Professional Documents
Culture Documents
Sensors 19 01661
Sensors 19 01661
Sensors 19 01661
Article
An Anti-Jamming Null-Steering Control Technique
Based on Double Projection in Dynamic Scenes for
GNSS Receivers
Hao Wang 1 , Qing Chang 1, * and Yong Xu 2
1 School of Electronic and Information Engineering, Beihang University, No. 37 Xueyuan Road,
Haidian District, Beijing 100191, China; buaahao@126.com
2 Institute of Unmanned Systems Research, Beihang University, No. 37 Xueyuan Road, Haidian District,
Beijing 100191, China; xuyong1518@163.com
* Correspondence: changqing@buaa.edu.cn; Tel.: +86-135-2053-8417
Received: 16 March 2019; Accepted: 4 April 2019; Published: 7 April 2019
Abstract: When a global navigation satellite system (GNSS) receiver suppresses interference in
a dynamic scene, the direction of the interference signal arriving at the receiver may change rapidly.
The null formed by the spatial filtering based on the array antenna will become wider and shallower,
and the anti-jamming performance will deteriorate. A null-steering control technique based on a
dual projection algorithm is proposed in this paper, which can effectively increase the depth of the
null. In this paper, the dynamic model between the interference and the receiver is established first.
Based on the model, the rate of change of the arrival direction of the interference is analyzed, and the
phenomenon of the spatial filtering null becoming wider and shallower is simulated and verified.
Then, the double projection algorithm is introduced to effectively deepen the null. The simulation
results show that the proposed method can effectively increase the null depth by 30 to 50 dB,
which significantly improves the anti-jamming performance of the spatial filtering in dynamic scenes.
1. Introduction
In complex electromagnetic environments or in dealing with battlefield electronic
countermeasures, a satellite navigation receiver is confronted with interference intrusions [1,2].
The spatial domain filtering technique based on the array antenna can form nulls in the direction
of interference to suppress interference, which is a mainstream anti-jamming technique [3–6]. In
practice, there is often a relative motion between the interference and the receiver. The incident
direction of the interference signal arriving at the receiver is constantly changing. When the relative
motion causes the interference direction to change rapidly, such as in a missile flight scene or when
there is a dynamic interference source, the null formed by the spatial filtering algorithm will not
be able to eliminate the interference normally. This is because the arrival angle of the interference
signal changes rapidly, and the signal sampled each time is also a dynamically changing signal;
therefore, calculating the anti-jamming vector based on the sample covariance matrix derived from
the sampled signal also forms a widened null. The depth of the null will also decrease, and the
interference cannot be effectively eliminated, resulting in a sharp deterioration of the anti-interference
performance. A method of interference-null control based on double projection is proposed in this
paper, which can significantly increase the depth of the null, and ensure the anti-jamming performance
of the spatial filtering in dynamic scenes. Regarding the null control technique, a curved array
split into two subarrays was proposed in Ref. [7] to improve the pattern null depth and width for
controlled reception pattern antenna systems. A method of projecting the autocorrelation matrix into
the interference space was proposed in Ref. [8] to improve the anti-interference performance in the
presence of coherent interference. In Ref. [9], a broadened model based on maximizing output power
and constraining interference was proposed to improve the signal-to-interference and -noise ratio
of the receiver. In Ref. [10], a novel statistical space–time null widening method was proposed to
suppress interference in dynamic conditions, and Refs. [11,12] proposed the implementation of an
adaptive null-steering beamformer for flexible broad null control. However, the simulation of dynamic
scenes is lacking in these papers. In the actual dynamic scene, the generated null will naturally widen.
The problem we face is actually how to deepen the null.
In this paper, we first establish a dynamic model and analyze which parameters are related to
the rate of change of the interference direction. Then, the phenomenon that the null of the spatial
filtering becomes wider and shallower in dynamic scenes is verified. Finally, the null control algorithm
based on a double projection is introduced to improve the anti-interference performance of the spatial
filtering in dynamic scenes by deepening the depth of the null. The simulation results show that the
proposed method has better ability to control for interference nulls. The first projection can increase
the depth of the null by at least 20–30 dB. After double projection, it can continue to increase by more
than 20 dB.
The outline for the remainder of this paper is as follows. Section 2 presents the dynamic model
and signal model. The analysis of the spatial filtering performance in static and dynamic scenes is
presented in Section 3, and the null control algorithm based on double projection is presented in
Section 4. Simulation results and conclusions are presented in Sections 5 and 6, respectively.
v · t · sin(π − ϕ0 )
θ (t) = arctan (1)
d0 + v · t · cos(π − ϕ0 )
where represents the relative motion velocity, and represents time, and we take the derivative of θ (t)
with respect to t and get
d0 · v · sin ϕ0
θ 0 (t) = (2)
(d0 − v · t · cos ϕ0 )2 + (v · t · sin ϕ0 )2
When the initial parameters of the motion are determined, d0 and v must be positive,
◦ ◦
and 0 < ϕ0 < 180 , so θ 0 (t) > 0. Therefore, θ (t) will gradually increase with time t, however,
it can be obtained from Equation (2) that as time t increases, θ 0 (t) will gradually decrease, so the rate of
◦ ◦
change of θ (t) will become increasingly small. In special cases, when ϕ0 = 0 or ϕ0 = 180 , θ 0 (t) = 0,
and at this time, θ (t) ≡ 0.
Regarding the relationship between the rate of change of the interference direction and the relative
motion speed v between the interference and the receiver, the initial distance d0 and the relative motion
direction ϕ0 , we separately take the partial derivatives of θ 0 (t) with respect to v, d0 , and ϕ0 .
First, we take the partial derivative of θ 0 (t) with respect to v, and it can be obtained from Equation (2),
ϕ0
θ d0
Figure 1. Schematic
Figure 1. Schematic diagram
diagram of
of interference
interference and
and receiver
receiver dynamic
dynamic motion
motion model.
model.
◦ ◦ ∂θ 0 (t)
We analyze Equation (3) and find that, since 0 < ϕ0 < 180 , the positive and negative of ∂v
When the initial parameters of the motion are determined, d0 and v must be positive, and
∂θ 0 (t)
depends on (d0 − vt). Therefore, when vt < d0 , ∂v > 0, if v increases, the rate of change of the
0° < φ0 < 180° , so θ′(t ) > 0 . Therefore, θ(t) will gradually ∂θ 0 (t)
increase with time t , however, it can be
interference direction will also increase; when vt > d0 , ∂v < 0, if v increases, the rate of change of
obtained from Equation (2) that as time t increases, θ′(t ) will gradually decrease, so the rate of
the interference direction will decrease. We verified the above conclusions through the simulation
change
Sensors 2019, θ(tx) FOR
of scenes.
19, will REVIEWincreasingly small. In special cases, when φ0 = 0° or φ0 = 180
become
PEER 4 of° 19
,
of specific Figure 2a shows the case of vt < d0 , as v increased, the slope of the curve also
θ′(t ) = 0 , and
increased. at this
Figure time, θthe
2b shows ≡ 0 . of vt > d0 , as v increased, the slope of the curve decreased.
(t ) case
Regarding the relationship between the rate of change of the interference direction and the
relative motion speed v between the interference and the receiver, the initial distance d0 and the
relative motion direction φ0 , we separately take the partial derivatives of θ′(t ) with respect to v ,
d0 , and φ0 .
First, we take the partial derivative of θ′(t ) with respect to v , and it can be obtained from
Equation (2),
∂θ′(t ) d0 sin φ0 ( d0 − vt )( d0 + vt )
= 2 (3)
∂v ( d0 + v 2 t 2 − 2d0 vt cos φ0 )2
∂θ′(t )
We analyze Equation (3) and find that, since 0° < φ0 < 180° , the positive and negative of
∂v
∂θ′(t )
depends on ( d0 − vt ) . Therefore, when vt < d0 , > 0 , if v increases, the rate of change of the
∂v
(a) ∂θ′(t ) (b)
interference direction will also increase; when vt > d0 , < 0 , if v increases, the rate of change
Figure2. 2.
Figure TheThe change
change of interference
of the the interference direction
direction with thewith ∂vthe relative
relative motion v.velocity
motion velocity v . (a) The
(a) The change of
of the interference
change
the of the direction
interference direction will
interference
in thedecrease.
direction
case where We
in the verified
vt case
< d0 ;wherethevt
(b) the d0 ;of
above
change conclusions
(b)
thethe through
change the the
ofdirection
interference simulation
interference
in the
of specific
case scenes.
where
direction in >Figure
vt thedcase
0. 2a shows the case of vt < d0 , as v increased, the slope of the curve also
where vt d0 .
increased. Figure 2b shows the case of vt > d0 0 , as v increased, the slope of the curve decreased.
Then, we take the partial derivative of θ (t) with respect to d0 and get
Then, we take the partial derivative of θ(t) with respect to d0 and get
∂θ 0 (t) v sin ϕ0 (vt − d0 )(vt + d0 )
θ =
(t ) v sin 2φ02( vt − d0 )( vt + d0 ) 2 (4)
∂d0 (=d0 2 + v t2 2− 2d0 vt cos ϕ0 )2 (4)
d0 (d0 + v t − 2d0 vt cos φ0 )
2
∂θ 0 (t)
Similarly,
Similarly, thethe
positive and negative
positive of Equation
and negative (4) depend
of Equation (4) on (vt − don
depend 0 ). When
(vt − d0d)0 .<When
vt, ∂vd > 0, ,
vt
∂θ 0 (t)0
if dθ
0 increases,
(t ) the rate of change of the interference direction will increase; when d0 > vt, ∂v < 0,
0 , if the
if d0 increases, increases,
d0 rate the of
of change rate
theofinterference
change of direction
the interference direction
will decrease. will3a,b
Figure increase;
showswhen the
v
θ(t )
d0 vt , 0 , if d0 increases, the rate of change of the interference direction will decrease.
v
Figure 3a,b shows the scene simulation results for d0 vt and d0 vt , respectively. We compared
the slopes of the curves and the simulation results were consistent with the above analysis.
Similarly, the positive and negative of Equation (4) depend on (vt − d0 ) . When d0 vt ,
θ(t )
0 , if d0 increases, the rate of change of the interference direction will increase; when
v
θ(t )
Sensors vt , 19, 1661 0 , if d0 increases, the rate of change of the interference direction will decrease.
d0 2019, 4 of 18
v
Figure 3a,b shows the scene simulation results for d0 vt and d0 vt , respectively. We compared
scene simulation results for d0 < vt and d0 > vt, respectively. We compared the slopes of the curves
the slopes of the curves and the simulation results were consistent with the above analysis.
and the simulation results were consistent with the above analysis.
(a) (b)
Figure
Figure 3. 3.The
Thechange
changeofofthe
theinterference
interferencedirection
directionwithwiththe
theinitial distanced0d. 0(a)
initialdistance . (a)
TheThe change
change ofof the
the
interference
interference direction in in
direction thethe
case where
case d0 <dvt; (b)
where vt ;the
(b)change of theofinterference
the change direction
the interference in the in
direction case
the
0
where d0 > vt.
case where d0 vt .
Finally, we take the partial derivative of θ 0 (t) with respect to ϕ0 and get
The positive and negative of Equation (5) depends on (d0 2 + v2 t2 ) cos ϕ0 − 2d0 vt, and we define
2d0 vt
ϕ0 = arccos( ) (8)
d0 2 + v2 t2
0 vt ∂θ 0 (t)
Therefore, when ϕ0 < arccos( d 2d
2 + v2 t2 ), ∂v > 0, if ϕ0 increases, the rate of change of the
0
0 vt ∂θ 0(t)
interference direction will increase; and when ϕ0 > arccos( d 2d
2 2 2 ), ∂v < 0, if ϕ0 increases, the rate
0 +v t
of change of the interference direction will decrease. The scene simulation results are shown in
Figure 4a,b.
Through the above analysis, we obtained the relationship between the rate of change of the
interference direction and the speed of the relative motion, the direction of the relative motion, and the
initial distance between the interference and receiver. Next, we simulated the corresponding scene to
verify the phenomenon that the nulling of the spatial filtering becomes wider and shallower under
dynamic scenes. First, we needed to introduce the basic principle of conventional spatial filtering.
φ0
x = [ x1 (n), x2 (n),..., xM (n)]H (9)
2d0 vt
φ0 = arccos( 2 ) (8)
Assuming that the anti-jamming weight vector is + v 2t 2
d0 calculated every L sampling moments, the
corresponding sampling signal is
Sensors 2019, 19, 1661 2d vt θ(t ) 5 of 18
Therefore, when φ0 arccos( 2 0 X2 2=)[,x , x ,...,x0 ], if φ0 increases, the rate of change of the
d0 + v t 1 2v L (10)
and the
As weight vector
showndirection
in corresponding
Figure antennatoand
5, anincrease; M when
array arraycomposed
was elements 2array
canMbe
of θ(as
d0expressed
vt elements. t ) The signal received
interference will φ arccos( ), 0 , if φ0 increases,
at the unit sampling time can be expressed as
0
d0 + v t
2 2 2
v
w = [w , w2 ,..., wM ] H
(11)
the rate of change of the interference direction1 will decrease. The scene simulation results are shown
H
in Figure x =signal
4a,b. that the power of the
We require [ x1 (nbe
), xminimal
2 ( n ), . . . ,and
x M (not
n)] equal to 0 [13], i.e., (9)
min w H Rx w
w
(12)
s.t. w H s = 1
Rx −1s
w= (13)
s H Rx −1s
It should be noted that regarding the autocorrelation matrix Rx , in practice, we use the sample
autocorrelation matrix and denote it with Rˆ x , and the anti-jamming weight vector w is then
statistically optimal [16]. Therefore
Rˆ x −1s
w= (b) (14)
(a) s Rˆ −1s
H
x
Figure
Figure
The 4. 4.
final TheThe change
change
output of
of the
of the the interference
interference
spatial isdirection
direction
filtering with with the relative
the relative motionmotion direction
direction φ0 .change
ϕ0 . (a) The (a) The
0 vt
of the interference direction in the case where ϕ0 < arccos( d 2d 2 2 2 ) ; (b) the change
2d vt of the interference
0 +v t
change ofin the
the interference
case where ϕdirection in the
2d0case w H X φ0 arccos( 2 0 2 2 ) ; (b) the change of the
vty =where (15)
direction 0 > arccos( d0 2 +v2 t2 ). d0 + v t
2d0 vt
interference direction in the case where φ0 arccos( ).
d0 + v2t 2
2
x1 (n)
Through the above analysis, we obtained the relationship between the rate of change of the
interference direction and the speed of the relativew1 motion, the direction of the relative motion, and
x ( n )
2
the initial distance between the interference and receiver. Next, we simulated
y (t ) the corresponding
w2
xM ( n )
wM
Schematicdiagram
Figure5.5.Schematic
Figure diagramof
ofthe
thespatial
spatialfiltering
filteringstructure
structure based
based on
on the
the antenna
antenna array.
array.
Assuming that the anti-jamming weight vector is calculated every L sampling moments,
the corresponding sampling signal is
X = [x1 , x2 , . . . , x L ] (10)
w = [ w1 , w2 , . . . , w M ] H (11)
Sensors 2019, 19, 1661 6 of 18
We require that the power of the signal be minimal and not equal to 0 [13], i.e.,
min w H Rx w
w (12)
s.t. w H s = 1
where Rx = E[XX H ], s is a constraint vector. From Equation (12), we construct the Lagrangian
function [14,15] and get
R x −1 s
w = H −1 (13)
s Rx s
It should be noted that regarding the autocorrelation matrix Rx , in practice, we use the sample
autocorrelation matrix and denote it with R̂x , and the anti-jamming weight vector w is then statistically
optimal [16]. Therefore
R̂x −1 s
w = H −1 (14)
s R̂x s
The final output of the spatial filtering is
y = wH X (15)
(a) (b)
Sensors 2019, 19, x FOR PEER REVIEW 8 of 19
Figure
Figure
Sensors 6.
The
2019,6.19, The power
power
x FOR spectral
spectral
PEER REVIEW density
density ofof
thethe signal.
signal. (a)(a)
TheThe power
power spectral
spectral density
density ofof
thethe signal
signal before8 of 19
before
interference suppression; (b) the power spectral density of the signal after interference suppression.
interference suppression; (b) the power spectral density of the signal after interference suppression.
Figure 8. Acquisition
Figure 8. Acquisition result.
result.
Figure 8. Acquisition result.
3.2. Dynamic Scene
3.2. Dynamic Scene
We set two dynamic scenes according to the dynamic model introduced above, a lower
dynamic Weand
seta two dynamic
higher scenes
dynamic, according
generated to the dynamic
corresponding satellite model introduced
and interference above,
signals, anda then
lower
dynamic and
suppressed a higher dynamic,
interference by spatialgenerated
filtering. corresponding satellite and interference signals, and then
suppressed interference by spatial filtering.
3.2.1. Lower Dynamic
Sensors 2019, 19, 1661 8 of 18
(a) (b)
(a) (b)
Figure
Figure 10.10.
TheThe power
power spectral
spectral density
density of theofsignal
the signal in thedynamic
in the lower lower dynamic
scene. (a)scene. (a) The
The power power
spectral
Figure
density of10.
spectral the The
density power
signal spectral
ofbefore
the signal density
before
interference of the signal
interference
suppression; in power
the lower
suppression;
(b) the (b) dynamic
the power
spectral scene.
of the(a)
spectral
density The after
density
signal power
of the
spectral density
signal after
interference of the signal
interference
suppression. before interference suppression; (b) the power spectral density of the
suppression.
signal after interference suppression.
Sensors 2019, 19, 1661 9 of 18
Figure 11. The null of the spatial filtering in the lower dynamic scene.
Figure
Figure 11.11.
TheThe null
null ofof the
the spatial
spatial filtering
filtering inin the
the lower
lower dynamic
dynamic scene.
scene.
(a) (b)
(a) (b)
Figure
Figure 14.14.
TheThe power
power spectral
spectral density
density of theofsignal
the signal in the dynamic
in the higher higher dynamic
scene. (a)scene. (a) The
The power power
spectral
Figure
spectral
density 14. The
density
of the power
ofbefore
signal spectral
the signal density of the
before interference
interference signal in
suppression;suppression;the higher
(b) the power(b) dynamic
the power
spectral scene.
spectral
density (a) The
of thedensity power
of the
signal after
spectral density of the
signal aftersuppression.
interference signal before
interference suppression. interference suppression; (b) the power spectral density of the
signal after interference suppression.
Sensors 2019, 19, x FOR PEER REVIEW 12 of 19
Sensors 2019, 19, x FOR PEER REVIEW 12 of 19
Sensors 2019,
Sensors 19,19,
2019, 1661
x FOR PEER REVIEW 1112ofof1819
Figure 15. The null of the spatial filtering in the higher dynamic scene.
Figure 15. The null of the spatial filtering in the higher dynamic scene.
Figure
Figure 15.15.
TheThe null
null ofof the
the spatial
spatial filtering
filtering inin the
the higher
higher dynamic
dynamic scene.
scene.
Figure 16. The null depth corresponding to the static scene, the lower dynamic scene and the higher
dynamic
Figure
Figure scene.
16.16.
TheThe null
null depth
depth corresponding
corresponding to to
thethe static
static scene,
scene, thethe lower
lower dynamic
dynamic scene
scene and
and thethe higher
higher
Figure
dynamic
dynamic 16. The null depth corresponding to the static scene, the lower dynamic scene and the higher
scene.
scene.
dynamic scene.
4. Interference-Nulling
4. Interference-Nulling Control
ControlAlgorithm
Algorithm
For the
For the dynamic
dynamic scene,
scene, aa shallow
shallow nulling
nulling will
will result
result in
in poor
poor spatial
spatial filtering
filtering performance.
performance.
Therefore we propose a dual projection method to deepen the nulling and improve
Therefore we propose a dual projection method to deepen the nulling and improve the performance the performance of
of the spatial filtering in dynamic
the spatial filtering in dynamic scenes. scenes.
First,we
First, wesolve
solvethethe anti-interference
anti-interference vector
vector normally
normally according
according to the conventional
to the conventional spatial
spatial filtering
filtering algorithm, and find the spatial spectral function of the corresponding
algorithm, and find the spatial spectral function of the corresponding anti-jamming vector, anti-jamming vector,
asshown
as shownininFigure
Figure 18,18, d represents
d represents the element
the element spacing,
spacing, M represents
M represents the number
the number of array
of array element,
λ wavelength. Then, we traverse φ ◦
from ◦
λ represents the signal wavelength. Then, we traverse φ from 0 –90 , calculate the valuevalue
element, represents the signal 0°–90°, calculate the of
of the
corresponding
the corresponding F(φ )then
F (φ), and , andfind thefind
then minimum
the minimum (φ).The
value of Fvalue F(φ ) .The corresponding
of corresponding φ is the
φ is the approximate
position of theposition
approximate interference
of thenulling. However,
interference the However,
nulling. nulling was wider
the andwas
nulling shallower,
wider and so we deepened
shallower, so
the nulling based on the double projection method.
we deepened the nulling based on the double projection method.
After
Afterφφis obtained, from
is obtained, Equation
from (17),(17),
Equation we can
we obtain the steering
can obtain vectorvector
the steering of the interference signal,
of the interference
signal, A = [a(φ1 ), a(φ2 ), . . . , a(φN )] (18)
A = [a(φ1 ), a(φ2 ),..., a(φN )] (18)
where N is the number for the interference. Since the generated nulling was not deep enough,
where N toisthe
according thespatial
number for the
filtering interference.
algorithm, Sincetothe
we hoped generated
increase nulling was
the proportion not deep enough,
of interference in the
according
signal. to the we
Therefore, spatial filtering
projected the algorithm,
signal into we hoped to increase
the interference theand
subspace proportion
deepenedofthe
interference
sampling ofin
the signal. Therefore, we projected the signal
the interference signal [19,20]. The projection matrix is into the interference subspace and deepened the
sampling of the interference signal [19,20]. The projection matrix is
−1
Pa = A(A HHA) −1 A H (19)
Pa = A( A A) AH (19)
Combined with Equation (10), we can obtain the deepened sampled signal,
Combined with Equation (10), we can obtain the deepened sampled signal,
D D==XX++ uP
XX uPaa X (20)
(20)
R− 1
xD s
wD = (22)
s H R− 1
xD s
Then, we consider the second projection to further deepen the depth of the nulling. Because the
interference signal subspace is orthogonal to the noise subspace, we can project the anti-jamming
weight vector wD to the noise subspace to further improve the orthogonality between wD and the
interference signal subspace. To do this, we need to obtain the corresponding projection matrix. First,
after the eigen decomposition of RxD , we can get
M
RxD = ∑ λi vi viH = Uj Λ j UjH + Un Λn UnH (23)
i =1
is a diagonal matrix of M − K smaller eigenvalues (The eigenvalues are arranged in descending order.
If the adjacent eigenvalues exceed one order of magnitude, the corresponding K can be used as the
boundary value of eigenvalues). Un = [vK +1 , vK +2 , · · · , v M ] represents the noise subspace formed by
the eigenvectors corresponding to the M − K eigenvalues [21,22]. According to the noise subspace Un ,
the corresponding projection matrix can be obtained,
−1
Pn = Un (Un H Un ) Un H (26)
Then, the anti-jamming weight vector wD is projected onto the noise subspace to obtain the
anti-jamming weight vector after double projection,
wDD = Pn wD (27)
The signal processing flow of the double projection algorithm was as follows, as shown in
Figure 19:
4. Perform eigen decomposition on RxD , construct a noise subspace U n , and calculate the
3.SensorsCalculate
2019, 19, xthe
FORprojection matrix Pa , project the sampled signal into the noise subspace and calculate
PEER REVIEW 15 of 19
corresponding matrix
the autocorrelation projection
RxD ,matrix Pn ;
then calculate the anti-jamming vector wD ;
4. 4. 5. Perform
PerformProject eigen
eigen
the decomposition
decomposition
anti-jamming onon
weight x DR,xDconstruct
Rvector , constructa anoise
wD to the noisesubspace
noise subspaceand
UU ,and
andcalculate
n ,nobtain calculate
wDD . the
the
corresponding projection matrix
corresponding projection matrix P ;
n P ;
n
5. Project the anti-jamming weight vector wD to the noise subspace and obtain wDD .
5. Project the anti-jamming weight vector wD to the noise subspace and obtain wDD .
w DD
x1 (t ) w DD
F (φ )
x2 (t ) w Pa RxD w D
x1 (t ) Un Pn
φ
…
…
F (φ )
x2x(Mt )(t )w Pa RxD w D Un Pn
φ
…
xM (t )
Figure 19. The schematic diagram of the signal processing flow of the double projection algorithm.
5. Simulation andschematic
Figure 19. The Test diagram of the signal processing flow of the double projection algorithm.
Figure 19. The schematic diagram of the signal processing flow of the double projection algorithm.
We simulated
5. Simulation and Testthe dynamic scene and deepened the nulling based on the dual projection
5.method
Simulation and Test
to verify the improvement effect of the proposed method on the spatial filtering
We simulated
performance. the dynamic scene and deepened the nulling based on the dual projection method
We the
to verify simulated the dynamic of scene and deepened on thethenulling based on performance.
the dual projection
The simulated sceneeffect
improvement the proposed
parameters were themethod
same as those spatial
in thefiltering
higher dynamics in the third
method to verify scene
The simulated the improvement
parameters effect
were the of the asproposed method ondynamics
the spatial filtering
section, and the maximum rate of change ofsame those in the
the interference higher
arrival angle was 0.09 in °/ms.
the third
First,
performance.
section, and the maximum rate of change of the interference arrival angle was 0.09 ◦ /ms. First, through
through the first projection, the sampling of the interference signal was deepened, and then the
the Theprojection,
first simulatedthe scene parameters were the same as those in the higher dynamics in the third
anti-jamming weightsampling
vector was of obtained.
the interference
Figuresignal was deepened,
20 shows and then
the corresponding the anti-jamming
nulling obtained by
section,vector
weight and thewas maximum
obtained. rate of20change
Figure shows of the
the interferencenulling
corresponding arrivalobtained
angle was by 0.09 °/ms.
different First,
nulling
different nulling control coefficient u , where u = 0 was the nulling depth corresponding to the
through
control the first
coefficient projection, the sampling
u = 0 was the
u, where algorithm, of
nulling the interference signal was deepened, and then the
original spatial filtering which wasdepth
–62.72corresponding
dB. The depth to the original
of the spatial
nulling afterfiltering
the first
anti-jamming
algorithm, weight vector
was −62.72 dB. was obtained. Figure 20 shows the corresponding nulling obtained by
projectionwhich was significantly The depth
deepened. When of the
u =nulling
10 , theafter the depth
nulling first projection
exceededwas –80 significantly
dB, and when
different nulling
deepened. When ucontrol
= depth coefficient
10, the u , where
nulling depth exceededu = 0−was
80thetheand
dB, nulling depth corresponding
u = 100, to the
u = 100 , the nulling exceeded –100 dB. Therefore, depthwhenof nulling could the
benulling
increased depth
by at
original
exceeded spatial
−100 filtering algorithm, which was –62.72 dB. The depth of the nulling after the first
least 20–30 dBdB. Therefore,
after the depth of
the first projection. nulling
It can could
be seen be increased
from Figure 21 thatby atasleast 20–30 dB after
u increased, the
the depth
projection
first was significantly
projection. It can be seen deepened.
from Figure Whenthat u =as
10u, the nullingthedepth exceeded –80 dB, gradually
and when
of the nulling gradually deepened, but the21magnitude increased,
of the increase depth of thedecreased.
gradually nulling Therefore,
u = 100 , the
deepened, nulling
but the depth exceeded
magnitude of the –100 dB. gradually
increase Therefore, decreased.
the depth ofTherefore,
nulling couldu be increased
could be as large byasat
u could be as large as possible, but the deepening effect on the nulling was gradually stabilized. In
least
possible,20–30 dB
but the after the first
deepeningset projection.
effect It
on50–100. can be seen from Figure 21 that as u increased,
the nulling was gradually stabilized. In this paper, we generally the depth
this paper, we generally u to
of the
set u tonulling
50–100.gradually deepened, but the magnitude of the increase gradually decreased. Therefore,
u could be as large as possible, but the deepening effect on the nulling was gradually stabilized. In
this paper, we generally set u to 50–100.
Figure
Figure 20.20.
TheThe corresponding
corresponding nulling
nulling forfor different
different nulling
nulling control
control coefficients.
coefficients.
Figure 20. The corresponding nulling for different nulling control coefficients.
Sensors
Sensors 2019,
2019,
Sensors 19,19,
2019, xx FOR
1661
19, FOR PEER
PEER REVIEW
REVIEW 16
16ofof
15 of1819
19
Figure
Figure 21.
Figure21. The
21.The corresponding
Thecorresponding nulling
correspondingnulling depths
nullingdepths
depths for
for different
different
for nulling
nulling
different control
control
nulling coefficients.
coefficients.
control coefficients.
WeWe tooku u
took
Wetook u ==50
= 50 and
50and solved
andsolved the
solvedthethe anti-jamming
anti-jamming
anti-jammingweightweight vector
weightvector
vectorfor for spatial
spatial
for spatial filtering.
filtering. The
filtering.The power
Thepower
power
spectral
spectral density
density after
after anti-jamming
anti-jamming is is shown
shown in in Figure
Figure 22b,22b,
and and
the the
power power spectral
spectral
spectral density after anti-jamming is shown in Figure 22b, and the power spectral density obtained density
density obtained
obtained by
by
the the original
byoriginal spatial
spatialspatial
the original filtering
filteringfiltering
is shownis shown
is inshown in
Figurein Figure
22a. 22a. By
By comparison,
Figure comparison, it
it can be seen
22a. By comparison, can be
that the
it can seen that
that the
interference
be seen the
interference
was better was better
suppressed, suppressed,
but the but
suppression the suppression
was still not was still
thorough not
enough.
interference was better suppressed, but the suppression was still not thorough enough. thorough enough.
(a) (b)
(b)
(a)
(c)
(c)
Figure
Figure
Figure 22.
22. The
22.The power
Thepower spectral
powerspectral density
spectraldensity
densityofof
ofthe signal. (a)
thesignal. (a) The
(a) Thesignal
The signalpower
signal powerspectral
power spectraldensity
spectral densityobtained
density obtainedby
obtained by
bythe
theoriginal
the originalspatial
original spatialfiltering;
spatial filtering;(b)
filtering; (b) the
(b) the signal
the signal power
signal power spectral density
spectraldensity obtained
densityobtained
obtainedbyby the
bythe spatial
thespatial filtering
spatialfiltering
filtering
after
after the
afterthe first
thefirst projection;
projection; (c)
firstprojection; (c) the
the signal
the signalpower
signal powerspectral
power spectraldensity
spectral density
density obtained by
obtained
obtained spatial
byby filtering
spatial
spatial after
filtering
filtering after dual
after
dual
projection.
dual projection.
projection.
Figure 23. The depth of the nulling corresponding to single projection and double projection.
Figure 23.23.
Figure The depth
The ofof
depth the nulling
the corresponding
nulling toto
corresponding single projection
single and
projection double
and doubleprojection.
projection.
Acknowledgments: This work was supported by the National Natural Science Foundation of China (No.61471021).
The author would like to thank the editor and anonymous reviewers for their thoughtful comments
and suggestions.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Lang, R.; Su, Z.; Zhou, K.; Mou, S. A Robust Signal Driven Method for GNSS Signals Interference Detection.
Chin. J. Electron. 2018, 27, 422–427. [CrossRef]
2. Amin, M.G.; Borio, D.; Zhang, Y.D.; Galleani, L.J.I.S.P.M. Time-Frequency Analysis for GNSSs: From
interference mitigation to system monitoring. IEEE Signal Process. Mag. 2017, 34, 85–95. [CrossRef]
3. Guo, Y.; Liao, G.; Feng, W. Sparse Representation Based Algorithm for Airborne Radar in Beam-Space
Post-Doppler Reduced-Dimension Space-Time Adaptive Processing. IEEE Access 2017, 5, 5896–5903.
[CrossRef]
4. Daneshmand, S.; Jahromi, A.J.; Broumandan, A.; Lachapelle, G. GNSS space-time interference mitigation
and attitude determination in the presence of interference signals. Sensors 2015, 15, 12180–12204. [CrossRef]
5. Delikaris-Manias, S.; Vilkamo, J.; Pulkki, V. Signal-Dependent Spatial Filtering Based on Weighted-Orthogonal
Beamformers in the Spherical Harmonic Domain. IEEE/ACM Trans. Audio Speech Lang. Process. 2016, 24, 1511–1523.
[CrossRef]
6. Fernandez-Prades, C.; Arribas, J.; Closas, P. Robust GNSS Receivers by Array Signal Processing: Theory and
Implementation. Proc. IEEE 2016, 104, 1207–1220. [CrossRef]
7. Byun, G.; Choo, H.; Kim, S. Improvement of Pattern Null Depth and Width Using a Curved Array With Two
Subarrays for CRPA Systems. IEEE Trans. Antennas Propag. 2015, 63, 2824–2827. [CrossRef]
8. Li, W.X.; Li, Y.P. An effective method on increasing the null depth of beam forming via virtual array
transformation. In Proceedings of the International Conference on Information Science & Engineering,
Hangzhou, China, 4–6 December 2010.
9. Chen, L.W.; Zheng, J. A Broadened and Deepened Anti-Jamming Technology for High-Dynamic GNSS Array
Receivers. IEICE Trans. Commun. 2016, 99, 2055–2061. [CrossRef]
10. Zhang, B.; Tan, Q.; Pan, H.; Ma, H.; Sun, X.-L. Robust anti-jamming method for high dynamic global
positioning system receiver. IET Signal Process. 2016, 10, 342–350. [CrossRef]
11. Leng, S.; Ser, W. Adaptive null steering beamformer implementation for flexible broad null control.
Signal Process. 2011, 91, 1229–1239. [CrossRef]
12. Dan, L.; Wu, R.; Wang, W. Robust widenull anti-jamming algorithm for high dynamic GPS. In Proceedings
of the IEEE International Conference on Signal Processing, Beijing, China, 21–25 October 2012.
13. Li, Q.; Wang, W.; Xu, D.; Wang, X. A Robust Anti-Jamming Navigation Receiver with Antenna Array and
GPS/SINS. IEEE Commun. Lett. 2014, 18, 467–470. [CrossRef]
14. Lu, Z.; Chen, H.; Chen, F.; Nie, J.; Ou, G. Blind adaptive channel mismatch equalisation method for GNSS
antenna arrays. IET Radar Sonar Navig. 2018, 12, 383–389. [CrossRef]
15. Dai, X.; Nie, J.; Chen, F.; Ou, G. Distortionless space-time adaptive processor based on MVDR beamformer
for GNSS receiver. IET Radar Sonar Navig. 2017, 11, 1488–1494. [CrossRef]
16. Igambi, D.; Yang, X.; Jalal, B. Robust Adaptive Beamforming Based on Desired Signal Power Reduction and
Output Power of Spatial Matched Filter. IEEE Access 2018, 6, 50217–50228. [CrossRef]
17. Xu, H.; Cui, X.; Lu, M. An SDR-Based Real-Time Testbed for GNSS Adaptive Array Anti-Jamming Algorithms
Accelerated by GPU. Sensors 2016, 16, 356. [CrossRef] [PubMed]
18. Pazos, S.; Hurtado, M.; Muravchik, C. On Sparse Methods for Array Signal Processing in the Presence of
Interference. IEEE Antennas Wirel. Propag. Lett. 2015, 14, 1165–1168. [CrossRef]
19. Chen, Z.; Xie, F.; Zhao, C.; He, C. An Orthogonal Projection Algorithm to Suppress Interference in
High-Frequency Surface Wave Radar. Remote Sens. 2018, 10, 403. [CrossRef]
20. Hanna, M.T.; Seif, N.P.A.; Ahmed, W.A.E.M. Hermite-Gaussian-like eigenvectors of the discrete Fourier
transform matrix based on the direct utilization of the orthogonal projection matrices on its eigenspaces.
IEEE Trans. Signal Process. 2006, 54, 2815–2819. [CrossRef]
Sensors 2019, 19, 1661 18 of 18
21. Wu, Q.; Zheng, J.-S.; Dong, Z.-C.; Panayirci, E.; Wu, Z.-Q.; Qingnuobu, R. An Improved Adaptive Subspace
Tracking Algorithm Based on Approximated Power Iteration. IEEE Access 2018, 6, 43136–43145. [CrossRef]
22. Suleiman, W.; Pesavento, M.; Zoubir, A.M. Performance Analysis of the Decentralized Eigendecomposition
and ESPRIT Algorithm. IEEE Trans. Signal Process. 2016, 64, 2375–2386. [CrossRef]
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).