new antivelocity deception jamming technique using pulses with adaptive initial phases
TRANSCRIPT
New Antivelocity Deception
Jamming Technique using
Pulses with Adaptive Initial
Phases
JINDONG ZHANG
DAIYIN ZHU, Member, IEEE
GONG ZHANG, Member, IEEE
Nanjing University of Aeronautics and Astronautics
An effective electronic counter-countermeasures (ECCM)
technique for a moving target indication/pulsed Doppler
(MTI/PD) radar system against velocity deception jamming is
proposed. A novel radar signaling strategy is outlined based
on variations of the initial phases of the transmitted pulses in
pulse repetition interval (PRI) domain. It makes the Doppler
spectrum of the false targets, created by a digital radio frequency
memory (DRFM) repeat-back jammer, form a notch around the
Doppler frequency of the true one. The penalty function and
corresponding algorithm for designing adaptive initial phases
are given. We also present an approach of the multi-channel
matched filter processing to estimate the manner, by which a
DRFM repeat jammer operates, and the parameters of the false
targets. The working flow of an MTI/PD radar for countering
velocity deception jamming is investigated. Simulation results
show the effectiveness of the above methods.
Manuscript received April 30, 2011; revised October 23, 2011 and
May 15, 2011; released for publication October 9, 2012.
IEEE Log No. T-AES/49/2/944536.
Refereeing of this contribution was handled by S. Hary.
This work was supported by the Chinese National Natural Science
Foundation under Contracts 61201367, 61071163, 61071165, and
61100195, the Natural Science Foundation of Jiangsu Province
under Contract BK2012382, and the Defense Industrial Technology
Development Program under Contract B2520110008.
Authors’ address: College of Electronic and Information
Engineering, Nanjing University of Aeronautics and Astronautics,
29 Yudao Street, 210016 Nanjing, China, E-mail: ([email protected]).
0018-9251/13/$26.00 c° 2013 IEEE
I. INTRODUCTION
Moving target indication (MTI) and pulsed
Doppler (PD) radars are two classes of radar systems
designed with coherent processing for detecting
targets in the midst of noise, clutter, and jamming,
which could be stronger than the target signal by
several orders of magnitude. With the capability of
separating moving targets from relatively stationary
clutter, MTI and PD techniques are widely used in
modern radar systems, and are playing an important
role in air surveillance of civilian systems and the
detection of low flying aircraft or missiles [1].
The electronic countermeasures (ECM) has
undergone rapid development with the military radar
over the past two decades [2—3]. The countermeasures
mainly intend to mask useful information, create
deceptive targets more attractive than the true ones,
or aim at saturating the target extraction and tracking
algorithms. The emergence of digital radio frequency
memory (DRFM) makes the jammer, which is referred
to as the repeat-back jammer, capable of listening,
storing, and repeating back the radar’s transmitted
signal to deceive hostile radar systems [4]. In MTI
and PD radar systems, the user relies on multiple
radar signal transmissions for coherent detection.
Based on this property, a DRFM repeat jammer
constructs a series of replicas of the transmitted radar
signal multiplied by a certain Doppler frequency to
deceive the radar system. Therefore DRFM induces
serious threats to the survival of the MTI and PD
radars in the modern battlefield [5].
The most
effective electronic counter-countermeasures (ECCM)
scheme for a radar system against the DRFM repeat
jammer-type systems is pulse diversity [6—9]. The
pulse-diversifying method being considered currently
is to transmit orthogonal pulses from one pulse
repetition interval (PRI) to another. Since the pulse is
varying in the PRI domain, the jammer cannot adapt
easily and the signal transmitted by the DRFM repeat
jammer is approximately orthogonal to the signal that
the radar system is transmitting at the current PRI;
a matched filter with the transmitted radar signal
of the current PRI would weaken the DRFM repeat
jammer signal. Random signal radar (RSR) has been
demonstrated to be quite effective in this application.
Liu and Gu have studied and implemented the random
binary phase-coded pulse radar [10], the random pulse
radar [11], the random position pulse radar [12], and
etc.
The pulse diversifying method has been proved
to be useful for target detection without Doppler
processing. However during MTI filtering in MTI
radar or range-Doppler processing in PD radar,
pulse diversity leads to range sidelobe variation,
severely degrading the integrated signal-to-jammer
ratio (SJR) and therefore limiting the ability to
1290 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 49, NO. 2 APRIL 2013
suppress interference for the radar system. In [8]
several methods have been proposed to mitigate this
performance loss. Code selection of the pseudonoise
(PN) codes with the capability of forming matched
filter output with minimal sidelobe variation improved
the performance by approximately 7 dB. The pulse
compression filter could succeeded in eliminating
sidelobe energy in the filtered output, but suffer
from a 2.5 dB signal-to-noise ratio (SNR) loss
when compared with the matched filter. Combing
the advantages of the matched filter and pulse
compression filter, the digital equalization filtering
method was successful in equalizing differences
created by variable PN codes.
To overcome the demerit of pulse diversity
and reserve the original coherent two-dimensional
processing in range-Doppler domain, pulses with
different initial phases can be utilized. We begin by
examining the merit of pulses with random initial
phases for countering the DRFM repeat-back jammer.
SJR gain obtained by Doppler processing is studied.
Although the Doppler spectrum of the false targets is
suppressed, it will be evenly spread over all Doppler
cells and will affect the true target detection.
In this presentation a novel radar signaling strategy
is outlined based on pulses with adaptive initial
phases. The Doppler spectrum of the false targets
could form a notch around the Doppler frequency of
the true target after Doppler processing. The adaptive
initial phases whose spectrum forms a notch in a
certain band are designed by an iterative algorithm.
The penalty function of a weighted integral of spectral
energy in this band is constructed. The algorithm
applied to compute the adaptive initial phases is to
minimize the penalty function with the phase-only
vector. This constrained optimization problem can be
successfully solved by the phase-only version of the
conjugate gradient and Newton’s method [13]. The
presented technique requires that the manner by which
a DRFM repeat jammer operates (i.e., at the current
PRI, it transmits the radar signal used for the certain
previous or some other past PRI) and parameters of
the false targets for adaptive initial phases design
are both available to the radar. For this purpose we
propose a scheme of multi-channel processing. We
also design a working flow for countering three kinds
of velocity deception jamming, including velocity gate
pull off, multiple false targets, and two blinking false
targets [14, 15]. The detailed introduction of these
deception jammings is given.
Section II introduces the received signal model
under the DRFM repeat-type velocity deception
jamming. Section III evaluates the performance of
pulses with random initial phases. Section IV proposes
a new signaling method using pulses with adaptive
initial phases. The phase-only conjugate gradient
method is introduced in optimizing the penalty
function. A scheme of the multi-channel processing
is proposed to estimate the jammer operation manner
and the parameters of the false targets. The flow
diagram of an MTI/PD radar for countering velocity
deception jamming is given. Section V shows the
simulation results. Section VI is the concluding
remarks.
II. ECHO SIGNAL MODEL UNDER VELOCITYDECEPTION JAMMING
An MTI/PD radar transmits a coherent signal that
is produced by modulating a sinusoid with a pulsed
waveform. Assuming that the signal is composed
of N pulses during coherent processing interval
(CPI). The target in the antenna beam is created as
a point scatter with the specified constant radar cross
section (RCS), and located at a given range with some
Doppler component to simulate velocity from one PRI
to the next. The Doppler frequency of the target is
fd = 2v=¸, where v is its relative velocity to the radar
system and ¸ is the signal wavelength. The signal
returned from a moving target with such Doppler
frequency can be represented as
sn(t) = ¾Tej2¼fdnTr¹(t¡ ¿d ¡nTr) (1)
where ¾T, ¹(t) and Tr denote the target reflectivity
coefficient, complex envelope of the transmitted pulse
and PRI, respectively, ¿d is the delay time, and n
is the pulse number. If the radar system transmits
pulses with a high repetition frequency, there are only
minor changes in the overall operating environment
during the CPI, and the target could be considered
stationary. Therefore we assume that ¾T, ¿d, and fdremain constant over the N slow-time intervals.
The DRFM repeat jammer requires several PRIs
to recognize the radar transmitted pulses, and then
utilizes the captured pulses of the previous PRIs to
replicate multiple false targets. The pulse generated
by the DRFM repeat jammer for deceiving the radar
system at the nth PRI can be expressed as
qn(t) =
PXp=1
¾pe2¼f
p
fnTr¹(t¡ ¿pf ¡nTr) (2)
where P is the number of false targets, and ¾p, ¿pf ,
and fpf denote the amplitude, delay time, and Doppler
frequency of the pth false target, respectively.
The signal received by the radar system can be
written as
rn(t) =qn(t)+ sn(t) + nn(t)
=
PXp=1
¾pe2¼f
p
fnTr¹(t¡ ¿pf ¡nTr)
+¾Tej2¼fdnTr¹(t¡ ¿d¡ nTr) + nn(t) (3)
where nn(t) is the additive white noise at the nth PRI.
ZHANG, ET AL.: NEW ANTIVELOCITY DECEPTION JAMMING TECHNIQUE 1291
Matched filtering is a signal processing strategy for
detecting target, and improving range resolution and
SNR, when using waveforms such as binary phase
encoded PN codes. The M sequence is a typical PN
code and its waveform correlation function can be
given by
ÂM(¿) =
Z¹¤M(t)¹M(t¡ ¿)dt
=1
L
L¡1Xl=¡(L¡1)
Âsub(¿ ¡ lT)Âcode(l) (4)
where
¹M(t) =1pL
L¡1Xk=0
ckrect
μt¡ kTT
¶
Âcode(l) =
L¡1¡kXk=0
c¤kck+l =½L, l = 0
·pL, l 6= 0
Âsub(¿) =T¡ j¿ jT
, j¿ j · T
and T is the subpulse duration, ck is the kth code, and
L is the code length.
The matched filter output, which represents the
correlation function between the received signal and
transmitted pulse, can separate the targets in different
range cells. For simplicity only the case that the
true and false targets are in the same range cell, i.e.,
¿pf ¼ ¿d, is considered, and the noise effect is alsoignored here. The matched filter output at the delay
time ¿d of the nth PRI can be written as
y(n) =
Z¹¤M(t¡ ¿d ¡ nTr)rn(t)dt
=
PXp=1
¾pej2¼f
p
fnTr +¾Te
j2¼fdnTr : (5)
To discriminate the targets with different velocities
in the same range cell, Doppler processing based
on the discrete Fourier transform (DFT) can then be
applied. The DFT result is given by
Y(k) = DFT[y(n)]
=
PXp=1
¾pej(¼=N)(N¡1)(k¡Nfp
fTr)sin¼(k¡Nfpf Tr)sin
¼
N(k¡Nfpf Tr)
+¾Tej(¼=N)(N¡1)(k¡NfdTr) sin¼(k¡NfdTr)
sin¼
N(k¡NfdTr)
: (6)
We can observe that the true and false targets appear
at the same time in Y(k), and cannot be distinguished
from each other. Obviously the enemy could easily
use a DRFM repeat jammer to jam the MTI/PD radar
systems with erroneous replicas of the transmitted
radar pulses.
Fig. 1. Scheme for transmitting pulses with random initial phases
in CPI with lag 2.
III. ECCM USING PULSES WITH RANDOM INITIALPHASES
Now we consider that the radar system is to
transmit a new or modified pulse with a random initial
phase at each PRI. According to (3) the received
signal of the radar system at the nth PRI can be
expressed as
rn(t) =
PXp=1
¾pejÁn¡i e
j2¼fp
fnTr¹(t¡ ¿d ¡ nTr)
+¾TejÁnej2¼fdnTr¹(t¡ ¿d¡ nTr) (7)
where Án, which is the initial phase of the nth
transmitted pulse, is assumed to be uniformly
distributed in [¡¼,¼] and statistically independent ofthat of the other pulses.
The pulse emitted by the jammer is assumed to
lag i PRI behind the pulse transmitted by the radar. It
means that at the nth PRI the DRFM repeat jammer
replicates the pulse that the radar has transmitted
at the (n¡ i)th PRI. For n· i the initial phase ofthe pulse transmitted by the jammer may be that of
the pulse transmitted by the radar during pervious
CPI, or randomly generated. For ease of treatment
extra i pulses with initial phases Á1¡i,Á2¡i, : : : ,Á0 aretransmitted during a CPI and we use the next N pulses
for processing. Figure 1 illustrates the scheme for
transmitting pulses with random initial phases in a
CPI.
For pulses with random initial phases, the matched
filter output at the delay time ¿d of the nth PRI can be
expressed as
y(n) =
PXp=1
¾pej(Án¡i¡Án)ej2¼f
p
fnTr +¾Te
j2¼fdnTr : (8)
In the above equation the initial phase of the
true target Án is canceled by the matched filter of
the current PRI and the residual phase Án¡i¡Ánis induced to the false targets. In the following
description this characteristic can be used to
discriminate the true and false targets in Doppler
1292 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 49, NO. 2 APRIL 2013
spectrum for countering velocity deception jamming.
Considering that the received pulse can be modeled
as the sum of two orthogonal blocks in the
range deception jamming environment in [6], the
matched-filtering operation with regard to the received
pulse of the current PRI brings the possibility of
separating the true target from the false ones in range
domain.
After matched filtering DFT processing is then
applied and the corresponding Doppler spectrum is
given by
P(k) = EfY(k)Y¤(k)g
=
N¡1Xn=0
N¡1Xm=0
E
(¾T exp[j(2¼=N)(NfdTr ¡ k)(n¡m)]
+
PXp=1
¾pej(Án¡i¡Án)e¡j(Ám¡i¡Ám)
¢ exp[j(2¼=N)(Nfpf Tr ¡ k)(n¡m)])
+crossterms
¼N¡1Xn=0
N¡1Xm=0
¾T exp[j(2¼=N)(NfdTr ¡ k)(n¡m)] +PXp=1
¾p
¢(N¡1Xn=0
N¡1Xm=0
exp[j(2¼=N)(Nfp
f Tr ¡ k)(n¡m)]
¢ E[ej(Án¡i¡Án)e¡j(Ám¡i¡Ám)])
(9)
where
Y(k) = DFT[y(n)]
=
N¡1Xn=0
"PXp=1
¾pej(Án¡i¡Án)ej2¼f
p
fnTr +¾Te
j2¼fdnTr
#
¢ e¡j(2¼=N)kn (10)
crossterms
= 2Re
(E
"ÃN¡1Xn=0
¾Tej2¼fdnTr e¡j(2¼=N)kn
!
¢ÃN¡1Xm=0
PXp=1
¾pej(Ám¡i¡Ám)ej2¼f
p
fmTr e¡j(2¼=N)km
!¤#)
= 2Re
(N¡1Xn=0
N¡1Xm=0
PXp=1
¾T¾pej2¼fdnTrE[ej(Ám¡Ám¡i)]
¢ e¡j2¼fp
fmTr exp[¡j(2¼=N)k(n¡m)]
)= 0: (11)
With conclusions in the Appendix, the expectation is 0in the crossterms of (11), which can be neglected.The phase Án introduced by the transmitted pulse
of the current PRI in (8) is canceled out, and theDoppler spectrum P(k) focuses towards the true target.The random phase Án¡i caused by the DRFM repeatjammer cannot cancel out Án, and the energy of thefalse targets will be spread over all Doppler cells.Therefore random initial phases result in the focusedtrue target and suppressed false targets. However in[8], when using variable pulses, changing sidelobemagnitude and phase generated after matched filteringleads to a spreading effect, thereby creating falsesignals in the Doppler spectrum.According to properties in the Appendix, (9) can
be simplified to
P(k) =
N¡1Xn=0
N¡1Xm=0
¾T exp[j(2¼=N)(NfdTr¡ k)(n¡m)]
+
PXp=1
¾p
ÃN¡1Xn=0
ej0
!(12)
where
N¡1Xn=0
N¡1Xm=0
¾T exp[j(2¼=N)(NfdTr¡ k)(n¡m)]¯̄̄̄¯k=NfdTr
=N2¾T (13)
is the power of the true target in the Doppler cellk =NfdTr, and
¾p
ÃN¡1Xn=0
ej0
!¯̄̄̄¯k½[1,N]
=N¾p (14)
is the pth false target power level in any Doppler cellk ½ [1,N]. We find that the energy of the false targetis evenly spread over all Doppler cells.From (13) and (14) the SJR after Doppler
processing can be obtained by
SJRp =N2¾TN¾p
=N¾T¾p=N ¢SJR0,p (15)
whereSJR0,p =
¾T¾p
is the SJR of the true and pth false target beforeDoppler processing. By using pulses with randominitial phases, the power of the false target can besuppressed 10log10N dB in the Doppler spectrum,where N denotes the number of the pulses in a CPI.For 10 pulses the Doppler processing gain is 10 dB.Although the false target is considerably suppressed,the energy of the false targets will be spread overthe entire Doppler spectrum. In the circumstance ofa high-power jammer or the jamming signal enteringvia antenna mainlobe, significantly degraded SJR willaffect the true target detection performance.
ZHANG, ET AL.: NEW ANTIVELOCITY DECEPTION JAMMING TECHNIQUE 1293
Fig. 2. Doppler spectrum of true and false targets formed by
pulses with adaptive initial phases.
IV. ECCM USING PULSES WITH ADAPTIVE INITIALPHASES
A. Adaptive Initial Phases Design
To further improve the target detection
performance in velocity deception jamming
environment, we present a new signaling strategy
and the jamming pulses generated by the DRFM
repeat jammer will adaptively form a notch around
the Doppler frequency of the true target. The width
and depth of this notch can be adjusted adaptively
according to the requirement of target detection. The
Doppler spectrum of the true and false targets formed
by the pulses with adaptive initial phases is sketeched
in Fig. 2.
According to (2) and (8) the matched filter output
of the pth false target created by the DRFM repeat
jammer can be rewritten as
ypn (n) = ¾pej2¼f
p
fnTr ej(Án¡i¡Án) (16)
and the Doppler spectrum of the pth false target can
be expressed as
Pp(k) =
ÃN¡1Xn=0
ypn (n)e¡j(2¼=N)kn
!¢ÃN¡1Xm=0
ypn (n)e¡j(2¼=N)km
!¤
=
N¡1Xn=0
N¡1Xm=0
gng¤m exp[j(2¼=N)(Nf
p
f Tr ¡ k)(m¡ n)]
(17)
where gn = ej(Án¡i¡Án) = ejÃn .
A penalty function is constructed to penalize
the Doppler spectrum Pp(k) of the pth false target
in certain Doppler frequency band where the true
target exists. Assuming the true target’s Doppler
frequency is between f1 and f2 and the weight of the
pth false target is ´p > 0, the penalty function can be
written as
J(g) =
PXp=1
´p
0@ k2Xk=k1
Pp(k)
1A= gHRPg (18)
where
g= [g1,g2, : : : ,gN]T = (ejÃ1 ,ejÃ2 , : : : ,ejÃN )T
is the residual phase vector, (¢)T and (¢)H denotethe transposition and conjugate transposition, k1 =
round(Nf1Tr)¡ 1, k2 = round(Nf2Tr)+1, the round(¢)operation returns a number rounded to a specified
number of decimal places, and
RPmn =
PXp=1
´p
0@ k2Xk=k1
exp[j2¼((m¡ n)=N)(Nfpf Tr¡ k)]1A(19)
is Hermitian Toeplitz [16]. The weight ´p may be
adjusted arbitrarily in accordance with the power of
the false target.
The minimization problem of the Doppler
spectrum of the false targets in certain frequency band
can be summarized as follows:
min J(g) = gHRPg
s.t. gn = ejÃn :
(20)
This optimization problem of minimizing J as a
function of residual phase vector g can be solved bythe new phase-only version of the conjugate gradient
method with quadratic convergence [13]. The gradient
of J is given by the expression
rJ = 2Im[diag(RPggH)] (21)
where diag(¢) denotes diagonal entries of the squarematrix.
Therefore, with the phase-only conjugate gradient
method proposed in [13] and the phase-only gradient
in (21), the optimized value g̃ of the penalty function
can be first derived. Then we can revert to calculate
the adaptive initial phases of pulses fÁngNn=1. Basedon the above description, the adaptive initial phases
design method can be summarized as follows.
Step 1: According to the Doppler frequency
of the true target fd, the false targets fpf , set the
frequency border f1 and f2.
Step 2: Set the weight ´p and calculate the
matrix RP .
Step 3: The phase-only conjugate gradient
algorithm is applied to optimize the problem in (20)
and g̃ is obtained.Step 4: Setting random phases Á¡i,Á¡i+1, : : : ,Á¡1,
the adaptive initial phase fÁng can be calculated byÁn = Án¡i+argfg̃ng:
By using pulses with adaptive initial phases, the
operation of matched filtering and Doppler processing
1294 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 49, NO. 2 APRIL 2013
makes the radar system capable of extracting the
true target while suppressing the false targets in the
Doppler frequency band of the true one.
B. Information Extraction by Multi-Channel Processing
Note that the adaptive initial phases design method
requires that the MTI/PD radar system know how a
DRFM repeat jammer is operating. Moreover, the
radar should have the information on ranges and
velocities of the false targets. In this section, we
discuss the multi-channel processing approach for
prior information extraction. The corresponding
scheme of the multi-channel processing is plotted in
Fig. 3.
We assume that a radar system transmits
pulses with random initial phases and employs the
multi-channel matched filtering for preprocessing.
When the ratio of the power of the jammer signal and
that of the target signal is less than 10log10N dB, the
true and false targets will be detected at the same time
in the two matched filter channels with regard to the
current and previous i PRI. However, if the power
of the jammer signal is very high, the true target
will be buried in the Doppler spectrum of the false
targets. Therefore, the radar system cannot determine
if a true target exists at that moment. Fortunately,
in the matched filter channel of the previous i PRI,
the false targets will not be buried in the Doppler
spectrum of the true one, and the manner by which
a DRFM repeat jammer operates and the parameters
of the false targets can be easily estimated. With these
prior information, the radar system can determine the
frequency border f1 and f2, and the lag PRI i.
In the above section the multi-channel processing
approach for pulses with random initial phases can be
used for prior information extraction. In this section
the multi-channel processing scheme for pulses with
adaptive initial phases is further discussed and the
information extraction capability of the true and false
targets by this method is also evaluated.
The matched filter output of the n¡ l PRI can bewritten as
yl(n) = ¾Tej(Án¡Án¡l)ej2¼fdnTr +
PXp=1
¾pej(Án¡i¡Án¡l)ej2¼f
p
fnTr
(22)
where l = 0,1, : : : ,M, M denotes the maximum
possible lag PRI number. The corresponding Doppler
spectrum can be expressed as
Pl(k) =
N¡1Xn=0
N¡1Xm=0
E
8<:¾Tej(Án¡Án¡l)e¡j(Ám¡Ám¡l) exp[j(2¼=N)(NfdTr¡ k)(n¡m)]+
PXp=1
¾pej(Án¡i¡Án¡l)e¡j(Ám¡i¡Ám¡l) exp[j(2¼=N)(Nfpf Tr¡ k)(n¡m)]
9=; (23)
Fig. 3. Multi-channel processing scheme for pulses with adaptive
initial phases.
where
Pl(k)jl 6=i =N¾T+PXp=1
¾p (24)
means that the Doppler energy of the true and false
targets is evenly spread over all Doppler cells and no
target will be detected in the l 6= ith channel, andPl(k)jl=i = PiT(k) +PiF(k)
=
N¡1Xn=0
N¡1Xm=0
¾Tej(Án¡Án¡i) exp[¡j(Ám¡Ám¡i)]
¢ exp[j(2¼=N)(NfdTr¡ k)(n¡m)]
+N2
0@ PXp=1
¾p±(k¡Nfpf Tr)1A (25)
indicates that the false target energy will be
accumulated by Doppler processing while the energy
of the true target will be disturbed by the phase
ej(Án¡Án¡i) in the ith channel.A Doppler energy function ET is constructed to
evaluate the Doppler spectrum of the true target in
the Doppler frequency band where the false targets
exist. For the true target, a set of P Doppler frequency
bands where the false targets exist are identified,
where the pth band is between kp1 and kp2, with a
correspondence weight ´p, and the Doppler energy
ET generated by the true target in P bands is
ET =
PXp=1
´p
0@ kp2Xk=kp1
PiT(k)
1A : (26)
It is assumed that NfdTr¡ kp1 =Nfpf ¡ k1 and NfdTr¡kp2 =Nf
pf ¡ k2; the above equation can be written as
ET = J(g): (27)
ZHANG, ET AL.: NEW ANTIVELOCITY DECEPTION JAMMING TECHNIQUE 1295
The above equation shows that by using the pulses
with adaptive initial phases, matched filtering
operation with regard to the previous PRI makes
it capable of extracting the false targets while
suppressing the true target in the Doppler spectrum.
This enables the radar to measure the parameters
of the true and false targets simultaneously, and to
monitor the DRFM repeat jammer operation at the
true time.
C. Antivelocity Deception Jamming Working Flow
As aforementioned, three types of velocity
deception jamming usually appear in modern jamming
techniques.
Velocity gate pull off (VGPO): The VGPO
jammer initially creates a false target with the same
Doppler frequency as the true target. The magnitude
of the false target is much higher than that of the
true target and the radar receiver’s automatic gain
control is forced to reduce the gain. The jammer then
increases the Doppler frequency, making the target
seem even farther out of the Doppler cell where the
true target exists. Finally the jammer switches off,
leaving the radar with a broken lock.
Multiple false targets (MFT): Multiple targets
are created simultaneously with different Doppler
frequencies.
Two blinking false targets (TBFT): In blinking
jamming mode, two jammers of the same operation
mode are switched on and off according to a
prescribed patten or almost randomly.
In Fig. 4 the working flow of the MTI/PD radar
for countering velocity deception jamming is given.
When the MTI/PD radar begins to work, it continues
to transmit pulses with random initial phases and
to utilize the multi-channel matched filters for
processing. According to the multi-channel processing
result with regard to the current and previous PRI, the
radar system will judge if the DRFM repeat jammer
exists. If the target is only detected in the matched
filter channel with regard to the current PRI, and no
false targets are detected, then the existence of the
true target can be confirmed. If the false targets are
detected in the channels matched with the previous
PRI, the radar system has to design adaptive initial
phases by using the information of the false targets,
and transmits pulses with them. The multi-channel
processing will also be applied for the true and false
targets information extraction simultaneously. With
this information the jamming mode can be identified
and the radar system will choose the corresponding
antijamming means.
For VPGO if a false target begins to move out
of the Doppler cell of the true target in the matched
filter channel with regard to the previous PRI, the
false target could be identified and recorded by the
radar system. In the following process constant pulses
are transmitted until the jammer shuts down, and the
Fig. 4. Working flow of MTI/PD radar for countering velocity
deception jamming.
pulses with adaptive initial phases are retransmittedfor the next VGPO process.For MFT and TBFT the false targets are very close
to the true one in the Doppler spectrum. Therefore,the pulses with adaptive initial phases which form anotch around the true target in the Doppler spectrumare continuously transmitted and the detection of thetrue target will not be affected.
V. SIMULATION RESULTS
The radar system simulated in this work is basedon modern MTI/PD radars. The carrier frequency is3.3 GHz, the pulse repetition frequency is 40 kHz,and the number of pulses in a CPI is 512. A targetwith specified RCS exists at a given range withthe Doppler frequency of 5 kHz. A false target iscreated by the DRFM repeat-back jammer at thesame range with the Doppler frequency of 5.2 kHzto pull the velocity gate off. The Doppler spectrumof pulses with fixed initial phases in the case ofjammer-to-signal ratio (JSR) of 7 dB and 14 dB areshown in Figs. 5(a) and (b). Both the true and thefalse targets appear in the Doppler spectrum and theycannot be distinguished. The Doppler spectrum iscalculated using fast Fourier transformation (FFT)with length 8192, and SNR is 20 dB.The Doppler spectrum of the pulses with random
initial phases in the case of JSR 7 dB and 14 dBis shown in Figs. 5(c) and (d). Because the powerof jamming signal in the Doppler spectrum canbe suppressed 10log10 512 = 27 dB in theory, thetrue target can be detected in the case of JSR 7 dB,
1296 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 49, NO. 2 APRIL 2013
Fig. 5. Doppler spectrum of pulses with fixed and adaptive initial
phases. (a) Doppler spectrum in case of JSR = 7 dB. (b) Doppler
spectrum in case of JSR = 14 dB. (c) Doppler spectrum of pulses
with random initial phases in case of JSR = 7 dB. (d) Doppler
spectrum of pulses with random initial phases in case of
JSR = 14 dB.
however in the case of JSR 14 dB, the energy ofjamming signal is spread over all Doppler cells andthe true target is buried.The adaptive initial phases are calculated using the
phase-only conjugate gradient method and diagonalloading equal to 10¡5 of the identical diagonal matrixelement of RP . The Doppler spectrum in the caseof JSR 14 dB, 20 dB, and 40 dB is shown by usingthe pulses with adaptive initial phases in Fig. 6(c). Anotch from 3 kHz to 7 kHz in the Doppler spectrum isformed and the true target which does not get buriedcan be detected. The average spectrum outside thenotch is 29.0 dB, 34.5 dB and 46.9 dB higher thanthat in the notch, respectively, in three cases. Theconvergence of the phase-only conjugate gradientmethod used to calculate the adaptive initial phases isplotted in Fig. 6(d). The quadratic convergence of thepenalty function versus the iteration number is evidentin this figure. After 40 iterations, approximatelyoptimal phases are derived.The other methods for countering repeat radar
jammers are applied here to compare with ourmethod in this paper. In Soumekh’s work in [6],a manipulation of the transmitted signal or itsparameters in the fast-time domain was given. Themodulation and/or the rate of the linear frequencymodulated (LFM) signal are varied from one PRIto another. Both the multi-tone phase-modulated(MT-PH) LFM signal and slope-varying (SV) LFMsignal are considered and comparative results ofthese signals are given in Figs. 6(a) and (b). In thecase of JSR 14 dB and 20 dB, the true target can bedetected in the Doppler spectrum. However, the truetarget will be buried in the Doppler spectrum withincreased JSR. Therefore, our conclusion here is that
Fig. 6. Doppler processing output using (a) MT-PM LFM signal,
(b) SV LFM signal, (c) pulses with adaptive initial phases in case
of JSR 14 dB, 20 dB, and 40 dB, (d) convergence of phase-only
conjugate gradient algorithm.
Fig. 7. Multi-channel processing output using pulse with adaptive
initial phases. (a) Channel 1. (b) Channel 2. (c) Channel 3.
(d) Channel 4.
our method circumvents the pulse diversity method inhigh jamming power environment.To measure ranges and velocities of the true
and false targets simultaneously, the multi-channelprocessing approach which utilizes the received signalto correlate with the transmitted pulses of the currentand previous PRIs is used. At the current PRI, theDRFM repeat-back jammer transmits the pulse thatthe radar used 2 PRI before. The Doppler spectrumof the four processing channels is shown in Fig. 7. Inchannel 1 the Doppler spectrum of the true target canbe detected with a surrounding notch. In channel 3 theDoppler spectrum of the jammer signal can also bedetected with a surrounding notch. In other channels,no obvious notches appear and no target can bedetected.
ZHANG, ET AL.: NEW ANTIVELOCITY DECEPTION JAMMING TECHNIQUE 1297
Fig. 8. (a) Doppler spectrum of true and multiple false targets.
(b) Doppler spectrum of MT-PM LFM Signal and SV LFM
signal. (c) Doppler spectrum of true and multiple false targets
using pulses with adaptive initial phases. (d) Detection probability
of true target versus SJR.
Considering the jamming mode of multiple false
targets, 6 false targets appear in 4» 6 kHz in theDoppler spectrum in Fig. 8(a). We design the initial
phases which form a notch in 4» 6 kHz and transmitthe pulses with them. The Doppler spectrum of the
true and multiple false targets using pulses with
adaptive initial phases, MT-PM LFM signal and SV
LFM signal is shown in Figs. 8(b) and (c). A notch
is formed and the true target can be detected by
using pulses with adaptive initial phases. However,
the true target is buried in the Doppler spectrum by
using MT-PM LFM signal and SV LFM signal. The
detection probability of the true target versus SJR
is plotted in Fig. 8(d). Obviously the performance
for pulses with adaptive initial phases is stable with
increased SJR.
VI. CONCLUSIONS
An effective scheme has been presented for MTI
and PD radars for countering DRFM repeat-type
jammers. Although the phase-only conjugate gradient
method has been presented by Smith in [13], this
scheme has novel features. The most obvious
advantage is that the MTI/PD radar can sense the
jammer signal and then change its transmitted pulses
with adaptive initial phases to adapt to the varying
interference. The jammer signal and target signal
can be focused, respectively, in different correlation
channels. According to the ranges and velocities
of the false targets and the manner the DRFM
repeat-back jammer operates, the receiver calculates
the adaptive initial phases and then sends it to the
transmitter. In [16] it is referred to as “cognitive
radar,” originally proposed by Professor Haykin [17].
Another advantage is that original range-Dopplerprocessing for PD radar and MTI filtering for MTIradar can be reserved. Multiple processing channelsare used to replace the original single channel.Although this will increase the system complexity forradar, with the high speed digital signal processorsnow used, the receiver can succeed in finishingthe work easily. The receiver is also required tocalculate the initial phases. It has been verifiedthat computational and memory requirements areboth linear in the number of the pulses, and theconvergence of the phase-only conjugate gradientmethod is quadratic. As a result the initial phases arerapidly calculated, permitting the real-time adaptationto dynamic jamming environment.
APPENDIX
If f'n,n= 1,2, : : : ,Ng is the random variableuniformly distributed in [¡¼,¼] and statisticallyindependent, the random variable ½n = e
j'n will alsobe statistically independent and satisfy
E[½n] =1
2¼
Z ¼
¡¼ej½nd½n = 0
E[j½nj2] =1
2¼
Z ¼
¡¼ej½ne¡j½nd½n = 1
E[½n½¤m] =
½E[½n]E[½
¤m] = 0, n 6=m
E[j½nj2] = 1, n=m
= ±(n¡m):Therefore the variable gn = e
j½n¡m¡½n = ½n¡M½¤n has the
following properties:
E[gn] = E[½n¡M]E[½¤n] = 0
E[jgnj2] = E[j½n¡M j2]E[j½¤nj2] = 1E[gng
¤l ] = E[½n¡M½
¤n½l¡M½
¤l ] = ±(n¡ l):
REFERENCES
[1] Meikle, H.
Modern Radar Systems.
Norwood, MA: Artech House, 2001.
[2] Li, N. and Zhang, Y.
A survey of radar ECM and ECCM.
IEEE Transactions on Aerospace and Electronic Systems,
31, 3 (July 1995), 1110—1120.
[3] Morris, G. V. and Kastle, T. A.
Trends in electronic counter-countermeasures.
In National Telesystems Conference Proceedings (NTC’91),
Atlanta, GA, Mar. 1991, pp. 265—269.
[4] Roome, S.
Digital radio frequency memory.
IEE Electronics and Communications Engineering Journal,
(Aug. 1990), 147—153.
[5] Szymanski, M. B., Lothes, R. N., and Wiley, R. G.
Radar Vulnerability to Jamming.
Upper Saddle River, NJ: Prentice-Hall, 1996.
[6] Soumekh, M.
SAR-ECCM using phase-perturbed LFM chirp signals
and DRFM repeat jammer penalization.
IEEE Transactions on Aerospace and Electronic Systems,
42, 1 (Jan. 2006), 191—205.
1298 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 49, NO. 2 APRIL 2013
[7] Akhtar, J.
An ECCM scheme for orthogonal independent
range-focusing of real and false targets.
In Proceedings of the IEEE Radar Conference (ICR’2007),
Apr. 2007, pp. 846—849.
[8] Lin, K.
Anti-jamming MTI radar using variable pulse-codes.
M.S. thesis, Dept. of Electrical Engineering and
Computer Science, Massachusetts Institute of Technology,
2002.
[9] Garmatyuk, D. and Narayanan, R.
ECCM capabilities of an ultrawideband bandlimited
random noise imaging radar.
IEEE Transactions on Aerospace and Electronic Systems,
38, 4 (Oct. 2002), 1243—1255.
[10] Guosui, L., et al.
A contrast experiment of ECCM of random to
pseudo-random binary phase coded CW radar.
In Proceedings of the 3rd International Conference on
Signal Processing, (ICSP’96), vol. 2, Beijing, China, Oct.
1996, pp. 1621—1624.
[11] Jinshan, W. and Guosui, L.
Pulse random signal radar.
In Proceedings of the CIE International Conference on
Radar, (ICR’96), Beijing, China, Oct. 1996, pp. 523—526.
[12] Guosui, L., et al.
The present and the future of random signal radars.
IEEE Aerospace and Electronic Systems Magazine, 12, 10
(Oct. 1997), 35—40.
[13] Smith, S. T.
Optimum phase-only adaptive nulling.
IEEE Transactions on Signal Processing, 47, 7 (July
1999), 1835—1843.
[14] Yang, Y., Zhang, W. M., and Yang, J. H.
Study on frequency-shifting jamming to linear frequency
modulation pulse compression radars.
In Proceedings of the International Conference on Wireless
Communications and Signal Processing (WCSP’09),
Nanjing, China, Nov. 2009, pp. 132—136.
[15] Dong, W. F., et al.
The deceptive effect of blinking decoys on ARMs.
In ICAIC’11 Proceedings, vol. 2, 2011, 338—347.
[16] Lindenfeld, M. J.
Sparse frequency transmit and receive waveform design.
IEEE Transactions on Aerospace and Electronic Systems,
40, 3 (July 2004), 851—859.
[17] Haykin, S.
Cognitive radar: A way of the future.
IEEE Signal Processing Magazine, 23, 1 (Jan. 2006),
30—40.
ZHANG, ET AL.: NEW ANTIVELOCITY DECEPTION JAMMING TECHNIQUE 1299
Jindong Zhang was born in Nantong, China, in 1981. He received his B.S,M.S., and Ph.D. degrees from the Nanjing University of Science and Technology
(NJUST), Nanjing, China, in 2004, 2006, and 2010, respectively, all in electronic
engineering.
In 2010, he joined the College of Electronic and Information Engineering,
Nanjing University of Aeronautics and Astronautics (NUAA), where he is now an
assistant professor. His current research interests include radar signal a processing
and adaptive radar waveform design.
Daiyin Zhu (M’04) was born in Wuxi, China, in 1974. He received his B.S.degree in electronic engineering from Southeast University, Nanjing, China, in
1996 and his M.S. and Ph.D. degrees in electronic engineering from Nanjing
University of Aeronautics and Astronautics (NUAA), Nanjing, in 1998 and 2002,
respectively.
From 1998 to 1999, he was a guest scientist with the Institute of Radio
Frequency Technology, German Aerospace Center, Oberphaffenhofen, where he
worked in the field of SAR interferometry. In 1998, he joined the Department
of Electronic Engineering, NUAA, where he is currently a professor. He has
developed algorithms for several operational airborne SAR systems. His current
research interests include radar imaging algorithms, SAR/ISAR autofocus
techniques, SAR ground moving target indication (SAR/GMTI), and SAR
interferometry.
Gong Zhang (M’07) received his Ph.D. degree in electronic engineering from the
Nanjing University of Aeronautics and Astronautics (NUAA), Nanjing, China, in
2002.
From 1990 to 1998, he was a member of technical staff at No724 Institute
China Shipbuilding Industry Corporation (CSIC), Nanjing, China. Since 1998, he
has been with the College of Electronic and Information Engineering at NUAA,
where he is currently a professor. His research interests include radar signal
processing and classification recognition.
Dr. Zhang is a member of Committee of Electromagnetic Information,
Chinese Society of Astronautics (CEI-CSA) and a Senior Member of the Chinese
Institute of Electronics (CIE).
1300 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 49, NO. 2 APRIL 2013