TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 12, No. 12, Decembe
r
2014, pp. 82
1
7
~ 822
8
DOI: 10.115
9
1
/telkomni
ka.
v
12i12.46
40
8217
Re
cei
v
ed O
c
t
ober 1
0
, 201
3; Revi
se
d Septem
ber 10,
2014; Accept
ed Octo
ber 1
9
, 2014
Chaotic Binary Modulation Excitation Sequences for
Multichannel Ultrasonic Ranging System
Yao Zhenjin
g, Yang Jingsong
Dep
a
rtment of Disaster Prev
e
n
tion Instrume
nt, Institute of
Disaster Prev
e
n
tion,
Sanh
e He
bei,
065
20
1, Chin
a
A
b
st
r
a
ct
Ultraso
nic cro
sstalk often causes fals
e time-
o
f-flig
ht (TOF
) in distance me
asur
ement. T
h
e
excitatio
n
se
q
uenc
es w
i
th g
ood c
o
rre
latio
n
char
acteri
sti
cs, i.e., sharp
autoc
orrel
a
tio
n
an
d flat cr
o
ss-
correlati
on, ca
n hel
p avoi
d crosstal
k betw
een
mu
lticha
n
nel u
l
trason
ic
sensors. A crosstalk eli
m
in
a
t
io
n
meth
od
by usi
ng the
opti
m
a
l
bin
a
ry excitati
on se
que
nc
es
mo
du
lated w
i
t
h
cha
o
tic cod
e
s, w
h
ich inclu
d
es
chaotic b
i
nary
amplit
ude sh
ift keying (c-BA
SK), chaot
ic bi
nary ph
ase shi
ft keying (c-BPSK) and cha
o
tic
bin
a
ry frequ
en
cy shift keying
(c-BF
SK), is
prop
osed
in
th
is pap
er. T
he selecti
on of th
e opti
m
a
l
cha
o
tic
initia
l v
a
lu
es w
i
th the
b
e
st ec
ho c
o
rrel
a
tio
n
character
i
stics is
rea
l
i
z
e
d
by ado
ptin
g
a
g
e
netic alg
o
rith
m.
A
squar
e
w
a
ve is u
s
ed
a
s
the
ca
rrie
r
o
f
the
ch
ao
ti
c
b
i
n
a
r
y
mo
du
la
tio
n
meth
od
s in
o
r
d
e
r
to
re
du
ce
ha
rdwa
re
cost. H
o
w
e
ve
r, it co
ul
d s
u
ffer
fro
m
th
e d
e
cr
e
a
se
d
en
erg
y
ef
fici
ency
ca
us
e
d
by
mis
m
atch
i
ng s
p
ectru
m
w
i
t
h
ultr
aso
n
i
c
r
a
n
g
i
n
g
sy
ste
m
. T
h
e
par
a
m
eters
of th
e c
h
aoti
c
bi
nar
y
mod
u
l
a
t
i
o
n
meth
ods
a
r
e c
onf
ig
ur
ed
to
make their central fr
equency and bandwidt
h m
a
tc
hed with
that of
the
ultr
asonic rang
ing system
very
well.
Experiments have been conducted us
ing
an ultras
onic r
anging system t
hat consists of
eight-channel
SensC
o
mp
60
0 seri
es i
n
stru
me
nt-gra
de
el
ectrostati
c se
n
s
ors excite
d
w
i
th chaotic
b
i
nary s
e
q
uenc
es.
Experimental r
e
sults show that
the c-BFSK outperfor
m
s the c-BASK and
c-BPSK in term
s of the ener
gy
efficiency, e.g.,
0.350
8
vs. 0.336
5
and 0.
32
79
, respectivel
y
. And the
opti
m
i
z
e
d
c-BF
SK
sequ
enc
e
also h
a
s the be
st echo correl
a
tion char
acteris
t
ics
amo
ng the
chaotic b
i
n
a
ry mo
du
latio
n
seq
uenc
es.
Ke
y
w
ords
:
u
l
trasonic cr
oss
t
alk, bin
a
ry
mo
du
latio
n
, c
haotic s
e
ri
es, ener
gy
effici
ency, corre
lati
on
character
i
stic
Copy
right
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Ultrasonic
sensors have been exten
s
ivel
y applie
d on rescue
robot for
distance
measurem
ents,
thanks
to
t
heir low price
and simp
le
hardwa
r
e interface. To obtain 360-degree
panorama
distance informa
t
ion, multichannel ultrasoni
c se
nso
r
s
co
mposing a
rin
g
are
require
d in
a rescue rob
o
t
[1]. One problem with these simult
aneo
usly working
senso
r
s in a
ring is ultrason
ic
crosstalk, in
which o
ne ultrasoni
c se
nsor re
ceiv
es e
c
ho tran
smitted by anothe
r ultrasoni
c
sensor
[2]. Generally, ultrasonic receiver can
n
o
t disti
nguish whether the
received echo from its own
transmissio
n
or not, so the
incorre
ct time
-of-flight (TO
F
) measurem
ents often occur.
The effective
cro
sstalk
elimination method is
to giv
e
each
ultrasonic se
nso
r
a unique
excitation signature in the transmission
and t
hen id
entify
the signature usin
g a co
rrelation
technique in the receiver. Jörg and Berg [3] were
the first to give
a recogni
za
ble signature
of
each sen
s
or i
n
the transmission using p
s
eudor
andom
code and frequency modu
lation.
And th
en
in the receivin
g circuit, the ident
ification of the transmitted sou
r
ce
se
nsor
wa
s by
a matched filter.
Subsequently, some rese
arche
r
s have
applied di
fferent code
s
and modulation schem
es to
constru
c
t excitation sequences a
s
a transmission signature to solve the ultr
asonic
cro
sst
alk
problem. Barker code
s [4] were used
to a
v
oi
d crosstalk in ultrasoni
c
system, although
the
available Barker
code
s lim
it their application. In
ultrasonic distance m
easureme
n
t system, G
o
lay
codes [5-7]
were a
pplied
to restrain
crosstalk and
increa
se the
signal-to-noi
se ratio. But the
realization
co
mplexity of Golay code
s re
stricts t
heir a
pplication. The binary-code
d frequen
cy shift
keying (BFSK) signal and
binary-code
d phase shi
ft keying (BPSK) signal [8]
were applied
to
drive multiple
piezoelectric ultrasonic se
nsors wi
th na
rrow b
a
ndwid
t
h. In reference [9] and [10],
BPSK modul
ation was both used to construct the tr
a
n
smission signals of
ultrasonic system.
But
they adopted
different code
s to mod
u
late, i.e., Al
va
r
e
z
et al
. [9] used
compleme
ntary sequ
ence
s
codes a
nd Iwa
s
aw
a
et al
. [10] applied M sequen
ce. Chaotic code
s having sharp
autocorrelatio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 12, Decem
ber 20
14 : 8217 – 82
28
8218
function and f
l
at cross-correlation function are
se
ns
itive to small changes in the
initial conditions
and in p
a
ra
meter values. Therefore, Fortuna
et
al
. [11] exploited chaot
ic pulse
po
sitio
n
modulation (CPPM) to e
x
cite the
ultr
asonic se
nso
r
to eliminate crosstalk and improve the
efficiency of ultrasonic system. Yao
et al
. [12] proposed ch
aotic pu
lse po
sition–width modula
t
ion
(CPPWM)
se
quence
s
to construct short
digital exci
tation sequen
ce
s to trigger ultrasoni
c sen
s
o
r
s.
Meng
et al
. [13] used a gen
etic algorithm (GA) to
optimize sho
r
t CPPM and pse
udorandom P
P
M
triggering se
q
uence
s
in o
r
d
e
r to minimize the maximal side-lobe
of autocorrelatio
n
and the p
e
a
k
of cross-co
rre
lation function
.
As we kno
w
, ultrasonic senso
r
s
work like band-pass filters and have bell-shaped
magnitude spectrum. If the spe
c
trum
of the ex
cited seq
uence
does not m
a
tch that of
th
e
ultrasonic se
nsor, some of
the excita
tio
n
energy
can
not be transmitted by the
ultrasonic system
which de
crea
ses the energ
y
efficiency.
Pollatowski
and Ermert [14] used transm
i
tter signals with
a con
s
tant amplitude level and nonline
a
r frequen
cy
modulation to match the spectrum of th
eir
ultrasonic system.
Meng
et al
. [15]
adopted spe
c
trum optimization of a
CPPM excitation
sequen
ce to
improve en
ergy efficiency. Yao
et al
. [12] used n
ondominated
sorting ge
n
e
tic
algorithm II to
optimize both pulse perio
ds and dut
y ratio of the CPPWM excitation seque
nce
s
in
order to maximize echo en
ergy
simultaneously on the
basis of a be
st correlation chara
c
teristics.
To our kno
w
l
edge, not many researche
r
s di
scuss rej
e
cting cro
sst
alk in a multichannel
ultrasonic
ra
nging system
based
on the binary ex
citation seque
nces m
odulated with ch
a
o
tic
cod
e
s t
o
dri
v
e elect
r
o
s
t
a
t
i
c sen
s
o
r
s.
M
o
r
eover, fewer
research
er
s
debate t
he parameters
configuration
method for binary modulation exci
tation sequen
ces to make their
central frequen
cy
and bandwidt
h
matched that of the ultrasonic syst
em. This paper aims to exploit
the nove
l
crosstalk
elimination method by
applying chaotic binary modulat
ion excitation sequence
s
which
includes
c
h
aotic
binary amplitude shift k
e
ying (c
-B
ASK), c
haotic
binary phas
e
shift k
e
ying (c-
BPSK), and
chaotic bi
nary frequency
shift keying
(c
-BFSK).
A square
wave i
s
used
as t
he
carrie
r
wave of
the chaotic binary modulati
on methods in order to reduce
hardware
cost.
However, it could suffer from the decre
ased en
er
gy
efficiency cau
s
ed by mism
atching spe
c
trum
with ultrasoni
c system. To improve energy effi
ciency, the parameters of the chaotic bi
nary
modulation methods are
co
nfigured to make t
heir
central frequen
c
y and band
width matched th
at
of the ultrasonic system
very well. The
chaotic initia
l val
ues are optimized by applying GA to
achieve the
optimal corre
lation c
h
a
r
a
c
t
e
ri
st
i
cs.
T
he
u
l
t
r
a
s
o
n
i
c sy
stem that co
nsi
s
ts of eig
h
t-
cha
nnel
Se
n
s
Comp
6
00 seri
es
i
n
stru
ment-g
rad
e
electrostati
c sen
s
o
r
s wa
s
desig
ned a
n
d
recomme
nde
d
the be
st ch
aotic bina
ry modulatio
n
a
ppro
a
ch
fo
r electrostati
c ultrasoni
c se
nso
r
via experime
n
ts.
The remaind
e
r of this pap
er is organize
d as
follows.
Section II presents the prin
ciple of
chaotic binary modulation
excitation s
equence.
Th
e correlation
chara
c
teristics and ene
rgy
efficiency are explained in section III. Se
ction IV
introd
uces the paramete
r configuration methods
of the binary
modulation and the GA-based optimiz
ation of the excitation sequences. Section V
shows the experiments an
d discussion, follo
wed by
the conclusio
n
s in section VI.
2.
The Principle of Ch
aotic
Bina
r
y
Modulation Excita
tion Sequen
c
e
2.1. Chao
tic
Co
d
e
s
Chaotic code
s had been u
s
ed to construct ex
citation
sequence because of th
eir good
correlation ch
aracteristics (i.e. sharp autocorrelati
on function and flat cross-corre
lation functio
n
).
Moreover, they are sen
s
itive to small ch
anges in
the i
n
itial conditions and in p
a
rameter values. In
this paper, the Ulam–von Neumann tra
n
sformation [
16] was used
to generate chaotic cod
e
s as
follows
:
,
2
,
1
,
1
,
1
,
2
1
2
1
i
y
y
y
i
i
i
(
1
)
Binary chaotic codes were
generated by the following
formula,
,
2
,
1
,
0
1
0
0
sgn
i
y
y
y
i
i
i
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Cha
o
tic Bina
ry Modul
ation
Excitation Se
quen
ce
s for
Multicha
nnel
Ultra
s
oni
c…
(Yao Zhenji
n
g
)
8219
2.2.
Binar
y
Modulation Sche
me
The bi
nary m
odulatio
n techniques incl
u
de
bi
nary am
plitude shift keying
(BASK),
binary
freque
ncy shi
ft keying (BF
SK) and bina
ry pha
se sh
ift keying (BPS
K), which ha
ve two states of
amplitude, freque
ncy a
n
d
pha
se, respectively.
In the propo
sed chaoti
c
b
i
nary mo
dula
t
ion
approac
h
, the variation of amp
litude,
frequenc
y
and
phas
e
are on t
h
e basis of
chaot
ic codes.
2.2.1.
Determinatio
n of carrier s
i
gnal
In traditional
approa
ch, a
sinu
soi
dal si
g
nal is g
ene
ral
l
y used a
s
th
e ca
rri
er
sign
al of the
binary mod
u
l
a
tion techni
q
ue. Since th
e hard
w
a
r
e i
m
pleme
n
tatio
n
of a square wave is m
u
ch
easi
e
r tha
n
a
sinu
soid
al wave, the following
squ
a
re wave
q
(
t
) is
a
dopted a
s
the
carrie
r si
gnal
of
cha
o
tic bin
a
ry modulation
seq
uen
ce
s,
,
2
,
1
,
0
,
1
2
1
0
2
1
1
n
T
n
t
T
n
T
n
t
nT
t
q
c
c
c
c
,
(3)
Whe
r
e
c
T
is the period of the
squ
a
re
wave.
The sq
ua
re wave
t
q
can be repre
s
e
n
ted a
s
a Fou
r
ier
se
ries,
t
T
k
k
t
q
c
k
1
2
2
sin
1
2
2
2
1
0
.
(
4
)
From th
e ab
ove expressi
on, we
can f
i
nd that the
squ
a
re
wave
is a
co
mpo
s
ite of a
dire
ct-curre
nt compo
nent
and odd
harm
oni
c
co
mpone
nts. In all of th
e odd ha
rm
onic
comp
one
nts,
the ene
rgy of the funda
mental ha
rm
onic i
s
the hi
ghe
st. Becau
s
e the ult
r
a
s
onic
system
works as a pa
ss-ba
nd filter, only the
odd ha
rm
onic
comp
on
ents withi
n
the pass-ban
d of
the ultrasoni
c syste
m
are receiv
ed. Therefore, to
obtain the highe
st ene
rgy efficiency,
the
spe
c
tru
m
of fundam
ental h
a
rmo
n
ic mu
st
matc
h with t
hat of the ultraso
n
ic
syste
m
.
2.2.2. c-BAS
K
In BASK, the amplitude of
a fi
xed-frequency carrier wave
i
s
changed with
each
symbol
of bas
e
-band s
i
gnal. For t
he
c
-
BASK, the binary
c
h
aotic
c
o
des were
us
ed
as
the bas
e
-band
signal. Mathematically, the form
for c-BASK sequence can be
written as:
t
T
k
k
t
c
t
X
c
k
BASK
1
2
2
sin
1
2
2
2
1
0
,
(
5
)
Whe
r
e
t
c
is binary cha
o
tic code
s used to cha
nge t
he a
m
plitude of carri
er si
gnal.
And the value
of
t
c
is either 1
or 0.
2.2.3. c-BPS
K
In BPSK, the
phase
of a
const
ant amplit
ude and
frequency
ca
rrier signal
alters between
zero and
.
The symb
ols “1” and
“0” are rep
r
e
s
e
n
ted by zero
and
of carrie
r
sig
nal,
respectively. With c-BPSK, the chaotic
informat
ion is contai
ned i
n
the
instantaneous phase of
the modulate
d
carrie
r sig
n
a
l. The c-BP
SK is given by the followin
g
formula:
s
c
k
BPSK
iT
c
t
T
k
k
t
X
-
1
1
2
2
sin
1
2
2
2
1
0
(
6
)
Whe
r
e
s
T
is the symbol
width of bas
e-band
signal. I
n
c-BPSK, the base-band signal is the
binary chaoti
c
co
de
s.
2.2.4. c-BFSK
BFSK transm
i
ts the inform
ation usin
g di
fferent
ca
rrie
r
frequen
cie
s
to represe
n
t symbol
states.
The
a
m
plitude
rem
a
ins un
ch
ang
ed. In BFSK,
the symb
ols “1” a
nd
“0
” a
r
e
rep
r
e
s
e
n
ted
by
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 12, Decem
ber 20
14 : 8217 – 82
28
8220
the carrie
r freque
nci
e
s
1
f
and
2
f
, res
p
ec
tively. For the
c
-BFSK, the value of
binary cha
o
tic
cod
e
s, eith
er 1 or 0, is d
e
termin
ed u
s
ing (1)
and
(2). Mathe
m
at
ically this is
written
by the
following:
0
,
1
2
2
sin
1
2
2
2
1
1
,
1
2
2
sin
1
2
2
2
1
2
0
1
0
t
c
t
f
k
k
t
c
t
f
k
k
t
X
k
k
BFSK
(
7
)
3.
The Cor
r
elation Chara
c
te
ri
stics and E
n
erg
y
Efficienc
y
3.1. Correla
tion
Char
acteristi
cs
Correl
ation chara
c
te
risti
c
s
[15] in
clud
e t
he a
u
tocorrel
ation fun
c
tion
and
cro
s
s-co
rrel
a
tion
function. In the ultrasoni
c rangin
g
syste
m
, the autoc
orrelation fun
c
tion of the
i
th
e
c
ho
s
e
qu
en
c
e
is defined as
follows:
M
i
m
m
R
m
x
x
m
R
ii
m
N
n
i
n
i
m
n
ii
,
,
2
,
1
,
0
0
1
0
,
(
8
)
Whe
r
e
M
is the cha
nnel
nu
mber of ult
r
a
s
oni
c sy
stem
,
i
n
x
and
i
m
n
x
are t
he
n
th an
d (
m
+
n
)th
sampli
ng
dat
a poi
nt of the
i
th
ec
ho s
e
quenc
e
, res
p
ec
tively,
N
i
s
t
he total
num
ber of
sampl
e
s i
n
the ech
o
se
q
uen
ce.
The definitio
n
of the cro
s
s-
correl
ation fu
nction of the
i
th and
j
th
ec
ho
s
e
qu
e
n
c
e
s
is given
as
follows
:
j
i
M
j
M
i
m
m
R
m
x
x
m
R
ij
m
N
n
j
n
i
m
n
ij
,
,
1
,
,
1
,
0
0
1
0
(
9
)
Whe
r
e
j
n
x
is
the
n
th samplin
g
data point of the
j
th echo
seque
nce.
The excitation seque
nces with bet
ter corr
elation cha
r
acte
ristics, i.e., sharp
e
r
autocorrelatio
n
and flatter cross-co
rrelati
on,
have better crosstalk rejection ability.
3.2. Energ
y
Efficienc
y
Becau
s
e
the ultrasoni
c se
nso
r
h
a
s a
b
and-pa
ss
sp
e
c
trum, the
ex
citation e
nerg
y
can b
e
transmitted by the ultraso
n
ic sy
st
em when sp
ect
r
a of the excita
tion seq
uen
ce
s match
with that
of the ultraso
n
ic sen
s
or. T
he sp
ectrum
of the
triggered se
que
nce whi
c
h mismat
che
s
with tha
t
of
the ultrasoni
c system can
decre
a
s
e the
energy effici
ency. The “e
nergy efficie
n
c
y” mea
n
s th
e
ratio between
the energy of
the echo
a
n
d
that of the excitation sequ
ence
[15].
The ene
rgy e
fficiency
is defined a
s
:
T
R
E
E
,
(
1
0
)
N
i
i
R
R
X
R
E
1
2
1
,
(
1
1
)
N
i
i
T
T
Y
R
E
1
2
1
,
(
1
2
)
Whe
r
e
T
E
and
R
E
are the
ene
rgies
of the e
x
citation
an
d
ech
o
sequ
e
n
ce
s, re
sp
ect
i
vely;
T
R
and
R
R
are the
equivalent re
sista
n
ce of the tran
sm
itting
and re
ceivin
g circuit
s
, re
spectively;
i
X
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Cha
o
tic Bina
ry Modul
ation
Excitation Se
quen
ce
s for
Multicha
nnel
Ultra
s
oni
c…
(Yao Zhenji
n
g
)
8221
and
i
Y
are the
i
th sampli
ng
data of exci
tation and
e
c
ho
se
que
nces, re
sp
ectiv
e
ly;
N
is the
numbe
r of sa
mples.
Be
c
a
us
e
T
R
and
R
R
a
r
e
different
, the efficie
n
cy ratio
exp
r
e
s
sed
as follo
ws is u
s
ed
to
comp
are the energy efficie
n
cy
of two excitation sequ
ences:
2
1
,
(
1
3
)
Whe
r
e
1
and
2
are the en
erg
y
efficiencie
s
of tw
o excitation se
que
nce
s
, re
spe
c
tively.
4.
The Parame
ter Co
nfigur
ation Metho
d
of
the Ch
aotic Bina
r
y
Modulation
Excitation
Sequenc
e
The spe
c
tru
m
of cha
o
tic binary mo
d
u
lation sequ
ence with
square carrier wave is
different with
that of sinu
so
id ca
rrie
r
wave. T
he ch
aoti
c
bina
ry mod
u
lation sequ
e
n
ce
with squ
a
re
carrie
r wave
has
a direct
-current compo
nent an
d ha
rmonic
co
mpo
nents. T
o
imp
r
ove the e
n
e
r
gy
efficien
cy
of excitation se
quen
ce,
the
harm
oni
c
co
mpone
nts wi
th
highe
r
e
n
e
rgy sho
u
ld be
adju
s
ted i
n
th
e pa
ss-b
and
of the ult
r
a
s
o
n
ic
syste
m
, which
can
en
sure th
e
spe
c
t
r
um
of excitat
i
on
seq
uen
ce ma
tche
s with tha
t
of the ultrasonic
system.
The power spectral density of the c-BASK ex
citation sequence is ex
pressed as f
o
llows:
0
2
2
2
1
2
1
2
1
2
8
1
1
2
1
2
sin
1
2
1
2
sin
1
2
8
8
1
sin
8
BASK
k
c
c
s
c
s
c
s
c
s
c
s
s
s
s
f
k
f
f
k
f
k
T
f
k
f
T
f
k
f
T
f
k
f
T
f
k
f
k
T
f
fT
fT
T
f
P
(14
)
And the power s
p
ec
tral dens
i
ty of the c
-
BPSK exc
i
tation s
e
quenc
e
is
the following:
0
2
2
BPSK
1
2
1
2
sin
1
2
1
2
sin
1
2
2
2
1
k
s
c
s
c
s
c
s
c
s
T
f
k
f
T
f
k
f
T
f
k
f
T
f
k
f
k
T
f
P
(
1
5
)
From
(14
)
a
nd (1
5), we
can find th
at the cha
n
ge tend
ency
of amplitud
e amon
g
freque
ncy ra
nge
s
c
s
c
f
f
k
f
f
k
1
2
,
1
2
at different value
s
of
k
is simila
r. The highe
st
amplitude
ap
pears in freq
uen
cy ran
ge
s
c
s
c
f
f
f
f
,
. In other wo
rds, the mo
st energy of the
c-BASK and
c-BPSK sequences i
s
fo
cused i
n
fre
quency range
s
c
s
c
f
f
f
f
,
. Therefo
r
e, the
bandwidth of
c-BASK and
c-BPSK are about
s
s
T
f
B
B
2
2
BPSK
BASK
.
For the c-B
ASK and c-BPSK excitation se
quences, the
carri
er frequency
c
f
and
band
width
BPSK
BASK
B
B
are dete
r
min
e
d
by the ce
ntral fre
quen
cy
of the ultra
s
onic
sen
s
o
r
and
the symbol wi
dth of base
-
b
and sig
nal
s
T
, r
e
sp
ectively. The ca
rrie
r
fre
quen
cy is set
to the central
freque
ncy of
the ultra
s
o
n
ic
sen
s
o
r
gene
rally. It is a
s
sume
d that the
symbol
widt
h is
,
2
,
1
,
n
f
n
T
c
s
,
n
f
T
B
B
B
c
s
2
2
BPSK
BASK
sonar
, where
sonar
B
is the bandwi
d
th of th
e
ultrasoni
c se
nso
r
.
The po
we
r sp
ectral d
e
n
s
ity of the c-BFSK excitation seque
nce is gi
ven as follo
ws:
0
2
2
1
1
2
2
2
2
2
2
2
1
1
2
1
1
1
2
1
2
1
2
8
1
1
2
1
2
1
2
8
1
1
2
1
2
sin
1
2
1
2
sin
1
2
8
1
2
1
2
sin
1
2
1
2
sin
1
2
8
2
1
BFSK
k
s
s
s
s
s
s
s
s
s
s
f
k
f
f
k
f
k
f
k
f
f
k
f
k
T
f
k
f
T
f
k
f
T
f
k
f
T
f
k
f
k
T
T
f
k
f
T
f
k
f
T
f
k
f
T
f
k
f
k
T
f
P
(
1
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 12, Decem
ber 20
14 : 8217 – 82
28
8222
To ma
ke mo
st of the en
e
r
gy of the
c-BF
SK sequ
e
n
ce to
be fo
cuse
d in the freque
ncy
band of the ul
traso
n
ic
sen
s
or, the param
eter setti
ng
s
are dete
r
min
ed by the followin
g
formul
as:
2
1
2
2
2
1
1
1
sonar
2
1
2
1
2
1
1
,
1
min
,
2
2
s
s
s
s
s
s
s
c
T
T
f
T
n
f
T
n
f
B
f
f
f
f
f
f
f
f
f
,
(
1
7
)
W
h
er
e
1
s
T
and
2
s
T
are
the
symb
ol wi
dths of t
he
“1” an
d
“0
”
symbol
s, re
spe
c
tively. Consi
deri
n
g
that the n
u
m
ber
of the
sq
uare
carrie
rs
sh
o
u
ld
be int
egral
withi
n
a
symbol
wi
dth,
1
n
an
d
2
n
are
positive integ
e
rs. Also, the symbol wi
dth
s
of
the “1” a
nd “0
” are a li
ttle bit different.
5.
The GA-ba
s
e
d Optimiza
tion of the
Ch
aoti
c Bin
a
r
y
Modulation
Excitatio
n
Sequen
c
e
Given the length of the chaotic bin
a
ry
m
odulation
excitation se
quen
ce, sym
bol width
and carrier f
r
equ
en
cy, a GA is use
d
to optimiz
e the ch
aotic i
n
itial values
to get the best
correl
ation
chara
c
te
risti
c
s (i.e., sha
r
p
e
st a
u
to
correlation
and
flattest cro
s
s-correl
ation).
The
pro
c
ed
ure is
pre
s
ente
d
in the followi
ng steps.
Step 1: The
initial paren
t populatio
n
Q
P
A
is produ
ce
d
rand
omly, whe
r
e
P
is
th
e
popul
ation si
ze and
Q
is the length of float chao
tic initial values. Let
P
= 100,
M
Q
(co
r
respon
din
g
to
M
chaotic i
n
itial values for
M
chan
nel ul
traso
n
ic
syste
m
), and th
e
maximum
gene
ration n
u
m
ber i
s
set to
100.
Step 2: The
obje
c
tive-fun
ction valu
es
of individual
s are
ord
e
re
d
and the
n
m
appe
d to
fitness value
s
. Th
en
a
col
u
mn ve
ctor o
f
fitnes
s valu
es i
s
retu
rne
d
. An o
b
je
ctive functio
n
ObjV
defined a
s
fol
l
ows:
x
c
a
R
R
ObjV
ma
-
max
-
,
max
:
(
1
8
)
δ
,...N
m
M
i
m
R
R
ii
a
1
0
,
,
,
1
,
max
:
max
-
(
1
9
)
j
i
M
j
M
i
N
,...
,
,
m
m
R
R
ij
c
,
,
1
,
,
1
,
1
2
2
1
0
,
max
:
max
-
(20
)
Whe
r
e
max
-
a
R
is
th
e ma
xima
l s
i
de
-
l
o
b
e
a
m
ong
M
auto
c
orrelation fun
c
tions,
max
c
R
is the
maximal
pea
k amon
g
2
1
-
M
M
cro
s
s-co
rrelat
ion functio
n
s.
Step 3: The selectio
n prob
ability of individual
s is
set to 0.9, and th
e sele
cted i
n
dividual
s
are return
ed to the new p
o
pulation.
Step 4: Th
e
cro
s
sove
r a
n
d
mutation
o
perato
r
s a
r
e
use
d
to
gene
rate the
ne
w
child
ren
popul
ation.
Step 5: The
o
ffspring
po
pul
ation is comb
ined
with the
curre
n
t gen
eration po
pulati
on an
d
sele
ction
is p
e
rform
ed to
set the in
divid
uals for th
e n
e
xt gene
ratio
n
. Since
all t
he p
r
eviou
s
and
curre
n
t be
st i
ndividual
s a
r
e add
ed to
th
e pop
ulation,
elitism i
s
e
n
sured.
Re
peat
Step 2 to Ste
p
5
until the maximum gen
erati
on numb
e
r i
s
rea
c
he
d.
6.
Experiments
and Discu
s
s
ion
6.1. Experimenta
l
Setup
The eight-cha
nnel ultrasoni
c ranging
system
was u
s
ed
in our experiments. Each channel
of the ultrasonic ranging
system
has the same hardwa
re re
alization. Fig
u
re 1 sho
w
s the
hardwa
r
e
real
ization schem
atic diagram f
o
r one
cha
n
n
e
l ultrasonic ranging syste
m
. The cha
o
tic
binary modul
ation sequen
ce was
sent
from the fie
l
d
p
r
o
g
r
a
mmin
g
g
a
t
e
a
r
r
a
y (
F
PG
A)
. Af
te
r
power ampli
f
ying, the u
l
traso
n
ic
se
nso
r
wa
s triggered to tran
smit ultra
s
ou
nd. In our
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Cha
o
tic Bina
ry Modul
ation
Excitation Se
quen
ce
s for
Multicha
nnel
Ultra
s
oni
c…
(Yao Zhenji
n
g
)
8223
experim
ents,
a SensCom
p
600 se
ri
e
s
instru
ment-grade ele
c
tro
s
t
a
tic se
nsor was u
s
ed a
s
b
o
th
transmitter a
nd re
ceiver.
After band-pa
ss filterin
g,
automatic gai
n
amplification
and sha
p
ing,
the
polarity co
rre
l
ation betwe
e
n
the binary
ech
o
se
quen
ce an
d a ref
e
ren
c
e e
c
h
o
sequ
en
ce was
carrie
d o
u
t. The referen
c
e
ech
o
seq
uen
ce
wa
s
re
cod
ed fro
m
a
n
a
c
rylic bo
ard p
l
ace
d
4
0
cm i
n
front of the ul
traso
n
ic
se
nsor. He
re it sh
ould
be n
o
te
d that the em
itted pulse
se
quen
ce a
nd i
t
s
ech
o
sequ
en
ce a
r
e
different owi
ng to
the filt
ering
effect of th
e ultra
s
oni
c
sen
s
o
r
, so the
correl
ation ch
ara
c
teri
stics
betwe
en the excitati
on se
quen
ce an
d its own e
c
ho
is poo
r. That is
why we did not
u
s
e
th
e emission
seq
uen
ce as
th
e
refere
nce to
cal
c
ul
ate the
correlatio
n
cha
r
a
c
teri
stics. Actually, si
milar correlati
on
processin
g
method
wa
s also ad
opte
d
by Jörg
et al
.
[17], where the echo in
ste
ad of
the emi
tted pulse
se
quen
ce
wa
s
use
d
as th
e referen
c
e. L
a
stly,
the di
stan
ce
cal
c
ulatio
n
was i
m
plem
ent
ed if th
e
e
c
h
o
sequ
en
ce
wa
s
re
cog
n
ized to
be
fro
m
its
own sen
s
o
r
transmi
ssion.
Figure 1. The
hard
w
a
r
e re
alizatio
n sche
matic diag
ra
m for one cha
nnel ultra
s
o
n
i
c
ran
g
ing
sy
st
em
6.2.
The Spec
tru
m
of the Ultr
asonic Rang
ing Sy
stem
The p
u
rp
ose
of this
exp
e
rime
nt is to
figure
out t
he u
s
abl
e freque
ncy b
a
n
d
of the
Senscom
p
600 elect
r
o
s
ta
tic ultrasoni
c rangin
g
system. In this
experim
ent, each excitation
s
e
quenc
e
included ten
50%-duty-
c
y
c
l
e rec
t
angle puls
e
s
with
s
a
me frequency. The f
r
equency
rang
e wa
s fro
m
20 kHz to 100 kHz, the interval
wa
s 0
.
5 kHz. The e
x
citation se
qu
ences
were the
binary se
que
nce
s
.
Th
en a
ll
the corre
s
p
ondin
g
e
c
ho
seq
uen
ce
s were sa
mple
d whi
c
h refle
c
ted
from an a
c
ryli
c boa
rd pla
c
e
d
40 cm in fro
n
t of ultrasoni
c se
nsor.
Thro
ugh cal
c
ulating
the en
ergie
s
of
the ex
citation
se
quen
ce
s a
nd
ech
o
sequ
en
ce
s, the
spe
c
tru
m
of the ultra
s
o
n
ic
rangi
ng
syste
m
is sh
ow
n i
n
Figure 2. From this expe
riment, it can
be
found that
ultrasoni
c rangi
ng sy
st
em h
a
s
its
own cen
t
ral freq
uen
cy
. The central
freque
ncy
an
d
the frequ
en
cy band of the
ultrasonic
sy
stem are abo
ut 55 kHz a
n
d
[40, 70]
kHz, re
spe
c
tively.
T
h
e
nearer i
s
the excitation seq
uen
ce f
r
equ
en
cy to
the ce
ntral fre
quen
cy, the less is the e
c
ho
sign
al attenu
ation. The
further i
s
the
ex
citation
se
qu
ence fre
que
n
c
y to the
cent
ral frequ
ency,
the
more i
s
the e
c
ho
sign
al attenuatio
n.
Figure 2. The
Spectru
m
of the ultrasoni
c rangin
g
syst
em
6.3.
Experimenta
l
Results a
n
d Discus
s
io
n
For ea
ch cha
o
tic
bi
nary
m
odulatio
n
seq
uen
ce,
ei
ght cha
o
tic co
de seri
es
were use
d
to
con
s
tru
c
t
eig
h
t ch
ann
els o
f
excitation
seque
nces,
re
spe
c
tively. T
he le
ngth
of t
he
cha
o
tic
bi
nary
modulatio
n e
x
citation se
q
uen
ce is
set
to 2 ms. Usin
g the GA optimization
algo
rithm, after 1
0
0
generations of
selection,
crossover
and mutation,
the optim
ized results for the
c-BASK, c-BFSK
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
()
Freq
u
e
n
c
y
k
H
z
A
m
p
litu
d
e
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 12, Decem
ber 20
14 : 8217 – 82
28
8224
and c-BPSK
sequences were
4735
.
0
c
BASK
ObjV
,
3379
.
0
c
BFSK
ObjV
an
d
3666
.
0
c
BPSK
ObjV
,
r
e
spec
tively.
6.3.1.
The Spec
tru
m
and Echo Analy
s
is of Chao
tic Bina
r
y
Modulation Sequenc
e
s
Figure 3(a), 4(a) and 5(a) show that
the c-BASK, c-BPSK and c-BFSK excitation
seq
uen
ce
s
whe
n
the
chaotic initial
value
s
a
r
e
optimized based on G
A
,
respe
c
tively.
In
subfig
ure
s
(b
) of Figu
re 3
-
5, the bla
ck lines
a
r
e th
e sp
ectra of
the ch
aotic
b
i
nary mo
dula
t
ion
excitation se
quen
ce
s in F
i
gure 3
(
a
)
, 4(a) and 5
(
a
)
, and the re
d lines a
r
e the
spe
c
tru
m
of the
ultrasoni
c ra
nging
syste
m
. As sh
own i
n
Figu
re
3(
b
)
, there
is
si
gnifica
nt am
ount of
spe
c
tral
distrib
u
tes in
the frequen
cy band [1, 40] kHz as
well as in the freque
ncy ba
nd [40, 70] kHz,
whi
c
h is th
e pass-ban
d of
the ultrasoni
c ra
ngin
g
sy
stem. In othe
r wo
rd
s, the
spe
c
tru
m
of c-
BASK excitati
on sequence mismat
ches
with that of the ultrasoni
c ranging system, t
he
ex
ci
t
a
ti
on
energy in freq
uency band
[1, 40] kHz ca
nnot be tr
an
smitted by the ultrasonic
system. Compare
d
to the c
-
BASK, c
-
BPSK in
Figure 4(b) pres
ents
that th
ere is
more
spec
tral dis
t
ributes
in the pas
s
-
band of the u
l
trasonic
rang
ing system (i.
e
., [40,
70] kHz), but
som
e
excitation energy is
still in
frequency ba
nd [1, 40] kHz whi
c
h i
s
no
t in the
pass-band of the u
l
trasoni
c rang
ing
system. As
sh
o
w
n
i
n
F
i
g
u
re
5
(
b
)
,
there is more e
n
e
rgy of c-BF
SK sequen
ce
distribute
s
i
n
the freque
ncy
band of
the ul
trasonic
rangi
ng sy
stem than that of
c-B
ASK and
c-
B
PSK sequences, at the same
time, there is less ene
rgy
of c-BFSK
seq
uen
ce
di
stributes i
n
th
e out-b
and
o
f
the ultraso
n
ic
rangi
ng
syste
m
. Therefo
r
e,
in three
chao
tic bina
ry mo
dulation ex
citation se
que
n
c
e
s
, the c-BF
SK
excitation seq
uen
ce mo
st spectrally matc
he
s with the
ultrasoni
c ra
n
g
ing sy
stem.
Figure 3(c), 4(c) and 5(c) illustrate the
correspondi
ng echo s
equences of c-BASK, c
-
BPSK and
c-BFSK excitation
sequences,
respectivel
y
. All the echo
sequences were
reflect
ed
from an ob
st
acle pla
c
e
d
4
0
cm in front
of the ultrasonic
sen
s
o
r
. The sa
mple
perio
d wa
s
1
μ
s
.
The c-BFSK method in Figure 5c p
r
odu
ce
s the
highe
st echo
amplitude o
f
0.9 V at s
o
me
sampli
ng tim
e
and mo
st e
c
ho a
m
plitud
e of [0.6, 0.8] V caused by
the matche
d
spe
c
tra b
e
tween
the excitatio
n
se
que
nce a
n
d
ultra
s
o
n
ic rangin
g
syste
m
. As in
dicated in
Figu
re
4(b
)
a
nd
5(b
)
,
the
echo amplitude in c-BASK
and c-BPSK
are less th
an that of the
c-BFSK in Figure 3(b). Among
three
ch
aotic bina
ry mod
u
lation
seq
u
ences, th
e c-BFSK sh
ows the
be
st result, i.e., e
c
h
o
amplitude of
0.9 V due to excellent m
a
tchin
g
sp
ect
r
a between t
he excitation
sequ
en
ce a
n
d
ultrasoni
c ra
n
g
ing sy
stem as sho
w
n in
Figure 5(b
)
.
Figure 3. Optimized
c-BAS
K sequence: (a) opti
mized c-BASK sequence, (b
) the
spectrum of
(a), (c) the
co
rre
sp
ondi
ng e
c
ho
seq
uen
ce
Figure 4. Optimized
c-BPS
K sequence: (a) opti
mized c-BPSK sequence, (b
) the
spectrum of
(a), (c) the
co
rre
sp
ondi
ng e
c
ho
seq
uen
ce
0
500
10
00
1
500
2
000
0
0.
2
0.
4
0.
6
0.
8
1
Ti
m
e
(
μ
s)
Am
p
l
i
t
ud
e
(
V
)
0
20
40
60
80
10
0
0
50
0
10
00
15
00
20
00
25
00
30
00
35
00
F
r
e
que
nc
y (
k
H
z
)
A
m
p
lit
u
d
e
0
50
0
1000
1
500
20
00
-0.
8
-0.
6
-0.
4
-0.
2
0
0.
2
0.
4
0.
6
0.
8
Ti
m
e
(
μ
s)
A
m
p
lit
u
d
e (V
)
0
500
10
00
1
500
2
000
0
0.
2
0.
4
0.
6
0.
8
1
Ti
m
e
(
μ
s)
A
m
p
lit
u
d
e (V
)
0
20
40
60
80
100
0
50
0
100
0
150
0
200
0
F
r
e
que
n
c
y (
k
H
z
)
A
m
p
lit
u
d
e
0
50
0
1
000
15
00
2000
-0.
8
-0.
6
-0.
4
-0.
2
0
0.
2
0.
4
0.
6
0.
8
Ti
m
e
(
μ
s)
A
m
p
lit
u
d
e (V
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Cha
o
tic Bina
ry Modul
ation
Excitation Se
quen
ce
s for
Multicha
nnel
Ultra
s
oni
c…
(Yao Zhenji
n
g
)
8225
Figure 5. Optimized
c-BFS
K
seque
nce: (a) o
p
ti
mize
d c-BFSK sequ
ence, (b) the
spe
c
tru
m
of
(a), (c) the
co
rre
sp
ondi
ng e
c
ho
seq
uen
ce
To quantitati
v
e comparison
,
the echo energies
of
the c-BASK, c-BFSK and c-BPSK
excitation
se
quen
ce
s
co
rresp
ondi
ng to
Figu
re 3
-
5
a
r
e li
sted in
T
able 1.
Fro
m
Table
1, we
can
see that the echo energy of t
he c-BFSK excitation si
gnal is be
tter t
han that of the c-BPSK and c-
BASK signal
s. It can be
also found
that the energi
es of the thr
ee excitation s
equences were
different. Both the ec
ho and the exc
i
tation energ
ies
of the c
-
BASK s
e
quenc
e
s
are the s
m
allest.
The e
nergy
efficien
cie
s
o
f
the thre
e chaot
ic bina
ry
modul
ation
seq
uen
ce
s a
r
e al
so
cal
c
ulated based Equation (
10)-(13). T
he energy efficienci
es of the c-BASK, c-BFSK and
c
-
BPSK excitation sequences corresponding to Figure
3-5 are illustrated in Table 1. It can be
found that th
e ene
rgy effi
cien
cy of the
c-BFSK ex
citation sequ
en
ce i
s
bette
r than that of th
e c-
BPSK and c-BASK sequences.
Table1. The ec
ho energies
of c
-
BASK,
c
-
BFSK and c
-
BPSK exc
i
t
a
tion s
e
quenc
es
Excitation Seque
nces
BASK
BFSK
BPSK
Excitation Seque
nces Energ
y
(
μ
s
V
2
)
498.0000
1022.0000
1003.0000
Echo Energ
y
(
μ
s
V
2
)
167.5800
358.4800
328.8600
Energ
y
efficiency
0.3365
0.3508
0.3279
6.3.2.
The Cor
r
elation Chara
c
te
ristic An
aly
s
is of Chao
tic
Binar
y
Mod
u
lation Sequ
ences
Figure 6-7
show the correl
ati
on charact
e
risti
c
of c-B
ASK
without optimization and after
optimizatio
n, i.e., correlati
on c
haracte
ri
stic in
clud
es
the auto
c
or
re
lation functio
n
s of two echo
s
e
quenc
e
s
and the crossc
orrelati
on func
tion. A
s
s
h
own in
Figu
re 6-7, the optimiz
e
d c-BASK
seq
uen
ce
s
h
a
ve lo
wer si
d
e
-lob
e of
ech
o
auto
c
o
r
rela
tion fun
c
tion
s than th
at of t
he u
noptimi
z
e
d
c-BASK sequences, i.e., 0.
33 vs 0.58.
Moreover,
the c-BASK sequence afte
r optimization also
has lower peak of echo
crossc
orrel
a
tion function than that
of the c-BASK sequence
without
optimizatio
n, i.e., 0.39 vs 0.51. Th
e
comp
ari
s
on
results b
e
twee
n the o
p
timized
c-BFSK
seq
uen
ce
s a
nd the
uno
ptimized
c-BFS
K
seq
uen
ce
s are
p
r
esent
ed in
Figu
re
8-9. F
r
om
th
ese
figure
s
, we can find that the side
-lob
e of echo a
u
tocorrelation fun
c
ti
ons of the opt
imized
c-BFS
K
seq
uen
ce
s i
s
abo
ut 0.1
4
l
o
we
r th
an th
at of the
un
o
p
timized
c-B
F
SK se
que
nces. An
d the
p
eak
of ech
o
cro
s
scorrelatio
n
functio
n
of th
e opt
imized
c-BFSK se
que
nce
s
i
s
ab
out
0.1 lower th
an
that
of
the
unoptimi
z
ed
c-BFSK se
q
uen
ce
s.
Fig
u
re
s 10-11
demon
strate
the co
rrela
t
io
n
characteri
stic of c-BPSK without optimization and
after optimization. As indicat
ed in Figure 10-
11, the side-l
obe of echo
autocorrelation functions of
the
optimized c-BPSK se
quences is about
0.1 lowe
r than that of the unopti
m
ized
c-BP
SK seque
nces. And the
peak of e
c
ho
crosscorrelati
on function of the optimized
c-
BPSK
sequences is
about
0.08 l
o
wer than that
of
the unoptimi
z
ed c-BPSK sequences.
Compari
s
on
with the opt
imized
c-BA
SK and
c-BPSK sequences, the correlation
characteri
stic of the optimiz
ed
c-BFSK sequence i
s
lower t
han that of c-BASK and c-BPSK, i.e.,
the optimized
c-BFSK
seq
uen
ce h
a
s th
e lowest
side
-lobe
of echo
autocorrelati
on fun
c
tion
s
and
the lowest
p
eak
of e
c
ho
crosscorrel
a
tion f
unctio
n
. In other word
s, the o
p
t
imized
c-BF
SK
seq
uen
ce h
a
s
the be
st echo
co
rrel
a
tion ch
ara
c
te
ri
stics amo
ng
the cha
o
tic b
i
nary mod
u
la
tion
seq
uen
ce
s.
0
500
10
00
1
500
2
000
0
0.
2
0.
4
0.
6
0.
8
1
Ti
m
e
(
μ
s)
A
m
p
lit
u
d
e (V
)
0
20
40
60
80
100
0
50
0
100
0
150
0
200
0
250
0
300
0
F
r
e
que
n
c
y (
k
H
z
)
A
m
p
lit
u
d
e
0
50
0
1000
1
500
20
00
-0.
8
-0.
6
-0.
4
-0.
2
0
0.
2
0.
4
0.
6
0.
8
1
Ti
m
e
(
μ
s)
A
m
p
lit
u
d
e (V
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 12, Decem
ber 20
14 : 8217 – 82
28
8226
(a)
(b)
(c
)
Figure 6. The correlation charac
teristi
c
of c-BASK without optim
ization: (a) the normali
z
ed
autocorrelatio
n
of echo
seq
uen
ce 1, (b
) the nor
mali
ze
d autocorrelat
i
on of
ech
o
seque
nce 2, (c)
the norm
a
lize
d
cro
sscorrel
a
tion of ech
o
seq
uen
ce 1 a
nd 2
(a)
(b)
(c
)
Figure 7. The c
o
rrelation charac
teris
t
ic
of
c
-
BASK after optimiz
ation: (a) the normaliz
ed
autocorrelatio
n
of echo
seq
uen
ce 1, (b
) the nor
mali
ze
d autocorrelat
i
on of
ech
o
seque
nce 2, (c)
the norm
a
lize
d
cro
sscorrel
a
tion of ech
o
seq
uen
ce 1 a
nd 2
(a)
(b)
(c
)
Figure 8. The
correlatio
n chara
c
te
risti
c
of c-
BFSK wit
hout optimiza
t
ion: (a) the n
o
rmali
z
e
d
autocorrelatio
n
of echo
seq
uen
ce 1, (b
) the nor
mali
ze
d autocorrelat
i
on of
ech
o
seque
nce 2, (c)
the norm
a
lize
d
cro
sscorrel
a
tion of ech
o
seq
uen
ce 1 a
nd 2
(a)
(b)
(c
)
Figure 9. The
correlatio
n chara
c
te
risti
c
of
c-BFSK after optimi
z
atio
n: (a) the no
rmalize
d
autocorrelatio
n
of echo
seq
uen
ce 1, (b
) the nor
mali
ze
d autocorrelat
i
on of
ech
o
seque
nce 2, (c)
the norm
a
lize
d
cro
sscorrel
a
tion of ech
o
seq
uen
ce 1 a
nd 2
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
a
l
ize
d
au
t
o
co
rre
lat
i
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
a
l
ize
d
au
t
o
co
rre
lat
i
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
r
m
ali
z
ed
c
r
os
s
c
or
re
l
a
t
i
on
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
alized
au
t
o
c
o
rr
elat
i
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
alized
au
t
o
c
o
rr
elat
i
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
alize
d
cr
os
s
c
o
r
r
e
lati
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
alized
au
t
o
c
o
rr
elat
i
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
a
l
ize
d
au
t
o
co
rre
lat
i
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
r
m
ali
z
ed
c
r
os
s
c
or
re
l
a
t
i
on
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
a
l
ize
d
au
t
o
co
rre
lat
i
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
alized
au
t
o
c
o
rr
elat
i
o
n
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
x 1
0
4
0
0.
2
0.
4
0.
6
0.
8
1
S
a
m
p
lin
g
d
a
t
a
N
o
rm
alize
d
cr
os
s
c
o
r
r
e
lati
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.