Professional Documents
Culture Documents
Sensors 22 06086
Sensors 22 06086
Article
Detection of Electronic Devices Using FMCW Nonlinear Radar
Kyuho Cha † , Sooyoung Oh † , Hayoung Hong, Hongsoo Park and Sun K. Hong *
Abstract: Nonlinear radars can be utilized to detect electronic devices, which are difficult to detect
with conventional radars due to their small radar cross sections (RCS). Since the receiver in a nonlinear
radar is designed to only receive harmonic or intermodulated echoes from electronic devices, it is
able to separate electronic devices from non-electronic scatters (clutter) by rejecting their echoes at
fundamental frequencies. This paper presents a harmonic-based nonlinear radar scheme utilizing
frequency-modulated continuous-wave (FMCW) signals for the detection of various electronic devices
at short range. Using a laboratory experiment setup for FMCW radar at S-band for Tx (C-band for Rx),
measurements are carried out to detect electronic devices of various sizes. The results show that the
detection of small electronic devices is possible with nonlinear FMCW radar when appropriate system
parameters are selected. Furthermore, we also discuss the maximum detectable range estimation for
electronic targets using the radar range equation for FMCW nonlinear radar.
Keywords: nonlinear radar; harmonic radar; FMCW; nonlinear RCS; nonlinear junction detection
1. Introduction
Radars have been extensively used for the remote detection of objects for military and
Citation: Cha, K.; Oh, S.; Hong, H.;
commercial applications since their first appearance in the 1930s [1,2]. Traditional radars
Park, H.; Hong, S.K. Detection of
use the same transmit and receive frequencies; that is, they detect the presence, distance,
Electronic Devices Using FMCW
and speed of targets by receiving the transmitted pulses that are reflected back from the
Nonlinear Radar. Sensors 2022, 22,
targets. In the sense that they transmit and receive the same frequencies, traditional radars
6086. https://doi.org/10.3390/
are based on “linear” detection. For traditional radars, difficulties arise when it comes to
s22166086
detecting targets with small radar cross sections (RCS) in the midst of clutter or other large
Academic Editor: Federico Alimenti objects. Targets of interest with small RCS may include small electronic devices, which are
Received: 4 July 2022
typically situated in an environment that is surrounded by other objects with larger RCS,
Accepted: 12 August 2022
rendering limitations in detecting them.
Published: 15 August 2022
To overcome this limitation of linear radars, “nonlinear” detection was proposed in
the 1970s [3]. Due to the nonlinearities of electronic components and circuit elements in
Publisher’s Note: MDPI stays neutral
electronic devices, it is possible to separate small electronic devices of interest from linear
with regard to jurisdictional claims in
scatterers, even in the presence of strong clutter and other objects with large RCS. Unlike
published maps and institutional affil-
linear radars, nonlinear radars use different transmit and receive frequencies. That is, by
iations.
tuning the receiver to only receive the nonlinear responses (harmonic or intermodulation)
that are excited by the nonlinear interaction at the semiconductor junctions in electronic
devices, nonlinear radars allow for the detection of electronic devices in the probed envi-
Copyright: © 2022 by the authors.
ronment [4]. Since natural linear scatterer and responses from other non-electronic objects
Licensee MDPI, Basel, Switzerland. do not produce nonlinear responses, it is possible to eliminate the linear response and only
This article is an open access article receive the desired nonlinear response from electronic devices. This allows for the effective
distributed under the terms and detection of small electronic targets that would otherwise be difficult to achieve with linear
conditions of the Creative Commons radars [5–12]. Nonlinear radars can be utilized in various applications. For military and
Attribution (CC BY) license (https:// security purposes, they can be used to detect concealed electronic devices [13,14]. They
creativecommons.org/licenses/by/ have also been applied for detecting vital signs for health monitoring [15–17] and insect
4.0/). tracking using harmonic tags [18–21].
Recent studies have presented nonlinear radar systems using various types of wave-
forms, i.e., pulsed RF, continuous wave (CW), and chirp and stepped frequency continuous
wave (SFCW) [8,22–24] for the detection of various nonlinear targets at various ranges.
Since the signal processing and detection methods depend on the type of waveform that
the radar transmits, it is important for radar systems to select an appropriate transmit
waveform for a given application. In the case of detecting targets at short range (up to
several meters), it is appropriate to transmit CW waveforms with modulation. An FMCW
waveform, which consists of periodically repeated linearly modulated chirp signals, is
appropriate for obtaining target information at short range [25–27]. The signal processing
of FMCW radar is based on analyzing the beat frequency [27,28], that is, the frequency
difference that is generated by mixing the transmitted and received signals through an
RF mixer. For this reason, FMCW waveforms can also be applied to nonlinear (harmonic)
detection at short range, where the radar would receive the harmonic version of the original
transmit signals, and the beat frequency is generated by mixing the received signal with
the frequency-doubled transmit signal.
In this work, we demonstrate the concept of an FMCW-based nonlinear radar for the
detection of various electronic targets. The proposed system is a second-harmonic-based
FMCW radar that uses S-band for transmit and C-band for receive with the aim of detecting
electronic devices at short range (up to several meters). Using S/C band for harmonic
radar may be advantageous compared to the previously reported UHF/L-bands [29] in
terms of a smaller system size due to shorter wavelengths. Furthermore, compared to
millimeter wave bands [14], utilizing S/C bands would allow for higher output power
which can increase the maximum detectable range. The organization of the paper is as
follows. In Section 2, the nonlinear radar equation and nonlinear RCS of various electronic
devices are considered. The range equation for nonlinear FMCW radar is then presented.
Section 3 presents the experimental apparatus that was used for the FMCW nonlinear radar
measurement. In Section 4, the measured results are presented and discussed, where it
shows that the targets are properly detected with their range information. The estimation
of maximum detectable range based on the derived nonlinear FMCW radar equation is
also presented. Section 5 concludes the paper with a quick summary and implications of
the results.
PG 1
2
Sensors 2022, 22, 6086 Pr = t t2 2 r
G 3 of
(2)15
4 R 4 R 4
Anillustration
Figure1.1.An
Figure illustrationofofa aharmonic-based
harmonic-basednonlinear
nonlineardetection.
detection.
The first term in Equation (2) describes the transmitted power density incident at the
The first term in Equation (2) describes the transmitted power density incident at the
target. The second term is the “apparent RCS” which refers to the apparent size of the target
target. The second term is the “apparent RCS” which refers to 0the apparent size of the
that is perceived by the radar [30]. Note that σ is changed to σ to indicate the apparent
target that is perceived by the radar [30]. Note that is changed to to indicate the
RCS. The third term represents the free space spreading loss. The last term describes the
apparent RCS. The third term represents the free space spreading loss. The last term de-
effective aperture of the Rx antenna. Hence, the total received power Pr is the product of
scribes the effective aperture of the Rx antenna. Hence, the total received power Pr is the
the abovementioned terms.
product of the2 abovementioned
Figure terms. of nonlinear radar from transmit to receive. Here,
illustrates each process
Figure 2 illustrates each process
we consider the nonlinear interaction of nonlinear radar
at the target byfrom transmit
treating to receive.
the target as anHere, we
effective
consider the nonlinear interaction at the target by treating the target
antenna as suggested by Gallagher et al. [29,30]. By applying the nonlinear interaction at as an effective an-
tenna as suggested
the target by Gallagher
into Equation et al. [29,30].
(2), derivation By applying
of the nonlinear theequation
radar nonlinear interaction
is possible at the a
through
target into Equation
few additional steps.(2),
Thederivation
transmitted of power
the nonlinear
densityradar equation
incident is possible
at the target can bethrough
expressed a
few additional
as the first termsteps. The transmitted
of Equation (2). Assuming power thedensity
target hasincident at thearea
an effective target
of Acan
tgt be
[m 2 ],ex-
the
2
power that
pressed as theis first
received
term byof Equation PinAssuming
the target,(2). , can be expressed
the targetbyhas
taking the product
an effective Atgt [mfirst
area of the ]
2022, 22, x FOR PEER REVIEW termpower
, the in Equation
that is (2) and Atgt
received by as:
the target, Pin , can be expressed by taking 4 of 15the product of
Pt Gt
the first term in Equation (2) and Atgt Pas: in = Atgt (3)
4πR2
PG
Pin = t t
Atgt (3)
4 R 2
The power that is excited from the nonlinear interaction by the target, Pout , can be
represented with a power series (assuming memoryless), i.e., [31]:
Pout = an ( Pin )
n
(4)
n =1
(a) th
(b)
Where a n is a scaling coefficient of the n harmonic that is produced by the nonlinear
Figure 2. (a) Transmit
Figureand (b)Transmit
2. (a) receive process
and (b)in nonlinear
receive detection.
process in nonlinear
th detection.
characteristics of the target. The power at the n harmonic, Poutn , can be extracted from
the power
As Pscn travels The series
backpower in Equation
that
the distanceis of R (4)
excited and
frombethe
towards expressed
nonlinear
Rx as:interaction
antenna, the incident the target, Pout , can be
byreceived
represented with a power series (assuming memoryless), i.e., [31]:
power density, Srn , can be obtained as: Pout = an ( Pin )
n
n
(5)
∞
out = ∑ an (gain
Pin )nof the target at the nth harmonic
Therefore, when the effective PscnP
transmission (4)
S rn = (7)
4 R
frequency is given as Gtgtn , the scattered 2 n =1
power from the target can be expressed as:
By rearranging an terms
wherethe is a scaling coefficient
in Equation (7) using nth harmonic
of theEquations that
(3)–(6) to is
beproduced
in line withby the nonlinear
Pscn = Gtgtthn Pout (6)
Equation (2) bycharacteristics
separating theoftransmit,
the target. The power
scattering, and the n harmonic,
at receive terms,
n
we Pobtain:
outn , can be extracted from the
power series in Equation (4) and be expressed as:
PG PGt 1 n
n −1
where is defined as the nonlinear RCS of the target, which describes the physical and
Sensors 2022, 22, 6086 4 of 15
Therefore, when the effective transmission gain of the target at the nth harmonic
frequency is given as Gtgtn , the scattered power from the target can be expressed as:
As Pscn travels back the distance of R towards the Rx antenna, the incident received
power density, Srn , can be obtained as:
Pscn
Sr n = (7)
4πR2
By rearranging the terms in Equation (7) using Equations (3)–(6) to be in line with
Equation (2) by separating the transmit, scattering, and receive terms, we obtain:
"
Pt Gt n−1
#
Pt Gt n 1
Sr n = Gtgtn an Atgt (8)
4πR2 4πR2 4πR2
The second term in Equation (8) is the apparent RCS σ0 which corresponds to the
nth harmonic response that is scattered from the target. The apparent RCS can then be
expressed by separating the radar-specific terms and target-specific terms as [29]:
( Pt Gt )n−1
σ0 = σn (9)
(4π )n−1 R2n−2
where σn is defined as the nonlinear RCS of the target, which describes the physical and
nonlinear characteristics of the target and can be expressed as [31]:
n
σn = an Gtgtn Atgt (10)
Since σ0 is the apparent RCS as seen by the radar, it has the unit of [m2 ]. But unlike
linear RCS its value depends on the transmit power, gain, and distance, which determines
the incident power level at the target as indicated in Equation (9). Depending on the
incident power level at the target, the nonlinear response from the target varies. On the
other hand, σn describes the physical and nonlinear characteristics of the target at the nth
harmonic, which is regarded as the true RCS of nonlinear targets at the corresponding
harmonic frequency. Therefore, the unit of σn depends on the harmonic order and is
expressed as [m2 / Wn−1 /m2n−2 ].
The total received power Prn can then be obtained by the product of Srn and the
effective area of the Rx antenna as:
"
Pt Gt n−1
#
λ2
Pt Gt n 1
Prn = Gtgtn an Atgt Gr (11)
4πR2 4πR2 4πR2 4π
By using Equations (9) and (10), Equation (11) can be expressed using σn as:
( Pt Gt )n Gr λ2 · σn
Prn = (12)
(4π )n+2 R2n+2
which is referred to as the nonlinear radar equation [7,29]. The derivation above implies that
one can obtain σn of a nonlinear target at the corresponding harmonic from the experimental
or calculated σ0 . Alternatively, Equation (12) can be directly used to obtain σn from Prn .
In our recent work, we have conducted a measurement to obtain the apparent and
nonlinear RCS of various electronic devices [30]. In the measurement, CW was used as
the transmitted signal at 3.1 GHz of the fundamental frequency with 43 dBm of transmit
power. Various types of electronic devices are selected as targets and located at 2.5 m from
the transmit and receive antennas. Figure 3 shows the targets of interest that were used
perimental or calculated . Alternatively, Equation (12) can be directly used to obtain
n from Prn .
In our recent work, we have conducted a measurement to obtain the apparent and
nonlinear RCS of various electronic devices [30]. In the measurement, CW was used as the
Sensors 2022, 22, 6086 transmitted signal at 3.1 GHz of the fundamental frequency with 43 dBm of transmit 5 of 15
power. Various types of electronic devices are selected as targets and located at 2.5 m from
the transmit and receive antennas. Figure 3 shows the targets of interest that were used in
theinexperiment, whichwhich
the experiment, are the
aretargets that were
the targets used for
that were usedthefor
FMCW nonlinear
the FMCW radar radar
nonlinear ex-
periment that are
experiment thatpresented in this
are presented in paper.
this paper.
Figure
Figure 4 shows
4 shows anan example
example of of
thethe received
received power
power of of harmonic
harmonic response,P
response, 2 , from
Pr2 , rfrom
one of the targets (i.e., laptop). It can be seen that −32.04 dBm is received at the second
one of the targets (i.e., laptop). It can be seen that −32.04 dBm is received at the second
Sensors 2022, 22, x FOR PEER REVIEW harmonic frequency of 6.2 GHz. Table 1 summarizes the types and physical cross–sections
6 of 15
harmonic frequency of 6.2 GHz. Table 1 summarizes the types and physical cross–sections
of the selected nonlinear targets and the received power level of the harmonic responses of
of the selected nonlinear targets and the received power level of the harmonic responses
each target.
of each target.
Table 1. Various nonlinear targets and their estimated nonlinear RCS based on the received power
[30].
Tablet
and Atgt of PC (15.7 × and
the2target, 24.7)(2) the increase−in
32.61
the electronic board size4.94 × 10−5also be
should
Laptop (35 × 22) −32.04 5.63 × 10−5
roughly proportional to the number of semiconductor devices, which may result in a
higher level of excited harmonic response (i.e., increase in the values of a n ).
The second harmonic nonlinear RCS, σ2 , of each measured target that is listed in Table 1
For a given target with 2 , the apparent RCS can vary depending on the incident
is calculated using Equation (9) from the measured received power. It is demonstrated that
power level as indicated in Equation (9). To validate
the values of the calculated nonlinear RCS are within this,the
wegeneral
obtain range
the values of atRCS
of nonlinear
t 4 R )] as
2
different
(secondincident power
harmonic) 10−8 –10
ofdensity −5 [m(2PG
levels /t W/m 2 by suggested
varying Ptinfor three targets
previous with
studies [7,32].
the largest, middle, and smallest 2 . Figure 5 shows the values of for these targets
at a distance of 2.5 m at three different incident power density levels, where it can be seen
that the value of indeed increased for all the targets as Pt increases.
The second harmonic nonlinear RCS, 2 , of each measured target that is listed in
Table 1 is calculated using Equation (9) from the measured received power. It is demon-
strated that the values of the calculated nonlinear RCS are within the general range of
Sensors 2022, 22, 6086 nonlinear RCS (second harmonic) of 10−8–10−5 [ m2 ( W m2 ) ] as suggested in previous stud- 6 of 15
ies [7,32]. Furthermore, it can be seen that the value of 2 and the received power is
roughly proportional to the physical size of the target. Such a relationship can be at-
tributed to two it
Furthermore, possible
can be reasons:
seen that(1) thethe increase
value of σ2 in
andthethe
physical size
received of theistarget
power roughlygenerally
propor-
follows the
tional to thesize of thesize
physical electronic boardsSuch
of the target. within the target,can
a relationship which in turn increases
be attributed to two possible
Gtgt
n
reasons: (1) the increase in the physical size of the target generally
and Atgt of the target, and (2) the increase in the electronic board size should also be follows the size of the
electronic boards within the target, which in turn increases Gtgtn and Atgt of the target,
roughly
and (2) proportional
the increase intothe theelectronic
number boardof semiconductor
size should also devices, whichproportional
be roughly may result to in the
a
higher level of excited harmonic response (i.e., increase in the
number of semiconductor devices, which may result in a higher level of excited values of a n ). harmonic
For a given target with
response (i.e., increase in the values 2 , the apparent
of an ). RCS can vary depending on the incident
powerForlevel as indicated
a given target in Equation
with σ2 , the (9). To validate
apparent RCS can this,vary
we obtain the values
depending on the at
ofincident
σ0 at
t t 4 R ) 2by varying Pt for three targets with
power level as indicated in Equation
different incident power density levels ( PG (9). To validate
2 this, we obtain the values of
different incident power density levels ( Pt Gt /4πR ) by varying Pt for three targets with
the largest, middle, and smallest 2 . Figure 5 shows the values of for these targets
the largest, middle, and smallest σ2 . Figure 5 shows the values of σ0 for these targets at a
atdistance
a distance of 2.5
of 2.5 m three
m at at three different
different incident
incident power
power density
density levels,
levels, where
where it canit can be seen
be seen that
that the value of
0 indeed increased for all the targets as Pt increases.
the value of σ indeed increased for all the targets as Pt increases.
Figure
Figure5.5.Measured
Measuredapparent
apparentRCS
RCSofoftargets
targetsvs.
vs.the
theincident
incidentpower
powerdensity.
density.
2.2.Estimation
2.2. EstimationofofMaximum
MaximumDetectable
DetectableRange
RangeforforFMCW
FMCWRadar Radar
Theradar
The radarequation
equationcan
canbebeused
usedtotoestimate
estimatethe themaximum
maximumdetectable
detectablerange.
range.Here,
Here,toto
estimate the maximum detectable range for a second harmonic nonlinear
estimate the maximum detectable range for a second harmonic nonlinear FMCW radar, FMCW radar,
Equation (12) is modified in the following manner. First, changing Pr2 into SSmin (the mini-
Equation (12) is modified in the following manner. First, changing Pr into min (the min-
mum receivable power) and equating with R, Equation (12) can be expressed as: 2
imum receivable power) and equating with R , Equation (12) can be expressed as:
!1/6
( Pt Gt )2 Gr λ2 2 · σ2
R= (13)
(4π )4 Smin
Smin can be expressed as kTe BF (SNR)min , where kTe B is the thermal noise in the
system, with k being the Boltzmann constant, Te being the temperature, and B being the
bandwidth of the receiver (second harmonic band) which needs to be the twice the transmit
bandwidth. F is the noise figure of the receiver and (SNR)min is the minimum signal to
noise ratio that is allowed for detection. Applying the aforementioned terms, the maximum
detectable range can be expressed as:
!1/6
( Pt Gt )2 Gr λ2 2 σ2
Rmax = (14)
(4π )4 kTe BF · (SNR)min
Next, Equation (14) is modified to account for the FMCW radar. To this end, frequency
modulation bandwidth and the chirp duration need to be considered, which results in the
processing gain (PGR) [33], i.e.,:
PGR = BTchirp (15)
Sensors 2022, 22, 6086 7 of 15
In general, loss exists throughout the radar system, which can be denoted by L.
Considering PGR and L, Equation (14) can be re-written as:
!1/6
2, 22, x FOR PEER REVIEW ( Pt Gt )2 Gr λ2 2 σ2 8 of 15
Rmax = 4
PGR (16)
(4π ) kTe BFL · (SNR)min
of 1 s . Figure 7which
showscan thealso be expressed
transmit FMCW as: waveform that is generated from Equation
(18) for one chirp duration. In the transmit mode, s (t ) is repeatedly transmitted in a con-
( Pt Gt )2 Gr λ2 2 σ2 Tchirp
tinuous manner. (SNR)min = 4 6
(17)
( 4π )
On the receiving end, the operating bandwidth needs to be the second harmonic kTe FLR max
(a)
(b)
Figure 6. NonlinearFigure
FMCW radar measurement
6. Nonlinear setup:
FMCW radar (a) Measurement
measurement setup:environment in anechoic
(a) Measurement environment in anechoic
chamber and (b) block diagram of setup.
chamber and (b) block diagram of setup.
(a) (b)
(a)
gain horn with a gain of 16.7 dBi. The transmit FMCW signal s(t) that was used in the
measurement is generated based on the following equation:
0.5 3.15
0 3.1
-0.5 3.05
Sensors 2022, 22, x FOR PEER REVIEW 9 of 15
-1 3
0 200 400 600 800 1000 0 200 400 600 800 1000
Time(ns) Time(ns)
(a) (b)
50
Normalized Amplitude
-50
-100
-150
-200
2.9 3 3.1 3.2 3.3
Frequency (GHz)
(c) (d)
Figure 7. Transmitted
Figure FMCW
7. Transmitted signalsignal
FMCW that was
thatused
was in theinmeasurement:
used (a) time
the measurement: (a)domain, (b) time‐
time domain, (b) time-
frequency domain,
frequency frequency
domain, domain
frequency (c) amplitude,
domain and (d)
(c) amplitude, andphase.
(d) phase.
4. ResultsOn andthe
Discussion
receiving end, the operating bandwidth needs to be the second harmonic
bandwidth of
4.1. Measured Second the transmitResponses
Harmonic signals, i.e.,
from6–6.4 GHz. Therefore, a C-band standard gain horn
Targets
antenna with a gain of 21 dBi is placed followed by two VHF-4600+ high pass filters that
The second harmonic FMCW responses are measured for the electronic devices with
cover a bandwidth of 5.2–10.5 GHz for eliminating fundamental responses sufficiently
various sizes as listed in Table 1. The time delay in the received harmonic responses in‐
and a ZX60-123SN-S+ low noise amplifier (LNA). A Tektronix TDS6154C fast-sampling
cludes not only the delay due to the target distance, but the delays that are caused by the
oscilloscope is used to directly sample the received signals in the time domain. A clock
system such as cable, filters, etc. Hence, it is essential for accurate target detection to sep‐
signal that is generated from the AWG is directly fed into the oscilloscope to act as a
arate unwanted time‐delay from the total delay. In order to remove the system time‐delay,
reference signal. In order to minimize the unwanted harmonics from the source reaching
a calibration measurement
the receiver, the Tx andisRxperformed
antennas are before taking with
positioned the target measurements.
a slight Figuresystem,
offset. In a practical 8
shows the spectrum of the received harmonic responses scattered from the
the received radar signals would pass through a mixer to be down-converted to baseband,targets that
werewhere
measured. It can be
the signals seen that
contain the each
beat target response
frequencies, is spread
which over
are then 6–6.4 GHz,
digitized for which
processing.
is theHowever,
second harmonic band of the transmit waveform. Also, the received
note that in our experiment we perform the frequency down-conversion power level ofin the
the target
signalresponses
processingdecreases in a sampled
with directly similar manner
data. to their nonlinear RCS in relation to
their size, as expected. Note that the power levels that are shown in Figure 8 may seem
lower4.since theyand
Results include the total cable loss of about 12 dB in the measurement setup. The
Discussion
plots4.1.
in Figure 8 are
Measured to show
Second that detection
Harmonic Responsesoffrom
target responses was properly achieved in
Targets
the measurement.
The second harmonic FMCW responses are measured for the electronic devices with
various sizes as listed in Table 1. The time delay in the received harmonic responses includes
-40 only the delay due to the target distance, but
not -40 the delays that are caused by the system
-50 -50
such as cable, filters, etc. Hence, it is essential for accurate target detection to separate
-60 -60
unwanted
-70
time-delay from the total delay. In order
-70
to remove the system time-delay, a
calibration
-80 measurement is performed before-80 taking the target measurements. Figure 8
shows
-90 the spectrum of the received harmonic-90responses scattered from the targets that
-100 measured. It can be seen that each target
were -100
response is spread over 6–6.4 GHz, which
-110
is the second harmonic band of the transmit-110 waveform. Also, the received power level
-120 -120
of the
5.8 target
6 responses
6.2 decreases
6.4 in a similar manner
6.6 5.8 to6their nonlinear
6.2 6.4RCS in
6.6 relation to
Frequency (GHz) Frequency (GHz)
their size, as expected. Note that the power levels that are shown in Figure 8 may seem
(a) (b)
-40 -40
-50 -50
-60 -60
-70 -70
arate unwanted time-delay from the total delay. In order to remove the system time-delay,
a calibration measurement is performed before taking the target measurements. Figure 8
shows the spectrum of the received harmonic responses scattered from the targets that
were measured. It can be seen that each target response is spread over 6–6.4 GHz, which
Sensors 2022, 22, 6086 is the second harmonic band of the transmit waveform. Also, the received power level 9ofof 15
the target responses decreases in a similar manner to their nonlinear RCS in relation to
their size, as expected. Note that the power levels that are shown in Figure 8 may seem
lower since they include the total cable loss of about 12 dB in the measurement setup. The
lower since they include the total cable loss of about 12 dB in the measurement setup. The
plots in Figure 8 are to show that detection of target responses was properly achieved in
plots in Figure 8 are to show that detection of target responses was properly achieved in
the measurement.
the measurement.
(a) (b)
(c) (d)
(e) (f)
(g)
Figure 8. Amplitude
Figure spectrum
8. Amplitude of the
spectrum measured
of the target
measured responses.
target (a) laptop,
responses. (b) (b)
(a) laptop, tablet PC PC
tablet 2, (c)
2, tablet
(c) tablet
PC PC
1, (d) cellcell
1, (d) phone, (e) diode-terminated
phone, antenna,
(e) diode-terminated (f) radio
antenna, 2, (g)
(f) radio radio
2, (g) 1. 1.
radio
Since
Since thethe target
target responses
responses areare
thethe second
second harmonic
harmonic version
version of the
of the transmit
transmit FMCW FMCW
waveform,
waveform, their
their phasemust
phase mustexhibit
exhibitaa nonlinear characteristic
characteristicasasa afunction
function of of
thethe
frequency.
fre-
As an example, the phase of the response from the laptop in comparison
quency. As an example, the phase of the response from the laptop in comparison to the to the transmit
waveform
transmit is shown
waveform in Figure
is shown 9. In Figure
in Figure 9a, the9a,
9. In Figure phase of the transmit
the phase waveform
of the transmit is shown,
waveform
while in Figure 9b the phase of the harmonic response from the laptop is
is shown, while in Figure 9b the phase of the harmonic response from the laptop is shown, shown, over their
overrespective frequency
their respective bands. bands.
frequency It can be seenbe
It can that thethat
seen phase
the change in the harmonic
phase change response
in the harmonic
response is twice that in the fundamental waveform as expected. Other target responses
are also confirmed to exhibit the same phase behavior.
Since the target responses are the second harmonic version of the transmit FMCW
waveform, their phase must exhibit a nonlinear characteristic as a function of the fre-
quency. As an example, the phase of the response from the laptop in comparison to the
Sensors 2022, 22, 6086 transmit waveform is shown in Figure 9. In Figure 9a, the phase of the transmit waveform 10 of 15
s 2022, 22, x FOR PEER REVIEW is shown, while in Figure 9b the phase of the harmonic response from the11laptop
of 15 is shown,
over their respective frequency bands. It can be seen that the phase change in the harmonic
response is twice that in the fundamental waveform as expected. Other target responses
is twice that in the fundamental waveform as expected. Other target responses are also
are also confirmed to exhibit the same phase behavior.
doubled to match the bandwidth
confirmed to exhibitwith the received
the same signals before mixing. Figure 10 illus-
phase behavior.
trates the process of extracting the beat signal and shows an example of the extracted beat
signal from a measured target response. After extracting the beat signals, the Fourier
Transform of sb (t ) is taken, which provides a compressed target echo, and the target
distance information can be calculated by using the following relationship:
cTs
R= fb (19)
2B
where c is the wave velocity, and f b is the baseband (beat) frequency.
Figure 11 shows the compressed echo as a function of the range for the measured
targets. Since the range is fixed at 2.5 m for all the targets, the peak of the Fourier Trans-
form of the beat signals occurs at 6.82 MHz, which is approximated to be 2.57 m using
(a) (b)
Equation (19). The detected range seems to be in a good agreement with the actual target
range of 2.5 Figure
m. 9. Phase
Figure plots
9. Phase of the
plots (a) transmit
of the andand
(a) transmit (b) (b)
receive (from
receive thethe
(from laptop) waveforms.
laptop) waveforms.
Range accuracy can be improved by considering the beat frequency error [27], which
4.2.4.2. Extraction
Extraction of Range
of Range Information
Information
refers to the difference between the ideal beat frequency and the measured beat frequency.
The maximum In beatIn
the the FMCW
frequency
FMCW radar, radar,
error the( fthe
) can
rangerange information
expressed
information is obtained
as:is obtained fromfrom the beat
the beat signal, sb (t s) b,(t),
signal,
which is extracted by mixing the transmit and receive signals. The beat frequency occurs
which is extracted by mixing the transmit and receive signals. The beat frequency occurs
due to the time delay between f =the f s transmit
Ns and receive frequencies that(20) is caused by the
due to the time delay between the transmit and receive frequencies that is caused by the
target distance. Here, we perform the mixing through signal processing by multiplying
where N s target
is thedistance.
number Here,of datawe performover
sampled the one
mixing through
period of beat signal
signalprocessing
and f isbythe multiplying
the transmit and receive signals and filtering out the lower band. Fors the second harmonic
sample rate. theTherefore,
transmit and the receive
range signals
error and filtering
(resolution) is out the lower
expressed as band. For the second harmonic
FMCW radar that was considered here, the frequencies of the transmit signal must be
FMCW radar that was considered here, the frequencies of the transmit signal must be
doubled to match the bandwidth cT f cT with the
f received signals before mixing. Figure 10
illustrates the process = s
ofRextracting = thes beats signal and shows an example of the extracted
(21)
2B 2B N s
beat signal from a measured target response. After extracting the beat signals, the Fourier
Transform
As shown in Equations (t) isand
of sb(20) taken,
(21),which provides relationship
it is a trade-off a compressed target echo,
between B , Tsand
, the target
distance information can be calculated by using the following relationship:
f s , and N s . Therefore, it is important to determine the appropriate parameters for a de-
sired range resolution in a given system. The parameterscT that
s were used in our measure-
R= × fb (19)
ment give the beat frequency error of 1 MHz, which corresponds 2B to the range resolution
of 0.375 m.
where c is the wave velocity, and f b is the baseband (beat) frequency.
(a) (b)
Figure 10. (a) Process
Figure of
10.extracting the
(a) Process ofbeat signal the
extracting andbeat
(b) the extracted
signal and (b)beat
the signal.
extracted beat signal.
Figure 11 shows the compressed echo as a function of the range for the measured tar-
gets. Since the range is fixed at 2.5 m for all the targets, the peak of the Fourier Transform of
the beat signals occurs at 6.82 MHz, which is approximated to be 2.57 m using Equation (19).
The detected range seems to be in a good agreement with the actual target range of 2.5 m.
(a) (b)
Sensors 2022, 22, 6086
(a) (b) 11 of 15
Figure 10. (a) Process of extracting the beat signal and (b) the extracted beat signal.
(a) (b)
(c) (d)
(e) (f)
(g)
Figure 11. 11.
Figure Compressed
Compressed target echo
target as aasfunction
echo of the
a function range
of the for for
range each measured
each measuredtarget. (a) laptop,
target. (a) laptop,
(b) (b)
tablet PC 2, (c) tablet PC 1, (d) diode-terminated antenna, (e) cell phone, (f) radio 2, (g) radio
tablet PC 2, (c) tablet PC 1, (d) diode-terminated antenna, (e) cell phone, (f) radio 2, (g) radio 1. 1.
4.3. Nonlinear
RangeDetection
accuracyincanthebe
Presence of Linear
improved Scatterer the beat frequency error [27], which
by considering
refers to the difference between the ideal beat
To test the proposed nonlinear radar’s ability frequency andlinear
to suppress the measured beat
scatter, the frequency.
detection
The maximum
measurements beat
were frequency
carried error
out with (∆f ) can targets
nonlinear expressed as:presence of a linear scatterer.
in the
Figure 12 illustrates the measurement setup where the laptop and diode-terminated an-
∆ f = f s /Ns (20)
tenna are used as nonlinear targets and a rectangular metal plate (19 cm × 22 cm) as a
linear scatterer.
where Here,
Ns is the the metal
number of dataplate is located
sampled at a period
over one distance
of of 1.5signal
beat m from
andthe antennas,
f s is the sample
andrate.
the nonlinear targets were placed at a distance of 2 m
Therefore, the range error (resolution) is expressed as from the antennas. The received
signals are processed in the same manner as described in Section 4.2, which can be seen
in Figure 13. Figure 13a,b, respectively, showcTs ∆ f the cT fs
compressed
s target echo of the laptop
∆R = = × (21)
and diode-terminated antenna that were measured 2B in the N
2B presence
s of the metal plate, as a
function of the range. The peak of the Fourier Transform of the beat signals occurs at 5.66
MHz, which approximately corresponds to 2.12 m. The detected range seems to be in good
agreement with the actual range of 2 m of the nonlinear targets. On the other hand, there
is no peak that occurs around 1.5 m, indicating that the linear scatterer is not detected.
Figure 13c shows the resulting compressed signal when the metal plate and antenna with
an open circuit are present. Note that the antenna with the open circuit is the same antenna
Sensors 2022, 22, 6086 12 of 15
experimentally obtained nonlinear RCS in Table 1. The thermal noise (kTe B) is estimated to
be −87.9 dBm using the standard room temperature. An overall receiver loss (L) of 5 dB
is assumed. The noise figure (F) is set to 2 dB. Figure 14 shows the estimated maximum
detectable range for two representative targets (i.e., laptop, radio 1) for various transmit
power levels (33–42 dBm). The minimum allowable signal to noise ratio (SNRmin ) is set at
12 dB as indicated by the red dashed line in the figure. It can be seen that the maximum
detectable range for the laptop is approximately 16 m for a transmit power of 33 dBm.
Furthermore, it is estimated that at least 39 dBm of transmit power is required to detect
radio 1 at distances that are greater than 10 m. This estimation seems to support the
measured detection results, where the smallest target (radio 1) is detected with a good level
of signal to noise ratio at a 2.5 m range. However, the values of the nonlinear RCS can
change according to each target’s physical characteristics and nonlinearities of the included
electronic circuit. Therefore, this range estimation can be utilized when designing an FMCW
Sensors 2022, 22, x FOR PEER REVIEW 14 of 15
nonlinear radar system while considering the aforementioned elements, thereby allowing
for the selection of appropriate system parameters for detecting various electronic targets.
(a) (b)
Figure 14. Estimation
Figure of the
14. Estimation ofmaximum
the maximumdetectable range
detectable for: (a)
range for:the
(a)laptop and (b)
the laptop andradio 1 based
(b) radio on on
1 based
the presented FMCW
the presented nonlinear
FMCW radarradar
nonlinear parameters.
parameters.
5. Conclusions
5. Conclusions
In paper,
In this this paper, we demonstrate
we demonstrate a seconda second harmonic-based
harmonic-based FMCW nonlinear
FMCW nonlinear radar con- radar
concept for the detection of electronic devices. With a laboratory experimental
cept for the detection of electronic devices. With a laboratory experimental apparatus, apparatus,
FMCWFMCW nonlinear
nonlinear radar
radar measurement
measurement is presented
is presented forforvarious
variouselectronic
electronicdevices.
devices.The
The re-
results
indicate the successful detection of small electronic devices with nonlinear
sults indicate the successful detection of small electronic devices with nonlinear FMCW FMCW radar
when appropriate system parameters are selected. Here, the detection
radar when appropriate system parameters are selected. Here, the detection of only sta- of only stationary
targets
tionary is considered.
targets SinceSince
is considered. FMCW FMCWradarradar
is widely utilized
is widely for detecting
utilized moving
for detecting targets, it
moving
targets, it is also possible to detect electronic devices in motion in the midst of linear scat-with
is also possible to detect electronic devices in motion in the midst of linear scatterers
further
terers with processing. The proposed
further processing. FMCW nonlinear
The proposed FMCW nonlinearradar can be utilized
radar can be in applications
utilized in
such as security, military, and search and rescue, where obstructed or concealed electronic
applications such as security, military, and search and rescue, where obstructed or con-
devices need to be detected in the presence of linear scatterer.
cealed electronic devices need to be detected in the presence of linear scatterer.
Author Contributions: Conceptualization, K.C., S.O. and S.K.H.; Data curation, K.C., S.O. and S.K.H.;
Author Contributions: Conceptualization, K.C., S.O., and S.K.H.; Data curation, K.C., S.O., and
Formal analysis, K.C. and S.O.; Funding acquisition, S.K.H.; Investigation, H.H.; Methodology, K.C.,
S.K.H.; Formal analysis, K.C. and S.O.; Funding acquisition, S.K.H.; Investigation, H.H.; Methodol-
S.O., H.P. and S.K.H.; Project administration, S.K.H.; Validation, K.C. and S.O.; Writing—original
ogy, K.C., S.O., H.P., and S.K.H.; Project administration, S.K.H.; Validation, K.C. and S.O.; Writing—
draft, K.C. and S.O.; Writing—review & editing, K.C., S.O., H.H., H.P. and S.K.H. All authors have
original draft, K.C. and S.O.; Writing—review & editing, K.C., S.O., H.H., H.P., and S.K.H. All au-
read and agreed to the published version of the manuscript.
thors have read and agreed to the published version of the manuscript.
Funding: This research was supported by the Challengeable Future Technology Research and
Funding: This research was supported by the Challengeable Future Technology Research and De-
Development
velopment Program Program (912902601)
(912902601) of Agency
of Agency for Defense
for Defense Development
Development in 2020.in 2020.
Institutional
Institutional Review
Review Board
Board NotNot
Statement:
Statement: Applicable.
Applicable
Informed
Informed Consent
Consent NotNot
Statement:
Statement: Applicable.
Applicable
DataData Availability
Availability DataData
Statement:
Statement: sharing
sharing is noisapplicable.
no applicable.
Conflicts
Conflicts The The
of Interest:
of Interest: authors
authors declare
declare no conflict
no conflict of interest.
of interest.
Reference
1. Skolnik, M. Role of Radar in Microwaves. IEEE Trans. Microw. Theory Tech. 2002, 50, 625–632.
2. Skolnik, M. Radar in the Twentieth Century. IEEE Aerosp. Electron. Syst. 2000, 15, 27–46.
3. Harger, R.O. Harmonic radar systems for near-ground in- foliage nonlinear scatterers. IEEE Trans. Aerosp. Electron. Syst. 1976,
Sensors 2022, 22, 6086 14 of 15
References
1. Skolnik, M. Role of Radar in Microwaves. IEEE Trans. Microw. Theory Tech. 2002, 50, 625–632. [CrossRef]
2. Skolnik, M. Radar in the Twentieth Century. IEEE Aerosp. Electron. Syst. 2000, 15, 27–46.
3. Harger, R.O. Harmonic radar systems for near-ground in- foliage nonlinear scatterers. IEEE Trans. Aerosp. Electron. Syst. 1976,
12, 230–245.
4. Mazzaro, G.J. Nonlinear Junction Detection vs. Electronics: System Design and Design and Improved Linearity. In Proceedings
of the 2020 IEEE International Radar Conference (RADAR), Washington, DC, USA, 28–30 April 2020.
5. Flemming, M.A.; Mullins, F.H.; Watson, A.W.D. Harmonic Radar Detection Systems. In Proceedings of the IEEE International
Conference on Radar (Radar-77), London, UK, 25–28 October 1977; pp. 552–554.
6. Bahr, A.; Frank, V.; Sweeney, L. Radar Scattering from Intermittently Contacting Metal Targets. IEEE Trans. Antennas Propag. 1977,
25, 512–518. [CrossRef]
7. Mazzaro, G.J.; Martone, A.F.; Ranney, K.I.; Narayanan, R.M. Nonlinear Radar for Finding RF Electronics: System Design and
Recent Advancements. IEEE Trans. Micro. Theory. Tech. 2017, 65, 1716–1726. [CrossRef]
8. Faia, J.M.; He, Y.; Park, H.S.; Wheeler, E.D.; Hong, S.K. Detection and Location of Nonlinear Scatterers Using DORT Applied with
Pulse Inversion. Prog. Electromagn. Res. Lett. 2018, 80, 101–108. [CrossRef]
9. Hong, S.K.; Park, H.S. Embedded Resonances for Discrimination of Multiple Passive Nonlinear Targets Applicable to DORT. Prog.
Electromagn. Res. 2019, 81, 31–42. [CrossRef]
10. Opitz, C.L. Metal-detecting radar rejects clutter naturally. Microwaves 1976, 15, 12–14.
11. Steele, D.W.; Rotondo, F.S.; Houck, J.L. Radar System for Manmade Device Detection and Discrimination from Clutter. U.S.
Patent 7,830,299, 9 November 2010.
12. Mazzaro, G.J.; Ranney, K.I.; Gallagher, K.A.; McGowan, S.; Martone, A.F. Simultaneous-Frequency Nonlinear Radar: Hardware
Simulation; Technical Report; Army Research Lab Sensors and Electron Devices Directorate: Adelphi, MD, USA, 2015.
13. Phelan, B.R.; Sherbondy, K.D.; Ranney, K.I.; Narayanan, R.M. Design and Performance of an Ultra-Wideband stepped-Frequency
Radar with Precise Frequency Control for Landmine and IED Detection. Proc. SPIE. 2014, 9077, 53–64.
14. Li, Z.; Yang, Z.; Song, C.; Li, C.; Peng, Z.; Xu, W. E-Eye: Hidden Electronics Recognition Through Mmwave Nonlinear Effects. 2018.
In Proceedings of the 16th ACM Conference on Embedded Networked Sensor Systems, Shenzhen, China, 4–7 November 2018.
15. Chioukh, L.; Boutayeb, H.; Deslandes, D.; Wu, K. Noise and Sensitivity of Harmonic Radar Architecture for Remote Sensing and
Detection of Vital Signs. IEEE Trans. Micro. Theory Tech. 2014, 63, 1847–1855. [CrossRef]
16. Chioukh, L.; Boutayeb, H.; Deslandes, D.; Wu, K. Dual-band linear antenna array for harmonic sensing applications. IEEE Ant.
Wirel. Propag. Lett. 2016, 15, 1577–1580. [CrossRef]
17. Mishra, A. Intermoudation-Based Nonlinear Smart Health Sensing of Human Vital Signs and Location. IEEE Access 2019,
7, 158284–158295. [CrossRef]
18. Psychoudakis, D.; Moulder, W.; Chen, C.; Zhu, H.; Volakis, J.L. A Portable Low-Power Harmonic Radar System and Conformal
Tag for Insect Tracking. IEEE Ant. Wirel. Propag. Lett. 2008, 7, 444–447. [CrossRef]
19. Milanesio, D.; Saccani, M.; Maggiora, R.; Laurino, D.; Porporato, M. Design of an Harmonic Radar for the Tracking of the Asian
Yellow-Legged Hornet. Ecol. Evol. 2016, 6, 2170–2178. [CrossRef] [PubMed]
20. Maggiora, R.; Saccani, M.; Milanesio, D.; Porporato, M. An Innovative Harmonic Radar to Track Flying Insects: The Case of Vespa
Velutina. Sci Rep. 2019, 9, 11964. [CrossRef] [PubMed]
21. Tsai, Z.M.; Jau, P.H.; Kuo, N.C.; Kao, J.C.; Lin, K.Y.; Chang, F.R.; Yang, E.C.; Wang, H. A High-Range-Accuracy and High-
Sensitivity Harmonic Radar Using Pulse Pseudorandom Code for Bee Searching. IEEE Trans. Microw. Theory Tech. 2013,
61, 666–675. [CrossRef]
22. Jankiraman, M. Design of Multifrequency CW Radars; SciTech Publishing Inc.: Raleigh, NC, USA, 2007.
23. Mazzaro, G.J.; Gallagher, K.A.; Martone, A.F.; Sherbondy, K.; Narayanan, R.M. Short-Range Harmonic Radar: Chirp Waveform,
Electronic Targets. Proc. SPIE. 2015, 9461, 946108.
24. Ranney, K.I.; Gallagher, K.A.; Sherbondy, K.; Martone, A.F.; Mazzaro, G.J.; Narayanan, R.M. Instantaneous, Stepped-Frequency,
Nonlinear Radar. Proc. SPIE. 2015, 9461, 946122.
25. Stove, A.G. Linear FMCW Radar Technique. IEE Proc. F (Radar Signal Process.) 1992, 139, 343–350. [CrossRef]
26. Richards, M.A. Fundamentals of Radar Signal Processing, 2nd ed.; McGraw-Hill Education: New York, NY, USA, 2005.
27. Jankiraman, M. FMCW Radar Design, kindle ed.; Artech House: Washington, DC, USA, 2018.
28. Hyun, E.; Lee, J.H. Method to Improve Range and Velocity Error Using De-interleaving and Frequency Interpolation for
Automotive FMCW Radars. Int. J. Signal Process. Image Process. Pattern Recognit. 2009, 2, 11–21.
29. Gallagher, K.A. Harmonic Radar: Theory and Applications to Nonlinear Target Detection, Tracking, Imaging and Classification.
Ph.D. Dissertation, The Pennsylvania State University, State College, PA, USA, 2015.
30. Oh, S.Y.; Cha, K.H.; Hong, H.Y.; Park, H.S.; Hong, S.K. Measurement of Nonlinear RCS of Electronic Targets for Nonlinear
Detection. J. Korean Inst. Electromagn. Eng. Sci. 2022, 22, 447–451. [CrossRef]
31. Gallagher, K.A.; Mazzaro, G.J.; Martone, A.F.; Sherbondy, K.; Narayanan, R.M. Derivation and Validation of the Nonlinear Radar
Range Equation. Proc. SPIE 2016, 9829, 98290P.
Sensors 2022, 22, 6086 15 of 15
32. Dardari, D. Detection and Accurate Localization of Harmonic Chipless Tags. EURASIP J. Adv. Signal Process. 2015, 2015, 77.
[CrossRef]
33. Kwak, Y.K. Radar System Engineering; Kyomoon: Paju, South Korea, 2020.