-FFT [14] D. Tarchi, G. Guglieri, M. Vespe, C. Gioia, F. Sermi, V. Kyovtorov,
“Search and rescue: surveillance support from RPAs radar”, in Proc.
outputs, instead of performing full 2D FFT. In order to verify IEEE Navigation Conf. 2017. DOI: 10.1109/euronav.2017.7954216.
the effect of the proposed scheme, the simulation results are [15] M. Caris, W. Johannes, S. Stanko, N. Pohl, “Millimeter wave radar for
shown such as RMSE and detection performance in the case perimeter surveillance and detection of MAVs”, in Proc. IEEE Radar
Symposium, 2015. DOI: 10.1109/irs.2015.7226314.
of two kinds of amplitudes. Especially, it can be seen that [16] C. Wang, H. Liu, G. Han, X. Jing, “Automotive FMCW
RMSEs of two conditions become almost similar when SNR radar-enhanced range estimation via a local resampling Fourier
is larger than 10 dB region. In addition, experiments were transform”, International Journal of Advanced Robotic Systems,
vol. 13, 2017. DOI: 10.5772/62179.
conducted on moving and stationary targets in an anechoic [17] S. Kim, D. Oh, J. Lee, “Joint DFT-ESPRIT estimation for TOA and
chamber in order to verify the implementation of the DOA in vehicle FMCW radars”, IEEE Antennas and Wireless Propag.
proposed scheme. Furthermore, it was practically confirmed Lett., vol. 14, pp. 1710–1713, 2015. DOI:
10.1109/lawp.2015.2420579.
that the multiple targets are well detected even in the
[18] Y. Jin, Y. Ju, J. Lee, “Development of real-time LCA system based on
clutter-vulnerable environment even though it used only automotive radar”, in Proc. (ICONS 2016), 2016. DOI:
2 chirps among the total 40 chirps. 10.1109/radar.2014.7060385.

66