Co
m
pu
ter Sci
ence a
nd Inf
or
mat
i
on
Tec
h
no
lo
gies
Vo
l.
2
, No
.
1
,
Ma
rch
2021
,
pp.
1
6
~
25
IS
S
N:
27
22
-
3221
,
DOI: 10
.11
591
/
csi
t.v
2i1
.p
1
6
-
25
16
Journ
al h
om
e
page
:
http:
//
ia
esprime
.com/i
ndex.
php/csit
An opti
mized
encryption
algorith
m and F
fun
ctio
n
with
Dynami
c sub
stitution f
or cr
eating
S
-
box an
d P
-
box
entries
for Bl
owfish Alg
orithm
Rekha
C, Kri
s
hna
m
urt
hy G
N
Depa
rtment
o
f
C
om
pute
r
Scie
n
ce a
nd
Engi
n
ee
rin
g
,
BNM
Instit
u
te of
T
ec
hno
log
y
,
Benga
luru
,
Karn
at
ak
a, I
ndia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
pr
26
, 20
2
0
Re
vised
Jun
17
, 20
2
0
Accepte
d
J
ul
19
, 2
0
2
0
In
the
fi
el
d
of
cr
y
ptogr
aph
y
,
the
r
e
has
bee
n
a
m
assive
amount
of
e
nhanc
emen
t
in
m
ani
pu
la
t
ing
the
plainte
x
t
whi
ch
is
unr
ea
dab
le,
le
ss
pron
e
to
cr
ac
ker
s
and
hac
ker
s,
aga
in
m
ani
pul
at
ing
th
is
unre
ada
b
le
form
to
ge
t
bac
k
pl
ai
n
t
ext
in
som
e
wa
y
.
Th
e
Blowf
ish
al
go
ri
thm
is
a
bloc
k
ci
ph
er,
has
complex
in
struct
ure
in
gene
ra
ti
ng
P
-
b
ox
and
S
-
box
ent
ri
es
using
enc
r
y
pt
ion
a
l
gorit
hm
.
B
y
sim
pli
f
y
ing
th
e
struct
ure
of
enc
r
y
pt
ion
al
gori
thm
as
wel
l
as
F
fu
nct
ion
with
d
y
nami
c
subs
ti
t
uti
on,
thi
s
ca
n
i
m
prove
the
p
erf
orm
anc
e
b
y
g
e
n
era
t
ing
P
-
box
and
S
-
box
ent
ri
e
s
of
blowfish
algorithm.
In
thi
s
pape
r,
th
e
propo
sed
m
et
hod
sim
pli
fie
s
the
str
uct
ure
to
produ
c
e
P
-
box
and
S
-
b
ox
en
tri
es
in
ord
er
to
red
u
ce
computat
ion
al
cost
and
dem
onstrat
es
the
per
form
anc
e
of
blowfish.
The
appr
oa
ch
c
onside
rs
diffe
r
e
nt
sec
ur
ity
aspe
ct
s
namel
y
EQ
ana
l
y
sis,
KS
ana
l
y
sis,
AV
an
aly
s
is,
Ent
rop
y
,
Float
ing
Freque
nc
y
ana
l
y
sis
and
cor
r
el
a
ti
on
of
horiz
on
ta
l
l
y
a
dja
c
ent
p
ixels i
n
an
en
cr
y
pt
ed
im
age
.
Ke
yw
or
d
s
:
Bl
ow
fis
h
Correl
at
ion
c
oe
ff
ic
ie
nt
Entr
op
y
Floati
ng freq
ue
ncy
P
-
box
S
-
box
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
B
Y
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Re
kh
a
C
,
Dep
a
rtm
ent o
f C
om
pu
te
r
Scie
nce a
nd E
ng
i
ne
erin
g,
BNM I
ns
ti
tute
of Tech
nolo
gy,
Be
ng
al
uru,
Ka
rn
at
a
ka,
India
.
Em
a
il
:
rek
hac
1976
@g
m
ai
l.com
1.
INTROD
U
CTION
Crypto
gr
a
phy
play
s
a
n
im
portant
ro
le
in
ne
twork
sec
ur
it
y
[
1
]
,
that
tra
nsfers
se
ns
it
ive
inf
or
m
at
ion
acro
s
s
i
ns
ecu
re
net
wor
ks
us
i
ng
e
nc
ryptio
n
a
nd
dec
ryptio
n
pr
ocess.
In
cry
pto
gra
ph
y
the
Ke
y
are
co
nf
i
de
ntial
ity,
integrity
,
an
d
a
uth
e
ntica
ti
on
[
2
]
,
[
3
]
.
Crypt
ogra
phic
al
gorithm
is
cat
ego
rised
int
o
tw
o
di
fferent
ty
pes
sy
m
m
e
tric
and
asy
m
m
et
ri
c
key
cry
ptogra
ph
y.
I
n
sym
m
e
tric
key
cry
ptogra
phy,
only
one
key
is
us
e
d
t
o
e
ncr
ypt
a
nd
de
crypt
the
i
nfor
m
at
ion
.
T
he
key
pl
ay
s
a
m
ajor
r
ole
i
n
sym
m
e
t
ric
key
c
rypto
gr
a
phy.
De
pendin
g
on
the
ba
sis
of
op
e
rati
on
sym
m
et
ric
key
cry
ptogra
phy
is
div
ide
d
i
nto
t
wo
ty
pes
stream
ci
ph
e
r
a
nd
blo
c
k
ci
ph
e
r
.
A
bl
ock
ci
phe
r
is
the
one
w
he
re
a
bl
ock
of
pl
ai
ntext
is
c
on
ver
te
d
i
nto
ci
pherte
xt
blo
c
k
of
sam
e
le
ng
t
h
[
4
]
.
O
ne
e
xam
ple
of
sy
m
m
e
tric
bloc
k
ci
pher
is
blowfis
h.
A
blow
Fish
is
a
16
rou
nd
Feist
el
netw
ork
w
hich
op
e
r
at
es
on
plai
ntext
wit
h
64
bit
bl
ocks
c
onve
rted
to
ci
pherte
xt
of
64
bi
t
blo
c
ks
,
us
in
g
a
key
wh
ic
h
is
ra
ng
i
ng
f
r
om
32
bits
to
448
bits
us
e
d
in
the
e
nc
ryptio
n
a
n
d
decr
y
ption
of
plainte
xt.
Bl
owfis
h
al
gorith
m
includes
tw
o
proce
dures:
a
key
-
exp
a
ns
i
on
pr
oc
edure
a
nd
a
dat
a
-
enc
ryptio
n
procedu
re.
Data
encr
y
ption
f
un
ct
ion
ta
ke
plac
e
via
16
rou
nd
Fiest
el
netw
ork
as
sho
wn
in
Fig
ur
e
1,
each
r
ound
ha
ving
per
m
utati
on
a
nd
s
ubst
it
ution,
us
in
g
F
-
functi
on
as
s
ho
wn
in
Figure
2,
with
key
dep
e
ndent.
All
op
e
rati
on
pe
rfor
m
ed
are
X
ORed
an
d
ad
di
ti
on
s
on
32
-
bit
words
a
nd
a
dd
i
ti
on
al
op
e
rati
ons
are
four
in
de
xe
d
a
r
ray
data
l
ooku
ps
pe
r
r
ound.
Key
e
xp
a
ns
i
on
pr
ocedure
c
on
ver
ts
44
8
-
bit
ke
y
into
f
ew
sub
key ar
r
ay
s o
f
41
68 b
y
te
s
[4
]
,
[
5
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t. Sci.
I
nf. Tec
hnol.
An o
ptimized e
ncry
ption al
gorit
hm
and
F fu
nction w
it
h
Dy
namic s
ubsti
tuti
on
f
or
cre
atin
g
…
(
Rekh
a
C
)
17
The
key
ex
pansi
on
proce
dure
us
es
a
key
to
gen
e
rati
ng
P
-
box
a
nd
S
-
box.
In
it
ia
li
zat
ion
of
18
-
P
ar
ray
us
in
g
key
ta
ke
n
as
P
0
from
P
-
ar
ray
is
XO
R
ed
with
first
32
bits
of
the
ke
y.
P1
XO
Re
d
with
sec
ond
32
bits
of
the
key.
Re
peat
this
cy
cl
e
un
ti
l
al
l
P
-
ar
ray
X
O
Re
d
with
key
bi
ts.
Af
te
r
init
ia
li
zat
ion
,
pass
tw
o
va
lues
of
P
va
lue
(P
0
an
d
P1
as
L0
a
nd
R0
)
to
the
f
unct
ion
Encr
y
pt
as
m
e
ntion
e
d
i
n
Fi
gure
1,
w
hich
ge
ner
at
e
t
wo
di
ff
e
ren
t
encr
y
pted
key
values
L
17
a
nd
R
17.
The
ou
t
pu
t
L
17
a
nd
R
17
of
en
crypti
on
f
unct
io
n,
is
then
co
pied
to
P0
an
d
P1
.
Repeat
t
his step un
ti
l al
l 1
8
-
P
v
al
ue
e
ntri
es are
ge
ner
at
e
d
c
on
ti
nu
ously
in or
der by
re
placi
ng the
outpu
t.
In
the
sam
e
way
S
-
bo
x
e
ntries
are
init
ia
li
zed
with
fi
xed
stri
ng
li
ke
‘
pi
’
va
lue
or
ze
r
o,
th
en
pa
ss
tw
o
va
lues
of
S
-
bo
x
(S0
a
nd
S
1
as
L0
a
nd
R0
)
to
the
e
nc
ryptio
n
f
un
ct
i
on
as
s
how
n
in
Fig
ur
e
1,
ge
ner
at
es
tw
o
di
ff
e
ren
t
encr
y
pted
valu
es
li
ke
L1
7
a
nd
R17
w
hich
is
cop
ie
d
to
S
0
a
nd
S1.
T
his
ste
p
Co
ntin
ues
ti
ll
,
rep
la
ci
ng
the
ou
t
put
by
c
hangin
g
c
on
ti
nu
ously
in
or
der
al
l
e
ntri
es
of
f
our
S
-
boxe
s
of
Bl
owfi
sh
al
gorithm
.
The
f
un
ct
i
on
F
w
orks
ta
kes
32
bit
val
ue
a
s
a
n
in
put,
and
it
div
i
ds
in
to
four
8
bit
dat
a.
Eac
h
f
our
8
bit
da
ta
is
us
e
d
for
substi
tuti
on
.
first
8
bit
is
us
e
d
to
get
32
bit
valu
e
f
r
om
S
-
box
0,
sec
ond
8
bit
data
is
us
ed
t
o
get
32
bit
valu
e
from
S
-
bo
x
1,
t
hir
d
8
bit
data
is
use
d
to
get
32
bit
value
from
S
-
box
2
a
nd
final
ly
la
st
8
bit
for
S
-
box
3.
T
he
n
first
32
bit
val
ue
i
s
add
e
d
with
sec
ond
32
bit
val
ue
,
the
outp
ut
is
XO
Re
d
wit
h
thir
d
32
b
it
value,
t
he
outp
ut
is
then
a
dde
d
with
four
t
h
val
ue
wi
ll
get
final
32
bi
t
value
as
sho
wn
i
n
Fig
ur
e
2.
To
pe
rfor
m
this
init
ia
li
zation
and
ge
ner
at
io
n
of
P
-
box
a
nd S
-
box t
akes m
or
e ti
m
e b
eca
us
e
of 16
-
rou
nd of e
nc
ryptio
n
al
gorit
hm
.
Figure
1.
Enc
r
ypti
on
al
gorith
m
f
or
b
l
owfish
Figure
2
.
I
nter
nal ope
rati
ons
of F fu
nctio
n
The
blowfis
h
al
gorithm
is
fa
st
and
us
e
fu
l
bl
ock
ci
phe
r.
m
any
i
m
ple
m
entat
ion
ha
s
bee
n
cond
ucted
ei
ther
th
rou
gh
so
ft
war
e
or
ha
rdwar
e
.
But
ve
ry
few
im
ple
mentat
ion
has
be
en
pr
opos
e
d
to
generate
e
ntries
of
P
-
bo
x
a
nd
S
-
box
f
or
blowfis
h
al
gorithm
.
A
ne
w
sec
ret
k
ey
a
s
bee
n
pro
pose
d
for
blo
c
k
ci
pher
,
blowfis
h,
wh
ic
h
is
a
Feist
el
net
work
with
bl
oc
k
siz
e
64
bit
a
nd
32
-
bit
to
448
-
bit
a
var
ia
ble
key
[
5
]
.
T
he
al
gorithm
i
m
ple
m
ented
with
a
com
plica
te
d
init
ia
li
zat
i
on
a
nd
la
rge
da
ta
caches
of
32
-
bit
m
ic
ro
pro
cesso
r.
Im
plem
en
te
d
a
nove
l
VLSI
arch
it
ect
ure
for
blowfis
h
al
go
rithm
wh
ic
h
is
base
d
on
pa
rtia
l
pipe
li
ned
str
uc
ture
[
6
]
.
The
auth
or
in
this
pa
per
has
us
ed
t
wo
di
ff
ere
nt
te
ch
niques
of
m
od
ifie
d
Feist
el
functi
on,
fir
st
is
it
erati
ve m
et
ho
d
a
nd,
sec
ond
is
pa
rtia
lly
pip
el
ine
d
te
c
hniqu
e.
A
f
our
sta
ge
pip
el
ine
d
a
r
chite
ct
ur
e
is
use
d
al
on
g
with
t
wo
it
erati
on
s
that
inc
reases
t
he
area
occupied
with
increasin
g
t
hro
ughput
of
the
al
gorithm
wh
en
c
om
par
ed
with
the
tw
o
sta
ge
pip
el
ini
ng
w
it
h
8
it
erati
on
s
t
hat
will
reduce
the
area
occupie
d
wit
h
reduce
d
t
hro
ughp
ut.
Disc
us
ses
ge
neral
optim
iz
at
ion
pri
nciples
of
de
sig
ning
t
he
al
gorithm
s,
a
nd
perform
ance
analy
zes
of
RC
4,
SEAL,
RC
5,
Bl
owfish
,
and
Khufu/
Khafr
e
on
the
I
ntel
Pe
ntium
with
res
pe
ct
to
th
os
e
ge
ner
al
optim
iz
e
d
pr
i
nciples
[
7
]
.
P
resen
te
d
,
a
on
e
rou
nd
VLSI
arch
it
ect
ure
of the
BLO
WFIS
H,
w
hich
is ba
sed
on
the
l
oop
-
f
old
in
g
te
c
hn
i
qu
e
com
bin
ed
w
it
h
sec
ur
e
dif
fer
e
nt
m
od
es
(ECB,
CB
C2,
CFB
2
a
nd
OF
B2
)
of
operati
on
[
8
]
.
T
he
a
rch
it
ect
ure
us
e
s
a
prot
otype
c
hip
to
im
pl
e
m
ent
by
us
i
ng
$0.
35$
/spl
m
u/
CMO
S
te
ch
nolo
gy.
Pr
ese
nted
D
RIL
arc
hitec
tu
re,
wh
ic
h
is
a
four
-
ti
er
arc
hitec
tur
e
involvin
g
both
software
an
d
hard
war
e
des
ign
s
,
f
or
im
pl
e
m
enting
blowfis
h
al
gorit
hm
us
ing
arc
hi
te
ct
ur
al
featur
e
s
li
ke
in
ner
lo
op
pi
pelinin
g
an
d
lo
op
fo
l
ding
with
dy
nam
ic
reco
nfi
gurati
on
[
9
]
.
The
m
ai
n
ob
j
ect
ive
of
the
resea
rc
h
w
hich
is
present
ed
in
this
pap
e
r
is
to
dev
el
op
a
n
al
gorit
hm
wit
h
lo
w
-
powe
r,
hi
gh
t
hro
ughput
,
real
-
tim
e,
reli
able
and
secu
re
c
r
ypto
syst
em
,
t
hat
can
be
ac
hieve
d
th
rou
gh
hardw
a
re
i
m
ple
m
entat
ion
s
[
10
]
.
Im
ple
m
ent
a
new
se
cret
-
key
'
Bl
ow
-
C
AS
T
-
F
ish'
blo
ck
ci
pher
that
us
e
s
go
od
featu
res
of
bo
t
h
CA
ST
-
12
8
an
d
Bl
ow
fis
h
al
gorithm
s
us
ing
VHDL
im
plem
entat
ion
[
11
]
.
Pro
posed
a
m
od
ifie
d
the
Bl
ow
fis
h
al
gorithm
by
enh
a
ncin
g
it
s
perform
ance
in
te
rm
s
of
s
pe
ed,
Th
r
oughpu
t,
P
ow
e
r
c
ons
um
ption
a
nd
Av
al
a
nch
e
e
ffec
t
[
12
]
.
Au
t
hor
has
pro
po
s
ed
a
way
to
enh
a
nce
t
he
pe
rfor
m
ance
of
t
he
Bl
owfish
cry
ptogra
phy
al
go
rithm
by
introd
ucin
g
par
al
le
l
pr
oces
sing
te
c
hniq
ue
and
m
aking
m
od
i
ficat
ion
s
t
o
the
Fiest
el
(F)
functi
on
of
Bl
owfish
by
c
ombini
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t. Sci.
I
nf. Tec
hnol.
,
V
ol. 2, N
o.
1, M
arch 2
021
:
1
6
–
25
18
the Bl
owfish
a
nd
t
he
Ru
nge
-
ku
tt
a (R
K)
Me
thod.
T
he
F
functi
on
of Bl
owfish has
been
m
od
ifie
d
with
diff
e
re
nt
form
ulae
and
the
ou
tc
om
e
of
a
series
of
RK
-
Bl
ow
fis
h
al
gor
it
h
m
s
wer
e
co
m
par
ed
with
th
e
Bl
owfish
al
gorithm
.
The
blowfis
h
a
lgorit
hm
has
been
wi
dely
us
e
d
in
netw
ork
s
ecur
it
y
m
e
tho
d
to
en
han
ce
t
he
secur
it
y
by
i
m
ple
m
enting
thr
ough
s
oft
wa
re
or
hard
war
e
bas
ed
on
var
i
et
y
of
as
pects
li
ke
s
peed,
sec
ur
it
y,
portabil
it
y
et
c.
The
sc
ope
of
th
is
work
inclu
de
s
ge
ner
at
io
n
of
P
-
b
ox
a
nd
S
-
box
e
ntries
us
in
g
a
m
od
ifie
d
fiest
el
networ
k,
wh
i
c
h
is
si
m
ple
struct
ur
e
,
in
orde
r
to
reduce
th
e
com
pu
ta
ti
onal
co
st
of
ge
ner
at
in
g
P
-
bo
x
an
d
S
-
box
entries
in
bl
owfis
h
al
gorithm
.
The
m
ai
n
m
otivatio
n
of
pro
po
se
d
w
ork
is
to
des
ign
a
nd
im
ple
m
entat
ion
of
ge
ner
at
in
g
P
-
bo
x
a
nd
S
-
bo
x
entries
by
reducin
g
th
e
num
ber
of
rounds
instea
d
of
16
-
r
ounds
of
e
ncr
y
ptio
n
al
gorithm
to
ov
e
rco
m
e
the
lim
it
a
ti
on
of
bl
owfish
al
gorithm
.
In
this
pa
per,
a
sim
ple
P
-
bo
x
a
nd
S
-
box
ge
ner
at
in
g
a
lgorit
hm
to
ov
e
rco
m
e
the co
m
pu
ta
ti
ona
l co
st i
s d
e
sign
e
d
a
nd im
ple
m
ented
usi
ng
m
od
ifie
d
fiest
e
l netw
ork.
2.
THE
PROPO
SED
APP
ROAC
H
In
t
his
sect
ion,
an
ap
proac
h
f
or
ge
ner
at
in
g
P
-
bo
x
a
nd
S
-
box
f
or
blowfis
h
al
gorithm
is
pr
ese
nted
.
The
blowfis
h
a
lgorit
hm
ta
kes
P
-
ar
ray
val
ues,
init
ia
li
zed
by
m
ast
er
key
K,
S
-
bo
x,
i
niti
al
ized
by
Pi
or
ze
r
o
val
ue,
will
be
ge
ne
rated
t
hro
ugh
m
od
ifie
d
e
ncr
y
ption
al
go
rithm
pr
oce
dure.
T
he
encr
y
ption
pro
cedure
m
od
ifie
d
by
reducin
g
num
ber
of
rou
nds,
9
it
erati
on
s
with
9
-
rou
nd
s
,
in
ste
ad
of
9
it
erati
on
with
16
-
r
ounds
in
the
proce
dure
.
By
reducin
g t
he
num
ber
of
r
ounds
in
t
he e
nc
ryptio
n
proce
dure
we
ca
n r
ed
uce
t
he t
im
e
as
well
r
ed
uctio
n
in
t
he
com
pu
ta
ti
on
al
cost
of
bl
owfis
h
al
gorithm
.
The
al
gorithm
tak
es
16
-
byte
Ke
y
K
(K0
K
1
K
2
K
3
K
4
K5
K
6
K
7
K8
K
9
K
10
K
11
K
12
K13
K
14
K
15
K16
)
,
as
a
n
in
pu
t
to
gen
e
rate
al
l
18
P
-
val
ues.
First
4
w
ord
K0K
1K2
K
3
from
key
is
stored
i
n
P
0
an
d
s
econde
4
w
ord
K4K
5K6
K
7
from
key
is
ta
ken
in
P
1
sim
il
ar
ly
al
l
key
values
are
store
d
in
P
-
val
ues
li
ke
[K0 K
1 K
2
K3
]
=
P(0)
[K4 K
5 K
6
K7
]
=
P(1)
[K8 K
9 K
10
K11
]
=
P(2)
[K1
2
K
13 K
14 K1
5] = P
(
3)
[K0 K
1 K
2
K3
]
=
P(4)
.......
. P
(
18)
Af
te
r
init
ia
li
zat
ion
,
pa
ss
t
wo
values
P
0
an
d
P1
each
of
32
-
bit,
(as
le
ft
hal
f
L
0
a
nd
ri
gh
t
half
R0
)
t
o
m
od
ifie
d
al
gor
it
h
m
. I
n t
he
al
gorithm
,
L1
=
L0 is
XOR
ed
with P
0
a
nd
R1 =
(R0 i
s XORed
w
it
h P
1) XORe
d wit
h F
-
f
un
ct
io
n wit
h i
nput L
1
a
nd S
-
bo
x
T
he
n
swa
p
L
1
and
R
1
an
d
c
onside
red
as
i
nput
for
ne
xt
it
er
at
ion
.
Re
pe
at
these
ste
ps
t
o
ge
t
the
ou
t
pu
t
with
tw
o
val
ue
s
P0
a
nd
P1,
f
or
this
the
m
od
if
ie
d
enc
ryptio
n
al
gorit
hm
ta
ke
s
9
it
erati
ons
in
ste
ad
of
16
it
er
at
ion
s.
The
flo
wc
har
t
of
t
he
pro
pos
ed
m
od
ifie
d
e
ncr
y
ption
al
go
rithm
is
giv
e
n
in
F
ig
ure
3
a
nd
the
al
go
rithm
fo
r
pro
po
se
d
e
ncr
y
ption
al
gorith
m
1
is
giv
en
in
sect
ion
4.1
.
Th
e
F
f
unct
ion
ta
kes
32
bit
valu
e
as
an
in
pu
t
a
n
d
div
i
ds
into
f
our
8
bit
value.
Eac
h
f
our
8
bit
value
is
use
d
for
s
ubs
ti
tuti
on
from
each
f
our
S
-
box
an
d
fir
st
4
bit
from
each
8
bit
val
ue
is
co
ns
ide
r
ed
f
ro
m
wh
ic
h
S
-
bo
x
we
s
ho
uld
get
the
val
ue.
T
he
fl
ow
c
har
t
of
the
pro
po
s
ed
m
od
ifie
d
F
-
f
unct
ion i
s
giv
e
n i
n
Fig
ure
4
a
nd the alg
ori
thm
2
is
giv
e
n
i
n
se
ct
ion
2
.
1.
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t. Sci.
I
nf. Tec
hnol.
An o
ptimized e
ncry
ption al
gorit
hm
and
F fu
nction w
it
h
Dy
namic s
ubsti
tuti
on
f
or
cre
atin
g
…
(
Rekh
a
C
)
19
2.1.
Prop
os
ed
m
odi
fied encr
yp
ti
on
algorit
hm
an
d
d
ynamic
subst
i
tu
ti
on
i
n F
-
fu
nc
tio
n
The
m
ai
n
go
al
of
t
his
w
ork
i
s
to
pro
vid
e
m
or
e
sec
uri
ty
,
m
ini
m
iz
es
the
tim
e
ta
ken
for
ge
ner
at
i
ng
P
-
bo
x
a
nd S
-
box
ent
ries an
d com
pu
ta
ti
on
al
cost. T
he pr
op
os
e
d
al
gorithm
inclu
des red
uc
ti
on
of
c
om
pu
t
at
ion
al
cost
by
re
du
ci
ng
the
num
ber
of
it
erati
ons
i
n
the
enc
rypti
on
al
gorithm
with
9
r
ounds
and
9
it
erati
ons,
w
he
re
or
i
gin
al
al
go
rithm
us
es
16
r
ound
a
nd
9
it
er
at
ion
s
i
n
orde
r
to
gen
e
rate
P
-
bo
x
a
nd
S
-
bo
x
e
ntries.
A
nd
al
so
pro
vid
es
bette
r
secu
rity
b
y u
sing dy
nam
ic
su
bs
ti
tuti
on in
m
od
i
fied F f
unct
ion
.
Figure
3
.
Flo
w
char
t
of the m
odifie
d
encr
y
ption al
gorithm
Figure
4
.
Flo
w
char
t
of the m
odifie
d F
un
ct
io
n F
3.
RESU
LT
S
AND
DI
SCUS
S
ION
In
this
sect
ion,
The
or
i
gin
al
bl
owfish
al
gorit
hm
[
5
]
a
nd
m
od
ifie
d
bl
owfis
h
al
gorithm
are
app
li
ed
on
the
im
age
A
rm
s.b
m
p
wit
h
t
he
sam
e
key.
T
he
com
par
iso
n
ar
e
m
ade
on
both
or
i
gin
al
an
d
m
od
ifie
d
al
go
r
it
h
m
s
by
m
aking
us
e
of
a
valanc
he
eff
ect
,
e
ncr
y
pt
ion
qu
al
it
y,
ke
y
sensiti
vity
and
sta
ti
sti
cal
a
naly
sis.
The
or
iginal
i
m
age
of
Ar
m
s.b
m
p
in
Fig
ur
e
5,
is
e
ncr
y
pted
and
dec
rypted
by
ap
plyi
ng
or
i
gin
al
blowfis
h
al
gorithm
are
sh
ow
n
in.
Sam
e
or
igi
nal
im
age
is
e
ncr
y
pted
an
d
decr
y
pted
us
i
ng
m
od
ifie
d
blowfis
h
al
gor
it
hm
.
The
Fig
ure
6
a
nd
Figure
7
s
how
s
the
res
ult
of
encr
ypte
d
an
d
dec
rypte
d
im
age
us
in
g
or
iginal
al
gorith
m
and
Fig
ur
e
8
a
nd
Figure
9
s
hows
the
resu
lt
of e
ncr
y
ption an
d decry
ption i
m
a
ges by a
pp
ly
in
g
m
od
ifie
d al
gorithm
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t. Sci.
I
nf. Tec
hnol.
,
V
ol. 2, N
o.
1, M
arch 2
021
:
1
6
–
25
20
Fi
gure
5
.
O
rigi
nal
im
age ar
m
s
bm
p
Figure
6
.
Enc
r
ypte
d
im
age
of
arm
s
bm
p
us
in
g
or
i
gin
al
alg
ori
thm
Figure
7
.
Dec
r
ypte
d
im
age
of
ar
m
s
bm
p
us
in
g
or
i
gin
al
alg
ori
thm
Figure
8. Enc
r
ypte
d
im
age
of
arm
s
bm
p
us
in
g
m
od
ifie
d
al
gor
it
h
m
Figure
9. Dec
r
ypte
d
im
age
of
ar
ms
bm
p
us
in
g
m
od
ifie
d
al
gor
it
h
m
3.1.
Encr
yption
Q
ua
li
t
y
te
st
The
e
ncr
y
ptio
n
qu
al
it
y
(EQ)
te
st
in
\
ci
te
{13
sw
f},
\
ci
te
{1
5s
w
f}
m
easur
e
s
the
qu
al
it
y
of
enc
ryptio
n
wh
ic
h
is
base
d
on
t
he
dev
ia
ti
on
bet
wee
n
th
e
plainte
xt
im
age
a
nd
ci
phe
r
te
xt
im
age.
Th
e
m
or
e
de
viati
on
of
ci
ph
e
rtext
c
ompare
d
to
plaint
ext, b
et
te
r
is
th
e
enc
ryptio
n
al
gorithm
.
EQ
te
st
is
the
a
ver
a
ge
num
ber
o
f
c
ha
nge
s
to
eac
h
gr
ey
le
vel
L
bet
ween
or
i
gin
al
a
nd
e
nc
rypted
im
ages.
T
he
m
at
he
m
atical
form
ula
an
d
t
he
ste
ps
t
o
pe
rfor
m
EQ
te
st i
s
giv
e
n
as:
=
∑
[
(
′
)
−
[
(
)
]
]
255
256
Let
F
an
d
F'
is
represe
nted
as
or
i
gin
al
im
ag
e
and
e
nc
rypte
d
im
age
of
siz
e
M*N
pix
el
s
with
L
gr
ey
le
vels.
At
posit
ion (
x,
y)
,
the
grey
levels
of th
e F
a
nd F'
,(0
≤
x ≤
M
-
1,
0 ≤
y
≤ N
-
1) is
repre
sented
as
F
(x,y)
a
nd
\
(F
'
\
)$
(
x,
y)
\
in
L$ ran
gi
ng
fro
m
$
0
to
255$.
The
ste
ps
a
re:
Com
pu
te
HL
(
F)
t
he
num
ber
of
occ
urren
ce
s
of
eac
h
gray
le
vel
L
in
the
ori
gin
al
im
ag
e
an
d
HL
(
\
(F
'
\
))
denotes t
he num
ber
o
f occ
urr
ences
of each
grey
level i
n t
he
en
c
rypted
im
a
ge.
Com
pu
te
the a
ver
a
ge n
um
ber
of c
hanges t
o ea
ch
gr
ey
le
vel L
us
in
g
a
bove
g
ive
n
m
at
hem
at
ic
al
f
or
m
ula.
Encr
y
ption
Q
ualit
y
te
st
is
cal
culat
ed
us
i
ng
both
or
i
gin
al
an
d
m
od
ifie
d
blowfis
h
al
gorithm
is
sh
ow
n
in
Ta
ble1.
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t. Sci.
I
nf. Tec
hnol.
An o
ptimized e
ncry
ption al
gorit
hm
and
F fu
nction w
it
h
Dy
namic s
ubsti
tuti
on
f
or
cre
atin
g
…
(
Rekh
a
C
)
21
Table
1.
C
om
par
iso
n
of enc
ryption q
ualit
ie
s o
f
original a
nd
m
od
ifie
d
bl
owfish
al
gorithm
for diffe
re
nt
r
ounds
Nu
m
b
e
r
o
f
Ro
u
n
d
s
Origin
al
Blo
wf
ish
Mod
if
ied
Blo
wf
ish
2
1
9
0
.351
1
6
7
.570
4
2
0
1
.898
2
0
2
.28
6
8
10
12
14
16
2
0
0
.132
2
0
3
.484
2
0
2
.796
2
0
3
.273
2
0
2
.046
1
9
2
.414
1
9
6
.507
1
8
0
.179
2
0
3
.601
1
9
9
.390
2
0
2
.859
2
0
2
.781
3.2.
Key Sensi
ti
vit
y
Te
s
t
A
Key
of
16
-
c
har
act
er
with
128
bits
is
us
ed
for
e
ncr
ypti
on
and
dec
ryptio
n.
The
Key
se
nsi
ti
ve
te
st
[
5
],
[1
4
]
h
a
s
been c
arr
ie
d o
ut as
fo
ll
ow
s
Applyi
ng the
16
-
c
har
act
e
r 128
-
bit key,
K
ey
1,
t
o
E
ncr
y
pt a
n
im
age A
rm
s.b
m
p
by origi
na
l al
gorithm
.
Chan
ge
a
ny
r
a
ndom
ly
sel
ect
e
d
on
e
bit
from
Key1.
T
hen
f
r
om
this
m
od
ifi
ed
key,
Key2,
encr
y
pt
the
s
ame
i
m
age
by
a
pply
ing
to
ori
gin
al
al
gorithm
.
Ex
:
ADF37
8
465E
262AB1F
5DE
C94A0AF
25
F
27,
from
this
ke
y1
we
ha
ve
ra
ndom
ly
sel
e
ct
ed
F
as
sho
wn
in
bold
an
d
cha
nged
to
B
a
s
A
DF378
465E
262A
B1
F
5DEC9
4A0
A25
B
27
wh
ic
h
is
key2.
Apply t
hese
tw
o keys t
o m
od
ifie
d blo
wf
is
h
a
lgorit
hm
u
sing
the sam
e i
m
ag
e an
d
the
n
c
om
par
e
d.
The
res
ult
is
show
n
i
n
belo
w
Table
2,
c
om
par
iso
n
of
both
ci
ph
e
red
im
age
s
w
hich
a
re
e
nc
rypted
by
or
i
gin
a
l
as w
el
l as m
odifie
d
al
gori
t
hm usi
ng these t
w
o diff
e
re
nt k
ey
s
The
res
ult
is
99.
384781
of
di
ff
e
ren
ce
,
w
hen
we
e
ncr
y
pt
t
he
im
age
us
i
ng
or
i
gin
al
blow
fish
al
gorithm
w
it
h
key1
a
nd
wh
e
n
we
e
nc
rypt
the
sam
e
i
m
age
usi
ng
sam
e
or
igi
nal
al
gorithm
with
key2,
in
te
rm
s
of
gr
ey
sc
al
e
values
where
there
is only
on
e b
it
d
i
ff
e
ren
ce
in
the
se tw
o k
ey
2.
The
res
ult
is
99.
539299
of
di
ff
e
ren
ce
w
he
n
we
e
ncr
y
pt
the
im
age
us
in
g
m
od
ifie
d
bl
ow
fish
al
gorithm
with
key1
a
nd
wh
e
n we e
nc
rypt the
sam
e i
m
age u
s
ing
sam
e algo
r
it
h
m
w
it
h
ke
y2
.
Table
2
. C
om
par
iso
n
of k
ey
s
ensiti
vity
o
f
or
iginal an
d
m
odifie
d blo
wf
is
h al
gorithm
for diffe
re
nt ro
unds
Nu
m
b
e
r
o
f
Ro
u
n
d
s
Origin
al
Blo
wf
ish
Mod
if
ied
Blo
wf
ish
2
9
5
.86
9
6
.04
9
4
9
9
.56
7
9
9
.57
8
6
8
10
12
14
16
9
9
.45
9
9
9
.61
2
9
9
.59
0
9
9
.59
9
9
5
.58
8
9
9
.41
0
9
9
.58
3
9
9
.65
2
9
9
.60
2
9
9
.58
8
9
9
.58
0
9
9
.62
8
3.3.
Avalanche E
ffec
t
The
a
valanc
he
eff
ect
i
n
[
5
]
,
[
13
-
1
4
]
m
eans
if
the
re
is
a
cha
ng
e
in
one
bit
in
the
plain
te
xt
then
t
here
will
be
nu
m
ber
of
bits
cha
nge
s
in
the
ci
ph
e
r
te
xt.
To
c
om
pu
te
a
valanc
he
e
ffec
t
we
need
t
o
change
on
e
bit
from
the
plai
n
te
xt
(
i
m
age
Ar
m
s.bm
p)
,
nam
ed
as
an
im
age
BArm
s.b
m
p,
an
d
then
e
ncr
y
pt
this
im
age
us
in
g
both
or
i
gin
al
blowfi
sh
a
nd
m
od
ifie
d
blowfis
h
al
gorithm
s.
He
re
the
pr
opos
e
d
a
lgorit
hm
is
com
par
ed
with
ori
gin
a
l
al
gorithm
at
diff
ere
nt
r
ounds
al
ong
with
f
our
cases
.
T
he
Table
3
pr
ovides
w
hich
al
gorithm
giv
es
bette
r
avalanc
he
e
ff
e
ct
.
Ca
se
1:
Com
par
in
g
A
valanc
he
e
ff
e
ct
f
or
e
ncr
y
pted
im
ag
e
of
A
rm
s.b
m
p
a
nd
B
1Arm
s.b
m
p
with
sam
e
key
1
us
in
g or
i
gin
al
and m
od
ifie
d
a
lgorit
hm
.
Ca
se
2:
Com
par
in
g
A
valanc
he
e
ff
e
ct
f
or
e
ncr
y
pted
im
ag
e
of
A
rm
s.b
m
p
a
nd
B
1Arm
s.b
m
p
with
sam
e
key
2
us
in
g or
i
gin
al
and m
od
ifie
d
a
lgorit
hm
.
Ca
se
3:
C
om
par
in
g
Av
al
a
nch
e
eff
e
ct
f
or
enc
r
ypte
d
im
age
of
A
rm
s.b
m
p
wit
h
key1
an
d
key
2
us
in
g
ori
gin
a
l
an
d
m
od
ifie
d
al
gor
it
h
m
.
Ca
se
4:
Com
par
in
g
A
valanc
he
ef
fect
f
or
e
ncr
y
pted
im
age
B1MA
ND.bm
p
with
key1a
nd
key2
us
i
ng
or
i
gin
al
and m
od
ifie
d
a
lgorit
hm
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t. Sci.
I
nf. Tec
hnol.
,
V
ol. 2, N
o.
1, M
arch 2
021
:
1
6
–
25
22
Table
3
. C
om
par
iso
n
of avala
nch
e
ef
fect
of
or
i
gin
al
a
nd m
od
i
fied blo
wf
is
h
al
go
rithm
f
or d
if
fer
e
nt ro
unds
with
four cases
Nu
m
b
e
r
o
f
Ro
u
n
d
s
Cas
e 1
Cas
e 2
Cas
e 3
Cas
e 4
Org
Mod
Org
Mod
Org
Mod
Org
Mod
2
1123
2527
770
1588
2015
1612
1630
2013
4
1149
2390
761
1547
2049
1620
389
3328
6
1137
2416
772
810
1131
1652
2473
1155
8
992
2552
716
1570
1131
1652
1162
2440
10
2340
1186
1513
778
2437
1151
1976
1623
12
2335
1192
882
1428
1168
2851
1069
2471
14
2379
1182
825
1474
1953
1646
2425
1152
16
1206
2333
800
804
1999
1141
2439
1135
3.4.
Co
rrel
at
i
on
c
oeffici
ent
The
c
orrelat
ion
coe
ff
ic
ie
nt
i
s
determ
ined
r
el
at
ion
sh
i
p
bet
ween
horiz
on
t
al
ly
adj
acent
pix
el
s
in
a
n
i
m
age [
9], [1
1]. Th
e ste
ps
for
determ
ining
th
e correlat
ion
of h
ori
zo
ntal ad
ja
cent p
ixels i
n an im
age A
rm
s
.b
m
p
is as f
ollo
ws
Fr
om
the
or
igi
nal im
age an
d t
heir
e
ncr
y
pted
i
m
age,
sel
ect
N
pairs o
f horiz
on
ta
ll
y ad
j
ace
nt p
i
xels.
Sele
ct
pix
el
s
r
andom
ly
and
pi
xels
ad
j
acent
to
them
fr
om
t
he
both
or
i
gina
l
i
m
age
(A
rm
s.b
m
p)
an
d
the
ir
encr
y
pted
im
a
ges usin
g b
oth
or
i
gin
al
alg
ori
thm
as w
el
l as
m
od
ifie
d
al
gor
it
h
m
.
Using
$r
_{x
y}
$
f
or
m
ulae
to
fin
d
co
rr
el
at
io
n
coe
ff
ic
ie
nt,
wh
e
re
x
a
nd
y
represent
gr
e
y
scal
e
values
of
horizo
ntall
y ad
j
acent
pi
xels in
an
im
ag
e.
=
(
,
)
√
(
)
√
(
)
wh
e
re
D
(X)
a
nd
D(Y)
re
prese
nts
the
var
ia
nc
e
of
x
a
nd
y
val
ues
a
nd
C
OV(
X,Y)
is
c
ov
a
ria
nce
of
x
a
nd
y and i
s
giv
e
n by
(
,
)
=
1
∑
(
−
(
)
)
(
−
(
)
)
=
0
Wh
e
re
N
repre
sents
num
ber
of
horizo
ntal
a
dj
acent
pix
el
s
sel
ect
ed
rand
om
l
y,
E
(X)
an
d
E(
Y
)
represe
nts
the
m
ean
valu
es
of
x
a
nd
y
va
lues.
T
his
te
st
is
ca
rr
ie
d
out
for
a
bout
ra
nd
om
l
y
sel
ect
ed
horizo
ntall
y
ad
j
acent
pix
el
s
f
ro
m
th
e
or
igi
nal
im
a
ge
A
rm
s.BM
P
a
nd
e
ncr
y
pted
im
ages.
The
n
us
i
ng
a
bove
e
qu
at
io
ns
c
orrelat
i
on
c
oeffici
ent
will
be
c
om
pu
te
d
a
nd
is
as
sho
wn
in
bel
ow
Fig
ure.
10,
Fi
gure.
11,
a
nd
Fig
ur
e
.
12.
T
he
co
rr
el
at
ion
coeffic
ie
nt
of
or
i
gin
al
im
age
is
0.0
7205
3
a
nd
for
ci
pher
i
m
age
wh
ic
h
is
e
ncr
ypte
d
by
blowfis
h
al
gor
it
h
m
is
0.005
616
a
nd
for
m
od
ifie
d
a
lgorit
hm
is
-
0.
429036.
I
n
bot
h
ori
gi
nal
an
d
m
od
ifie
d
al
gor
it
h
m
the
correla
ti
on
coeffic
ie
nts
for
p
la
intext
im
ag
e w
it
h
t
hat
of
c
iph
e
rtext im
ages ar
e
far apa
rt
.
Figure
10
.
C
orrelat
ion
distrib
ution o
f
t
wo
horizo
ntall
y ad
j
acent
pi
xels fo
r or
igi
nal im
age ar
m
s
BMP
Figure
11
.
C
orrelat
ion
distrib
ution o
f
t
wo
horizo
ntall
y ad
j
acent
pi
xels fo
r
e
ncr
ypte
d
im
age
us
in
g or
i
gin
al
al
gorithm
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t. Sci.
I
nf. Tec
hnol.
An o
ptimized e
ncry
ption al
gorit
hm
and
F fu
nction w
it
h
Dy
namic s
ubsti
tuti
on
f
or
cre
atin
g
…
(
Rekh
a
C
)
23
Figure
12
.
C
orrelat
ion
distrib
ution o
f
t
wo ho
rizo
ntall
y adj
a
cent p
i
xels fo
r e
ncr
y
pted
im
a
ge usin
g
m
od
if
ie
d
al
g
ori
thm
3.5.
Entropy
The
e
ntr
op
y
[1
5
]
of
an
i
nfor
m
at
ion
data
is
m
easur
e
d
i
n
bits
per
c
ha
racter.T
he
e
ntropy
is
c
al
culat
ed
as
the
av
era
ge
am
ount
at
w
hich
i
nfor
m
at
ion
data
is
pr
oduce
d
by
a
truly
rand
om
so
ur
ce
of
da
ta
.
To
cal
culat
e
this
m
ean,
the
in
di
vidual
in
f
or
m
a
ti
on
a
re
wei
ghte
d
wi
th
t
he
prob
a
bili
ti
es
of
t
heir
occ
urre
nc
e.
T
he
m
at
hem
at
ic
al
form
ula f
or cal
culat
ing
entr
op
y i
s
(
)
=
∑
(
)
log
2
1
(
)
2
−
1
=
0
=
wh
e
re
p(
)
be
t
he
pro
ba
bili
ty
of
occurre
nce
of
a
c
har
act
er
and
e
ntr
op
y
is
pr
ese
nted
i
n
bi
ts.
Af
te
r
evaluati
ng t
h
e
above e
quat
io
n 4, we
obtai
n i
ts entr
opy as
H(m
)
= 8
, w
hich i
s co
r
res
pondin
g
to
a tr
uly ra
ndom
so
urce.
Give
n
an
in
form
at
ion
so
urce
that
ge
ne
rates
ra
ndom
m
essages,
in
ge
ner
al
it
s
entr
opy
value
is
le
ss
er
tha
n
the
ideal
one
.
Howe
ver,
whe
n
t
he
m
essages
are
e
ncr
ypte
d,
their
e
ntropy
s
hould
be
8.
If
the
outp
ut
of
a
blo
c
k
ci
ph
e
r e
m
i
ts
with
entr
op
y
le
ss
than
8,
t
her
e
e
xists
certai
n
de
gr
ee
of
predict
abili
ty
, which
end
a
nger
it
s
se
cur
it
y
.
Let
us
c
onside
r
t
he
ci
ph
e
rtex
t
of
a
n
im
age,
encr
y
pted
usi
ng
t
he
ori
gin
al
as
wel
l
as
m
od
ifie
d
al
go
rith
m
,
the
nu
m
ber
of
occ
urren
ce
of
eac
h
ci
phe
rtext
blo
ck
an
d
t
he
prob
a
bili
ty
of
oc
currence
is
c
om
pu
te
d.
T
he
obta
ine
d
inf
or
m
at
ion
e
nt
ropy
is
ve
ry
m
uch
cl
ose
t
o
t
he
the
or
et
ic
al
value
of
8.
T
his
m
eans
that
le
a
ka
ge
of
t
he
i
nform
at
ion
data
in
the
e
nc
ryptio
n
proces
s
is
ne
gligible
and
the
e
ncr
y
pt
ion
syst
e
m
is
secur
e
a
gainst
entr
op
y
at
ta
ck.
T
he
entr
op
y i
s calc
ulate
d usin
g
e
quat
ion.
4
a
nd is
li
ste
d
in
belo
w
Tab
le
4.
Table
4
. C
om
par
iso
n
of entr
opy f
or d
i
ff
e
ren
t
roun
ds
of
ori
gi
nal a
nd m
od
ifi
ed
blowfis
h
al
gorithm
Nu
m
b
e
r
o
f
Ro
u
n
d
s
Origin
al
Blo
wf
ish
Mod
if
ied
Blo
wf
ish
2
6
.74
6
.86
4
7
.03
7
.03
6
8
10
12
14
16
6
.96
7
.06
7
.07
7
.06
7
.05
7
.07
7
.07
7
.05
7
.07
7
.06
7
.06
7
.07
3.6.
Floating
Freq
uency
The
floati
ng
f
r
equ
e
ncy
[
1
6
]
how
m
any
dif
fe
ren
t
c
har
act
e
rs
are
to b
e fou
nd
in
any g
ive
n
64
-
c
har
act
e
r
long
se
gm
ent
in
ci
phe
rtext.
F
reque
ncy
anal
ysi
s
is
base
d
on
ce
rtai
n
le
tt
er
s
an
d
c
om
bin
at
ion
s
of
le
tt
ers
occ
ur
with
var
yi
ng
f
r
equ
e
ncies.
T
he
f
reque
ncy
f
un
ct
ion
c
onside
rs
seq
ue
nces
of
char
act
e
rs
i
n
t
he
64
c
har
act
e
rs
l
ong
segm
ent
and
c
ounts
ho
w
m
any
diff
e
ren
t
c
harac
te
rs
are
to
be
found
in
t
his
64
-
c
har
act
e
r
lo
ng
segm
ent.
The
n
the
segm
ent
is
sh
if
te
d
one
c
har
a
ct
er
to
t
he
ri
ght
and
the
cal
c
ulati
on
is
re
peated
.
This
proce
dure
res
ults
in
a
s
um
m
ary
of
the
ci
ph
e
rte
xt
w
hich
ide
n
ti
fy
the
places
w
it
h
high
a
nd
lo
w
i
nfo
rm
ation
densi
ty
.
De
pe
ndin
g
on
the
str
uctu
re
and
c
on
te
nt
of
the
data
in
the
segm
ent,
enc
rypted
im
ages
(bm
p
file
s)
val
ue
s
obta
ine
d
usua
ll
y
li
e
between
5
a
nd
20,
as
s
how
n
i
n
Fi
gure
13
an
d
the
floati
ng
fr
e
qu
e
ncy
f
or
t
he
e
n
crypte
d
i
m
ages
us
i
ng
ori
gin
al
a
nd
m
od
ifie
d
al
gorithm
is aa shown i
n
Fi
gu
re
s
14 and
15.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t. Sci.
I
nf. Tec
hnol.
,
V
ol. 2, N
o.
1, M
arch 2
021
:
1
6
–
25
24
Figure
13. Fl
oa
ti
ng
fr
e
qu
e
ncy
for ori
gin
al
im
age a
rm
s
b
m
p
Figure
14
.
Floa
ti
ng
fr
e
qu
e
ncy
for
e
ncr
y
pted
i
m
age u
si
ng original al
go
rithm
Figure
15
.
Floa
ti
ng
fr
e
qu
e
ncy
for
e
ncr
y
pted
i
m
age u
si
ng m
od
ifie
d al
gorith
m
4.
CONCL
US
I
O
N
To
e
nhance
th
e
secu
rity
feat
ur
es
of
blow
fish
al
gorithm
,
the
pro
pose
d
m
e
thod
has
bee
n
desig
ne
d
a
nd
i
m
ple
m
ented
to
cre
at
e
S
-
box
and
P
-
box
val
ue
s
of
blowfis
h
al
gorithm
us
ing
m
od
ifie
d
e
nc
ryptio
n
al
gorit
hm
and
m
od
ifie
d
F
f
un
ct
ion
with
dyna
m
ic
su
bs
ti
tuti
on.
T
he
m
ai
n
m
otivati
on
be
hind
f
or
pr
opos
e
d
al
gorithm
is
to
reduce
the
ti
m
e
fo
r
ge
ner
at
in
g
s
-
box
and
P
-
box
val
ue
s
by
re
duci
ng
the
num
ber
of
r
ounds,
9
it
era
ti
on
s
with
9
rounds,
instea
d
of
9
it
erati
on
s
with
16
r
ounds
in
the
e
ncr
y
ption
al
go
r
it
h
m
for
blow
fish
al
gorithm
. F
ro
m
the
res
ults,
the
pro
po
se
d
m
od
i
fied
e
ncr
y
ptio
n
al
go
rithm
per
f
or
m
s
bette
r
i
n
al
l
the
aspec
ts
when
com
pa
red
with
the
or
i
gin
al
blowfis
h
al
go
rithm
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t. Sci.
I
nf. Tec
hnol.
An o
ptimized e
ncry
ption al
gorit
hm
and
F fu
nction w
it
h
Dy
namic s
ubsti
tuti
on
f
or
cre
atin
g
…
(
Rekh
a
C
)
25
REFERE
NCE
S
[1]
Kum
ar,
M.
Ana
nd
and
S.
K
art
hi
ke
y
an.
“
Inve
st
ig
at
ing
th
e
Eff
ic
i
e
nc
y
of
B
lowfish
and
R
ej
ind
ael
(
AES)
Algorit
hm
s.”
Inte
rnational
Jo
urnal
of
Comput
er
Net
work
and
I
nformation
Se
cu
rity
,
vol
.
4
,
pp
.
2
2
-
28,
2
012
.
[2]
Behrouz
A
.
Foro
uza
n
.
“
Cr
y
p
togr
aph
y
and
Netwo
rk
Secur
i
t
y
”
,
Tat
a
Mc
Gr
aw
-
Hill
,
2nd
edition
,
200
8.
[3]
Anand
Kum
ar
M
and
Dr.
S.
K
art
hik
e
y
an
.
“
Inve
stiga
t
ing
th
e
Ef
fic
i
ency
of
B
lo
wfish
and
Rejin
dae
l
Algor
it
hm
s
.”
Inte
rnational
Jo
urnal
of
Comput
er
Net
work
and
I
n
formation
Se
cu
rity
,
pp.
22
-
28,
2
012.
[4]
W
il
li
am Sta
ll
ing
s
. “
Cr
y
ptogr
aph
y
and
N
et
work S
ec
uri
t
y
.
”
F
if
th
E
dit
ion, Pears
on
Publ
ic
a
ti
on
,
Pre
nti
c
e
ha
ll,
2013
.
[5]
B.
Schneier.
“
Descri
pti
on
o
f
a
n
e
w
var
ia
bl
e
-
l
ength
ke
y
,
64
-
b
it
b
lo
ck
ci
ph
er
(blowf
ish
).
”
Inte
rnational
Workshop
on
Fast
Soft
war
e
E
ncry
pti
on
,
Sprin
ger
,
B
erlin,
Heid
el
ber
g
,
vo
l
809
,
pp
191
-
204,
199
4.
[6]
Kart
higaikum
ar,
P.
and
K.
Baska
ran
.
“
Parti
a
lly
P
i
pel
in
ed
VLSI
Im
ple
m
ent
a
ti
on
of
Blowfish
En
cr
y
p
ti
on/De
cr
y
pti
on
Algorit
hm
.
”
Int.
J.
Imag
e
Gr
aph
.
vol.
10
,
pp
.
327
-
341,
2010
.
[7]
Schnei
er
B
.
,
W
h
it
ing
D.
“
Fast
software
en
cr
y
pt
io
n:
Design
ing
en
c
r
y
pt
ion
al
go
rit
h
m
s
for
opt
imal
software
spee
d
o
n
the
Int
el
Pent
iu
m
proc
essor.”
In
:
Bi
ham
E
.
(
Eds)
Fast
Soft
ware
E
ncry
pti
on
.
FSE
1997.
Lect
ure
N
ote
s
in
Compute
r
Sci
en
ce
,
Springe
r,
Be
r
li
n
,
He
ide
l
ber
g
,
vo
l
1267
,
2
42
-
259,
1997
.
[8]
Yeong
-
Kang
Lai
and
Yu
-
Chuan
Shu.
“
A
novel
VLSI
arc
hitec
tur
e
for
a
var
i
able
-
l
engt
h
ke
y
,
64
-
bi
t
Blowfish
bloc
k
ci
pher
.
”
1999
I
EE
E
Workshop
on
Signa
l
Pr
oce
ss
ing
Syst
e
ms
.
SiP
S
99
.
Design
and
Im
ple
menta
ti
on
(
Cat.
No.
99
TH
8461)
,
Ta
ip
ei
,
T
ai
wan
,
pp.
568
-
577
,
19
99
.
[9]
Sudarshan
T
.
S.
B.
,
Mir
R
.
A.,
Vijay
a
la
kshm
i
S.
“
DRIL
–
A
Flex
ibl
e
Archi
t
ectur
e
for
Blowfish
Enc
r
y
pt
ion
Us
in
g
D
y
namic
R
ec
onf
igura
ti
on
,
Rep
lic
at
ion
,
Inne
r
-
Loo
p
Pipel
ini
ng
,
Lo
op
Folding
Tech
nique
s.”
In
:
Srik
ant
han
T.,
Xue
J.
,
Chang
CH.
(
Eds
)
Advanc
es
in
C
omputer
Syste
ms
Archi
t
ec
ture
.
A
CSAC
2005.
Le
c
ture
Note
s
in
Computer
Sci
en
ce
,
Springer,
B
erlin,
Heidelbe
rg
,
vo
l 3740,
pp
.
625
-
6
39,
2005
.
[10]
P.Kart
higaiKum
ara
,
K.B
aska
ran
.
“
An
AS
IC
implementation
of
l
ow
power
and
high
throug
hpu
t
blowfish
cr
y
pt
o
al
gorit
hm
.”
M
ic
r
oel
e
ct
ronics
Jou
rnal
,
vol
.
41
,
no
.
6,
pp.
347
-
355,
June
2010.
[11]
Krishnam
urth
y
G.N,
Dr.
V.
R
a
m
aswam
y
,
Leel
a
G.H
and
As
halatha
M.
E
.
“
Blow
-
CAS
T
-
Fis
h:
A
New
64
-
bit
B
loc
k
Ciphe
r
.”
IJCSNS
Inte
rnationa
l
Jo
urnal
of
Comput
er
Scien
ce
and
Net
work
Se
curity
,
vo
l
.
8
,
no
.
4,
p
p.
282
-
290,
Apri
l
2008.
[12]
K
rishnamurth
y
G.N,
Dr.
V.
Ra
m
aswam
y
and
Mrs
.
Le
e
la
G
.
H.
“
Perform
anc
e
en
hanc
ement
of
Bl
owfish
al
gorit
h
m
b
y
m
odif
y
ing
i
ts
f
unct
ion
.”
Innov
ati
v
e
Al
gorithm
s
and
Te
chni
q
ues
in
Au
tomat
ion,
Industrial
El
e
ct
ronics
and
Tele
communic
a
t
ions
,
pp.
241
-
24
4
,
2007
.
[13]
V
Ramasw
am
y
Krishnam
urth
y
G
N.
“
Enc
r
y
pt
io
n
qua
li
t
y
a
na
l
y
s
is
and
se
cur
ity
e
val
ua
ti
on
of
blo
w
-
ca
stfish
using
digi
tal
images.
”
Comm
unic
ate
d
t
o
Inte
rnat
ional
J
ournal
of
Comp
utat
ional Sc
ie
nc
e
,
20
08
.
[14]
Shail
aja
S1,
Dr
Krishnam
urth
y
G
N2
.
“
Com
par
i
son
of
B
lowfish
and
C
ast
-
128
Al
gorit
hm
s
Us
ing
Enc
r
y
pt
ion
Qua
l
ity
,
Ke
y
Sensit
ivi
t
y
and
C
orr
elati
on
Coeff
icient
Ana
l
y
sis.
”
Ame
rican
Journal
o
f
Enginee
ring
Re
s
earc
h
(
AJ
ER
)
,
vol
.
3
,
no.
7
,
pp
-
161
-
16
6,
2014
.
[15]
Yue
W
ua,
Jos
ep
h
P.
Noonan
a,
S
os
Agaia
nb
.
“
Shannon
En
trop
y
b
ase
d
Randomnes
s
Mea
surem
ent
a
nd
Te
st
for
Im
ag
e
Enc
r
y
pt
ion
.
”
Inf
orm
ati
on
Scienc
es
,
2011
.
[16]
Sonal
Kulshresht
ha,
Vikr
am
Verm
a,
Ric
h
a
Kalr
a
.
“
Anal
y
t
ical
Vie
w
of
Cr
y
ptogr
ap
hic
T
ec
hniqu
es
t
hrough
Cr
y
p
tool
.”
Jo
urna
l
of Te
le
c
om
m
unic
at
ions
,
vol.
10
,
no
.
2
,
pp
.
22
-
26
,
2011
.
Evaluation Warning : The document was created with Spire.PDF for Python.