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.
3
,
June
202
0
,
pp. 3
125
~
3135
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
3
.
pp3125
-
31
35
3125
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
New mod
ificatio
n on f
eistel
DES
algorith
m based on
mu
lti
-
level keys
Suh
ad
M
u
haj
er
Ka
re
em
1
,
A
bdul
Monem
S. Rahma
2
1
Depa
rtment of
Com
pute
r
Scie
n
c
e, Unive
rsi
t
y
of
Basra
h,
Ir
aq
2
Depa
rtment of
Com
pute
r
Scie
n
ce
,
Univer
si
t
y
of
T
e
chnol
og
y
,
Ir
a
q
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
J
un
7
, 2019
Re
vised
Sep
2
8
,
2019
Accepte
d
Dec
6,
2019
The
data
en
cr
y
p
ti
on
standa
rd
(D
ES)
is
one
of
th
e
m
ost
comm
on
s
y
m
m
et
ric
enc
r
y
pt
ion
a
lgo
rit
hm
s,
but
it
e
xper
ie
n
ce
s
m
an
y
prob
le
m
s.
Fo
r
exa
m
ple,
it
uses
onl
y
one
func
ti
on
(
XO
R)
in
the
enc
r
y
p
ti
on
pr
oce
ss
,
an
d
the
combina
ti
on
of
dat
a
is
finite
bec
ause
it
o
cc
ur
s
onl
y
twic
e
and
oper
ates
on
bit
s.
Thi
s
paper
pre
sents
a
new
m
odifi
ca
t
ion
of
the
DES
to
over
come
the
s
e
proble
m
s.
Thi
s
coul
d
be
done
t
hrough
addi
ng
a
new
l
eve
l
of
se
cur
i
t
y
b
y
inc
re
asing
the
ke
y
spa
ce
(usi
ng
thre
e
k
e
y
s)
during
the
16
rounds
of
the
stand
ard
en
cr
y
p
ti
on
al
gor
ithm
and
b
y
rep
l
ac
ing
the
pre
de
fine
d
XO
R
oper
ation
with
a
new
#
oper
at
io
n.
Our
proposed
al
gori
thm
uses
thre
e
k
e
y
s
inste
ad
of
on
e.
The
first
k
e
y
is
th
e
inpu
t
ke
y
used
for
encr
y
pt
ing
an
d
dec
r
y
pt
ing
oper
a
ti
ons.
The
se
con
d
ke
y
is
used
for
det
ermining
th
e
num
ber
of
bit
s,
whi
le
the
th
ird
ke
y
is
used
for
det
ermining
t
he
ta
ble
num
ber
s,
which
are
from
0
to
255.
Having
ev
al
ua
ted
the
complexi
t
y
of
ou
r
proposed
al
gor
it
hm
,
the
result
s
show
tha
t
it
is
the
m
ost
complex
co
m
par
ed
with
the
well
-
known
DES
and
othe
r
m
odifi
ed
al
g
orit
hm
s.
Consequent
l
y
,
in
ou
r
proposed
al
gorit
hm
,
the
attac
k
er
tr
y
a
num
ber
of
attempts
2
1173
at
m
ini
m
um
to
dec
r
y
pt
the
m
essage
.
T
his
m
ea
ns
that
th
e
proposed
D
ES
al
gor
it
hm
will
inc
r
ea
s
e
the
se
cur
ity
l
evel
of the
well
-
kno
wn DES
.
Ke
yw
or
d
s
:
Des
Inform
at
ion
secur
it
y enc
rypti
on
Mult
i
-
le
vel k
e
ys
Sy
m
m
e
tric
b
lo
ck
ci
phe
r
Copyright
©
202
0
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
ser
v
ed
.
Corres
pond
in
g
Aut
h
or
:
Suhad M
uha
j
er
Ka
reem
,
Dep
a
rtm
ent o
f C
om
pu
te
r
sci
e
nce
,
Un
i
ver
sit
y o
f B
asrah
,
Ba
srah, Ira
q.
Em
a
il
:
su
had_
al
thaher@y
ah
oo.c
om
1.
INTROD
U
CTION
Nowa
days,
the
rap
id
gr
ow
t
h
of
el
ect
ronic
da
ta
transf
er
re
d
via
un
se
cu
re
ne
tworks
m
akes
in
form
ation
secur
it
y
on
e
of
the
m
os
t
im
portant
chall
e
ng
e
s
.
Th
us
,
it
is
necessa
ry
to
a
pp
ly
e
ncry
ption
al
gorith
m
s
on
el
ect
ro
nic
data
to
pr
otec
t
the
data
from
a
tt
a
cker
s
[
1,
2].
More
ov
e
r,
in
a
ny
encr
y
ption
syst
e
m
sh
ared
by
two
par
ti
es,
t
he
se
nder
us
e
s
a
n
e
nc
rypt
io
n
al
gorit
hm
to
conve
rt
the
or
i
gin
al
te
xt
into
ci
ph
e
rtext
,
w
hile
the
rec
ei
ver
us
es t
he reve
rs
e proces
s fo
r
re
t
rievin
g
the
ori
gin
al
text
[3,
4].
Gen
e
rall
y,
al
gorithm
ic
encr
y
ption
is
cl
assif
ie
d
int
o
tw
o
c
at
egories:
symm
et
ric
-
key
an
d
as
ym
metri
c
-
key
encr
ypti
o
n.
Sy
m
m
et
ric
alg
ori
thm
s
are
of
two
ty
pes:
bl
ock
ci
pher
s
and
stream
ci
ph
ers
.
Stream
c
iph
e
r
s
encr
y
pt
each
bi
t
ind
ivid
ually
by
ad
ding
a
bit
fr
om
a
key
stream
to
a
plainte
xt
bit.
Bl
oc
k
ci
pher
s
enc
r
ypt
an
entire
blo
c
k
of
plainte
xt
bit
s
at
a
tim
e
wi
th
the
sam
e
k
ey
.
Bl
ock
ci
phers
c
ou
l
d
be
based
on
the
Feist
el
netw
ork,
s
uc
h
as
the
data
encr
y
ption
sta
nd
a
r
d
(
DES)
and
a
non
-
Fei
ste
l
network,
su
c
h
as
an
a
dvan
ce
encr
y
ption sy
stem
(
AES)
[5
-
7]
.
The
Feist
el
network
wa
s
in
ve
nted
by
Horst
Feist
el
and
has
bee
n
us
ed
in
m
any
bl
oc
k
ci
ph
e
r
desig
ns.
It
can
be
def
i
ned
as
a
c
omm
on
m
e
tho
d
of
c
onve
rting
a
f
un
ct
io
n,
wh
i
ch
is
usual
ly
cal
le
d
an
F
f
unct
ion,
into
a
pe
rm
uta
ti
on
.
The
pri
nc
iple
w
ork
of
F
ei
ste
l
networ
ks
is
to
c
om
bin
e
the
nu
m
ber
s
of
r
ounds
to
reit
erat
e
the
sam
e
op
er
at
ion
s,
s
uch
as
bit
-
shuf
fling
(
us
in
g
bo
xes
f
or
per
m
utati
on
s
cal
le
d
P
-
boxe
s),
sim
ple
no
n
-
li
near
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
.
3
,
J
une
2020
:
3
1
2
5
-
3
1
3
5
3126
functi
ons
(s
ubsti
tuti
on
boxe
s
or
S
-
boxe
s),
a
nd
a
l
ogic
al
operati
on,
wh
ic
h
is
pr
e
def
i
ned
as
XO
R
(u
si
ng tw
o
sta
t
es,
0 or
1) [8
-
10]
.
In
t
his
pa
pe
r,
we
f
oc
us
on
t
he
Feist
el
ne
t
work
ap
plied
in
the
DES
al
gorithm
.
It
is
the
earli
est
sy
m
m
e
tric
encr
ypti
on
al
gorithm
dev
el
op
e
d
by
IBM
in
1972
an
d
was
adopted
in
1977
as
the
Fe
der
al
Inform
at
ion
Pr
ocessin
g
Stan
da
rd
by the N
at
ion
al
Bu
reau
of Stand
a
rd
s
. The DES
alg
or
it
hm
is b
est
su
it
a
ble fo
r
i
m
ple
m
entat
io
n
in
ha
rdwa
r
e.
Conve
rsely
,
it
te
nd
s
to
dem
on
strat
e
slow
im
ple
m
entat
ion
in
so
ft
war
e
.
The
al
gorithm
acce
pts
64
-
bit
keys,
wh
e
re
on
ly
56
bits
ha
ve
previ
ou
sly
been
use
d
be
cause
the
rem
ai
nin
g
8 bit
s w
e
re
us
e
d for
e
rror dete
ct
ion
pur
po
se
s
[11
-
13]
.
The
pri
nciple
work
of
the
D
ES
al
go
rithm
c
an
be
s
umm
ari
sed
as
f
ollo
ws:
input
t
he
plain
te
xt
(
64
bits
)
of
the
al
gorith
m
,
wh
ic
h
is
processe
d
with
a
n
init
ia
l
per
m
utati
on
,
then
16
ro
un
ds
of
the
key
and
plai
ntext
are
app
li
ed
,
an
d
th
e
inv
erse
per
m
utati
on
is
the
f
inal
ste
p
in
the
al
go
rithm
.
The
structure
of
t
he
DES
al
go
rithm
is
base
d
on
t
he
F
ei
ste
l
network,
w
hich
div
i
des
the
in
put
plai
ntext
(64
bits)
blo
c
k
int
o
tw
o
halve
s:
le
ft
(
32
b
it
s)
and ri
gh
t
(32 b
it
s)
[
14,
15
]
.
The
co
re
w
ork
of
t
he
DES
F
ei
ste
l
is
the
F
funct
io
n,
w
hic
h
is
key
-
de
pe
ndent
an
d
con
si
sts
of
fou
r
ph
a
ses as
f
ollo
ws [16
-
18]
:
-
Ex
pan
si
on
pha
se:
The
32
-
bit
i
nput
w
ord
is
e
xp
a
nded
i
nto
48
bits
by
dupli
cat
ing
a
nd
reor
der
i
ng
t
he
bits
of
the wo
rd.
-
Key
m
ixing
phase:
T
he
resul
t
word
from
the
pr
e
vious
phase
us
es
X
O
R
with
a
r
ound
key
co
ns
tr
uc
te
d
by selec
ti
ng 48
b
it
s fr
om
the 56
-
bit key,
and i
n
eac
h rou
nd, a
d
if
fer
e
nt selec
ti
on
of b
it
s is
use
d.
-
Substi
tuti
on ph
ase:
Th
e
ste
p u
ses eig
ht S
-
bo
xe
s to
m
ap
the
48
bit f
or
pro
du
ci
ng
32
new b
i
t.
-
Perm
utati
on
phase:
The
32
-
bit
resu
lt
f
ro
m
the
S
-
boxes
i
s
reord
e
re
d
ac
cordin
g
to
a
f
ixed
per
m
utatio
n
cho
ic
e ta
ble.
In
t
he
undo
ste
p,
the
m
od
ifie
d
ri
gh
t
blo
ck
the
n
use
s
XO
R
operati
on
with
t
he
le
ft
blo
c
k,
and
t
he
res
ult
f
ro
m
this
ste
p
is
pro
vid
e
d
in
the
ne
xt
rig
ht
bl
ock
.
T
he
un
m
od
i
fied
ri
gh
t
blo
ck
is
f
ed
to
t
he
nex
t
le
ft
blo
ck
reg
i
ste
r.
The
sam
e
process
is
it
erated
sixtee
n
tim
es
fo
r
m
a
king
sixtee
n
r
ounds
of
DE
S
[19].
The w
ork of w
el
l
-
known
D
E
S alg
or
it
hm
in st
eps
ci
te
d
i
n [
16,
20
]
:
An
y
e
ncr
y
ptio
n
al
gorithm
de
pends
on
the
key
as
the
s
ign
ific
a
nt
el
em
ent
can
be
“def
i
ned
a
s
a
“nu
m
eric,
“al
ph
a
num
eric
t
ext
or
s
pecial
sy
m
bo
l
[11,
21]
.
Most
m
od
er
n
crypto
gr
a
ph
ic
al
go
rithm
s
“dep
en
d
on
f
unct
ion
s
“
with
tw
o
“st
at
es
(
0,
1)
f
or
“
encr
y
ption
an
d
dec
ryptio
n.
D
ES,
as
one
of
the
bl
ock
al
go
rithm
s
us
es t
he
cl
assic
al
log
ic
al
ope
r
at
ion
(XOR)
w
hich de
pends o
n
tw
o
sta
te
s: si
m
pl
y (0
, 1
) whi
ch
ha
s se
ver
al
weak
po
i
nts
le
ad
t
o
br
ea
k
t
he
D
ES,
s
uch
as
bein
g
sim
ple
wh
e
re
it
can
be
decip
her
e
d
easi
ly
by
at
t
ackers
.
Con
se
quently
researc
hers
ha
ve
at
te
m
pted
t
o
re
place
the
t
w
o
sta
te
s
with
four
“o
nes
(0,
1,
2,
3)
as
s
how
n
in
Figure
1
i
n
the
fo
ll
owin
g
sect
ion
s f
or
inc
rea
sing
k
ey
sp
ace [22].
I
n
this
pa
per,
we
f
oc
us
on
the
w
eak p
oi
nts
of
XO
R
by
re
plac
ing
it
wit
h
a
ne
w
#
ope
rati
on
with
var
ia
ble
bl
ock
bit
siz
es
(
n)
:
(1
or
2
or
4
or
8)
instea
d
of
one
blo
c
k
siz
e.
Ea
ch
blo
c
k
will
gen
e
rate
diff
e
r
ent
sta
te
s
ta
ble
s
base
d
on
a
ddit
ion
in
GF
(
2
n
).
T
he
overall
new
#
op
e
rati
on
is
m
anag
e
d
by
us
in
g
ad
diti
on
al
tw
o
keys.
T
his
w
ork
is
rep
eat
e
d
in
each
r
ound
of
D
ES
to
inc
rease
the
secu
rity
le
vel
of
t
he
al
gorithm
.
Our
re
su
lt
s
sho
w
tha
t
this
ne
w
m
od
ific
at
ion
on
DES
al
gorith
m
wil
l
increase
the
s
ecur
it
y
le
vel
of
the
e
ncr
yp
ti
on
by
incre
asi
ng
the
c
om
plexit
y
in
e
ach
r
ound
an
d
thus
the pr
otect
ion
of enc
rypte
d m
essages
w
il
l be
guara
nteed
.
Figure
1. Tr
ut
h t
ables f
or the
# op
e
rati
on
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
Ne
w modif
ic
ation o
n
fe
ist
el
DES
algorit
hm
base
d on m
ulti
-
le
vel
keys
(
Suhad
M
uhajer
K
are
em
)
3127
2.
RELATE
D
W
ORK
This
sect
ion
presents
the
overv
ie
w
of
the
relat
ed
li
te
ratur
e
on
var
i
ous
m
od
ific
at
ion
s
of
the
DE
S
al
gorithm
and
us
es
the
t
ru
t
h
ta
bles
in
the
key
distri
bu
ti
on.
I
n
20
09
[22],
re
searc
her
s
presente
d
th
e
w
ork
by
com
bin
in
g
the
cu
r
ve
sec
uri
ty
m
et
ho
ds
with
qu
a
ntu
m
crypto
gr
a
phy
c
on
ce
pts
t
o
inc
r
ease
the
s
ecu
rity
and
key
spa
ce
to
m
ake
the
enc
ryp
ti
on
op
e
rati
on
m
or
e
secur
e
an
d
r
obust
.
I
n
t
hi
s
w
ork,
the
pro
po
s
ed
m
od
ific
at
ion
fo
c
us
es
on
t
he
us
e
of
f
our
different
sta
te
s
(
0,
1,
2,
an
d
3)
in
ste
ad
of
two
(0
and
1).
This
is
to
m
ake
var
ia
t
ions
in
the
pola
rise
d
a
ng
le
s
t
hat
hav
e
bee
n
us
e
d
in
t
he
qu
a
nt
um
descr
ipti
on
e
ncode
d
in
these
f
our
ta
bles
in
add
it
io
n
to
the
ou
t
pu
t
de
scriptio
ns
that
ha
ve
use
d
po
la
rised
sta
t
e
an
gles
acc
ordin
g
to
the
ta
bles.
The
n,
m
anipu
l
at
ion
ci
phe
rs
c
onve
rt
the
plainte
xt
int
o
ci
phertext
by
cha
ngin
g
the
act
ual
sta
te
patte
r
n
of
eac
h
char
act
e
r usin
g a l
ogic
al
opera
tor (#).
Th
e
op
erator #
h
a
s the
foll
ow
i
ng f
ig
ure tr
uth
ta
bles.
The
w
ork
of
the
#
operati
on
involves
th
r
ee
in
pu
ts.
T
he
first
inp
ut
re
fer
s
to
the
ta
ble
nu
m
ber,
wh
ic
h
sho
uld
be
us
e
d
to
co
m
pu
te
the
resul
t
a
m
on
g
the
f
our
ta
bles.
T
he
oth
er
tw
o
in
puts
determ
ine
t
he
r
ow
and
c
olu
m
n
nu
m
ber
s in
the g
i
ven
ta
ble to give the
resu
lt
as
a cr
oss p
oin
t.
In
2010, [2
3] re
searche
rs
int
r
oduce
d
a
proposal
f
or
a
ne
w
m
et
h
od
t
o
im
pr
ove
the
perf
or
m
ance
of
the
DES
al
go
rith
m
.
This
i
m
pr
ov
em
ent
is
dem
on
strat
e
d
by
rep
la
ci
ng
the
prede
fine
d
XO
R
ope
rati
on
ap
plied
du
rin
g
the
16
r
ounds
in
the
sta
nd
a
r
d
al
gorithm
Feistel
with
a
new
#
operati
on
t
ha
t
de
pends
on
usi
ng
tw
o
keys.
Each
key
c
onsist
s
of
a
com
bina
ti
on
of
f
our
sta
te
s
(0,
1,
2,
a
nd
3)
instea
d
of
t
he
ordi
nar
y
tw
o
-
sta
te
keys
(0
and
1)
us
in
g
di
ff
ere
nt
tr
uth
t
ables
pro
po
se
d
i
n
[
22
]
Fi
gure
1
.
The
first
key
is
us
e
d
to
det
erm
ine
the
ta
bl
e
nu
m
ber
am
on
g
t
he
four
ta
bles,
and
t
he
seco
nd
key
is
us
e
d
in
the
enc
rypti
on
al
gorithm
.
This
re
placem
ent
ad
ds
a
ne
w
le
vel
of
sec
ur
it
y
to
the
al
gorithm
against
at
ta
c
ker
s
thr
ough
increasin
g
t
he
le
vel
of
c
om
plexit
y.
In
our
pro
po
se
d
m
et
ho
d,
we
operate
on
m
ulti
-
sta
te
s
a
s
com
bin
at
ion
of
0
-
or
1
-
or
4
-
or
8
-
bits
fo
r
re
prese
nting
(0
,
1,
2,
3,
…
255)
wh
il
e
the
pro
pose
d
in
[23]
op
erate
only
2
-
bits
to
rep
rese
nt
(0
,
1,
2,
3)
.
Con
se
qu
e
ntly
,
ou
r
propose
d
m
or
e
rand
om
ness
and
m
or
e
secu
rity
.
In
2017,
[5
]
the
aut
hors
ha
ve
pro
pose
d
a
new
m
od
ific
at
ion
on
DE
S
by
exte
nd
e
d
t
he
sta
ndar
d
bit
siz
e
from
64
–
bit
to
128
-
bit
f
or
both
plainte
xt
an
d
key
siz
e
in
order
to
i
ncr
eas
e
the
secur
it
y
of
al
go
rithm
.
This
is
do
ne
by
doubli
ng
the
s
iz
e
of
ta
bles,
f
un
ct
io
n
an
d
ke
ys.
“
By
”
incre
asi
ng
“
the
”
ov
e
ral
l
“
siz
e
”
in
“
ci
ph
e
r
”
wi
ll
“
m
ade
”
the
“
al
gorithm
”
stron
ge
r
a
gainst
br
ute
-
f
orce at
ta
ck
.
3.
PROP
OSE
D
I
MP
ROVEME
NT OF
THE
DA
T
A
E
NCR
YPTIO
N
ST
A
NDA
RD
The
DES
i
s
c
onside
red
unsec
ur
e
f
or
m
any
app
li
cat
io
ns
f
or
seve
ral
reas
on
s.
It
pr
im
aril
y
dep
e
nds
on
on
ly
a
sin
gle
bit
(
0
or
1).
Si
m
il
arly
,
it
us
es
only
one
functi
on
(
X
O
R),
as
it
does
not
co
ntain
e
nough
rand
om
ness
an
d
is vu
l
ner
a
ble
to
at
ta
cks
.
T
he
refor
e
,
to
over
com
e
these
pro
blem
s,
in
this
s
ect
ion
, w
e
i
ntr
oduce
a
new
m
et
hod
to
m
od
ify
the
DES
t
o
im
pr
ov
e
t
he
e
ncr
y
ption
pe
rf
or
m
ance
an
d
m
ake
the
al
gorithm
m
or
e
com
plex
again
st
at
ta
cks.
Thi
s
is
can
be
ac
hieve
d
by
m
a
king
a
m
od
ific
at
ion
on
a
bina
ry
functi
on
a
nd
key
gen
e
rati
on
us
i
ng
m
ulti
ple
keys
in
each
rou
nd
of
the
sta
ndar
d
DE
S
al
gorithm
instea
d
of
one.
Eac
h
key
is
gen
e
rated
i
nd
e
pende
ntly
.
Th
e
first
key
is
the
input
key
us
ed
for
e
nc
ryptin
g
an
d
de
crypti
ng
oper
at
ion
s
.
The
sec
ond
ke
y
is
gen
er
at
ed
rand
om
l
y
in
bin
ary
f
orm
at
ca
ll
ed
the
key
nu
m
b
er
of
bits
,
wh
ic
h
is
us
ed
t
o
de
te
rm
ine
the
num
ber
of
bits
(b
l
ock
bit
siz
e)
ta
ke
n
f
rom
key
and
m
essage.
T
he
t
hird
key
is
cal
le
d
ke
y
no.
of
table
,
whic
h
is
us
e
d
to
sel
ect
on
e
sta
te
ta
ble
a
m
on
g
differe
nt
sta
te
ta
bles
that
is
us
ed
to
app
ly
the
#
op
erati
on.
This
work
is
done
by
rep
la
c
ing
the
XO
R
f
un
ct
io
n
with
a
new
lo
gical
operati
on
cal
le
d
the
#
operat
i
on.
T
his
#
op
e
r
at
ion
needs
three
in
puts:
the
first
one
sp
eci
fies
th
e
sta
te
ta
ble
num
ber
,
wh
ic
h
s
hould
be
use
d
to
cal
culat
e
the
resu
lt
a
m
on
g
dif
fer
e
nt
sta
te
ta
bles.
The
oth
er
t
wo
inputs
ide
ntify
the
r
ow
a
nd
co
lum
n
nu
m
ber
s
in
the
s
pecifie
d
sta
te
ta
ble
wh
e
re
th
ei
r
cro
s
s
point
giv
es
the
resul
ts.
Howev
e
r,
t
he
num
ber
of
t
ables
with
m
or
e
sta
te
s
are
use
d
in
this w
ork
t
o
in
crease the
r
a
nd
om
ness
in the
al
gorithm
.
In
this
pa
per,
a
new
m
anipu
la
ti
on
of
the
bi
t
pr
ocess
has
been
int
rod
uce
d
beca
us
e
the
well
-
kn
own
DES
al
g
or
it
hm
is
ba
sed
on
X
OR,
w
hich
op
erates
on
ly
on
(0,
1),
w
her
ea
s
the
propose
d
a
lgorit
hm
us
es
a
ne
w
op
e
rati
on
(
#)
, w
hic
h
w
orks
on
dif
fer
e
nt
tr
uth
sta
te
ta
bles.
These
sta
te
ta
bl
es
are
ge
ner
at
ed
in
the
sam
e
m
ann
e
r
as
f
or
the
ta
bl
es
of
t
he
p
re
vi
ou
s
sect
io
n
s
how
n
i
n
Fig
ur
e
1,
ye
t
wit
h
m
or
e
spa
ces.
Th
ese
ta
bles
a
re
m
ai
nly
const
ru
ct
e
d
ba
sed
on
t
he
a
ddit
ion
operati
on
in
t
he
Galoi
s
fiel
d
GF
(
2
n
)
,
wh
e
re
n
is
th
e
value
de
pend
on
the
bl
ock
bit
siz
e
that
sp
eci
fy
by
the
key
number
of
bits
.
In
ou
r
wor
k,
four
va
r
ia
ble
blo
c
k
bit
siz
e:
1,
2,4
a
nd
8
are
us
e
d,
t
her
e
are
2
-
sta
te
ta
bles
(0
an
d
1)
for
GF
(
2
1
),
4
-
sta
te
ta
bles
(0,
1,
2,
a
nd
3)
f
or
GF
(
2
2
),
16
-
sta
te
ta
ble
s
(0,
1,
…
,
15)
f
or
G
F
(
2
4
)
,
an
d
256
–
sta
te
ta
bl
es
(0
, 1
,
…
,
25
5)
f
or
GF
(
2
8
),
the
sam
ples
of
theses
ta
bles
s
how
n
in
the
ne
xt
se
ct
ion
Ta
bles
1
to
6
.
T
he
fo
ll
ow
i
ng
e
xam
pl
es
il
lustrate
th
e
proce
ss
how
ge
ner
at
e
sta
te
ta
ble
s
base
d on bl
ock b
it
size a
nd se
le
ct
o
ne st
at
e from
the
m
.
Let
,
K
b
=
key
n
umber of
bits
,
and
K
c
=
key
no. of
table
, th
e
n:
-
K
b
= 00011
011010
10
10
… (
ge
ne
rated
r
an
dom
l
y fo
r
en
c
ryptio
n
a
nd
decr
y
ption);
-
At eac
h
r
ound,
take tw
o bit
s f
ro
m
the
K
b
and check i
t
:
-
If
K
b
=
00,
the
n
the
bl
ock
bit
siz
e=
1
a
nd
rec
al
l
2
-
sta
te
ta
bl
es:
K
c
sel
ect
ran
dom
ly
on
e
ta
ble
ei
ther
0
or
1
for
e
ncr
y
pting
and d
ec
ryptin
g.
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
.
3
,
J
une
2020
:
3
1
2
5
-
3
1
3
5
3128
-
If
K
b
=
01,
th
en
th
e
blo
c
k
bit
siz
e=
2
an
d
recall
4
-
sta
te
ta
bles:
K
c
select
rando
m
ly
on
e
ta
ble
am
on
g
(0,
1,
…, 4)
for enc
ryptin
g
a
nd
decr
y
pting
;
-
If
K
b
=
10,
th
en
the
blo
c
k
bi
t
siz
e=
4
and
r
ecal
l
16
-
sta
te
ta
bles:
K
c
sel
ect
ran
dom
ly
one
ta
ble
a
m
on
g
(0,
1,
…, 1
5)
f
or enc
ryptin
g
a
nd d
ec
ryptin
g;
-
If
K
b
=
11,
the
n
the
bl
ock
bit
siz
e=
8
a
nd
re
cal
l
256
-
sta
te
t
ables:
K
c
sel
ect
rando
m
ly
one
ta
ble
am
on
g
(0,
1,
…, 2
55) for e
ncr
y
pting
and d
ec
ryptin
g;
The o
ver
al
l
process
of the
#
op
e
rati
on fo
r
e
ach
rou
nd in
t
he
prop
os
ed
DE
S alg
or
it
hm
as sh
o
w
n
in
Fi
gur
e 2
.
Figure
2.
Proce
ss of the
# op
e
r
at
ion
3.
1
.
C
on
s
truc
tion o
f the s
t
ate t
ab
le
s
This
sect
io
n
s
how
s
am
ples
of
ta
bles
t
hat
c
onstr
ucted
base
d
on
the
ad
diti
on
m
at
he
m
atic
al
op
e
rati
on
in
Galois
Fiel
d
(GF
(2
n
)).
Tables
1,
2,
3
and
T
a
bles
4,
5,
6
represe
nt
the
ad
diti
on
in
GF
(
2
4
)
an
d
G
F
(2
8
)
consecuti
vely
.
Table
1.
Stat
e (
#0)
a
ddit
ion
i
n GF (2
4
)
0
0
0
0
0
0
0
1
0
0
1
0
0
0
1
1
0
1
0
0
0
1
0
1
0
1
1
0
0
1
1
1
1
0
0
0
1
0
0
1
1
0
1
0
1
0
1
1
1
1
0
0
1
1
0
1
1
1
1
0
1
1
1
1
#0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
0000
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
0001
1
1
0
3
2
5
4
7
6
9
8
11
10
13
12
15
14
0010
2
2
3
0
1
6
7
4
5
10
11
8
9
14
15
12
13
0011
3
3
2
1
0
7
6
5
4
11
10
9
8
15
14
13
12
0100
4
4
5
6
7
0
1
2
3
12
13
14
15
8
9
10
11
0101
5
5
4
7
6
1
0
3
2
13
12
15
14
8
9
10
11
0110
6
6
7
4
5
2
3
0
1
14
15
12
13
10
11
8
9
0111
7
7
6
5
4
3
2
1
0
15
14
13
12
11
10
9
8
1000
8
8
9
10
11
12
13
14
15
0
1
2
3
4
5
6
7
1001
9
9
8
11
10
13
12
15
14
1
0
3
2
5
4
7
6
1010
10
10
11
8
9
14
15
12
13
2
3
0
1
6
7
4
5
1011
11
11
10
9
8
15
14
13
12
3
2
1
0
7
6
5
4
1100
12
12
13
14
15
8
9
10
11
4
5
6
7
0
1
2
3
1101
13
13
12
15
14
9
8
11
10
5
4
7
6
1
0
3
2
1110
14
14
15
12
13
10
11
8
9
6
7
4
5
2
3
0
1
1111
15
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
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
Ne
w modif
ic
ation o
n
fe
ist
el
DES
algorit
hm
base
d on m
ulti
-
le
vel
keys
(
Suhad
M
uhajer
K
are
em
)
3129
Table
2.
Stat
e (
#3)
a
ddit
ion
i
n GF (2
4
)
#
3
0
1
2
……
13
14
15
0000
0
3
2
1
……
12
13
14
0001
1
2
3
0
……
15
12
13
0010
2
1
0
3
……
12
15
14
0011
3
0
1
2
13
14
15
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1100
12
13
12
……
0
3
2
1101
13
12
13
14
……
1
2
3
1110
14
15
14
13
……
2
1
0
1
111
15
14
15
12
……
3
0
1
Table
3.
Stat
e (
#15) a
dd
it
io
n
i
n GF (2
4
)
#15
0
1
2
……
13
14
15
0000
0
15
14
13
……
2
1
0
0001
1
2
3
0
……
15
12
13
0010
2
1
0
3
……
12
15
14
0011
3
4
5
6
9
10
11
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1100
12
11
10
9
……
6
5
4
1101
13
14
15
12
……
3
0
1
1110
14
13
12
15
……
0
3
2
1111
15
0
1
2
……
13
14
15
Table
4.
Stat
e (
#0)
a
ddit
ion
i
n GF(2
8
)
0000
0000
0000
0001
0000
0010
0000
0011
…
0001
0010
0001
0011
0011
0100
…
0111
1101
0111
1110
0111
1111
…
1111
1101
1111
1
110
1111
1111
#0
0
1
2
3
…
50
51
52
…
125
126
127
…
253
245
255
0000
0000
0
0
1
2
3
…
50
51
52
…
125
126
127
…
253
254
255
0000
0001
1
1
0
3
2
…
51
50
53
…
124
127
126
…
252
255
254
0000
0010
2
2
3
0
1
…
48
49
54
…
127
124
125
…
255
252
253
0000
0011
3
3
2
1
0
…
49
48
55
…
126
125
124
…
254
253
252
0000
0100
4
4
5
6
7
…
54
55
48
…
121
122
123
…
249
250
251
0000
0101
5
5
4
7
6
…
55
54
49
…
120
123
122
…
148
251
250
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1111
1010
250
250
251
248
249
…
200
201
206
…
135
132
133
…
7
4
5
1111
1011
251
251
250
249
248
…
201
200
207
…
134
133
132
…
6
5
4
1111
1100
252
252
253
254
255
…
206
207
200
…
129
130
131
…
1
2
3
1111
1101
253
253
252
255
253
…
207
206
201
…
128
131
130
…
0
3
2
1111
1110
254
254
255
252
253
…
204
205
202
…
131
128
129
…
3
0
1
1111
1111
255
255
254
251
252
…
205
204
203
…
130
129
128
…
2
1
0
Table
5.
Stat
e (
#170)
addit
ion
in GF (2
8
)
0000
0000
0000
0001
0000
0010
0000
0011
0
000
0100
0000
0101
…………
..
1111
1010
1111
1011
1111
1100
1111
1101
1111
1110
1111
1111
#170
0
1
2
3
4
5
…………
..
250
251
252
253
254
255
0
0
0
0
0
0
0
0
0
170
171
168
169
174
175
…………
..
80
81
86
87
84
85
0
0
0
0
0
0
0
1
1
2
3
0
1
6
7
…………
..
248
249
254
255
252
253
000
0
0
0
1
0
2
1
0
3
2
5
4
…………
..
251
250
253
252
255
254
0
0
0
0
0
0
1
1
3
4
5
6
7
0
1
…………
..
254
255
248
249
250
251
0
0
0
0
0
1
0
0
4
3
2
1
0
7
6
…………
..
249
248
255
254
253
252
0
0
0
0
0
1
0
1
5
6
7
4
5
2
3
…………
..
252
253
250
251
248
249
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
,
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
1
1
1
0
1
0
250
251
250
249
248
255
254
…………
..
1
0
7
6
5
4
1
1
1
1
1
0
1
1
251
250
251
248
249
254
255
…………
..
0
1
6
7
4
5
1
1
1
1
1
1
0
0
252
253
252
255
254
249
248
…………
..
7
6
1
0
3
2
1
1
1
1
1
1
0
1
253
252
253
254
255
248
249
…………
..
6
7
0
1
2
3
1
1
1
1
1
1
1
0
254
255
254
253
252
251
250
…………
..
5
4
3
2
1
0
1
1
1
1
1
1
1
1
255
254
255
252
253
250
251
…………
..
4
5
2
3
0
1
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
.
3
,
J
une
2020
:
3
1
2
5
-
3
1
3
5
3130
Table
6.
Stat
e (
#200)
addit
ion
in GF (2
8
)
0000
0000
0000
0001
0000
0010
0000
0011
0000
0100
0000
0101
…………
..
1111
1
010
1111
1011
1111
1100
1111
1101
1111
1110
1111
1111
#200
0
1
2
3
4
5
…………
..
250
251
252
253
254
255
0
0
0
0
0
0
0
0
0
200
201
202
203
204
205
…………
..
50
51
52
53
54
55
0
0
0
0
0
0
0
1
1
2
3
0
1
6
7
…………
..
248
249
254
255
252
253
0
0
0
0
0
0
1
0
2
1
0
3
2
5
4
…………
..
251
2
50
253
252
255
254
0
0
0
0
0
0
1
1
3
4
5
6
7
0
1
…………
..
254
255
248
249
250
251
0
0
0
0
0
1
0
0
4
3
2
1
0
7
6
…………
..
249
248
255
254
253
252
0
0
0
0
0
1
0
1
5
6
7
4
5
2
3
…………
..
252
253
250
251
248
249
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
1
1
1
0
1
0
250
251
250
249
248
255
254
…………
..
1
0
7
6
5
4
1
1
1
1
1
0
1
1
251
250
251
248
249
254
255
0
1
6
7
4
5
1
1
1
1
1
1
0
0
252
253
252
255
254
249
248
…………
..
7
6
1
0
3
2
1
1
1
1
1
1
0
1
253
252
253
254
255
248
249
…………
..
6
7
0
1
2
3
1
1
1
1
1
1
1
0
254
2
55
254
253
252
251
250
…………
..
5
4
3
2
1
0
1
1
1
1
1
1
1
1
255
254
255
252
253
250
251
…………
..
4
5
2
3
0
1
3.2.
Pr
oposed
data encr
yption
s
t
an
d
ard i
n st
e
ps
This
sect
io
n
pro
poses
the
pr
opos
e
d
DES
a
lgorit
hm
in
st
eps
as
s
how
n
her
e
unde
r.
M
odifie
d
ste
ps
hav
e
b
ee
n hi
gh
li
gh
te
d wit
h t
he
r
e
d
c
olour:
Algorit
h
m
1:
Pr
oposed
DES
usi
ng
m
ulti
-
lev
el key
s
Input:
Plaintext
m
e
s
sa
ge 64
bit
s
and key
64
bits called K.
Output
:
Pr
oduc
e
cipher
block
o
f
64
bits
Be
gin
Step
1
:
Key
is pr
ocessed to
produce
sixte
en (
48
-
bit)
sub_ke
y
s Ki
fr
o
m
K
as
fo
llows:
Ste
p1.
1:
The
8
pari
ty
bits are
rem
o
ved f
ro
m
the ke
y
using the
initial
Per
m
utati
on
table.
Ste
p
1.
2:
Split
K i
n
to
two
halves
:
Ci and
Di.
Ste
p
1.
3:
Each
half
of
the key
is shifted
by
one
or
t
wo bi
ts depen
di
ng
on
the r
ound.
Ste
p
1.
4:
The
halves
are
reco
m
bin
ed and
subj
ect to
co
m
pre
s
sion.
Ste
p
1.
5:
Perm
utati
on
is use
d
to
re
duce the
key
fro
m
56
bit
s to
48
bits.
Step
2:
Generat
e r
ando
m
ly
con
trol
key
32
b
its
in
binary
fo
r
m
called
K
bi
(nu
mber
of
bits)
for
1
6
round.
Step
3:
Us
e the
IP
table to
per
m
ute t
he bit
s of
plaintext
block (
64
bits).
Step
4:
The
block r
es
ult
fr
o
m
S
tep 3
is split int
o
two
32
-
bit
halves
(
L
0
,
R
0
)
left
and r
ight.
For
16
rounds,
co
m
pu
te
L
i
and
R
i
as
foll
o
ws
:
Step
4.
1:
L
i
=
R
i
-
1
Step
4.
2:
R
i
-
1
=
L
i
-
1
#
P
̀
,
where
P
̀
=
f
(
R
i
-
1
,
K
i
)
=
P
(
S
(
E
(
R
i
-
1
)
K
i
))
,
co
m
put
ed as
foll
ows
:
St
ep 4.
2.
1:
Expand
R
i
-
1
=
r
1,
r
2,
.
.
.
,
r
32
fr
om
32
t
o
48
bits
T
←
E
(
R
i
-
1
)).
St
ep 4.
2.
2:
Apply
the XOR oper
a
tion
T
,
T
̀
←
(
T
K
i
)
St
ep 4.
2.
3:
Output
of
Step
4.
2.
2
i
s fed
into
an S
-
box,
which sub
stitute
s key
bits and
reduce
s the 48
-
bit
block back
down t
o
32
bits.
T
̋
←
S
(
T
̀).
St
ep 4.
2.
4:
Output
of
Step
4.
2.
3
i
s subj
ec
t
to
a P
-
box
to
per
m
ut
e,
P
̀
←
(
P
(
T
̋
)).
a.
Co
m
put
e the
operation
#
in
R
i
-
1
=
L
i
-
1#
P
̀
as
foll
ows
:
b.
Split
K
bi
to
two
bits (
N
b
),
then t
es
t
as
foll
ow
s:
where
n
is the
block bi
t
siz
e s
elected fr
o
m
L
i
an
d
P
̀
1.
If
N
b
= 00
,
t
hen n
= 1;
go
to
b;
2.
If
N
b
= 01
,
t
he
n
n
= 2;
go
to
b;
3.
If
N
b
= 10
,
t
hen n
= 4;
go
to
b;
4.
If
N
b
= 11
,
t
hen n
= 8;
go
to
b.
c.
Dependi
ng
on
the pr
evious
st
e
p,
recall
sta
tes
t
ables
u
sing addi
tion
on
GF
(2
n
).
d.
Generate r
ando
m
key
(
K
c
i
)
for
se
lecti
ng
one st
ate table
that
create in
ste
p
(c)
.
e.
Co
m
put
e
R
i
by
apply
ing
th
e
operation
o
n
L
i
-
1
#
P
̀
acc
ording
to
three
inputs
(index
=
nu
m
be
r
of
sta
te t
able, row
=
L
i
-
1
,
and
colu
m
n
=
P
̀
).
f.
The
output
of
st
ep
(d)
as
the
cr
oss
p
oint
betw
e
en
the
row
and
colu
m
n
in
the
s
pecified
st
ate
table to
give t
he
result.
End
for
Step
5:
Exchan
ge f
inal
blocks
L
16
,
R
16
.
Step
6:
Tr
ans
po
se
the r
es
ult
s us
ing
inverse per
m
ut
ation
(
IP
-
1
).
End.
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
Ne
w modif
ic
ation o
n
fe
ist
el
DES
algorit
hm
base
d on m
ulti
-
le
vel
keys
(
Suhad
M
uhajer
K
are
em
)
3131
4.
SIMULATI
O
N
RESU
LT
S
Crypto
gr
a
phy
is
the
sci
ence
that
pr
ovi
des
secu
re
c
omm
un
ic
at
ion
ov
e
r
unsec
ure
cha
nn
el
s
.
The
m
essage
is
enc
rypted
usi
ng
t
he
key
by
app
ly
in
g
m
a
them
a
ti
cal
op
erati
on
s
to
pro
du
ce
the
ci
phe
rtext
.
Con
se
quently
,
with
ou
t
kn
owin
g
th
e
key,
an
at
ta
cker
a
s
a
thi
rd
pa
rty
can
not
cal
cul
at
e
the
m
essage
fro
m
the
ci
ph
e
rtext.
More
ov
e
r,
the
nu
m
ber
of
at
te
m
pts
to
est
i
m
a
te
the
key
by
t
he
at
ta
cker
s
ca
n
be
de
fine
d
as
br
ute
force
at
ta
cks.
As
an
il
lustrati
on,
one
of
the
al
go
rithm
s
that
is
vu
lnera
ble
to
this
ty
pe
of
at
ta
ck
is
the
DE
S
a
lgorit
hm
[2
4,
25
]
.
T
hus,
this
pap
e
r
intr
oduc
es
a
new
m
odific
at
ion
of
the
DES
al
go
rith
m
.
This
is
to
e
nh
a
nc
e
the
secu
rity
le
vel
by
inc
reasi
ng
t
he
key
s
pa
ce
us
in
g
m
ultip
le
co
ntr
ol
ke
ys
to
determ
ine
the
num
ber
of
bits
us
e
d
f
ro
m
the
blo
c
k
to
e
nc
rypt
into
t
he
s
pe
ci
fied
ta
ble.
H
ence,
it
will
be
ver
y
dif
ficult
to
est
im
a
te
the
key
.
This
sect
io
n
presents
th
ree
m
et
rics
evalua
ti
on
(c
om
plexity
,
encr
ypti
on
tim
e,
thro
ug
hout,
N
I
ST
te
sts
and
histo
gr
am
anal
ysi
s)
of
the
pr
opose
d
DES
al
gorithm
as
s
ho
wn
belo
w,
w
he
re
t
he
sim
ulatio
n
of
the
al
go
r
it
h
m
is
done
to
pe
rform
the
evaluatio
n
te
sts
on
I
nt
el
Core
i7
-
85
50U
@
2.00
G
H
z
pr
oce
ssor
usi
ng
Mi
cro
s
oft
visu
al
stud
i
o
c
# 201
7
.
4.1.
Securi
ty
c
omp
le
xity an
aly
sis
We
cal
culat
e
t
he
com
plexity
of
the
pro
pose
d
al
gorithm
by
com
pu
ti
ng
the
nu
m
ber
of
po
ssibil
it
ie
s
of
keys,
w
hich
th
e
at
ta
cker
nee
ds
to
dec
rypt
the
ci
ph
e
r
-
te
xt
with
64
bits
usi
ng
th
ree
key
s
with
four
blo
ck
s
of
(1,
2,
4,
or
8)
bit
siz
e
and
d
i
ff
e
ren
t
sta
te
ta
bles.
First,
we
com
pu
te
the
com
plexity
of
the
well
-
kn
own
DE
S
al
gorithm
us
in
g
a
prede
fine
d
XO
R
bi
nar
y
operati
on
(
0,
1)
;
thus,
c
om
pu
ti
ng
the
num
ber
of
possi
ble
key
s
us
e
d
in the e
nc
ryption an
d dec
rypt
ion
is
calc
ulate
d
as
foll
ows:
2
×
(
2)
8
×
32 ×
2
=
2
×
(2)
8
×
2
5
× 2
=
2
15
(1)
Seco
nd,
wh
e
n
us
i
ng
the
#
operati
on
n
the
m
od
ifie
d
DE
S
al
gorit
hm
[2
3]
with
f
our
sta
te
s
(0,
1,
2,
an
d
3)
a
nd
tw
o
bits
instea
d
of
one
bit,
the
num
ber
of
key
poss
ibil
it
ie
s
us
ed
i
n
the
enc
rypti
on
an
d
decr
y
ption i
s c
om
pu
te
d
as
foll
ow
s:
(2
2
)
16
× (
2
2
)
8
×
2
2
× 32 ×
2 =
2
32
× 2
32
× 2
2
×
2
5
× 2
=
2
72
(2)
Finall
y,
we
co
m
pu
te
the
com
plexity
of
the
pro
po
se
d
al
gor
it
h
m
us
in
g
th
r
ee
keys.
T
he
overall
c
om
plexity
of
our pro
po
s
ed
a
lgorit
hm
is as f
ollows:
((21)
32 × (2
2)16 ×
(24)8
×
(2
8)
4) ×
((21)
32
× (
22)16
×
(2
4)8
× (2
8)4 )8 ×
(2 ×
22 ×
24 ×
28)
×
32 ×
2
=
2128 ×
2102
4 × 2
15 ×
25 ×
21 =
2117
3
(3)
Table
7
s
ummari
ses
the
res
ults
base
d
on
c
om
pu
ti
ng
the
c
om
plexit
y
as
the
com
par
iso
n
betwee
n
ou
r
pro
po
se
d
al
go
rithm
with
the
well
-
kn
own
DES
al
g
ori
thm
and
the
m
od
ifie
d
al
gorithm
cited
in
[23].
The
fi
nd
i
ng
s
s
how
that
ou
r
pro
posed
al
gorithm
is
m
or
e
com
plex
than
the
oth
er
s.
Fig
ure
3
sho
ws
the
s
ecur
it
y
com
plexity
of
pro
po
se
d
DE
S
f
or
16
rou
nd
with
tw
o
al
gor
it
h
m
s
(w
el
l
-
known
DES
a
nd
DES
m
od
ifie
d
with
4
-
sta
te
s
ci
te
d
in
[
23
]
.
T
hese
r
esults
have
show
n
in
Ta
ble.
7
an
d
F
ig
ur
e
3
prov
e
d
that
our
propose
d
al
gorithm
has bee
n
m
or
e
com
plex
tha
n othe
rs.
C
onseq
uen
tl
y,
our pro
po
s
ed
b
ec
om
e strong
e
r
a
gains
t brute
-
force at
ta
cks.
4.2.
Encr
yp
ti
on
time
an
d
throu
ghput
As
a
nothe
r
m
e
tric
f
or
m
easur
ing
the
pe
rform
ance
of
t
he
a
lgorit
hm
,
the
e
ncr
y
ption
ti
m
e
is
com
pu
te
d
by
the
tim
e
req
uire
d
f
or
c
onve
rting
the p
la
in
t
ext
into
an
unr
ecognise
d
f
orm
.
The
thro
ug
hput
m
et
ric
as
app
li
ed
in this c
onte
xt
is cal
culat
ed
as
[26
]
:
Throug
hpu
t
=
plaintext
size
(
in
k
il
obyte
)
t
ota
l
en
crypti
on
ti
me
(
ms
)
⁄
(4)
Fr
om
these
Table
8
an
d
Fig
ure
4
,
the
or
igi
na
l,
m
od
ifie
d,
a
nd
the
pro
po
se
d
DES
al
gorith
m
s
are
equ
ival
ent
in
te
rm
s
of
com
pu
ta
ti
on
ti
m
e.
Howe
ver,
ou
r
pro
pose
d
m
et
hod
offe
rs
m
or
e
eff
ect
iv
e
resu
l
ts
r
el
at
ed
to
the
com
plexity
evaluati
on
a
gainst
at
ta
ck
s,
w
hich
ena
ble
s
our
D
ES
al
gorithm
to
be
m
or
e
diff
ic
ult
f
or
a
n
at
ta
cker
to
r
et
ri
eve th
e
origi
na
l
m
essage.
4.3. Nis
t
t
ests
analysis
The
ou
t
pu
t
of
t
he
enc
ryptio
n
al
gorithm
sh
ould
be
m
or
e
ra
ndom
”
and
un
predict
able.
Se
ve
ral
m
e
tho
ds
exist
for
com
pu
ti
ng
the
ra
ndom
ness,
su
c
h
as
NI
ST
(
“
Nati
on
al
“
In
sti
tute
of
Stan
dard
s
“
and
Tech
nolog
y
),
Dieha
rd
te
sts,
and
Te
stU
01.
In
this
paper
,
we
us
e
15
s
ta
ti
sti
ca
l
te
sts
from
NI
ST
s
ta
ti
sti
ca
l
fo
r
t
est
ing
the
“ra
ndom
ness”
of
t
he
bin
a
ry
”
seq
ue
nces,
as
show
n
in
T
able
9
.
This
a
nd
the
m
od
ifie
d
te
sts
are
cal
cu
la
te
d
ov
e
r
m
ulti
ple
cy
ph
e
r
-
te
xt
pr
od
uced
from
the
well
-
known
DE
S.
T
he
pro
bab
il
it
y
value
(
p
-
value
)
is
set
"
to a
value o
f 0.
01 to
c
onfirm
if th
e
outp
ut is
r
andom
. Th
e a
ve
rag
e
te
sts are
com
pu
te
d
a
nd
li
ste
d
in Ta
ble
9
.
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
.
3
,
J
une
2020
:
3
1
2
5
-
3
1
3
5
3132
If
the
te
st
re
su
l
ts
prov
i
de
a
p
-
value
“asy
m
pto
ti
cal
ly
app
r
oa
chin
g
1,
the
n
the
outp
ut
s
houl
d
ap
pear
t
o
hav
e
c
om
plete
ra
ndom
ness.
A
p
-
value
eq
ua
l
to
ze
ro
sig
nifies
that
t
he
outpu
t
is
non
-
ra
ndom
.
The
pa
ss
sta
tu
s
repre
se
nts
that
the
p
-
val
ue
of
t
hese
te
sts
is
gr
eat
er
t
ha
n
0.0
01
an
d
denotes
the
ou
t
pu
t
is
acce
ptable
(e.
g.
,
offer
s
good
ra
ndom
ness)
.
T
he
p
-
val
ues
of
m
os
t
of
the
te
sts
fro
m
the
pr
opose
d
DE
S
al
gorithm
are
gr
eat
er
tha
n
th
e
p
-
val
ues
of
the
well
-
known
DES
,
as
sho
w
n
in
Ta
ble
9
.
Con
se
quently
,
the
propose
d
DES
is bett
er tha
n t
he ori
gin
al
DE
S in
m
os
t t
est
s.
Table
7.
T
he
r
e
su
lt
s of sec
uri
ty
co
m
plexity
an
al
ysi
s
Alg
o
rith
m
The co
m
p
lex
ity
W
ell
-
k
n
o
wn
DE
S
2
15
=3
2
,768
Mod
if
ied
DE
S [
2
3
]
2
72
=4
,7
2
2
,366,4
8
2
Ou
r
p
rop
o
sed
DE
S
2
1
1
7
3
=1
.28
2
8
7
6
6
8
9
4
6
2
7
9
2
1
7
4
3
7
4
1
1
e
+3
5
3
F
igure
3.
The
s
ecur
it
y com
plexity
an
al
ysi
s
Table
8.
T
he
e
ncr
y
ption t
i
m
e and th
rou
ghput o
f or
igi
nal,
previo
us
ly
m
od
ifie
d,
a
nd pr
opose
d
DES al
gorithm
s
File sizes(k
b
)
Execu
tio
n
ti
m
es (
i
n
m
illiseco
n
d
s)f
o
r:
Origin
al DE
S algo
rith
m
Mod
if
ied
DE
S alg
o
rith
m
in[
1
1
]
Ou
r
p
rop
o
sed
DE
S
algo
rith
m
10
5
5
5
50
13
14
14
100
29
29
30
450
113
113
11
7
Av
erage ti
m
e
40
4
0
.25
41
.5
Thro
u
g
h
p
u
t
3
.81
5
3
.78
8
3
.67
4
Figure
4. The
ov
e
rall
co
m
pari
so
n o
f
the
ex
e
cution t
im
e and
th
r
ou
ghout
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
Ne
w modif
ic
ation o
n
fe
ist
el
DES
algorit
hm
base
d on m
ulti
-
le
vel
keys
(
Suhad
M
uhajer
K
are
em
)
3133
Table
9.
Re
s
ult o
f
Ru
nnin
g NIST
on the
ge
ne
rated
key
by DES a
nd the
p
r
opose
d DE
S
Test no
.
Statistical
Test N
a
m
e
W
ell
-
k
n
o
wn
DE
S
Prop
o
sed
DE
S
P
-
Valu
e
Statu
s
P
-
Valu
e
Statu
s
1
Ap
p
rox
i
m
ate
E
n
tr
o
p
y
0
.62
4
p
ass
0
.91
0
p
ass
2
Blo
ck
Fr
eq
u
en
cy
0
.63
9
p
ass
0
.84
9
p
ass
3
Cu
m
u
lativ
e
Su
m
s
0
.06
8
p
ass
0
.43
0
p
ass
4
FFT
0
.08
2
p
ass
0
.56
2
p
ass
5
Frequ
en
cy
0
.11
6
p
ass
0
.31
0
p
ass
6
Linear c
o
m
p
l
ex
ity
0
.88
4
p
ass
0
.62
3
p
ass
7
Lon
g
est Ru
n
0
.25
p
ass
0
.84
7
p
ass
8
No
n
Overlapp
in
g
T
e
m
p
late
0
.52
7
p
ass
0
.51
7
p
ass
9
Ov
erlapp
i
n
g
T
e
m
p
late
0
.48
0
p
ass
0
.78
7
p
ass
10
Ran
d
o
m
Excu
rsio
n
s
0
.59
1
p
ass
0
.75
7
p
ass
11
Ran
d
o
m
Excu
rsio
n
s Variant
0
.76
1
p
ass
0
.63
0
p
ass
12
Ran
k
0
.43
2
p
ass
0
.43
4
p
ass
13
Ru
n
s
0
.00
1
p
ass
0
.38
8
p
ass
14
Serial
0
.64
9
p
ass
0
.67
0
p
ass
15
Un
iv
ersal
0
.32
6
p
ass
0
.97
3
p
ass
4.4. Hist
ogra
m a
n
al
ys
is
A
histo
gr
am
is
us
ed
to
m
easur
e
the
secur
it
y
of
the
ori
gi
nal,
encr
ypte
d,
an
d
decr
y
pted
im
ages
us
i
ng
the w
el
l
-
know
n
a
nd pr
opos
e
d DES
. T
he
e
xp
erim
ental
r
esults
of
th
ree im
a
ges
a
re sh
own i
n
Fig
ures
5,
6 and 7
.
I
m
ag
e1
:
I
m
ag
e
2:
Figure
5. Re
su
l
ts an
d hist
ogra
m
o
f
well
-
kn
own
DES
an
d pro
pose
d
D
ES
f
or
ori
gin
al
im
age1
and
encr
y
pted
im
a
ge1
Figure
6.
Re
su
l
ts an
d hist
ogra
m
o
f
well
-
kn
own
D
E
S
and pr
opos
e
d DES f
or
:
or
i
gin
al
im
age2
and
encr
y
pted
im
a
ge2
I
m
ag
e3
:
Figure
7.
Re
su
l
ts an
d hist
ogra
m
o
f
well
-
kn
own
D
E
S a
nd
propose
d DES
for ori
gin
al
im
age3 an
d
e
nc
rypted
im
age3
(
a)
Or
ig
in
a
l i
m
a
g
e
1
(
b
)
E
n
cr
y
p
ted
i
m
a
g
e
1
b
y
W
ell
-
k
n
o
w
n
DE
S
(c
)
E
n
cr
y
p
ted
i
m
ag
e
1
b
y
P
r
o
p
o
s
ed
DE
S
(
d
)
His
to
g
r
a
m
o
f
o
r
ig
i
n
a
l
im
a
g
e
1
(
e
)
His
to
g
r
a
m
o
f
E
n
cr
y
p
ted
im
a
g
e
1
b
y
w
ell
-
k
n
o
w
n
DE
S
(
f
)
His
to
g
r
a
m
o
f
E
n
cr
y
p
ted
i
m
a
g
e
1
b
y
p
r
o
p
o
s
ed
DE
S
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
.
3
,
J
une
2020
:
3
1
2
5
-
3
1
3
5
3134
4.5
. Disc
ussio
n t
he
resul
ts
Our
pro
pose
d
DES
us
es
th
re
e
keys
thr
ough
16
rou
nd
s
with
each
rou
nd
usi
ng
th
ree
key
s,
inclu
ding
an
input
key
for
enc
ryptio
n
and
dec
r
ypti
on,
a
second
key
for
deter
m
ining
the
num
ber
of
bl
oc
ks
of
1
-
,
2
-
,
4
-
or
8
-
bits
then
gen
e
r
at
ing
the
nu
m
ber
of
sta
te
ta
bles,
an
d
a
thi
rd
key
to
sel
ect
on
e
sta
te
ta
ble
for
encr
y
ption
a
nd
decr
y
ption.
T
his
ap
proac
h
i
nd
ic
at
es
that
usi
ng
t
hese
co
e
ff
ic
ie
nt
par
am
et
ers
(i.e.
,
th
re
e
keys
with
dynam
ic
blo
c
k
siz
es)
in
each
r
ound
increases
the
com
plexity
of
ou
r
propose
d
DES
against
at
ta
ck.
Wh
e
n
c
om
par
ing
with
a
tripl
e
DES
,
wh
ic
h
works
by
us
i
ng
DE
S
three
ti
m
es,
then
we
us
e
only
three
keys
thr
ough
a
t
otal
of
48
r
ounds
.
More
over
,
our
propose
d
DE
S
m
anipu
la
te
s
the
bits
with
di
ff
ere
nt
sta
te
s
since
the triple DES m
anipu
la
te
s th
e b
it
s o
nly wit
h
two
sta
te
s (0 an
d 1).
So, the co
m
plexity
o
f
t
he
triple
D
ES i
s less
than o
ur prop
ose
d DES
.
Fr
om
the
histog
ram
analy
sis,
the
correla
ti
on
betwee
n
the
pi
xels
in
an
i
m
age
al
lows
us
to
sel
ect
three
diff
e
re
nt
im
ag
es.
I
n
im
age1
,
the
c
orrelat
ion
bet
ween
t
he
pix
el
s
is
high
wh
il
e
be
ing
le
ss
i
n
i
m
age2.
In
im
age3
,
li
tt
le
correla
ti
on
e
xists.
T
he
e
ff
e
ct
of
t
he
dyna
m
ic
blo
ck
siz
e
is
sel
ect
ed
eac
h
t
im
e
fo
r
e
ncry
ption
and
dec
ryptio
n.
As
sho
wn
in
the
previ
ous
fig
ures,
t
he
distrib
ution
of
the
pix
el
s
i
n
our
pro
pose
d
DE
S
is
equ
i
valent
to
a
unif
or
m
distribu
ti
on
s
ugge
sti
ng
that
it
is
stronge
r
com
par
e
d
to
t
he
or
iginal
DE
S
f
or
these
three ty
pes
of i
m
ages.
5.
CONCL
US
I
O
N
The
DES
is
on
e
of
t
he
m
os
t
popula
r
a
nd
earl
ie
st
encr
ypti
on
al
gorithm
s
that
has
bee
n
us
e
d
unti
l
ver
y
recently
.
The
DES
is
con
side
red
ins
e
cur
e
f
or
m
a
ny
app
li
cat
io
ns
due
to
it
s
m
any
weak
ne
sses.
Exam
ples
of
s
uch
wea
knesse
s
inclu
de
t
he
ke
y
le
ng
t
h,
u
sin
g
on
ly
one
f
un
ct
ion
,
co
ntaini
ng
le
ss
rand
om
ness,
et
c.
Th
us,
it
is
necessa
ry
to
in
crease
t
he
sec
uri
ty
of
this
al
gorithm
by
a
dd
i
ng
new
le
vels
of
sec
ur
it
y
to
m
ake
it
m
or
e
secur
e.
An
a
dd
it
io
nal
key
is
add
ed,
an
d
the
ol
d
X
OR
is
rep
la
ced
by
a
new
op
e
rati
on
ca
ll
ed
the
#
op
e
rati
on,
with
m
or
e
truth
ta
bles.
I
n
this
pa
pe
r,
this
change
is
su
ggest
e
d
to
giv
e
m
or
e
stren
gth
to
the
D
ES
al
gorithm
.
This
will
m
ake
it
m
or
e
powerfu
l
against
a
ny
ki
nd
of
s
noopin
g.
Using
m
ulti
-
keys
instea
d
of
on
e
key
inc
reases
the
reli
abili
ty
of
the
ke
y.
Using
a
va
riable
bl
ock
bit
siz
e
in
each
rou
nd
inc
rease
the
secu
rity
of
the
al
gorithm
.
H
oweve
r,
t
his
will
increase
the
ef
fici
ency
of
the
e
ncr
y
pt
ion
a
nd
dec
re
ase
the p
r
oba
bili
ti
e
s o
f
a
br
ea
k
a
ga
inst dif
fer
e
ntial
an
al
ysi
s f
rom
b
ru
te
for
c
e att
acks
. Th
e m
od
ific
at
ion
of
t
he
D
E
S
al
gorithm
add
s
com
plexity
i
n
com
pu
ti
ng
the
key
but
sa
ves
the
ti
m
e
t
aken
i
n
m
at
he
m
at
ic
al
co
m
puta
ti
on
,
as sho
wn in t
he
previ
ou
s
sect
ion
.
REFERE
NCE
S
[1]
P.
Pati
l.,
P.
Nara
y
ank
ar,
DG
.
Nar
a
y
an.,
and
Mee
n
a
S.
,
“
A Co
m
pre
hensive
E
va
lua
t
i
on
of
Cry
p
tographic
Algorit
hm
s:
DES,
3DES,
AE
S,
RS
A
and
Blo
wfish
,
”
Int
ernational
Conf
ere
nce
on
Information
Sec
urit
y
&
Priv
acy
(ICISP
2015)
,
11
-
12
Dec
ember
2015,
Nagpur, I
ndia
,
Proc
edi
a
C
omputer
Scienc
e
,
v
ol
.
78
,
pp
.
617
–
624,
2016
.
[2]
Sh.
Kanda
r
,
Dh
.
Ch
audhur
i
,
A.
B
hat
tacha
je
e
,
B
.
Chandra
,
“
Im
age
enc
r
y
p
ti
on
usi
ng
seque
nce
gen
era
t
ed
b
y
c
y
c
lic
group
,
”
Journal of
Information
S
ec
urit
y
and
Applicat
ions
,
vol
.
44
,
pp.
117
-
129
,
20
19
.
[3]
S.
Rani
and
H.
Kaur
,
“
Techni
cal
Surve
y
on
Cr
yptogra
ph
y
Algor
it
hm
s
for
Ne
two
rk
Secur
i
t
y
,
”
In
t
ernati
onal
Journ
al
of
Ad
vanc
ed
R
ese
arch
in
Comput
er
Scienc
e
and
S
oft
ware
Engi
ne
e
ring
,
vol
.
6
,
no
.
9
,
pp
.
204
-
209
,
Sep
.
2016
.
[4]
A.
M.
Abdulla
h
,
“
Advanc
ed
Encr
y
pt
ion
Standa
rd
(AES)
Algorit
hm
to
Enc
r
y
pt
an
d
Dec
r
y
p
t
Data
,
”
Cryptography
an
d
Net
work
Se
c
urity
,
2017.
[5]
B.
cru
z
and
K.
Dom
ingo,
“
Expa
nded
128
-
b
it
Data
Enc
r
y
pt
ion
Standa
rd
,
”
In
ter
nati
onal
Journ
al
of
Comput
er
Sci
en
ce and
Mo
bil
e
Computing
,
v
ol.
6
,
no
.
6
,
Au
g
.
2017
.
[6]
W
il
li
am
Stalling
s,
“
Cr
y
ptogr
aph
y
and
ne
twork
Secur
i
t
y
:
Princ
ipl
es
and
Prac
ti
c
e
,
”
Pear
son
Educat
i
on/Prent
ice
Hal
l,
5th
Ed
it
ion
,
201
0
.
[7]
M.
A.
Ham
ee
d
,
Ahm
ed
I.
Jabe
r
,
Jam
hoor
M.
Alobai
d
y
,
and
Al
aa A.
Haje
r
,
“
Desig
n
and
Sim
ula
t
io
n
DES
Algorit
h
m
of
Enc
r
y
pti
on
f
or
Inform
at
ion
Secur
ity
,
”
Ame
r
ic
an
Journal
of
Engi
nee
ring
R
ese
arch
(
AJ
ER
)
,
vol.
7,
no
.
4
,
pp.
13
-
22
,
2018
.
[8]
E.
Tha
m
biraja
,
G.
Ramesh,
“
A
Surve
y
on
Vari
o
us
Mos
t
Com
m
o
n
Enc
r
y
p
ti
on
Techni
ques,
”
In
te
r
nati
onal
Journal
of
Ad
vanc
ed
R
ese
arch
in
Comput
er
Scienc
e
and
S
oft
ware
Engi
ne
e
ring
,
vol
.
2
,
no
.
7
,
pp
.
226
-
223
,
J
ul
.
2012
.
[9]
H
N.
Shandi,
A.
S.
Raut,
Sh.
R.
V
i
dhal
e
,
R.
V.
Saw
ant
,
V.
A.
Kotkar
,
“
A
Revi
ew
of
Vari
ous
Enc
r
y
pt
ion
Techni
ques
,
”
Inte
rnational
Jo
urnal
Of
Eng
ineering
And
Computer
Sc
i
ence
,
vol
.
3
,
no
.
9
,
pp
.
80
92
-
8096
,
Sep
.
2
014.
[10]
S.
Alber
m
an
y
,
F.
R
adi
,
“
Surv
ey
:
Bloc
k
ci
ph
er
Methods
,
”
In
te
r
nati
onal
Journal
of
Ad
vancemen
ts
in
R
ese
arch
&
Technol
ogy
,
vo
l. 5, no. 11, pp.
11
-
22,
Nov
.
2016
.
[11]
G.
S
ingh
,
Supriy
a
,
“
A
Stud
y
of
Enc
r
y
pti
on
Alg
orit
hm
s
(RSA
,
DES,
3DES
and
AES)
for
Inform
at
ion
Secur
ity
,
”
Inte
rnational
Jo
u
rnal
of
Comput
er
Applications
,
vol.
67
,
no
.
1
,
pp
.
33
-
38
Apr
.
201
3
.
[12]
S.
D.
Riha
n
,
A.
Kh
alid
,
S.E
.
F
,
Os
m
an,
“
A
Perf
orm
anc
e
Com
pa
rison
of
Enc
r
y
p
t
ion
Algorit
hm
s
AES
and
DES
,
”
Inte
rnational
Jo
urnal
of Engi
ne
e
ring R
ese
arch
&
Technol
og
y
(
IJ
ER
T)
,
vol. 4, no.
12
,
pp.
151
-
154
,
De
c
.
2015.
[13]
M.
Kaur,
N.
Kaur,
B.
Singh
,
“
C
om
par
at
ive
Stud
y
Of
Diffe
ren
t
C
r
y
ptoghr
aphic
Algorit
hm
s
,
”
Inte
r
nati
onal
Journal
of
Ad
vanc
ed
R
ese
arch
in
Comput
er
Scienc
e,
vol.
8,
no
.
4
,
pp
.
352
-
354,
Ma
y
2017.
Evaluation Warning : The document was created with Spire.PDF for Python.