Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
10
,
No.
5
,
Octo
be
r
2020
,
pp.
4824
~
4834
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
5
.
pp
4824
-
48
34
4824
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
A
ne
w d
ynamic s
peech en
cryption
algorith
m based on
L
oren
z
chaotic
map ov
er
i
nte
rnet
p
rotoc
ol
Oba
id
a M.
A
l
-
H
az
aimeh
Depa
rtment
o
f
C
om
pute
r
Scie
n
ce a
nd
In
form
at
ion
Technol
og
y
,
Al
-
Bal
q
a
'
Appl
ie
d
Univer
sit
y
,
Jordan
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Feb
3
, 2
0
20
Re
vised
Ma
r
13
,
2020
Accepte
d
Ma
r
2
4
, 202
0
Thi
s
pape
r
int
r
oduce
s
a
d
y
n
a
m
ic
spee
ch
encr
y
pt
ion
al
gor
it
h
m
base
d
on
Lore
nz
ch
aot
i
c
m
ap
over
int
ern
et
proto
col
to
enha
nc
e
the
servic
es
of
the
re
al
-
t
ime
ap
pli
c
at
ions
such
as
inc
re
ase
s
the
sec
urity
l
evel
and
red
uc
es
la
t
ency
.
Th
e
pr
oposed
al
gorit
h
m
was
divi
ded
int
o
two
proc
es
ses:
d
y
namic
ke
y
gene
r
ation
proc
ess
using
128
-
bit
hash
val
ue
to
d
y
na
m
ic
al
l
y
al
t
er
the
initial
sec
re
t
ke
y
s,
and
enc
r
y
pt
io
n
and
de
cr
y
pt
ion
proc
ess
using
Lore
nz
s
y
stem.
In
the
p
roposed
al
gor
it
h
m
,
the
p
erf
orm
a
nce
eva
lu
at
ion
i
s
ca
rri
ed
out
through
eff
ic
i
en
t
sim
ula
t
ions
an
d
implementatio
ns
and
sta
ti
sti
c
al
ana
l
y
sis.
In
addi
ti
on
,
the
ave
rag
e
ti
m
e
de
lay
in
the
propo
sed
al
gorit
hm
an
d
som
e
of
the
ex
isti
ng
a
lg
orit
hm
s
such
as
AES
is
comp
are
d.
The
ob
ta
i
ned
result
s
conc
lud
ed
tha
t
,
the
proposed
d
y
namic
spe
e
ch
enc
r
y
pt
ion
al
gorit
hm
is
eff
ectua
l
l
y
se
cur
ed
ag
ai
nst
v
ari
ous
cr
y
p
ta
n
aly
s
is
atta
cks
and
h
as
useful
cr
y
p
tographic
prop
ert
i
es
such
as
c
onfusi
on
and
diffusio
n
for
bette
r
voic
e
comm
unicati
on
in the
voice
app
li
c
ations fi
el
d
in
th
e
In
te
rn
e
t.
Ke
yw
or
d
s
:
AMR
-
WB
-
G
.722.2
Chaos
-
the
or
y
Crypto
gr
a
phy
IP
tel
ep
hony
Lor
e
nz
m
ap
Copyright
©
202
0
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Ob
ai
da
M.
A
l
-
h
azai
m
eh,
Dep
a
rtm
ent o
f C
om
pu
te
r
Scie
nce a
nd Info
rm
at
ion
Tec
hnol
ogy,
Al
-
Ba
lqa'
A
pp
li
ed
U
niv
e
rsity
,
21163, J
orda
n.
Em
a
il
:
dr
_obai
da@ba
u.
e
du.jo
1.
INTROD
U
CTION
Vo
ic
e
a
ppli
cat
ion
over
I
nter
net
pr
oto
c
ol
(
i.e.,
VoIP
)
is
a
IP
te
le
phon
y
te
chnolo
gy
that
al
lows
the
voic
e
data
as
hum
an
voic
e
to
tra
ns
fe
rs
in
real
-
tim
e
ov
e
r
In
te
r
net
Pr
ot
oc
ol
(i.e.
,
IP
)
in
a
m
anner
that
e
m
ulate
s
the
tr
aditi
on
al
te
le
phone
ser
vice
(i.e.,
PST
N
)
[
1
,
2
]
.
In
ge
ne
ral,
the
vo
ic
e
ap
plica
ti
on
over
int
ern
e
t
prot
oc
ol
only
r
equ
i
res
a
n
pr
ogram
on
t
he
e
nd
-
po
i
nt
com
pu
te
r
(i.e.,
sen
de
r,
or
receive
r)
capa
ble
of
e
nc
od
i
ng
and d
ec
odin
g
t
ran
sm
it
t
ing
s
pe
ech a
nd Inter
ne
t acc
ess
[
3
,
4
]
.
Vo
ic
e
a
pp
li
cat
ion
over
inter
ne
t
pr
ot
oco
l
te
c
hnology
pro
vi
des
m
or
e
adv
a
ntages
wh
e
n
it
is
com
par
ed
to
the
tra
diti
onal
te
le
ph
one
se
rv
ic
e.
IP
te
le
phony
te
ch
no
l
ogy
is
cheape
r,
can
be
i
nteg
rated
with
oth
e
r
m
edia
serv
ic
es
,
porta
ble,
a
nd
al
lo
ws
f
or
m
or
e
e
ff
ic
i
ent
ba
ndwi
dth
util
iz
at
ion
.
T
he
refor
e
,
the
ser
vi
ce
pro
vid
e
rs
prefe
r
the
IP
te
le
phony
te
chnolo
gy
as
a
m
et
ho
d
to
re
du
ce
t
he
cost
over
e
xisti
ng
m
ultim
edia
ser
vices.
Mo
r
eov
e
r,
the
in
fr
ast
ru
ct
ure
of
the
IP
te
le
phony
is
c
ons
idere
d
as
a
s
olid
ec
onom
ic
al
gro
und
i
n
bu
il
ding
t
he
m
or
e
recent
rev
e
nue
-
gen
e
r
at
ing
ser
vices.
Ma
rk
edly
,
th
e
dep
loym
ent
of
IP
te
le
ph
on
y
te
ch
no
l
ogy
is
beco
m
ing
m
or
e
popula
r
an
d
is
con
si
der
e
d
as
an
integral
pa
rt
of
a
gl
ob
al
com
petit
ive
lan
dsc
ape
[
5
,
6
]
.
Desp
it
e
of
al
l
thes
e
po
sit
ive
feat
ures,
the
I
P
te
le
phony
te
ch
nolog
y
is
faci
ng
so
m
e
diff
ic
ulti
es
an
d
c
halle
nges
s
uc
h
as
se
cur
it
y
,
pack
et
loss
,
an
d
la
te
ncy.
Conseq
uen
tl
y,
m
or
e
adv
a
nced
te
c
hn
i
qu
e
s
or
stra
te
gies
are
warr
anted
to
com
petentl
y
m
anag
e
these
diff
ic
ulti
es,
w
hi
ch
are
e
xp
ect
ed
to
e
nsure
t
he
qu
al
it
y
of
t
he
IP
te
le
phony
te
chnolo
gy
se
rv
ic
es
(i.e.,
Q
oS
)
[
3
,
7
]
.
F
or
e
xam
pl
e,
the
th
reat
of
the
intr
ud
e
rs
ov
e
r
I
P
net
wor
ks
is
the
gr
eat
e
st
secur
it
y
chal
le
ng
e
in
IP
te
le
ph
ony
te
chn
ol
og
y.
The
la
te
r
will
be
a
great
co
nc
ern
si
nce
thes
e
offende
rs
uti
li
ze
var
io
us
s
ni
ff
in
g
too
ls t
o
com
prom
ise
the co
nv
ersati
on
of
the
IP
tel
epho
ny. To
m
anag
ing
t
he
secu
rity
ch
al
le
ng
es, cr
y
ptog
ra
phy
serv
e
s as a
v
al
uab
le
t
oo
l t
o
m
ai
ntain dat
a se
crecy
[
8
,
9
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A n
ew
dy
namic
sp
eec
h
e
ncry
pt
ion
algorit
hm
ba
s
ed
on l
or
en
z ch
aoti
c map
(
Obaida M
. Al
-
H
azaime
h)
4825
The
sci
e
nce
of
crypt
ology
co
ns
ist
s
of
t
wo
m
ajo
r
pa
rts:
cr
yptogra
ph
y
a
nd
cry
ptanaly
sis
.
The
sci
enc
e
of
c
ryptol
og
y
and
it
s
cry
ptog
raphic
pri
m
it
ives
are
cat
eg
or
i
zed
as
il
lustrat
ed
in
Fi
g
ure
1.
W
hile
cry
ptog
raphy
is
the
art
of
m
ai
ntaining
da
ta
secrecy
ag
ai
ns
t
una
uthor
iz
ed
acce
ss
(i.
e.,
int
ruders),
on
the
oth
e
r
sid
ed
,
cryptanaly
sis i
s abo
ut an
al
yz
ing an
d ha
nd
li
ng in
fr
i
ng
em
ent of sec
ur
e
co
m
m
un
ic
at
ion
[
10
,
11
]
.
Figure
1. Tax
onom
y of
c
rypt
ology
To
m
ai
ntain
data
secrecy
against
un
a
uthori
zed
acce
ss
ove
r
IP
net
works
the
crypto
gr
a
phy
is
widely
us
e
d.
H
oweve
r,
m
os
t
of
th
e
existi
ng
c
rypto
gr
a
phic
al
gorithm
s
are
buil
t
to
m
ai
nta
in
te
xt
data
s
ecrecy
.
Ther
e
f
or
e,
thes
e
existi
ng
al
go
rithm
s
are
not
su
it
able
f
or
voic
e
app
li
cat
io
ns
over
inter
net
prot
oco
l
beca
us
e
it
involves
e
xten
sive
com
pu
ta
ti
on
a
nd
co
nsum
ed
a
la
rg
e
am
ou
nt o
f
ti
m
e
(i.e.,
delay
)
[
5
,
12
]
.
The
f
ollow
i
ng
f
our
perform
an
ce
f
act
or
s:
e
ncr
ypt
ion
delay
,
the
secur
it
y
le
vel,
m
essage
delay
,
an
d
processi
ng
powe
r
a
re
m
ai
nl
y
us
e
d
to
m
easur
e
the
VoIP
se
cur
it
y.
T
he
sec
ur
it
y
le
vel
(i.e.
,
c
om
plexity
)
of
the
secu
rity
al
gorithm
seem
s
to
hav
e
a
n
im
pact
on
these
m
ea
su
rem
ent
facto
rs
[
8
,
13
,
14
]
.
Con
st
ru
ct
in
g
ne
w
s
peec
h
e
nc
ryptio
n
al
go
rithm
s
base
d
on
cha
os
t
heory
to
increase
t
he
s
ecur
it
y
le
vel
has
dr
a
w
n
for
m
any
sci
entist
s
and
resea
r
che
r
s
at
te
ntion
[
15
-
21
]
,
but
unf
ort
un
at
el
y
so
m
e
of
these
co
nst
ru
ct
ed
al
go
rithm
s
hav
e
a
nu
m
ber
of
pract
ic
al
pro
blem
s
in
te
rm
s
of
crypt
analy
sis
at
ta
cks
[
22
,
23
]
.
I
n
t
his
pap
e
r,
a
new
dynam
ic
sp
eec
h
enc
r
ypti
on
al
gorithm
us
ing
L
or
e
nz
c
ha
ot
ic
m
ap
ov
e
r
I
P
net
work
is
pro
posed
.
Wh
il
e
g
e
ner
at
in
g
a
cha
otic
key
st
ream
,
the
syst
e
m
par
a
m
et
ers
and
128
-
bit
hash
val
ue
of
the
plain
-
sp
eec
h
are
inc
lud
e
d
in
the
pr
opos
e
d
al
gorithm
to
obt
ai
n a
highes
t security
level.
The
rest o
f
t
his
pa
per
is organi
zed
int
o
5
sect
ion
s
inclu
din
g
the
intr
oductio
n
as
f
ollows:
s
ect
ion
2
ha
s
descr
i
bed
the
pro
posed
al
gorit
hm
based
on
L
or
e
nz
syst
em
.
Sect
ion
3
desc
ribes
sim
ulati
on
r
esult
a
nd
se
cur
it
y
analy
sis,
an
d
e
xp
la
in
s
the
procedu
res
use
d
to
te
st
the
pro
po
s
ed
al
gorith
m
.
In
sect
ion
4
,
a
c
om
par
iso
n
wit
h
existi
ng work i
s g
i
ven an
d fin
al
ly
co
nclusi
on is prese
nted
i
n sec
ti
on
s
5
.
2.
PROP
OSE
D ALGO
RITH
M
In
pr
act
ic
al
,
Lor
e
nz
c
hao
ti
c
m
ap
has
use
fu
l
c
rypt
ogra
ph
ic
prop
e
rtie
s
su
c
h
as
co
nfusi
on
a
nd
diffusi
on,
sens
it
ivit
y
to
init
i
al
con
diti
ons
and
par
am
et
ers,
an
d
un
-
pr
e
dicta
bili
ty
[
20
]
.
Ther
ef
ore,
Lor
e
nz
chao
ti
c
m
ap
is
us
e
d
in
this
pa
per
t
o
pro
po
se
a
new
sym
m
etr
ic
sp
eec
h
enc
r
ypti
on
al
go
rith
m
.
It
con
sist
s
of
tw
o
m
ajo
r
par
ts:
ke
ys
gen
e
rati
on
us
in
g
12
8
-
bit
has
h
value
of
plain
sp
ee
ch
fra
m
e
and
the
lo
gisti
c,
and
e
nc
rypti
ng
the
sp
eec
h
f
ra
m
e
us
ing
c
ha
otic
sequ
e
nces
w
hich
a
re
ge
ne
r
at
ed
by
L
or
e
nz
syst
e
m
.
T
he
overall
bl
ock
di
agr
am
of the
pro
pose
d
al
go
rithm
arch
it
ect
ur
e ca
n b
e il
lustrate
d
by
Fig
ur
e
2.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
24
-
48
34
4826
Figure
2
.
Bl
oc
k diag
ram
Be
low,
we
wi
ll
pr
ovide
a
n
i
ntr
oducto
ry
de
finiti
on
of
the
Lor
e
nz
syst
em
and
analy
ze
it
s
m
ai
n
po
sit
ive
featu
r
es
reg
a
rd
i
ng
t
o
crypto
gr
a
phy.
Then,
the
ste
ps
to
im
ple
m
e
nt
su
c
h
a
proc
ess
of
t
he
enc
r
ypti
on
par
ts
will
b
e
di
scuss. T
he decr
ypti
on
ste
ps
f
or c
om
plete
n
ess w
il
l al
so
be
il
lustrate
d.
2.1.
L
orenz
chaot
ic
s
ystem
Lor
e
nz
syst
em
is
a
dynam
ical
syst
e
m
wh
ic
h
was
st
ud
ie
d
first
by
E
dwar
d
Norto
n
L
or
e
nz
ar
ound
1960
[
24
]
,
T
his
ch
aotic
sy
stem
is
descr
i
bed
by
non
-
li
near
syst
em
of
ordi
nar
y
di
ff
ere
ntial
eq
ua
ti
on
s
(i.e., O
DEs
)
as
g
ive
n
i
n
(
1).
x
̇
=
a
(
y
−
x
)
,
y
̇
=
(
σ
−
z
)
x
−
y
,
z
̇
=
xy
−
bz
.
(
1)
The
real
num
ber
s
a
,
σ
,
b
are
cal
le
d
t
he
co
ntr
ol param
et
ers,
w
he
re
x
,
y
,
z
are v
ariab
le
s
cal
le
d
the
s
ta
te
var
ia
bles,
a
nd
x
̇
,
y
̇
,
an
d
z
̇
are
the
t
i
m
e
der
ivati
ve
s
of
x
,
y
,
an
d
z
.
U
su
al
ly
,
the
L
orenz
syst
em
(
1
)
is
com
m
on
ly
so
lve
d
nu
m
erical
ly
us
ing
R
K45
(Runge
-
K
utta
m
et
ho
ds)
for
gi
ven
c
ont
ro
l
par
am
et
ers,
and
i
niti
al
values
of
the
sta
te
va
riables
(i.e
.,
x
0
,
y
0
,
z
0
).
In
Fig
ure
3
a
cha
otic
at
tract
or
(i.e.
,
L
or
e
nz
butt
erf
ly
)
of
the
dy
nam
ic
a
l
Lor
e
nz sy
stem
(
1) is
pr
e
sente
d.
Figure
3. Cha
ot
ic
att
ractor
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A n
ew
dy
namic
sp
eec
h
e
ncry
pt
ion
algorit
hm
ba
s
ed
on l
or
en
z ch
aoti
c map
(
Obaida M
. Al
-
H
azaime
h)
4827
Fo
r
diff
e
re
nt
va
lues
of
pa
ra
m
et
er
σ
the
dynam
ic
al
syst
e
m
(
1
)
e
xh
i
bits
cha
os
.
By
ha
ving
posit
iv
e
Ly
apun
ov
ex
pone
nts
(i.e.
,
se
ns
it
ivit
y
to
ini
ti
al
conditi
on
s
),
c
ha
os
ca
n
be
rec
ognized
f
or
m
any
dyna
m
ic
al
syst
e
m
s,
and
i
n
par
ti
cular
(
1)
.
In
Fig
ure
4
,
the
m
axi
m
u
m
Ly
apun
ov
ex
pone
nt
of
(
1)
ve
rsu
s
val
ues
of
σ
are
plo
tt
ed. I
n
this
pap
e
r,
we ha
ve
f
ixe
d
a
and
b
to
be
a
=
10
,
b
=
8
/
3
, and
al
lo
w
v
al
ue o
f
σ
to d
y
nam
ic
ally vary
in
the
inte
rv
al
[
28
,
90
]
to
pr
ov
i
de
a
l
arg
e
e
nough
ke
y
-
sp
ace
to
res
ist
brute
-
force
at
ta
cks.
M
or
e
over
,
t
his
will
gu
a
ra
ntee
that
(1)
has
de
ns
e
(
chao
ti
c)
at
tract
or
s
,
wh
ic
h
is
i
m
po
rtant
in
te
r
m
s
of
crypt
ogr
aph
y;
a
ny
c
ha
ng
e
in
init
ia
l
con
diti
ons
will
cause
traject
or
ie
s
to
r
e
m
ai
n
in
the
sa
m
e
at
tract
or
set
,
thu
s
m
aking
it
diff
ic
ult
to
pr
e
dict
any
outc
om
e
without
knowing
the
e
xact
i
niti
al
conditi
ons
of
t
he
syst
e
m
as
well
as
t
he
it
erati
on
c
ounts
i
n
the num
erical
so
luti
on.
Figure
4
.
Maxi
m
u
m
l
ya
pu
nov
e
xpone
nts
2.2.
Ge
nera
ti
on
proce
ss
fo
r
t
he
initial
va
l
ues and
pa
r
am
eters
In
the
ge
ne
rati
on
process,
we
util
iz
e
the
128
-
bit
hash
valu
e
of
the
plain
-
sp
eec
h
fr
am
e
gen
e
rated
by
MD5
ha
sh
f
un
ct
ion
t
o
dyna
m
ic
al
l
y
chang
e
the
init
ia
l
secret
ke
ys
x
0
,
y
0
,
z
0
an
d
σ
0
.
In
the
f
ollow
i
ng
ste
ps,
the g
e
ne
rati
on
process
is d
e
sc
ribe
d
in
the
det
ai
ls.
Algorithm 1: Generation process
Step 1:
Compute a 128
-
bit hash value
K
from the plain speech frame
B
j
, and then
compute four
26
-
bit sequences as the following:
=
digits
(
0
−
floo
r
(
0
)
,
26
)
=
digits
(
0
−
floo
r
(
0
)
,
26
)
=
digits
(
0
−
floo
r
(
0
)
,
26
)
=
digits
(
0
−
floo
r
(
0
)
,
26
)
Step 2:
Set
=
mod
(
co
u
nt
(
,
1
)
,
13
)
.
Step 3:
Set
s
=
x
n
,
y
n
,
z
n
,
1
100
to construct the 128
-
bit string.
Step 4:
Compute
a
new
128
-
bi
t
ha
sh
-
va
lu
e
by
=
rot
L
(
XOR
,
)
,
th
en
di
vi
de
12
8
-
bi
t
ha
s
h
value into four 32
-
bit vectors:
=
1
,
2
,
3
,
4
.
Step 5:
Normalize
the
initial
s
ecret
key
t
o
the
inter
val
(
0
,
1
)
using
f
(
t
)
=
(
arct
an
(
t
)
+
π
/
2
)
/
π
as the following:
=
(
0
)
,
=
(
0
)
,
=
(
0
)
,
σ
=
(
0
)
.
Step 6:
Se
t
=
100
+
co
u
nt
(
,
1
)
,
=
1
,
2
,
3
,
4
to
co
n
st
ru
ct
a
fo
u
r
ch
ao
ti
c
nu
mb
er
s
1
,
2
,
3
,
4
∈
(
0
,
1
)
by
it
er
at
in
g
th
e
lo
gi
st
ic
ma
p
1
,
2
,
3
,
an
d
4
t
im
es
us
in
g
th
e
initial
-
values
,
,
,
σ
, respectively.
Step 7:
Finally,
the
following
f
ormula
is
us
ed
to
comput
e
the
new
fo
ur
chaotic
k
eys
̅
,
̅
,
̅
, and
σ
̅
:
[
̅
̅
̅
]
=
[
0
0
0
]
+
[
1
2
3
]
+
2
−
32
×
[
1
2
3
]
,
σ
̅
=
14
+
15
×
(
4
+
2
−
32
4
)
.
2.3.
Encr
yption proce
ss
In
th
e
enc
ryption
process
,
a
sever
al
it
erati
ons
ove
r
the
plain
sp
eec
h
f
ra
m
e
that
inv
olve
s
a
set
of
op
e
rati
on
su
c
h
as
substi
tut
ion
s
,
pe
rm
uta
ti
on
s,
tra
nsfo
r
m
at
ion
s,
an
d
trans
po
sit
io
ns
are
pe
rform
ed
to
com
plica
te
the
proce
ss
of
de
crypti
ng
c
i
pher
s
peec
h
fr
a
m
e
by
cy
be
r
-
i
ntr
ud
e
rs
(i.e.
,
un
a
utho
rized
a
ccess)
[
25
,
26
]
.
I
n
F
ig
ure
5
the
bl
ock
diag
ram
of
the
pro
po
s
ed
de
c
ryptio
n
process
is
presented
.
The
sp
eec
h
encr
y
ption al
gorithm
p
r
opos
e
d
in
this
pa
per
consi
sts o
f
the
fo
ll
owin
g
se
qu
entia
l st
eps:
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
24
-
48
34
4828
Algorithm 2: Encryption process
Step 1:
Arrange the 20
-
ms speech frame in
=
{
1
,
2
,
…
,
}
of length
=
132
, each element
is
the 8
-
bit representation of the bit 0 (BIT
-
0: FF81) or the bit 1 (BIT
-
1: 007F) of
the coder parameters which is codified on 8 bits.
Step 2:
Set
̅
=
floo
r
(
/
13
)
, then iterate the Lorenz system
(
1
)
using RK
-
45 method for
(
̅
+
1
)
times using the new initial
-
values
(
0
)
=
̅
,
(
0
)
=
̅
, and
(
0
)
=
̅
, and new
control parameter
=
σ
̅
, to get three chaotic sequences
,
, and
. Then
compute the following three sequences:
=
{
1
,
2
,
…
,
̅
+
1
}
,
=
{
1
,
2
,
…
,
̅
+
1
}
,
=
{
1
,
2
,
…
,
̅
+
1
}
for
=
1
,
2
,
…
,
̅
+
1
:
=
digits
(
−
floo
r
(
)
,
26
)
,
=
digits
(
−
floo
r
(
)
,
26
)
,
=
digits
(
−
floo
r
(
)
,
26
)
.
Step 3:
Set
̃
=
̅
if
(
/
13
)
is an integer. Otherwise,
̃
=
̅
+
1
,. Then, partitioning
the
array
into
̃
vectors as
=
1
,
2
,
…
,
̃
where,
=
13
(
−
1
)
+
1
,
…
,
13
,
̃
=
13
̅
+
1
,
…
,
,
=
1
,
2
,
…
,
̅
Step 4:
Construct the sequence
=
1
,
2
,
…
,
̃
with
=
, for
=
1
,
2
,
…
,
̅
, and
̃
=
digits
(
̃
,
length
(
̃
)
)
. Observe that
length
(
)
=
28
+
28
=
56
bit
=
7
×
8
bit
, and
length
(
̃
)
=
(
−
13
̅
)
×
8
bit
.
Ste
p 5:
Construct rotations
=
co
u
nt
(
,
1
)
,
=
1
,
2
,
…
,
̃
.
Step 6:
Finally,
the cipher speech frame
is obtained using the formula:
=
rot
L
(
,
)
,
=
1
,
2
,
…
,
̃
.
Figure
5. Dec
r
ypti
on
al
gorith
m
3.
SIMULATI
O
N RESULT
A
ND SEC
U
RIT
Y
ANALY
SIS
This
sect
io
n
di
scusses
the
pr
opos
e
d
s
peech
encr
y
ption
al
gorithm
,
wh
ic
h
sim
ulate
d
on
a
1.6
G
Hz
cor
e
(TM)
i5
,
1
-
TB
hard
di
sk
capaci
ti
es
and
4
-
GB
m
e
m
or
y
with
MA
TL
AB
softw
are.
I
n
this
se
ct
ion
,
as
an
e
xam
ple,
dif
fer
e
nt
s
pee
ch
sam
ples
of
aud
i
o
file
s
are
sel
ect
ed
with
diff
e
re
nt
sam
pling
rate
(i.e.
,
5000
sam
ples
per
se
cond
).
T
he
sim
ula
ti
on
res
ult
was
validat
e
d
by
a
pp
ly
in
g
t
he
sta
ti
sti
cal
te
sts
su
c
h
as
his
togra
m
analy
sis,
co
rrel
at
ion
,
ra
ndom
ness,
and
s
ensiti
vity
to
pro
ve
the
pe
rfor
m
ance
m
e
tric
s
[
22
]
.
Be
low,
sta
nd
a
rd
iz
e
d
te
sts are
giv
e
n.
3.1.
His
togr
am
a
n
al
ys
is te
s
t
To
eval
uate
the
qual
it
y
of
encr
ypte
d
s
pe
ech
sig
nals,
this
analy
sis
is
the
m
os
t
ac
cur
at
e
te
st.
Pr
act
ic
al
ly
,
the
pro
po
se
d
al
gorithm
is
l
ikely
to
encr
ypt
the
plain
s
pe
ech
sig
nals
in
a
rand
om
mann
e
r
(i.e.,
noisy
sig
nals).
Hist
og
ra
m
of
the
in
put
sp
eec
h
si
gnal
s
are
il
lust
rated
in
Fig
ure
6(a)
a
nd
Fig
ure
6(d
)
.
The
c
orres
pondin
g
histo
gr
a
m
of
the
e
nc
rypted
sp
eec
h
sig
nals
usi
ng
the
sec
ret
ke
ys:
x
0
,
y
0
,
z
0
,
σ
0
,
and
128
-
bit
ha
sh
val
ue
of
th
e
sp
eech
sig
na
ls
are
il
lustrate
d
in
Fi
g
ure
6
(b)
an
d
Fi
gure
6(
e
).
T
he
res
ults
in
Fig
ure
6
s
how
s
that
the
histog
ram
s
of
the
encr
y
pted
sp
eec
h
sign
al
s
are
f
ai
rly
un
ifo
rm
distrib
uted
an
d
total
ly
diff
e
re
nt
from
the
plain
sp
e
ech
sig
nals.
In
add
it
io
n,
t
he
resu
lt
in
dicat
es
that
the
pro
po
s
ed
al
gorith
m
can
be
dec
ry
pted
c
orrectl
y wit
h
t
he
c
orrect k
ey
s as ill
us
trat
ed
in
Fi
gure
6(
c
)
a
nd Fig
ure
6(f)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A n
ew
dy
namic
sp
eec
h
e
ncry
pt
ion
algorit
hm
ba
s
ed
on l
or
en
z ch
aoti
c map
(
Obaida M
. Al
-
H
azaime
h)
4829
(
a)
(d)
(
b)
(e)
(
c)
(f)
Figure
6
.
H
ist
ogram
analy
sis
,
(
a)
I
nput s
peec
h
si
gn
a
l
_1
,
(
b)
Encr
y
pted spee
ch
si
gn
al
_1
,
(
c)
Dec
rypted
sp
eec
h
sig
nal
_1
,
(
d)
I
nput s
pe
ech si
gn
a
l
_2
,
(
e)
E
ncr
y
pted
s
peech si
gnal
_2
,
(
f)
D
ecrypte
d
s
peech si
gnal
_2
3.2.
Co
rrel
ati
on
test
The
c
hao
ti
c
syst
e
m
was
fou
nd
to
pro
du
ce
a
st
ron
g
enc
ryptio
n
m
et
ho
d,
w
hi
ch
can
be
eff
ic
ie
ntly
ide
nt
ifie
d
by c
orrel
at
ion
m
et
ho
d
[
23
]
.
Th
e
correl
at
ion
m
et
ho
d i
s co
m
pu
te
d as
(
2)
.
=
c
ov
(
,
)
√
(
)
√
(
)
(
2)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
24
-
48
34
4830
w
he
re
cov
(
,
)
=
1
∑
(
−
(
)
)
(
−
(
)
)
=
1
,
(
)
=
1
∑
=
1
,
(
)
=
1
∑
(
=
1
−
(
)
)
2
,
wh
e
re
N
is
the
num
ber
of
sp
ee
ch
sam
ples.
In
pro
po
se
d
al
gorit
hm
,
the
correla
ti
on
c
oe
ff
ic
ie
nt
cov
(
x
,
y
)
betwee
n
the
pl
ai
n
a
nd
t
he
ci
phere
d
sp
e
ech
s
a
m
ples
are
ta
bula
te
d
in
Ta
ble
1.
The
resu
lt
in
Table
1
sh
ows
t
ha
t
the
pro
po
se
d
s
pe
ech
e
ncr
ypti
on
al
gorithm
is
eff
ic
ie
nt
e
nough,
bec
ause
t
he
correla
ti
on
t
est
has
been sat
isfie
d t
hat wo
uld
resis
t st
at
ist
ic
al
at
tack
s.
Table
1
.
C
orrel
at
ion
test
File Na
m
e
Origin
al
Encry
p
ted
Sa
m
p
le_
1
.wav
0
.98
1
5
3
0
.00
0
9
9
2
Sa
m
p
le_
2
.wav
0
.98
7
2
6
0
.00
0
6
4
2
Sa
m
p
le_
3
.wav
0
.99
1
1
4
0
.00
1
7
6
5
Sa
m
p
le_
4
.wav
0
.96
2
4
1
0
.00
1
3
5
7
3.3.
Rand
om
ness t
e
sts
To
m
easur
e
t
he
stren
gth
of
t
he
e
ncr
y
pted
sp
eec
h
f
ram
es
again
st
crypt
analy
sis
at
ta
ck
s,
we
ha
ve
perform
ed
a
diff
e
ren
t
s
uite
of
sta
ti
sti
ca
l
te
st
s
in
this
sect
ion
know
n
as
D
IEHARD
(i.e.,
18
-
te
sts)
an
d
NI
S
T
(i.e.,
15
-
te
sts)
te
st
su
it
e
[
23
,
27
]
.
For
giv
e
n
s
equ
e
nce,
the
se
su
it
es
are
desi
gn
e
d
m
ai
n
ly
to
m
easur
e
the
qual
it
y
of
the
r
an
dom
ness.
As
can
be
seen
from
T
able
2
an
d
Ta
ble
3,
the
p
-
va
lue
of
eac
h
te
st
ran
ges
from
0.01
to
0.99 (i.e
.,
s
ucc
ess r
a
nge),
whi
ch
m
eans th
e e
ncr
y
pted
se
que
nce is
rand
om
at
the 99%
of c
onfide
nce le
vel.
Table
2
.
DIEH
ARD
t
est
s
su
it
e
Test No.
p
-
v
alu
e
Res
u
lt
1
0
.50
8
9
5
5
Testin
g
pas
sed
2
0
.35
2
1
7
8
Testin
g
pas
sed
3
0
.98
5
1
3
1
Testin
g
pas
sed
4
0
.43
2
2
3
4
Testin
g
pas
sed
5
0
.57
7
1
6
5
Testin
g
pas
sed
6
0
.53
1
1
6
8
Testin
g
pas
sed
7
0
.57
9
9
6
2
Testin
g
pas
sed
8
0
.46
1
1
8
1
Testin
g
pas
sed
9
0
.53
4
2
7
5
Testin
g
pas
sed
10
0
.44
0
0
8
5
Testin
g
pas
sed
11
0
.53
7
7
1
5
Testin
g
pas
sed
12
0
.69
8
6
2
1
Testin
g
pas
sed
13
0
.70
0
9
1
3
Testin
g
pas
sed
14
0
.38
5
7
2
5
Testin
g
pas
sed
15
0
.70
0
9
8
2
Testin
g
pas
sed
16
0
.67
2
7
2
7
Testin
g
pas
sed
17
0
.70
0
1
1
6
Testin
g
pas
sed
18
0
.07
1
1
6
3
Testin
g
pas
sed
Table
3
.
NIST
te
sts suite
Test No.
p
-
v
alu
e
Res
u
lt
1
0
.64
4
5
6
9
Testin
g
pas
sed
2
0
.53
7
2
1
2
Testin
g
pas
sed
3
0
.82
2
9
5
3
Testin
g
pas
sed
4
0
.69
1
6
7
2
Testin
g
pas
sed
5
0
.28
7
1
8
5
Testin
g
pas
sed
6
0
.27
9
0
9
5
Testin
g
pas
sed
7
0
.51
3
4
4
8
Testin
g
pas
sed
8
0
.43
6
1
5
8
Testin
g
pas
sed
9
0
.21
4
3
6
7
Testin
g
pas
sed
10
0
.43
3
0
4
1
Testin
g
pas
sed
11
0
.20
7
4
1
5
Testin
g
pas
sed
12
0
.38
7
1
4
2
Testin
g
pas
sed
13
0
.59
7
8
2
4
Testin
g
pas
sed
14
0
.50
9
7
3
2
Testin
g
pas
sed
15
0
.48
4
7
2
7
Testin
g
pas
sed
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A n
ew
dy
namic
sp
eec
h
e
ncry
pt
ion
algorit
hm
ba
s
ed
on l
or
en
z ch
aoti
c map
(
Obaida M
. Al
-
H
azaime
h)
4831
3.4.
In
f
orm
ati
on
en
tr
opy
Inform
at
ion
en
tro
py
is
us
e
d
us
ua
ll
y
as
a
m
easur
e
of
disor
der,
or
rand
om
ness
in
encry
pted
s
peec
h
sign
al
[
28
]
. T
he
en
tr
opy f
un
ct
ion
(
)
of a s
ource
s
ca
n be c
om
pu
te
d
a
s
(3)
.
(
)
=
∑
(
)
log
2
1
(
)
2
−
1
=
1
(3)
wh
e
re
P
(S
i)
re
pr
ese
nts
Si
pr
obabili
ty
.
The
e
ntr
op
y
va
lue
f
or
the
en
crypte
d
sp
eec
h
sig
nals
sh
ould
i
deall
y
be
8.
In
oth
e
r
wor
ds,
if
the
e
ntr
op
y
value
of
t
he
encr
y
pted
s
pe
ech
sig
nals
is
le
ss
than
8,
th
en
the
predict
abili
t
y
threats
to
the
encr
y
pted
s
pe
ech
sig
nals
are
exists.
In
this
pa
per
,
ENT
te
st
su
it
e
is
us
ed
t
o
cal
culat
e
the
inf
or
m
at
ion
entr
opy
valu
e.
Pr
act
ic
al
ly
,
ENT
is
a
s
uite
com
po
sed
of
t
he
f
ollow
i
ng
three
te
sts:
ent
ropy,
serial
correla
ti
on,
an
d
chi
-
square
c
oeffici
en
t
as
sh
own
in
Table
4.
F
r
om
these
r
es
ults,
the
obta
ine
d
va
lues
are
evaluate
d
to
be
alm
os
t
as
the
theo
reti
cal
valu
e,
w
hich
m
eans
that
the
pr
opos
e
d
s
peech
e
nc
ryptio
n
al
go
rithm
is
eff
ic
ie
nt e
noug
h (i.e.,
secu
re)
upon the
cry
ptanaly
sis at
ta
ck
s (
i.e.
, e
ntr
op
y)
.
Table
4.
E
NT
test
su
it
e
No
.
ENT
test su
ite
Theo
retical
v
alu
e
Ob
tain
ed
valu
e
1
Entro
p
y
Clo
se to
8
7
.99
9
9
8
6
2
Serial
co
rr
elatio
n
Clo
se to
0
0
.00
0
3
8
1
3
Ch
i
-
sq
u
are
co
ef
f
ic
ien
t
Clo
se to
12
7
.5
1
2
7
.5019
3.5.
Ke
y sens
itivity tes
t
To
en
sure
the
secur
it
y
agains
t
br
ute
-
f
or
ce
a
tt
acks,
the
exc
el
le
nt
encr
ypte
d
sp
eec
h
qu
al
i
ty
m
us
t
be
sensiti
ve
in
e
xt
rem
e
way
wit
h
res
pect
to
th
e
secret
keys
[
24
,
29
]
.
I
n
the
propose
d
al
gorithm
,
any
sli
gh
tl
y
change (
i.e
.,
bi
t
flip)
on the fo
ll
ow
in
g
sensiti
vity
f
act
or
s: ha
sh
v
al
ues,
keys, an
d othe
r
enc
ryptio
n
keys such as
par
am
et
ers
that w
ou
l
d gen
e
rat
e a co
m
plete
ly
d
iffe
re
nt d
ec
ry
ption res
ults as
shown i
n
Fi
gu
re
7.
(
a)
(
b)
(
c)
(
d)
Figure
7. Key
sensiti
vity
,
(a
)
Inp
ut sp
eec
h si
gn
al
,
(
b)
Dec
ry
pted speec
h
si
gnal
w
it
h co
rr
e
c
t key
,
(c)
Dec
rypted
sp
eec
h
sig
nal
with inc
orrect
key h
a
sh val
ue
,
(
d)
Dec
rypte
d speec
h
si
gn
al
with inc
orrect
x
0
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
24
-
48
34
4832
4.
COMP
AR
I
S
ON WIT
H
E
X
I
STI
NG W
ORKS
Ther
e
are
seve
ral
ty
pes
of
del
ay
in
the
c
urre
nt
IP
te
le
phon
y
netw
orks
s
uc
h
as
processi
ng
(i.e
.,
c
oder
,
decode
r,
dig
it
al
sign
al
proc
essing),
se
rial
iz
at
ion
,
qu
e
uing,
propagati
on
,
an
d
net
wor
k
delay
as
sho
wn
i
n
Fig
ure
8.
So
m
e
of
them
are
fixed
w
hile
othe
rs
are
va
riable
as
il
lustrate
d
i
n
Table
5.
IP
t
el
ephon
y
ap
pli
cat
ion
s
are
delay
se
ns
i
ti
ve.
T
her
e
fore
,
the
IT
U
-
T
ha
s
rec
omm
end
e
d
that
t
he
m
axi
m
u
m
threshold
of
t
he
delay
tim
e
(i.e.,
∑
delay
so
urces
)
is
15
0
m
s
[
2
,
5
,
6
]
.
The
ada
ptive
m
ul
ti
-
rate
wideb
an
d
(
AMR
-
WB
G.722.
2)
sp
ee
c
h
cod
ec
is
us
e
d
i
n
this
pa
pe
r
to
colle
ct
N
s
pee
ch
sam
ples
as
on
e
f
ram
e
becau
se
it
s
upport
s
the
GS
M
fea
tures
su
c
h
as
disc
on
t
inuous t
ra
ns
m
i
ssion, c
om
fo
rt
no
ise
g
e
ner
at
i
on, a
nd
vo
ic
e a
c
ti
vity
d
et
ect
ion
[
30
]
.
Figure
8
.
Pac
ke
t flo
w
Table
5.
Delay
sour
ce
Delay s
o
u
rce
Fo
r
m
u
la
Descripti
o
n
Delay type
Co
d
er
=
(
2
+
1
)
∗
+
ℎ
Co
llect the sp
eech
sa
m
p
l
es as
f
ra
m
es.
Fix
ed
Decod
er
=
2
Play
b
ack o
f
the co
d
er
p
rocess
.
Fix
ed
Serialization
=
Ti
m
e
to
clo
ck
a
sp
eech
f
ra
m
e
on
to
a
n
etwo
rk in
terface
.
Fix
ed
Qu
eu
in
g
=
(
+
5120
)
Ti
m
e
the v
o
ice pac
k
et sp
en
d
s in
rou
ter.
Variable
Netwo
rk
=
∗
(
+
)
Ti
m
e
the v
o
ice pac
k
et sp
en
d
s to
go
f
ro
m
on
e po
in
t to an
o
th
er
.
Variable
Prop
ag
atio
n
C
au
sed
by
th
e ligh
t sp
eed in
cop
p
er
-
b
ased
netwo
rks
(
i.e
.
,
1
2
5
,00
0
m
iles
per secon
d
),
or in
f
ib
er
(i.
e.
,
1
8
6
,00
0
m
iles
per secon
d
)
Fix
ed
Secu
rity
C
au
sed
by
th
e encr
y
p
tio
n
and
decry
p
t
io
n
pro
cess
es.
Fix
ed
A
pract
ic
al
co
m
par
ison
on
th
e
aver
a
ge
tim
e
delay
betwee
n
adv
a
nced
e
ncry
ption
sta
nda
r
d
al
gorithm
and
pro
posed
a
lgorit
hm
in
the
sam
e
env
iro
nm
ent
and
c
ondi
ti
on
s
is
m
ade
in
this
sect
io
n.
In
e
nd
-
to
-
e
nd
delay
,
the
i
m
ple
m
ent
at
ion
res
ult
shows
that
the
a
ver
a
ge
tim
e
delay
to
encr
ypt
the
sp
eec
h
dat
a
us
in
g
the
propose
d
al
gorithm
is
2.287
701494
m
s
and
the
a
ver
a
ge
tim
e
delay
to
encr
ypt
the
sp
eec
h
data
us
in
g
AES
is
2.783
276282
4
m
s
as
illustrated
in
Table
6
and
Ta
ble
7
re
s
pecti
vely
.
In
ot
her
words,
a
colum
n
char
t
showi
ng
the AES al
gorithm
is slow
er t
han the
pro
pos
ed
s
peec
h
e
ncry
ption
al
gorith
m
is pr
ese
nted
in Fig
ure
9.
Table
6
.
E
nd
-
to
-
e
nd
delay
ti
m
e fo
r
f
ast
Eth
ern
et
us
in
g
t
he pr
opos
e
d
s
peec
h
e
ncr
y
ption al
gorithm
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A n
ew
dy
namic
sp
eec
h
e
ncry
pt
ion
algorit
hm
ba
s
ed
on l
or
en
z ch
aoti
c map
(
Obaida M
. Al
-
H
azaime
h)
4833
Table
7
.
E
nd
-
to
-
e
nd
delay
ti
m
e fo
r
f
ast
e
t
he
rn
et
us
in
g AE
S alg
or
it
hm
Figure
9. A
verage
delay
tim
e
5.
CONCL
US
I
O
N
In
this
pa
per
we
ha
ve
achie
ved
our
obj
ect
ive
of
desi
gnin
g
a
dynam
ic
sp
eech
e
ncr
y
ption
al
gorit
h
m
base
d
on
L
orenz
cha
otic
m
ap
over
i
nter
ne
t
pr
ot
oc
ol
to
enh
a
nce
the
real
-
ti
m
e
serv
ic
es.
The
pro
pose
d
al
gorithm
was
div
ide
d
int
o
two
processes:
key
gen
e
rati
on
proc
ess
us
i
ng
128
-
bit
has
h
value,
a
nd
enc
ryptio
n
and
decr
y
ptio
n
proce
ss
us
i
ng
Lo
re
nz
syst
e
m
.
To
evalua
te
the
secur
it
y
le
vel
of
the
pro
po
se
d
al
go
rithm
,
eff
ic
ie
nt
sim
ulati
on
s
a
nd
im
p
lem
entat
ion
s
a
nd
analy
sis
ha
ve
been
c
ondu
ct
ed.
T
he
sec
uri
ty
le
vel
is
increase
d
b
y
us
in
g
a
m
ulti
key
with
ever
y
new
sp
eec
h
pack
et
an
d
inc
reasin
g
a
key
l
eng
t
h
w
hich
pr
otect
s
the
propose
d
al
gorithm
agai
ns
t
va
rio
us
cry
ptanaly
sis
at
ta
cks.
In
a
ddit
ion
,
a
c
om
par
iso
n
bet
ween
t
he
aver
a
ge
ti
m
e
d
el
ay
in
the
propose
d
a
lgorit
hm
and
so
m
e
of
the
exi
sti
ng
encr
y
ption
al
gorit
hm
s
is
m
ade.
The
resu
lt
s
en
dorse
that
the
propose
d
al
gorithm
achieved
a
highe
r
secu
rity
le
vel
with
a
l
ow
es
t
aver
a
ge
dela
y
tim
e
and
it
is
a
n
excell
ent c
ho
ic
e f
or
voic
e co
m
m
un
ic
at
ion
i
n pr
act
ic
al
app
li
cat
ion
s in
t
he Inter
net.
REFER
E
NCE
S
[1]
B.
Goode
,
“
Voic
e
over
in
te
rne
t
p
rotoc
ol
(VoIP
),
”
Proce
ed
ings o
f
t
he
IE
EE,
vol
.
90
,
no
.
9
,
pp
.
1495
-
1517,
2002
.
[2]
H.
Oouch,
e
t
al
.
,
“
Stud
y
on
appr
opria
t
e
voice
da
ta
l
engt
h
of
IP
p
ac
ke
ts
for
VoIP
net
work
adj
ustm
ent
,
”
in
Global
Tele
communic
a
t
ions
Confe
ren
ce,
2002.
GL
OBECOM'02.
IEEE
,
v
ol.
2
,
pp
.
1618
-
1
622
,
2002
.
[3]
R.
Dantu
,
et al
.
,
“
Iss
ues
and
cha
l
l
enge
s in
sec
ur
in
g
VoIP
,
”
C
ompu
te
rs
&
S
ec
urity
,
vol.
28
,
no
.
8
,
pp
.
743
-
753
,
2009
.
[4]
O.
M.
Al
-
Ha
zai
m
eh,
“
Inc
re
ase
t
he
sec
ur
ity
le
v
el
for
r
ea
l
-
ti
m
e
ap
pli
c
at
ion
usi
ng
new
ke
y
m
an
ag
ement
solut
ion,
”
Inte
rnational
Jo
urnal
of
Comput
er
Scienc
e
Iss
ue
s (
IJCSI
)
,
vol.
9
,
no.
3,
p
p.
240
-
2
46
,
2012
.
[5]
S.
Na
and
S.
Yo
o,
“
Allowabl
e
pr
opaga
t
ion
de
lay
for
Vo
IP
ca
l
ls
o
f
acce
pt
able
qua
li
t
y
,
”
in
Proc
ee
d
ings
of
th
e
F
irst
Inte
rnational
W
orkshop on
Ad
va
nce
d
In
te
rne
t
Se
rvic
es
and
App
licati
ons
,
pp.
47
-
5
6
,
2002
.
[6]
O.
M.
Al
-
Haz
aim
eh,
“
New
cry
p
togra
phi
c
al
gori
t
hm
s
for
en
hanc
i
ng
sec
urity
of
voic
e
da
ta,
”
The
sis
,
Univer
siti
Utar
a
Malay
s
ia,
2010
.
[7]
M.
Za
ndi
,
et
al
.
,
“
Overv
ie
w
of
sec
urity
issues
of
VoIP
,
”
in
Proceedi
ngs
of
the
Thi
rd
IASTED
Euro
pean
Confe
renc
e
on
Inte
rne
t
and
Mult
imedi
a
Syst
ems and
Applica
ti
ons
,
pp
.
254
-
25
9
,
2007
.
[8]
J.
Seedor
f
,
“
Sec
urity
cha
l
le
ng
es
for
peer
-
to
-
pe
er S
IP,
”
IEEE
N
etwor
k,
vol
.
20
,
no
.
5
,
pp
.
38
-
45
,
2
006.
[9]
O.
M.
Al
-
Ha
za
i
m
eh
,
“
Com
bini
ng
audi
o
sam
ple
s a
nd
image
fra
m
e
s
for
enha
n
ci
ng
vide
o
se
cur
ity
,
”
Indian
Journal
o
f
Sci
en
ce and
Tec
hnology
,
vol
.
8
,
no.
10
,
p
.
940
,
2
015.
[10]
E.
Sch
ae
f
er,
“
An
int
roduc
ti
on
to
cr
y
ptogr
aph
y
and
Cr
y
p
ta
n
a
l
y
sis,
”
C
al
iforn
i
a'
s
Si
li
con
Val
l
e
y
:
San
ta
C
la
ra
Univ
ersity
,
pp.
4
-
5,
2009
.
[11]
M.
Mishra
and
V.
H.
Manka
r
,
“
Revi
ew
on
c
haotic
seque
n
ce
s
bas
ed
cr
y
ptogr
aph
y
and
cr
y
pt
anal
y
s
is,
”
Int
ernati
ona
l
Journal
of
Elec
t
ronics
Engi
n
ee
ri
ng,
vol
.
3
,
no
.
2
,
pp.
189
-
194
,
20
11.
Evaluation Warning : The document was created with Spire.PDF for Python.