TELKOM
NIKA
, Vol.13, No
.2, June 20
15
, pp. 510 ~ 5
1
7
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i2.1470
510
Re
cei
v
ed
Jan
uary 26, 201
5
;
Revi
sed Ma
rch 2
6
, 2015;
Acce
pted April 17, 2015
Sampling Deviation Real-Time Calibration Method for
Wideband Simulator
Song Bingbi
ng*
1
, Pan Minghai
2
, Hu Xiaohua
3
Ke
y
La
bor
ator
y of Ra
dar Ima
g
in
g an
d Micro
w
a
v
e Ph
ot
on
ic
s Ministr
y
of E
ducati
on D
epar
tment, Nanji
ng
Univers
i
t
y
of Aeronautics and Astronauti
cs, 29 Yudao St., Nanj
ing, 210016, China
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: coldic
eson
g
@
13
9.com
1
, panmh@
nu
aa.e
du.cn
2
, hu
xh
19
82@
126.com
3
A
b
st
r
a
ct
Hardw
a
re-i
n-th
e-lo
op si
mu
lati
on is a
n
effici
ent metho
d
fo
r research
on
radar syste
m
.
T
target
’
s
echo w
h
ich off
e
red by the si
mu
lator sh
oul
d
be synchro
ni
zed w
i
th radar o
n
freque
ncy, time, a
nd ran
g
e
bin.
How
e
ver, b
e
ca
use si
mul
a
tor
nee
ds to tak
e
i
n
to acc
ount
of
the re
quir
e
men
t
s of vario
u
s ty
pes
of rad
a
r, it
is
difficult to
ma
ke the cl
ock
of simul
a
tor s
y
nchro
n
i
z
e
d
w
i
th the cl
ock
of radar. T
o
s
o
lve th
e pr
obl
em,
synchro
nous
s
a
mpli
ng
dev
iat
i
on mod
e
l is establ
ishe
d. In
fluenc
e of s
a
mp
lin
g
devi
a
ti
on
on
imag
ing
is
ana
ly
z
e
d. A
n
e
ngi
neer
in
g met
hod
is put forw
ard to e
l
i
m
i
nat
e the sa
mpli
ng
devi
a
tion. T
h
is
meth
od
not o
n
ly
provi
des a refe
rence for si
mu
l
a
tion syste
m
, b
u
t also pr
ov
id
e
s
a reference f
o
r the desi
gn o
f
radar system.
Ke
y
w
ords
:
Sy
nchro
nous S
a
mp
lin
g Dev
i
ati
on, Rea
l
-T
i
m
e
Ca
li
brati
on, Ha
rdw
a
re-in-T
h
e-
Loo
p Si
mul
a
tio
n
1. Introduc
tion
Coh
e
re
nt rad
a
r mea
n
s tha
t
transmitting
sign
al
of rada
r system, lo
ca
l oscillator, coherent
oscillation, tri
gger
pul
se of
timer should
be offere
d b
y
the same
referen
c
e
clo
c
k. The p
h
a
s
e
among
tho
s
e
sig
nal
s
sho
u
ld b
e
fixed.
Co
heren
ce
i
s
the
foun
dat
ion to
achiev
e the f
r
eq
ue
ncy
hoppi
ng tech
nology and li
near freque
n
c
y modulatio
n
technol
ogy. Pulse-Dop
p
l
e
r rad
a
r
which is
widely u
s
ed i
s
co
herent ra
dar [1].
One of condi
tions for Pul
s
e-Doppl
er ra
dar’
s
sig
nal p
r
ocessin
g
is
makin
g
the sampling
sign
als
coh
e
rent amon
g p
u
lse
s
. Some
con
s
trai
nts
are
requi
re
d
for sam
p
lin
g paramete
r
s in
[2],[3]. It is r
equired that the sampli
ng clock an
d the trigger should sy
nchroni
ze to the pul
se
repetition fre
quen
cy (PRF
). The sy
n
c
h
r
onization pre
c
isi
on will di
rectly affect the perfo
rman
ce of
sign
al pro
c
e
s
sing [4].
For P
u
lse-Do
ppler rada
r,
fr
eque
ncy
-
ste
pping
ra
da
r
and
pul
se-co
m
pre
s
sion
ra
dar, i
n
orde
r to me
e
t
the requi
re
ment of sp
ectrum an
alysi
s
and ima
ge
pro
c
e
ssi
ng, it is re
qui
red t
o
maintain
coh
e
ren
c
e
amo
n
g
pul
se
s. Fo
r cohe
rent
radar, to m
a
i
n
tain the
co
here
n
ce am
o
n
g
pulses, sampl
i
ng time interval betwee
n
the nth sa
mp
l
e
point of the i-th pulse and
the nth samp
le
point of the i+1-th pul
se sh
ould be the p
u
lse repetitio
n time [3].
For unive
rsal
signal a
c
qui
sition sy
stem, it is
easy to meet the req
u
irem
ents of
coh
e
re
nt
among p
u
lse
s
wh
en the sampling
clo
c
k is offered by
the rada
r sy
stem. This sam
p
ling metho
d
is
calle
d extern
al syn
c
h
r
on
o
u
s
sam
p
ling.
If rada
r
syst
em do
es
not
provid
e a
sample
clo
c
k
fo
r
acq
u
isitio
n sy
stem by itself
, an internal
clo
ck in
th
e acq
u
isitio
n sy
stem is
used.
This
sampli
n
g
method i
s
called inte
rnal
synchrono
u
s
sampli
ng. Whe
n
the
i
n
ternal syn
c
hronou
s sam
p
l
i
ng
method is u
s
ed, it is diffic
u
lt to maintain cohe
re
nce among pl
use
s
, beca
u
se the internal cl
o
ck of
acq
u
isitio
n sy
stem nee
d to give con
s
id
erati
on to vario
u
s pul
se repe
tition frequen
cy [5].
Hard
wa
re
-in
-
the-loo
p
sim
u
lation is
an
efficient meth
od for resea
r
ch on
ra
dar
system.
The accu
ra
cy
of echo sim
u
lation ha
s a
direct
impa
ct
on the effectivenes
s and
accuracy of the
pro
c
e
ssi
ng result
s of ra
d
a
r sy
stem [6]. When
si
mu
lating echo
signal of ra
da
r, target’
s
echo
sho
u
ld be
synchro
n
ized
with ra
dar on fr
eq
ue
ncy, time, and rang
e
bin. Fre
que
ncy
synchro
n
ization
requi
re
s t
hat t
he
sa
me
clo
c
k
sou
r
ce
sh
ould
be
share
d
b
e
twe
en
simulato
r
an
d
rada
r; time synchroni
zatio
n
requi
re
s th
at the
simula
tor’s outp
u
t is syn
c
hroni
zed with PRF
of
rada
r; ran
ge
bin syn
c
hroni
zation r
equi
re
s that the ran
ge bin interva
l
of simulator
is the sam
e
a
s
that of rada
r, or it is so
m
e
multiple of
the r
ang
e bi
n interval of
rada
r. Ho
wev
e
r, in actu
all
y
,
simulato
r
nee
ds to
ta
ke i
n
to a
c
count
of
the requi
rem
ents
of vari
ou
s type
s
of ra
dar,
or ne
eds to
take i
n
to
account the
requ
ireme
n
ts
of v
a
riou
s types
of PRF. It i
s
difficult to m
a
ke th
e
clo
c
k
of
simulator
same as the
cl
ock of
radar. It will
reduce the sy
nchronization accu
racy between
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Sam
p
ling De
viation
Real
-Tim
e Calibrat
i
on Method fo
r Wide
ban
d Sim
u
lator (Son
g Bingbing
)
511
sampli
ng frequency
and P
R
F
when P
R
F is sampl
e
d by
simu
lator.
It will
destroy the
coherence
among p
u
lse
s
. It will have an influen
ce
on sig
nal pro
c
e
ssi
ng, su
ch
as SAR imag
ing.
In this pap
er,
synchro
nou
s sampli
ng d
e
v
iation model
is est
ablishe
d, and influe
nce
of
sampli
ng dev
iation on ima
g
ing is a
nalyzed. On this b
a
si
s, an en
gineeri
ng meth
od is p
u
t forwa
r
d
to eliminate t
he samplin
g deviation. Thi
s
metho
d
not
only provide
s
a refere
nce
for the de
sig
n
of
simulatio
n
sy
stem, but also provide
s
a
referen
c
e fo
r the de
sign of radar
system.
2.
Des
c
ription
of the pr
oble
m
Assu
me that
the pulse re
p
e
tition interva
l
(P
RI) of ra
d
a
r’s tran
smitting sig
nal is
Tr, and
the sam
p
ling
freque
ncy fo
r simul
a
tor i
s
Fs (F
s = 1/T
s
). On the b
a
s
is, the nu
m
ber of sampli
ng
points in a P
R
I is Np = T
r
/Ts. If the sampling fre
q
u
ency is not a
n
integer mul
t
iple of the pulse
repetition
fre
quen
cy, Np i
s
n
o
t an
inte
ger. It
will d
e
s
troy th
e
coh
e
ren
c
e
am
on
g plu
s
e
s
and
will
affect the radar imaging.
Detailed theoretical
analy
s
is i
s
com
p
l
e
ted in [4],[5]. In
this paper,
analysi
s
con
c
lusio
n
is qu
oted dire
ctly.
The IF ech
o
signal of the ra
dar can be ex
pre
s
sed a
s
0
exp(
2
)
IF
t
s
t
r
ect
f
t
(1)
The echo si
g
nal whi
c
h i
s
g
enerated by
simulator a
nd receive
d
by ra
dar is
0
(,
)
/
(,
)
e
x
p
[
2
(
)
]
D
d
in
R
c
s
i
n
r
ect
j
f
f
t
T
(2)
whe
r
e
(,
)
in
is the time for the nth sampl
e
poi
nt of the i-th pulse.
Let
(,
)
i
t
in
i
T
r
n
T
s
, then
0
/
,e
x
p
[
2
(
)
(
)
]
i
i
t
D
dr
s
t
iTr
n
Ts
R
c
si
n
r
e
c
t
j
f
f
i
T
n
T
T
(3)
If the sampli
ng freq
uen
cy is not an int
eger m
u
ltiple
of the pulse
repetition fre
quen
cy,
i
t
cha
nge
s
alon
g with
the
nu
mber of
pul
se, and
it
cha
nge
s p
e
rio
d
ically. The
mini
mum
cha
n
g
e
cycle i
s
M, an
d M can be gi
ven by the followin
g
equati
on:
M
Tr
N
T
s
Suppo
se
ik
M
m
，
then
(,
)
(
)
i
t
in
k
M
m
T
r
n
T
s
(4)
So, the signal
for the nth sa
mple point of
the i-th pulse can b
e
expre
s
sed a
s
0
,
e
xp
[
2
(
)((
)
)
]
i
Dd
t
si
n
j
f
f
k
M
m
T
r
n
T
s
(5)
In general, the most obvio
us way to implement
the
coh
e
re
nt pro
c
e
ssi
ng is
ca
lculatin
g
the discrete
Fouri
e
r tran
sf
orm (DFT
) of the output
of the nth sample point per
pulse [7-9]. The
result for DFT
can be exp
r
e
s
sed a
s
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 510 – 51
7
512
(
,
)
(
,
)
e
xp(
2
)
1
e
xp(
2
)
(
,
)
(
/
)
Dr
i
ds
d
c
r
k
S
f
n
s
i
n
j
fiT
jf
n
T
A
k
f
X
f
k
f
M
T
(6)
W
h
er
e
1
0
0
12
(
,
)
e
xp(
2
(
)
)
e
x
p(
)
i
M
dd
t
m
Ak
f
j
f
f
j
k
m
M
M
,
()
2
(
)
cd
X
ff
f
The result of above sho
w
s that the peri
od of
S(f,n)’
s freque
ncy
sp
ectru
m
is F
r
(Fr=1/Tr).
There are M l
i
nes in
ea
ch
cycle. Interva
l
of
lines i
s
Fr/M. The maximum pea
k
co
rre
sp
ond to t
he
Dop
p
ler
shift
(fd) of the
ech
o
. The lo
cation fo
r th
e k-th line o
f
the frequ
en
cy sp
ect
r
um
is
fd+
k
Fr/M.
However,
the
Doppler shift s
h
ould be
f
d
in
theo
ry. It is
ca
use
d
by
no-unif
o
rm
sampli
ng. No
-unifo
rm sam
p
ling
d
e
st
roy
s
the co
here
n
ce amo
ng pulses,
a
nd
l
ead
s
to seve
ral
s
p
ec
trum lines
appear [4, 5].
Acco
rdi
ng to the analysi
s
, simulatio
n
is
co
mpl
e
ted. T
he paramete
r
s for sim
u
lati
on are
as follo
ws. The tran
smitti
ng si
gnal
of rada
r i
s
linea
r freq
uen
cy
modulatio
n si
gnal (LFM
); the
band
width
of the sig
nal i
s
100M
Hz; the
PRF is
42KHz
; the pul
se
width i
s
3.2u
s; the velo
city of
the rada
r i
s
3
500m/s;
the
carrie
r frequ
e
n
cy i
s
3
5
.35
G
Hz; the
hei
ght of the
ra
dar is 40Km;
and
the
sa
mplin
g clo
ck
i
s
150
MHz. The re
sults
of
the
si
mulation are
sho
w
n
i
n
Fig
u
re 1.
Figu
re 1(a
)
sho
w
s the coherent accumulation re
sult for t
he situation that
the sampli
ng deviation
is
eliminated; a
nd Figu
re 1(b
)
sho
w
s t
he result cau
s
ed
by samplin
g deviation.
Figure 1(a
)
.
Coherent accu
mulation resu
lt for
that the samp
ling deviation
is eliminate
d
Figure 1(b
)
Coherent accu
mulation resu
lt
cau
s
e
d
by sa
mpling deviati
on
3.
Influence o
f
sampling de
v
i
ation on imaging
Accordi
ng to
the analysi
s
of
section II, the base
band si
gnal
which is
received by SAR
can b
e
expre
s
sed a
s
:
2
0
/
,e
x
p
[
(
/
)
]
e
x
p(
2
/
)
e
xp(
2
)
i
i
i
tt
D
rt
t
tt
iT
r
n
T
s
R
c
s
in
r
e
c
t
j
K
i
T
r
n
T
s
R
c
T
jR
j
f
(7)
If
0
i
t
, the sam
p
ling fre
que
n
c
y is an i
n
teger
multiple
of the pulse of the pul
se
repetition
fre
quen
cy, the
sampling
devi
a
tion i
s
not
e
x
ist, SAR ima
g
ing
can
be
completed
ea
sily,
and the ima
g
e
quality is g
ood. Ho
weve
r, if
0
i
t
, the sampling d
e
viation app
ears,
and the
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Sam
p
ling De
viation
Real
-Tim
e Calibrat
i
on Method fo
r Wide
ban
d Sim
u
lator (Son
g Bingbing
)
513
coh
e
re
nce a
m
ong pul
se i
s
de
stroyed.
The image
q
uality goes worse. The influen
ce on ima
g
ing
will be focused on in this part.
The se
co
nd
-o
rde
r
app
roxi
mation mod
e
l
of range is in
trodu
ced [10]:
2
0
1
()
2
td
c
r
R
Rf
t
f
t
(8)
0
R
is the initial ro
und-t
r
ip ra
ng
e,
dc
f
is Doppl
er center frequ
ency
，
r
f
is
Doppler rate.
Tak
e
the model into
,
D
s
in
:
2
2
00
/
,e
x
p
[
(
/
)
]
e
x
p
(
2
/
)e
x
p
(
2
)e
x
p
(
2
)
i
i
i
tt
D
rt
t
dc
r
t
iT
r
n
T
s
R
c
s
i
n
rect
j
K
i
T
r
n
T
s
R
c
T
jR
j
f
t
j
f
t
j
f
(9)
Usi
ng the RD imaging algo
rithm, the sig
nal after pul
se comp
re
ssi
o
n
in range di
mensi
on
will be get:
0
2
0
/
,
,
s
i
n
(
)
e
xp(
2
/
)
e
xp(
2
)
e
x
p
(
2
)
i
i
tt
R
dc
r
t
iT
r
n
T
s
R
c
si
n
t
c
B
j
R
T
jf
t
j
f
t
jf
(10
)
whe
r
e
i
t
is the
deviate time f
o
r
sampli
ng,
and it is pe
rio
d
ic. Its p
e
rio
d
is M. In a
ddition,
is
tf
.
As can
be se
en from the a
bove analy
s
is, the effects on imaging a
r
e
as follows:
(1) Th
e d
e
viation
woul
d
lead to
po
si
tion offset
o
f
the maxim
u
m p
eak in
ra
nge
dimen
s
ion.
i
t
is pe
riodi
c, so
the positio
n
of t
he maximum pea
k a
p
p
ears jitter. Be
cau
s
e
is
tf
,
the jitter is limited in a rang
e unit.
(2) B
e
cau
s
e
of the ad
ditional(1
) After
p
u
lse
co
mpression i
n
rang
e
dimen
s
ion, th
ere i
s
a
deviation (
i
t
)
at the m
a
ximum p
e
a
k
.
pha
se
(
0
exp
(
2
)
i
jf
t
), p
u
lse
comp
re
ssi
on i
n
a
z
i
m
uth
dimen
s
ion
wil
l
be affected.
4.
A metho
d
fo
r eliminating the samplin
g dev
i
ation in real time
From
the
abo
ve analy
s
is we can
se
e th
at the influ
e
n
c
e
of samplin
g deviatio
n
o
n
si
gna
l
pro
c
e
ssi
ng
cannot
be
igno
red. It i
s
ne
ce
ssary to
elimi
nate the
sam
p
ling
deviatio
n
. In thi
s
p
a
p
e
r,
a metho
d
i
s
p
u
t forward to
solve thi
s
pro
b
lem.
Th
rou
g
h
the
analy
s
i
s
of
pa
rt 2, th
e pe
riod
of
i
t
is
determi
ned
b
y
the relatio
n
s
hip
between
the sa
mp
lin
g
clo
c
k
of simulator and pulse
re
petition
freque
ncy of
rada
r. If the sampli
ng
clo
ck
and p
u
ls
e
repetition fre
quen
cy are a
l
ready give
n, the
perio
d of
i
t
ca
n be
get e
a
sily. However,
be
cau
s
e t
h
e
initial ph
ase
of the
sam
p
ling
clock i
s
rand
om
and
the
sam
p
ling
time i
s
rand
o
m
when
sim
u
l
a
tor
po
wers
u
p
, it is difficult
to elimi
nate t
he
sampli
ng dev
iation. If the sampling time
is fixed whe
n
simulato
r po
wer
up every
time, it is easily
to compe
n
sate the sampli
n
g
deviation b
y
using a gro
up of fixed paramete
r
s.
There i
s
a
fixed relatio
n
s
hip
bet
wee
n
samplin
g
freque
ncy
a
nd p
u
lse re
petition
freque
ncy:
M
Tr
N
T
s
, so the numbe
r of sampling
points in a pu
lse re
petition interval varie
s
in a
cycle. Th
e n
u
mbe
r
of sa
mpling poi
nts in a pul
se
repetition in
terval ca
n b
e
expre
s
sed
as
/
p
rs
N
TT
. Becau
s
e of
the sa
mpling
deviation,
p
N
is not an inte
ge
r.
()
p
f
rac
N
is the f
r
a
c
tion of
p
N
. There i
s
on
e mo
re
sa
mp
le every
M P
R
Is if
()
0
.
5
p
fra
c
N
, and o
ne le
ss
sam
p
le every
M
PRIs
if
()
0
.
5
p
fra
c
N
[4]. The P
R
I wh
o
s
e
sam
p
le
s’
numbe
r
cha
n
ges can b
e
fo
und e
a
sily. If
the
next PRI is
c
o
ns
idered as
the firs
t PRI when
s
i
m
u
lator p
o
wers up, the i
n
itial pha
se of
the
sampli
ng
clo
c
k is fixed eve
r
y time. On th
e ba
sis,
sam
p
ling d
e
viatio
n ca
n be
com
pen
sated
ea
sily
by a group of
fixed param
eters. Th
e impl
em
entation di
agra
m
is sho
w
n in Figu
re
2.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 510 – 51
7
514
Figure 2. The
implementati
on diag
ram fo
r finding the first PRI
If the initial p
hase of the sampling
clo
c
k is
fixed every time, the deviation that should b
e
comp
en
sated
for every PRI is fixed too. If the period
of
i
t
is M, supp
ose
1
0
t
,
2
tt
,
3
2
tt
,
4
3
tt
,
,
(1
)
M
tt
M
. The
dela
y
of the
sam
p
ling
clo
c
k fo
r the
first P
R
I
is
(1
)
/
(
*
)
s
M
MT
, for
se
con
d
P
R
I
i
s
(2
)
/
(
*
)
s
M
MT
, for the third P
R
I is
(3
)
/
(
*
)
s
M
MT
, and fo
r the
(M-1)-th P
R
I i
s
1/
(
*
)
s
M
T
. The delay
of the sampli
ng clo
c
k for
the M-th PRI
is ze
ro. In this way, it is ensu
r
e
12
3
(1
)
/
(
*
)
M
s
tt
t
t
M
M
T
.
Becau
s
e
the
delay is le
ss t
han o
ne
clo
c
k
cycle,
it
can
not be
co
mpl
e
ted by u
s
in
g
system
clo
ck. Th
e de
lay can be
ca
rrie
d
out by the pha
se
shif
ting module i
n
FPGA. Phase shift pre
c
ision
for FPGA ca
n achi
eve to 1/56 cy
cl
e of Mixed-Mo
de
Clo
ck Ma
nag
er’s V
C
O, so
the minimu
m of
i
t
that can be a
c
hieve
d
is T/56 of VCO.
5. Simulation
and
v
e
rificati
on
Hardware-in
-
t
he-lo
op sim
u
l
a
tion system
whi
c
h co
nsi
s
t
s
of rada
r an
d echo
simul
a
tor is
an effective
mean
s to ve
rify the co
m
pen
sati
on
m
e
thod a
nd a
nalysi
s
the i
n
fluen
ce of t
he
sampli
ng
dev
iation. The
system is a u
s
eful
sup
p
le
ment to the t
heoretical
an
alysis
and
di
gital
simulatio
n
. The co
nstructi
on of the hardwa
r
e
-
in-t
h
e
-l
oop sim
u
latio
n
system i
s
a
s
Figu
re 3.
The sy
stem
works at Ka
wavele
ngth. Its ban
dw
idth
is 100
MHz. T
he sy
stem co
nsi
s
ts of
microwave d
o
wn
-conve
r
si
on unit, mi
cro
w
ave u
p
-con
versio
n unit a
nd si
gnal p
r
o
c
e
ssi
ng unit
etc.
Microwave d
o
wn
-conve
r
si
on unit
re
cei
v
es the
RF
signal, an
d m
o
ves the f
r
eq
uen
cy sp
ectrum
from Ka wave
length to S wavelength. Th
e Micr
owave up-conve
r
sio
n
unit
co
nvert
s
the ba
se
ba
nd
ech
o
to Ka
wavelength, filters ha
rmo
n
ic wave,
a
nd
control
s
the
p
o
we
r of o
u
tp
ut. The
RF
si
gnal
from Mi
crowave up
-conv
ersi
on
unit
will b
e
p
our
ed into
the
receiver of t
he
rada
r. Si
gnal
pro
c
e
ssi
ng u
n
it sample
s IF signal, cha
nge
s IF
signa
l to zero inte
rmediate fre
q
uen
cy, generates
ech
o
of base
band u
s
in
g convolution
s
.
To verify the
impa
ct of
sa
mpling
deviat
i
on,
set the
work pa
rame
ters of the
sy
stem
a
s
follows: the tran
smitting
si
gnal of rada
r is LF
M; its
band
width
(B
W) i
s
1
00M
Hz; the P
R
F
is
42KHz; the p
u
lse
width
(P
W) i
s
3.2
u
s; t
he velo
city
of the aircraft whi
c
h
carrie
s the ra
dar
(Vt) is
3500m/
s
; the
carrie
r fre
que
ncy of
rada
r
(Fc) is
35.
35
G
H
z; th
e hei
gh
t of the aircra
ft (H) i
s
40K
m;
and the sam
p
ling freq
uen
cy of simulat
o
r (F
s) i
s
15
0
M
Hz. Be
cau
s
e N=Fs/PRF is not an inte
ger,
the
samplin
g
deviation a
ppea
rs, and
the
cohe
re
n
c
e amo
ng
pul
se i
s
de
stroy
ed. The
ima
g
e
quality goe
s worse.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Sam
p
ling De
viation
Real
-Tim
e Calibrat
i
on Method fo
r Wide
ban
d Sim
u
lator (Son
g Bingbing
)
515
Figure 3.
Blo
ck di
agram of
the con
s
tru
c
t
i
on of
hard
w
a
r
e-i
n
-the
-lo
o
p
simulation
system
In experim
ent
, the tran
smit
ting sig
nal i
s
injecte
d
into t
he si
mulato
r, and i
s
conve
r
ted to
low intermediate frequency. At the s
a
me time,
the
power i
s
a
d
j
u
sted
by the
auto gai
n co
ntrol
unit (AG
C
) in
microwave d
o
wn
-fre
que
ncy unit to ma
ke sure
the a
n
a
log-to
-di
g
ital co
nverte
r ca
n
get the be
st
SNR
(signa
l to noise ra
tio). Then, t
he si
gnal i
s
sampl
ed by
analo
g
-to
-
digi
tal
conve
r
ter. Th
e sam
p
ling freque
ncy is n
o
t an intege
r multiple of the pulse re
pet
ition freque
ncy,
so the
sampl
i
ng time is n
o
t fixed whe
n
simul
a
tor p
o
we
rs
up. Th
e pha
se of
signal is
ran
d
o
m
.
Signals of M
pulse
s with
samplin
g d
e
viation and
signal
s that
are corre
c
ted are sh
own in
Figure 4.
(a)
(b)
Figure 4. Signals of M
=
7 p
u
lse
s
; a) with
samp
li
ng devi
a
tion; b) sig
n
a
ls that are
correcte
d
The imagi
ng
results that
with sa
mplin
g dev
iation a
nd the re
sult
s that deviati
ons a
r
e
corre
c
ted a
r
e
sho
w
n in Fig
u
re 5. The
re
sults a
r
e
con
s
iste
nt with the analysi
s
in
part 3.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 510 – 51
7
516
(a)
(b)
(c)
(d)
(e)
(f)
Figure 5. The
imaging resu
lts that with sampling
d
e
viation and the
results that d
e
viations, gra
y
-
scale map aft
e
r: (a
). Gray-scale map aft
e
r
ra
nge
com
p
re
ssi
on with
sampli
ng devi
a
tion, Gray-
scale map aft
e
r ra
nge
com
p
re
ssi
on which deviation is
corre
c
ted, (c). Spectrum af
ter azim
uth
FFT with sam
p
ling deviatio
n
, (d). Spe
c
trum after
azi
m
uth FFT whi
c
h deviation is
corre
c
ted, (e
).
3D imag
e after azi
m
uth co
mpre
ssion
with sa
mpli
ng d
e
viation, (f). 3d image after
azimuth
comp
re
ssion
whi
c
h deviati
on is corre
c
te
d
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Sam
p
ling De
viation
Real
-Tim
e Calibrat
i
on Method fo
r Wide
ban
d Sim
u
lator (Son
g Bingbing
)
517
The si
mulatio
n
re
sults
sh
o
w
that the d
e
v
iation
lead
s
to positio
n of
target ap
pea
r jitter in
rang
e dime
nsio
n. The
azimuth
co
mpre
ssion
i
s
greatly affected by
the presen
ce
of
0
exp(
2
)
i
t
jf
. Becau
s
e of
the perio
dic of
i
t
, the peak after
azi
m
uth co
mpre
ssi
on al
so
appe
ars pe
ri
odic va
riation
and d
e
focus appe
ars.
It may cau
s
e
false
alarm. Usi
ng the m
e
tho
d
mentione
d in
this pa
per,
the i
n
fluen
ce
on
imagi
ng fo
r
sam
p
ling d
e
viatio
n is elimi
nat
ed
effectively. This metho
d
h
a
s a hig
h
value in the field
of enginee
rin
g
.
6. Conclu
sions
In this p
ape
r, problem
s
cau
s
e
d
by
sampling
devi
a
tion a
r
e
st
udied, the
sampling
deviation mo
del is e
s
tabli
s
he
d, and inf
l
uen
ce
of sa
mpling deviat
i
on on ima
g
i
ng is an
alyzed.
Becau
s
e
of t
he
sampli
ng
deviation, the
pea
k a
fte
r range
co
mpre
ssi
on a
ppe
ars jitter, a
nd t
he
azimuth
com
p
re
ssi
on
is greatly a
ffected
by the
p
r
ese
n
ce
of th
e a
d
d
itional
pha
se. Becau
s
e
of the
perio
dic of th
e samplin
g d
e
viation, the
minor l
obe
in
cre
a
ses sha
r
ply, and
defo
c
u
s
a
ppe
ars. On
this ba
si
s, a
n
engi
nee
rin
g
metho
d
is put forw
ard
to eliminate
the sa
mplin
g deviation.
This
method
not o
n
ly provid
es
a refe
ren
c
e
for
simula
tio
n
system, but also provide
s
a
refere
nce for
the desi
gn of rada
r sy
stem.
Referen
ces
[1]
Ye H
R
. Co
her
ent ra
dar
an
d
non-c
oher
ent r
adar
differ
ence
an
al
ysis
(in
C
h
in
ese) .
Mod
e
r
n El
ectroni
c
s
T
e
chni
que
. 2
0
10; 3: 17-1
9
.
[2]
Qi R, Co
akle
y F
P
, Evans B
G. Practical c
onsi
derati
on f
o
r ban
d-p
a
ss s
a
mpli
ng.
IEE Electron
Lett
..
199
6; 32: 186
1
-
186
2.
[3]
Ma BT
,
F
an HQ, F
u
Q. IF sampli
ng co
ndi
ti
ons for coher
e
n
t pulse ra
dar
(in Chin
ese).
Data Acqu
i.
Proces
. 200
9; 24: 114-
11
8.
[4]
He Y, Z
hang C
S
, Ding JH, et al. T
he impact
of time
s
y
nchr
o
n
izati
on error o
n
passiv
e
cohe
rent puls
e
d
radar s
y
stem.
Sci. Chin
a Inf.
Sci.
2010; 5
3
: 266
4–
267
4.
[5]
Z
hu YL, F
an HQ, Ma BT
, Lu Z
Q. Design of IF
signal acq
u
i
s
ition s
y
stem for puls
e
coh
e
r
ent radars (i
n
Chin
ese).
Systems E
ngi
ne
eri
ng an
d Electro
n
ics
. 200
9; 31(
3): 489-4
96.
[6]
He Z
H
, H
e
F
,
Don
g
Z
,
Li
ang
DN. Re
al-T
ime Ra
w
-
S
i
g
nal
Si
mulati
on A
l
gor
i
t
hm for InSAR
Hard
w
a
re-i
n-
the-Lo
op Simu
latio
n
Appl
icati
ons.
IEEE Geoscience and Re
m
o
te Sensing Letter
. 201
2; 9(1): 134-
138.
[7]
Kulp
a KS, Cze
k
ala Z
.
Maski
n
g
effect an
d its
remova
l in P
C
L rad
a
r.
IEE R
adar S
o
n
a
r Na
v.
2005;
152:
174
–1
78.
[8]
Griffiths HD, Baker CJ.
Passive
co
here
n
t l
o
cation
rad
a
r s
ystems Part 1:
performa
nce
p
r
edicti
on.
IEE
R
a
da
r So
na
r N
a
v
. 2005; 1
52: 153
–1
59.
[9]
Ho
w
l
an
d PE, Maksimiuk D,
Reitsma
G. F
M
radio b
a
se
d
bistatic rad
a
r.
IEE Radar So
n
a
r Nav.
20
05;
152: 10
7–
11
5.
[10]
Z
hang Y S, Lia
ng D N, Sun Z
Y,
Dong Z
.
Effect of time syn
c
hron
izatio
n er
ror on interfer
o
m
etric phas
e
of spaceb
orn
e
paras
itic InSAR s
y
stem (in C
h
in
ese).
Journ
a
l of Astrona
uti
cs.
2007; 28(
2)
: 370-37
4.
Evaluation Warning : The document was created with Spire.PDF for Python.