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.
9
, No
. 6, Dece
m
ber
20
19, p
p.
5615
~
5627
IS
S
N: 20
88
-
8708, DO
I:
10
.11
591/ijece
.v9i6
.
pp5615
-
56
27
5615
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Cryptog
ra
ph
ic a
daptati
on
of
th
e mi
dd
le square gen
erator
Ha
n
a Ali
-
P
ac
ha
1
,
N
aima
H
ad
j
-
S
aid
2
,
Adda
Ali
-
P
acha
3
,
Moham
ad A
fe
ndee
Moham
e
d
4
, Must
afa
Mam
at
5
1
,2,3
La
b.
of
Cod
i
ng
and
Se
cur
i
t
y
of
Inform
at
ion
(
LACOS
I),
Univer
sit
y
of
Sci
enc
e
s a
nd
T
ec
hno
log
y
of
Oran
,
Alg
er
ia
4,5
Facul
t
y
of
Inf
orm
at
ic
s a
nd
Co
m
puti
ng,
Univer
siti
Sult
an Za
in
a
l
Abidin
,
B
esut Cam
pus,
Malay
s
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
pr
5
, 2
01
9
Re
vised
Ju
l
19
,
201
9
Accepte
d
J
ul
28
, 2
01
9
Curre
ntly
,
cr
y
p
togra
ph
y
pl
a
y
s
a
m
aj
or
rol
e
in
var
ious
co
m
pute
r
and
te
chno
logi
c
al
ap
pli
c
at
ions.
W
it
h
the
high
num
ber
of
i
nt
ern
et
users
,
th
e
use
of
cr
y
p
togra
ph
y
t
o
provide
info
rm
at
ion
sec
ur
ity
h
as
be
come
a
prio
rity
.
Sever
al
appl
i
ca
t
i
ons
such
as
e
-
ma
il
s,
el
e
ct
roni
c
banki
ng,
m
edica
l
dat
ab
ase
s
and
e
-
c
om
m
erce
req
uir
e
the
exc
hang
e
of
p
riva
t
e
information.
W
hil
e
,
if
the
connect
io
n
is
not
sec
ure
,
thi
s
sensiti
ve
i
nform
at
ion
ca
n
be
attac
k
ed
.
The
best
-
known
cr
y
p
togra
ph
ic
s
y
stems
rely
on
the
generation
of
ran
dom
num
ber
s,
which
are
funda
m
ent
a
l
in
var
ious
cr
y
p
to
gra
phi
c
appl
i
c
at
ions
such
as
ke
y
gen
era
t
i
on
and
dat
a
en
cr
y
p
ti
on
.
In
what
foll
ows
,
we
want
to
use
pseudo
-
ran
dom
seque
nce
s
gen
er
at
ed
b
y
the
m
id
dle
square
gen
er
at
or.
In
thi
s
work,
it
m
ust
be
poss
ibl
e
to
es
ti
m
at
e
th
at
th
e
dat
a
produc
ed
h
as
ran
dom
cha
ra
cteri
sti
cs,
k
nowing
tha
t
th
e
al
gorit
hm
used
i
s
det
erministi
c
.
Overa
ll,
thi
s
pape
r
foc
uses
on
the
te
sting
of
ps
eudo
-
ran
dom
seque
nce
s
gene
rated
b
y
the m
iddl
e
square
g
ene
r
at
or
a
nd
it
s use
in
data
enc
r
y
p
ti
on
.
Ke
yw
or
d
s
:
Crypto
gr
a
phy
Entr
op
y
t
est
Mi
dd
le
s
qu
a
re
g
ene
r
at
or
Ra
ndom
n
ic
kna
m
es
V
on
n
e
um
ann
Copyright
©
201
9
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
:
Han
a
Ali
-
Pach
a
,
Lab. o
f
C
od
i
ng and Sec
uri
ty
of In
form
ation
(
LACO
SI)
,
Un
i
ver
sit
y o
f S
ci
ences a
nd Te
chnolo
gy
of Or
an
,
Po
B
ox
1505
Or
a
n
M’
Na
oue
r
Al
ger
ia
.
Em
a
il
:
han
a.ali
pach
a
@
un
i
v
-
ust
o.dz
1.
INTROD
U
CTION
The
de
velo
pme
nt
of
al
gorit
hm
s
gen
erati
ng
ps
e
udo
-
ra
ndom
nu
m
ber
s
is
ver
y
m
uch
relat
ed
to
that
of
crypto
gr
a
phy
[
1
-
8].
Es
pecial
ly
,
t
he
m
il
i
ta
r
ily
i
m
po
rtance
s
uch
as
c
omm
un
ic
at
ion
a
nd
m
on
it
or
ing
[9
-
10
]
of
this
sci
ence
ha
v
e
m
otivate
d
m
any
researc
he
s
thr
ough
out
histor
y.
But
t
he
re
is
no
ps
e
udo
-
ra
ndom
al
gorithm
that
can
esca
pe
from
sta
ti
s
tical
analy
sis,
especial
ly
beca
us
e
the
"seed"
m
us
t
theor
et
ic
al
ly
it
sel
f
be
rand
om
,
an
d
the
al
gorithm
us
ed
canno
t
be
init
ia
li
zed
by
it
sel
f.
The
current
cry
ptogra
phic
gen
e
ra
tors
are
th
us
obli
ge
d
to
include
el
e
m
ent
that
is
no
t
gen
e
rated
in
a
deter
m
inist
i
c
way.
On
e
th
us
m
ov
es
towa
rd
s
hybr
i
d
ge
ne
rators,
fou
nd
i
ng
a
r
obus
t
al
gorithm
of
pse
ud
or
a
nd
om
nu
m
ber
gen
e
rati
on
by
init
ia
li
zi
ng
it
sel
f
thr
ough
a
physi
cal
m
eans of
c
hance p
rod
uction.
On
the
oth
e
r
ha
nd,
the
im
age
bec
om
es
m
or
e
an
d
m
or
e
in
di
sp
e
ns
able
i
n
se
ver
al
fiel
ds
a
nd
esse
ntial
ly
in
com
m
un
ic
at
ion
betwee
n
pe
op
le
.
I
nd
ee
d,
t
he
ex
pone
ntial
dev
el
op
m
e
nt
of
c
omm
un
ic
ation
m
edia
on
t
he
one
hand,
a
nd
di
gital
storag
e
m
edia
on
the
oth
e
r
hand,
ha
ve
en
or
m
ou
sly
trans
form
ed
the
way
we
co
m
m
un
ic
at
e.
These
ne
w
te
c
hnologies
a
re
ba
sed
e
ssentia
ll
y
on
the
ef
fici
ent
e
xch
a
nge
a
nd
st
or
a
ge
of
m
ultim
edia
data
and
i
n
par
ti
cu
la
r digit
al
i
m
ages,
he
nc
e the
nee
d
f
or
i
m
age en
c
rypt
ion
al
gorithm
s.
In
w
hat
f
ollo
w
s,
we
disc
us
s
t
he
desig
n
a
nd
reali
zat
ion
of
t
he
m
idd
le
-
squ
are
ge
ne
rato
r
i
n
the
co
ntext
of
pr
oducin
g
pse
udo
-
ra
ndom
seq
uen
ces
.
The
si
m
plistic
of
m
idd
le
-
squa
re
gen
e
rato
r
can
be
ex
plo
it
ed
f
or
go
od
rand
om
sequ
ence.
I
n
or
der
t
o
eval
uate
these
sequ
e
nces
and
validat
e
our
gen
e
rato
r,
we
i
m
ple
m
ented
five
sta
ti
c test
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.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
5615
-
5627
5616
This
pa
per
is
structu
re
d
as
fo
ll
ows.
Sect
io
n
2
intr
oduces
so
m
e
basic
pri
nciples
of
ps
e
udo
-
ra
ndom
nu
m
ber
ge
ne
ra
tor.
Sect
i
on
3
di
scusses
the f
or
m
at
ion
of
m
idd
le
square g
e
ne
rator,
outl
inin
g
the p
r
ocedu
r
e
with
an
e
xam
ples.
S
ect
ion
4
pr
ese
nts
va
rio
us
ty
pe
s
of
te
sti
ngs
pur
posedly
to
e
valuate
th
e
qual
it
y
of
our
pr
opos
e
d
m
et
ho
d.
Sect
i
on
5
repo
rts
the
fin
dings
ac
cordin
g
to
va
r
iou
s
te
sts
pro
po
s
ed
earli
er
Sect
ion
6
c
oncl
ud
es
our
st
ud
ie
s
.
2.
PSEU
DO
-
R
A
NDOM
GE
N
ERATO
RS
The
nee
d
f
or
rando
m
num
ber
s
is
felt
in
m
any
app
li
cat
ion
s
of
crypto
gr
a
phy.
In
com
m
on
crypto
gr
a
phic
syst
e
m
s,
the k
e
ys (num
ber
s)
t
hat are use
d
m
us
t be
r
a
ndom
l
y gen
e
rated
. For
e
xam
ple, when o
ne
consults
on
th
e
In
te
rn
et
his
e
-
m
ail
accounts
or,
w
hen
one
carries
out
a
n
orde
r
by
I
nter
net
s
om
e
"sensit
ive"
inf
or
m
at
ion
(your
acce
ss
c
od
e
or
y
our
cre
di
t
card
nu
m
ber
)
m
us
t
rem
a
in
confide
ntial
to
ens
ur
e
a
uth
e
nt
ic
ity,
nobody
s
hould
be
able
t
o
ac
cess
the
acco
unts
or
orde
r
w
it
h
the
car
d.
T
o
ens
ure
these
functi
ons,
I
nt
ern
e
t
protoc
ols
hav
e
bee
n
pu
t
in
pl
ace.
They
al
l
ow
yo
u
to
ent
er
yo
ur
co
des
on
a
web
pa
ge
with
ou
t
t
he
risk
of
an
outsi
de
pe
rson
ha
ving
a
ccess
to
them
.
These
pr
otoc
ols
us
e
ra
nd
om
nu
m
ber
s
to
enc
rypt
dat
a
and
pr
e
ve
nt sp
yi
ng
.
2.1.
Def
ini
ti
on
of
a ran
do
m
seq
u
ence
In
m
at
he
m
at
ics,
a
rand
om
se
qu
e
nce,
or
ra
ndom
infin
it
e
sequ
e
nce,
is
a
s
equ
e
nce
of
sy
m
bo
ls
of
an
al
ph
a
bet
ha
vin
g
no
str
uctu
re,
no
re
gu
la
r
it
y,
or
identif
ia
ble
pr
e
dicti
on
r
ule
[
11
-
12
]
.
Su
ch
a
seq
uen
c
e
corres
ponds
to
the
intuit
ive
no
ti
on
of
nu
m
ber
s
dr
aw
n
at
ran
dom
.
A
sequ
e
nce
of
rand
om
nu
m
ber
s
is
a
sequ
e
nce
of
nu
m
ber
s
ra
ndom
ly
cho
se
n.
T
his
seq
uen
c
e
ha
s
the
pro
per
ty
that
we
canno
t
pr
edict
the
nu
m
ber
s
to co
m
e from
t
he
al
rea
dy
known
num
ber
s
, w
hateve
r
they
a
r
e
[
13
]
.
2.2.
Def
ini
ti
on
of
a pseud
o
-
ra
nd
om
seque
nce
The
ps
e
udo
-
ra
ndom
te
r
m
is
us
e
d
in
m
at
hem
at
ic
s
and
co
m
pu
te
r
sci
ence
to
de
sig
nate
a
seq
uen
ce
of
nu
m
ber
s
that
a
ppr
oach
es
a
st
at
ist
ic
ally
per
f
ect
hazard
[14
-
16
]
.
By
the
al
gorithm
ic
pr
oce
sses
us
e
d
to
cr
eat
e
i
t
and
t
he
s
ource
s
us
e
d,
t
he
se
qu
e
nce
ca
nnot
be
c
om
plete
l
y
con
si
der
e
d
a
s
truly
ra
ndom
.
A
ps
eu
do
-
r
andom
seq
uen
ce
(Pse
udo
Ra
ndom
Sequ
e
nce
i
n
E
ngli
sh
)
[
17]
is
a
seq
ue
nce
of
intege
rs
x
0
,
x
1
,
x
2
,
...
ta
king
it
s
val
u
es
in
the
set
M
=
{0,
1,
2,
...
,
m
-
1}.
The
te
rm
x
n
(n
>
0)
is
the
resu
lt
of
a
cal
culat
ion
(to
be
de
fine
d)
on
the
pre
vious
te
rm
(s)
.
The
first
te
r
m
x
0
is
called
the
se
ed
.
W
it
h
the
sam
e
init
ial
seed,
the
seq
uen
ce
of
ps
e
udoran
do
m
nu
m
ber
s
produce
d
by
the
ps
eu
dora
ndom
nu
m
ber
generator
is
dete
r
m
inist
i
c
and can
the
refor
e
b
e
r
e
pro
duced.
A
ps
e
udo
-
ra
ndom
nu
m
ber
ge
ner
at
or
is
a
n
al
gorithm
that
ge
ner
at
es
a
seq
uen
ce
of
num
ber
s
wit
h
certai
n
pro
per
t
ie
s
of
c
han
ce
.
The
pri
nciple
of
these
ge
ne
r
at
or
s
is
to
crea
te
from
an
init
ia
l
seed,
a
s
o
-
cal
le
d
ps
e
udo
-
ra
ndom
nu
m
ber
, whi
ch has
no appa
ren
t l
ogic
al
or
arit
hm
etic connecti
on w
it
h t
he
seed. T
his g
e
ner
at
e
d
nu
m
ber
is
then
us
ed
to
create
a
second
ps
e
udo
-
ra
ndom
num
ber
.
W
e
ca
n
thu
s
rec
ursivel
y
gen
erate
a
se
ries
of
nu
m
ber
s
that
do
not
ap
pea
r
to
ha
ve
a
ny
logi
cal
li
nk
in
t
he
ir
seq
uen
ce
,
but
w
hich
a
re
in
fact
al
l
obta
ined
by
a
determ
inist
ic
f
or
m
ula.
T
his
cl
ass
of
ge
ne
rators
is
easy
to
im
ple
m
ent
and
al
lo
ws
high
th
r
oughputs
wh
il
e
pro
du
ci
ng
s
uites
that
hav
e
go
od
sta
ti
sti
cal
pr
operti
es.
It
is
therefo
re
ver
y
su
it
able
for
ap
plica
ti
on
s
that
do
no
t
require
the
un
pr
e
dicta
bili
ty
of
the
su
it
es
(s
uch
as
dig
it
al
si
m
ulati
on
),
but
can
al
so
be
us
e
d
in
cry
ptogra
phic
app
li
cat
io
ns
prov
i
ded that ce
r
ta
in cr
it
eria are
m
e
t.
3.
MIDDLE
S
Q
UARE GE
NE
RA
TO
R
This
gen
e
rato
r
is
base
d
on
th
e
m
edian
sq
ua
re
m
e
t
ho
d,
kn
own
in
t
he
E
ng
li
s
h
li
te
ratur
e
al
so
as
m
idd
le
sq
ua
re,
wa
s
in
ven
te
d
by
the
Am
erican
-
H
ungar
ia
n
m
at
he
m
at
ic
ia
n
an
d
physi
ci
st
Jo
hn
V
on
Ne
um
ann
i
n
1946.
The
m
idd
le
s
quare
is
c
onsid
ered
as
t
he
fir
st
m
e
tho
d
of
autom
at
ic
generati
on
of
ps
e
udoran
dom
num
ber
s.
T
he
pri
nciple
of
this
m
et
ho
d
is
ver
y
sim
ple,
we
ge
ne
r
at
e
a
sequ
enc
e
of
num
ber
s
each
ha
ving
2k
dig
it
s
(ev
e
n
nu
m
ber
)
.
The
su
cc
esso
r
of
a
num
ber
in
this
se
qu
e
nc
e
is
obta
ined
by
raisi
ng
t
his
nu
m
ber
s
quare
d
a
nd
then ret
ai
ning t
he 2k m
idd
le
num
ber
s.
T
he pr
incipl
e of t
his
m
et
ho
d
is
d
esc
ribe
d by the
fo
l
lowing
ste
ps
:
Start wit
h a see
d (a
nu
m
ber
)
of
n
-
dig
it
(n
dig
i
ts),
Ra
ise
squar
e
d
t
o get a
nu
m
ber o
f 2n di
gits, a
dd zer
os
i
f nec
essary,
Take t
he
m
idd
le
n
num
ber
s as
the
nex
t
ra
ndom
n
u
m
ber
,
Re
peat 1
-
2
-
3
(t
he pr
ocess
).
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
-
8
708
Crypto
graphic
adaptati
on
of t
he mid
dle s
quare
g
e
ner
ato
r
(Han
a
Ali
-
P
ac
ha)
5617
Exam
ple
:
If
w
e
gen
e
rate
4
-
di
git
nu
m
ber
s
f
r
om
35
67,
we
ge
t
72
34
as
the
nex
t
val
ue
sinc
e
the
3
56
7
sq
ua
re
is
e
qu
al
to
1272
3489,
c
on
ti
nuin
g
in
t
he
sam
e
way,
the
ne
xt
num
ber
will
be
3307
,
3567,
7234,
3307,
9362
,
6470,
8609,
11
48,
31
79,
1060,
1236,
5276,
836
1,
9063,
1379,
9016,
28
82,
30
59,
3574,
7734,
81
47
,
3736,
9576,
6997,
95
80,
77
64,
2796,
81
76,
8469,
7239
,
4031,
24
89,
1951,
8064,
280,
78
4,
61
46,
7733
,
7992,
8720,
384,
1474,
1726,
9790,
84
41,
2504,
27
00,
2900,
4100,
8100,
61
00,
2100,
4100
,
a
nd
so
on
.
The
res
ulti
ng
s
equ
e
nce
retu
r
ns afte
r
46 iterat
ion
s
in
a
p
e
rio
dic or
bit
.
4.
TE
ST A GE
N
ERATO
R
Determ
ining
w
hethe
r
a
gen
e
r
at
or
is
rand
om
or
not
is
a
tric
ky
pro
blem
.
Indeed,
the
re
is
no
unive
rsal
te
st,
that
can
s
ay
with
certai
nty
that
a
generator
is
rand
om
[18
-
19]
.
T
he
pr
in
ci
ple
is
to
sho
w
that
it
is
no
t
biased
by
study
ing
the
pro
pe
rtie
s
of
the
num
ber
s
it
gen
erates.
I
n
pr
a
c
ti
ce,
a
ran
dom
gen
e
rator
pro
du
ce
s
a
sequ
e
nce
of
nu
m
ber
s
with
pro
per
ti
es
of
unpre
dicta
bili
ty
[
20
-
22
]
an
d
ind
e
pe
nd
e
nce,
and
fo
ll
ows
a
certai
n
dis
trib
ution
(unif
or
m
in
cryptography,
Ga
ussi
an
in
te
le
co
m
m
un
ic
at
ion
s,
et
c.).
The
e
va
luati
on
of
the
r
andom
qu
al
it
y
of
a
ge
ner
at
or
th
us
passes
th
rou
gh
the
con
tr
ol
of
the
prop
e
rtie
s
of
the
seq
ue
nc
e
that
it
gen
e
rates.
This
is
achieve
d
thr
ough
sta
ti
sti
cal
te
sts
that
com
par
e
the
per
f
or
m
ance
of
the
gen
e
rato
r
s
tud
ie
d
c
om
par
ed
to
tho
se
, th
e
or
et
ic
al
.
The
pur
pose of
stat
istica
l
te
sts
is to
m
easur
e the quali
ty
o
f
a
rand
om sequ
e
nc
e.
W
e can
co
nc
lud
e that
a
su
it
e
ge
ner
at
ed
by
a
PR
NG
is
rando
m
and
of
good
qual
it
y,
if
it
sat
isfie
s
these
te
sts.
Th
eref
or
e
,
a
sta
ti
sti
cal
te
st
can
in
no
way
guara
ntee
that
a
giv
e
n
se
qu
e
nce
is
ra
nd
om
.
The
on
ly
inf
or
m
at
ion
tha
t
a
sta
ti
st
ic
al
test
can
pro
vid
e
is
that
the
seq
ue
nce
seem
s
ran
do
m
.
Seve
ral
sta
ndar
ds
e
xist
to
evaluate
a
nd
c
erti
fy
the
qual
i
ty
of
ps
e
udor
a
ndom
nu
m
ber
ge
ne
rators.
W
e
will
pr
ese
nt
so
m
e
te
s
ts
us
ed
to
eval
uate
the
perform
a
nce
of
our ge
ner
at
or
.
4.1.
Entropy
tes
t
An
e
ntr
op
y
ca
lc
ulate
s
the
am
ou
nt
of
in
f
orm
ation
co
ntained
i
n
a
file
.
The
file
is
co
ns
ide
red
a
s
a seq
ue
nce
of
words
of
1 o
r 8
bits. T
he
e
ntr
opy i
s calc
ulate
d
as
sho
wn b
el
ow
:
(
)
=
−
∑
(
)
.
2
(
2
−
1
=
0
(
)
)
W
he
re
X
is
the
stud
ie
d
s
ource
,
P
i
is
the
pro
ba
bili
ty
of
ap
pe
aran
ce
of
the
word
i
of
n
bit
s.
T
he
com
pu
t
at
ion
of
the
ent
ropy
m
akes
t
he
m
inim
u
m
nu
m
ber
of
bits
per
w
ord
c
onta
inin
g
al
l
the
in
f
orm
at
ion
.
F
or
e
xam
ple,
if
the
e
ntr
opy
is
6
bits/
w
ord
for
8
-
bit
w
ords
the
n
2
bits
car
ry
re
dund
ant
in
form
at
io
n
a
nd
the
file
co
ul
d
theo
reti
cal
ly
b
e com
pr
esse
d
t
o
th
ree
quarter
s of its
or
igi
nal size
.
4.2.
Mean,
st
an
d
ar
d
de
viation
and
au
t
o
-
correl
at
i
on
fa
c
t
or
This
is
the
si
m
plest
te
st
po
s
sible.
It
con
sist
s
of
cal
culat
ing
t
he
m
ean,
the
va
riance
a
nd
the
autoc
orrela
ti
on
facto
r
of
the
ps
e
udo
-
ra
ndom
sequ
e
nce.
L
e
t
x
i
,
f
or
i
=
1,
2,
.
.
,
n,
be
a
s
e
qu
e
nc
e
ob
t
a
i
ne
d
f
r
om
a
ps
e
ud
o
-
r
a
n
d
om
num
be
r
ge
ne
r
a
t
or
.
T
h
e
s
e
qu
e
nc
e
of
u
i
=x
i
/
n,
f
or
i
=
1,
2,
.
.
,
n
i
s
a
se
qu
e
nc
e
o
f
ps
e
ud
o
-
r
a
nd
om
num
be
r
s
di
s
t
r
i
bu
t
e
d
u
ni
f
or
m
ly
i
n
t
he
i
nt
e
r
va
l
[
0,
1]
.
F
or
a
r
a
nd
om
se
qu
e
nc
e
,
t
he
s
e
f
a
c
t
or
s
t
e
nd
t
o
w
a
r
ds
i
de
a
l
va
l
ue
s
,
w
hi
c
h
a
r
e
t
hu
s
s
uf
f
i
c
i
e
nt
t
o
c
om
pa
r
e
w
i
t
h
th
e
c
a
l
c
ul
a
t
e
d
va
l
ue
s
f
or
t
he
f
ol
l
ow
i
ng
U
.
I
de
a
l
l
y
,
w
e
m
us
t
f
i
nd
th
e
t
hr
e
e
va
l
ue
s
be
l
ow
:
a)
Av
e
ra
ge:
=
1
.
∑
=
1
=
1
2
.
b)
Var
ia
nce:
=
1
.
∑
(
2
−
2
)
=
1
12
.
c)
Au
t
o
C
orrelat
ion
:
=
1
.
∑
(
∗
+
1
)
=
1
4
−
1
.
4.3.
Spectr
al
te
st
The
i
dea
is
to
visu
al
ly
re
pr
es
ent
the
s
eq
ue
nc
e
of
pse
ud
o
-
r
andom
nu
m
bers
in
1
dim
e
ns
i
on
(
1D),
2D
and
3D.
F
or
3D
Re
pr
ese
ntati
on
,
th
ree
c
onse
cutive
values
will
be
t
he
c
oord
i
nates
of
a
point
i
n
s
pace.
We
l
ook
at
wh
et
her
t
he
po
i
nts
are
eve
nly
distrib
uted
in
a
c
ub
e
.
B
y
t
ur
ni
ng
t
he
c
ub
e
a
s
s
ho
w
n
i
n
F
i
gu
r
e
1
,
o
ne
s
e
e
s
a
n
un
de
s
i
r
a
bl
e
e
f
f
e
c
t
a
pp
e
a
r
,
t
ha
t
i
s
t
he
pl
a
ns
of
M
a
r
s
a
gl
i
a
[20]
.
I
t
i
s
c
l
e
a
r
be
l
ow
t
ha
t
t
he
poi
nt
s
a
r
e
l
oc
a
te
d
on
pl
a
ns
.
I
n
f
a
c
t
,
a
l
l
l
i
ne
a
r
c
on
gr
u
e
nt
i
a
l
ge
ne
r
a
t
or
s
(
L
C
G
)
s
uf
f
e
r
f
r
om
t
hi
s
e
f
f
e
c
t
(
t
hi
s
i
s
du
e
to
t
he
f
a
c
t
t
ha
t
w
e
do
no
t
ge
ne
r
a
t
e
a
l
l
r
e
a
l
s
,
bu
t
o
nl
y
f
r
a
c
t
i
on
s
)
.
T
he
sm
a
ll
e
r
t
he
i
nt
er
-
pl
a
na
r
di
s
t
a
nc
e
,
t
he
be
t
t
e
r
t
he
ge
ne
r
a
t
or
.
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.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
5615
-
5627
5618
Figure
1. Ma
rs
aglia
shots
4.4.
Pok
er
tes
t
The
idea
of
thi
s
te
st
is
to
com
par
e
t
he
the
or
e
ti
cal
fr
eq
uen
ci
es
of
ha
nds
in
poke
r
with
t
he
fr
e
qu
e
ncies
ob
s
er
ved
by
si
m
ula
ti
ng
these
ha
nd
s
(a
ha
nd
is
a
set
of
ca
r
ds
)
.
A
re
su
lt
ca
n
be
c
on
si
der
e
d
as
a
n
orde
re
d
li
st
of
4 digit
s.
T
he
re
are in
all
10
4
.
The
t
heoreti
cal
proba
bili
ti
es o
btained
are
as
f
ollows:
a)
F
or
4
di
f
f
e
r
e
nt
di
gi
t
s
(
e
g.
15
74
)
,
t
he
n
um
be
r
of
p
os
s
i
bl
e
c
a
s
e
s
i
s
10
*
9
*
8
*
7,
10
f
o
r
t
he
f
i
r
s
t
nu
m
be
r
,
9
f
or
t
he
ne
xt
,
a
nd
s
o
o
n
t
he
p
r
ob
a
bi
l
i
ty
i
s
t
he
r
e
f
or
e
(
1
0
*
9
*
8
*
7)
/
1
00
0
0.
b)
F
or
a
pa
i
r
,
t
y
pe
A
B
C
C
(
e
g.
4
84
9)
,
w
e
ha
ve
10
*
9
*
8
*
1
w
a
y
s
t
o
m
a
ke
it
,
t
o
m
ul
t
i
pl
y
by
t
he
nu
m
be
r
o
f
w
a
y
s
t
o
pl
a
c
e
t
he
pa
i
r
a
m
on
g
t
he
4
p
os
s
i
bl
e
pl
a
c
e
s
:
T
he
pr
o
ba
bi
l
i
ty
i
s
:
(
4
2
)
∗
10
∗
9
∗
8
10
4
.
c)
F
or
a
do
ub
l
e
p
a
i
r
,
t
y
pe
A
A
B
B
(
e
x:
73
37
)
,
w
e
ha
ve
10
w
a
y
s
t
o
c
ho
os
e
t
he
f
i
r
s
t
nu
m
be
r
,
t
he
n
9
w
a
y
s
t
o
c
ho
os
e
t
he
s
e
c
on
d
o
ne
(
w
hi
c
h
m
us
t
be
di
f
f
e
r
e
nt
)
,
i
n
a
l
l
10
*
9
=
90
c
ho
i
c
e
s
,
t
o
m
ul
ti
pl
y
by
t
he
nu
m
be
r
o
f
w
a
y
s
t
o
pl
a
c
e
t
he
s
e
nu
m
be
r
s
a
m
on
g
t
he
4
p
os
s
i
bl
e
pl
a
c
e
s
(
1
2
∗
(
4
2
)
=
3)
,
t
he
pr
ob
a
bi
l
i
ty
i
s
=
1
2
∗
(
4
2
)
∗
10
∗
9
10
4
.
d)
F
or
t
h
r
e
e
i
de
n
t
i
c
al
di
gi
t
s
,
t
yp
e
A
A
A
B
(
e
x
:
55
1
5)
,
t
he
r
e
a
r
e
1
0
*
9
w
a
y
s
t
o
m
a
ke
i
t
,
t
o
m
ul
t
i
pl
y
by
t
he
nu
m
be
r
of
w
a
y
s
t
o
pl
a
c
e
t
he
t
hr
e
e
di
gi
t
s
am
on
g
t
he
4
po
s
s
i
bl
e
pl
a
c
e
s
,
th
e
pr
ob
a
bi
l
i
ty
is
=
(
4
3
)
∗
10
∗
9
10
4
.
e)
Finall
y, f
or
f
our
ide
ntica
l digit
s (
e
x: 44
44), i
t i
s
worth
10
10
4
=
0
.
001
.
No
te
t
hat the
s
um
o
f
these
f
i
ve
proba
bili
ti
es g
ives
1.
5.
RESU
LT
S
AND INTE
RP
RETATIO
N
The
resu
lt
s
a
r
e
m
ade
un
de
r
a
PC
TO
S
H
IB
A,
on
w
hich
is
instal
le
d
W
i
ndows
7
(
32
bits),
R
AM:
4.00
GB,
AMD
E
-
45
0
A
PU
proces
sor
with
Ra
deon
(
TM)
HD
G
raphics
1.65
G
Hz.
T
he
functi
ons
devel
op
e
d
us
in
g
M
ATL
AB
[
23
-
24]
al
low
us
to
ge
ne
rate
pse
ud
o
-
r
andom
sequ
e
nc
es
an
d
a
naly
ze
their
pe
rform
ance.
Figure
2
i
ll
us
tr
at
es
the
se
que
nce
of
ps
e
udo
rand
om
nu
m
ber
s
acc
ordin
g
t
o
the
Von
Ne
um
ann
m
edial
sq
ua
re
m
et
ho
d,
we
t
ake
as
par
am
et
er
n
represe
nting
t
he
le
ngth
of
t
he
s
e
qu
e
nce
n
=
6553
6,
a
nd
t
he
seed
x0 =
236589
741.
Af
te
r gene
ra
ti
ng
the
s
uite,
we
a
pp
li
ed
stat
ist
ic
al
t
est
s to
evaluate i
t.
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
-
8
708
Crypto
graphic
adaptati
on
of t
he mid
dle s
quare
g
e
ner
ato
r
(Han
a
Ali
-
P
ac
ha)
5619
Figure
2. Ra
nd
om
data su
it
e
5.1.
Te
st 1:
mea
n
,
stand
ard devi
at
i
on
an
d
aut
o
-
c
orrela
tion f
actor
Table
1
il
lustra
te
s sev
e
ral te
sts wit
h t
he
sam
e seed
and
of di
ff
ere
nt
value n
:
Table
1
.
Me
a
n,
sta
ndar
d dev
ia
ti
on
a
nd au
t
o
-
c
orrelat
ion
fac
tor
Test
Seed
n
Mean
Au
to
-
Co
rr
elatio
n
Variance
1
5746
5000
0
.52
1
7
4
8
0
.27
2
4
0
4
0
.05
2
6
8
9
2
1
0
0
0
0
0
.52
1
7
9
7
0
.27
2
3
6
3
0
.05
2
5
2
0
3
2
0
0
0
0
0
.52
1
8
2
1
0
.27
2
3
4
3
0
.05
2
4
3
5
4
6
5
3
3
6
0
.52
1
8
3
8
0
.27
2
3
2
9
0
.05
2
3
7
6
5
2
3
6
5
8
9
7
4
1
5000
0
.50
9
8
3
4
0
.26
0
2
5
8
0
.08
2
7
5
8
6
1
0
0
0
0
0
.5
0
7
5
8
8
0
.25
7
6
2
9
0
.08
3
0
9
7
7
2
0
0
0
0
0
.50
4
4
8
1
0
.25
4
2
4
8
0
.08
3
0
5
4
8
6
5
3
3
6
0
.50
4
0
3
2
0
.25
4
1
0
5
0
.08
3
5
2
3
9
6
5
3
3
6
0
.49
4
6
6
0
0
.24
4
0
5
6
0
.08
3
4
6
2
10
2
3
6
5
8
9
7
4
1
5000
0
.50
9
8
3
4
0
.26
0
2
5
8
0
.08
2
7
5
8
11
2
3
6
6
8
4
7
4
1
6000
0
.50
1
7
2
8
0
.25
1
2
0
7
0
.08
3
3
0
0
12
2
8
9
7
8
4
7
4
1
7000
0
.49
6
6
6
4
0
.24
7
9
7
7
0
.08
4
1
1
9
13
2
3
6
5
8
9
7
4
1
5000
0
.50
9
8
3
4
0
.26
0
2
5
8
0
.08
2
7
5
8
14
6000
0
.50
9
9
8
4
0
.25
9
9
6
0
0
.08
3
1
8
8
15
7000
0
.51
1
1
3
0
0
.26
0
5
9
8
0
.08
3
0
1
2
No
te
t
hat
Test
9
c
orres
ponds
to
t
he
sta
nda
r
d
se
quence
of
the
co
ntin
uatio
n
of
Test
8.
A
ccordin
g
t
o
the
resu
lt
s
obta
in
ed
an
d
afte
r
sever
al
te
sts,
it
has
been
fou
nd
that
with
n
suffici
entl
y
la
rg
e,
and
with
su
f
fici
ently
la
rg
e
see
d
(
of
la
r
ge
di
git
or
>
6)
the
cal
c
ulate
d
val
ues
gr
a
du
al
ly
te
nd
towa
r
ds
the
i
deal
as
sh
ow
n
in Figu
re
3
(
Te
st 8).
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.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
5615
-
5627
5620
Figure
3
.
The
gr
a
phic
al
curv
e o
f
the
thr
e
e c
al
culat
ed value
s
5.2.
Te
st 2:
th
e
spe
ctra
l
test
This
te
st
ai
m
s
to
a
naly
ze
the
di
stribu
ti
on o
f
points
i
n
2D
a
nd 3
D
s
pace,
Fi
gure 4
il
lust
rates
the v
is
ual
represe
ntati
on
of
the
se
qu
e
nc
e
gen
e
rated
(T
est
8)
.
I
n
the
3D
represe
ntati
on,
by
ro
ta
ti
ng
the
cub
e,
on
e
visu
al
ly
no
ti
ces
the
ab
sence
of
t
he
Ma
rsag
li
a
pla
nes
a
nd
the
points
a
re
unif
or
m
ly
distribut
ed,
this
im
pli
es
that
the test
ed
s
uite
is ra
ndom
.
Figure
4.
Distri
bu
ti
on
of
value
s in 2D a
nd 3D
5.3.
Te
st 3:
freq
ue
ncy tes
t
Re
cal
l
that
this
te
st
ev
al
uates
wh
et
her
the
se
qu
e
nce
is
ra
ndom
by
co
m
par
ing
the
cal
c
ulate
d
P
-
val
ue
with
th
e
sig
nifi
cance
t
hr
es
hold
α
(
ta
king
α
=
0.0
1).
I
f
P
-
val
ue
>
α
,
the
n
t
he
seq
ue
nce
is r
andom
oth
er
wi
se
it
is
no
t
ra
ndom
.
The
sta
ndar
di
zed
P
-
value
c
olu
m
n
corresponds
to
the
norm
a
li
zed
sequ
ence
(in
m
od
256).
The res
ults are
pr
ese
nted
in T
able 2.
Table
2
.
Fr
e
quency te
st res
ults
Test
n
Seed
P
-
v
alu
e
P
-
v
alu
e
(no
r
m
alized
)
Interpretatio
n
1
5000
2
3
6
5
8
9
7
4
1
0
.92
5
1
0
.78
7
8
+
2
1
0
0
0
0
0
.98
8
5
0
.56
3
7
+
3
2
0
0
0
0
1
0
.38
0
6
+
4
6
5
5
3
6
1
.54
0
3
0
.04
4
7
+
No
te
t
hat
f
or
al
l
four
te
sts,
the
P
-
va
lue
is
gr
eat
er
tha
n
α
,
so
we
ac
cept
the
null
hy
po
t
hesis
wh
i
c
h
sta
te
s that "t
he se
quence
is ra
ndom
"
.
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
-
8
708
Crypto
graphic
adaptati
on
of t
he mid
dle s
quare
g
e
ner
ato
r
(Han
a
Ali
-
P
ac
ha)
5621
5.4.
Te
st 4:
e
nt
r
opy
test
In
Ta
ble
3
we
find
t
he
ent
ropy
value
s
f
or
the
norm
al
iz
ed
sequ
e
nces
with
dif
fe
ren
t
n,
the
highe
r
the
e
ntropy,
th
e
m
or
e
ra
ndom
the
s
eq
ue
nce
i
s.
I
n
the
i
deal
c
ase
the
entr
opy
val
ue
is
e
qual
to
8.
N
ote
that
eac
h
tim
e the v
al
ue of
n is i
nc
rease
d,
t
he value
of
the en
t
ropy als
o
inc
reases
.
Table
3
.
E
ntr
opy t
est
r
es
ults
Test
n
Seed
Entro
p
y
(
n
o
r
m
aliz
ed
)
1
5
000
2
3
6
5
8
9
7
4
1
7
.96
2
8
2
1
0
0
0
0
7
.97
9
3
3
2
0
0
0
0
7
.98
9
9
4
6
5
5
3
6
7
.99
3
4
5.5.
Te
st 5:
poker
test
The
pu
rpose
of
this
te
st
is
to
com
par
e
the
theo
reti
cal
prob
a
bili
ti
es
Pt
of
po
ker
hands
with
the
pro
bab
il
it
ie
s
ob
s
er
ved
Po
by
si
m
ulati
ng
these
ha
nds
(a
s
et
of
car
ds).
N
oting
here
that
the
Po
a
re
obta
ine
d
by d
i
vid
in
g
t
he
observe
d fr
e
quencies
of eac
h
case
on t
he
t
otal n
um
ber
n.
5.5.1.
For 4's h
and
This ti
m
e w
e cal
culat
e the the
or
et
ic
al
pr
ob
a
bi
li
t
ie
s,
with
k
=
4
s
how
n
i
n
Ta
ble 4
.
Table
4
.
T
he
oret
ic
al
pr
oba
bili
ti
es f
or
han
d p
ok
e
r of 4
4
dif
f
erent car
d
s
1
pair on
ly
2
d
istin
cts
p
airs
3
iden
tical car
d
s
4
iden
tical car
d
s
n
Pt
0
.50
4
0
.43
2
0
.02
7
0
.03
6
0
.00
1
Po
0
.46
0
0
0
.44
0
0
0
.06
0
0
0
.03
0
0
0
.01
0
0
100
0
.00
9
2
0
.98
8
8
0
.00
1
2
0
.00
0
6
0
.00
0
2
5000
0
.00
4
6
0
.99
4
4
0
.00
0
6
0
.00
0
3
0
.00
0
1
1
0
0
0
0
0
.00
0
2
3
0
.99
7
2
0
.00
0
3
0
.00
0
1
0
.00
0
1
2
0
0
0
0
7
.01
9
*
1
0
-
4
0
.99
9
1
9
.15
5
*
1
0
-
5
4
.57
7
*
1
0
-
5
1
.52
5
*
1
0
-
5
6
5
5
3
6
5.5.2.
For 3
hand
The
sam
e
pr
inciple
is
us
ed
to
cal
culat
e
the
t
heoreti
cal
pr
ob
abili
ti
es,
bu
t
this
tim
e
with
k
=
3.
In
thi
s
case t
he
te
st i
s
app
li
ed
to
t
he
s
ta
nd
a
rd seque
nc
e or the
num
ber
of
dig
it
s
reduced
to 3
sho
w
n
in
Ta
ble 5
.
Table
5
.
T
he
oret
ic
al
pr
oba
bili
ti
es f
or
hand p
ok
e
r
of
3
3
dif
f
erent car
d
s
1
pair on
ly
3
iden
tical car
d
s
n
Pt
0
.72
0
0
0
.27
0
0
0
.01
0
0
Po
0
.64
0
0
0
.36
0
0
0
.02
0
0
100
0
.69
8
2
0
.28
9
8
0
.01
2
0
5000
0
.70
4
2
0
.28
4
6
0
.01
1
2
1
0
0
0
0
0
.70
1
4
0
.28
6
4
0
.01
2
2
2
0
0
0
0
0
.70
3
2
0
.28
4
4
0
.01
2
4
6
5
5
3
6
5.6.
Encr
ypting
ima
ges
Using
the
MA
TLAB
softwa
r
e,
we
hav
e
devel
op
e
d
an
ap
pl
ic
at
ion
to
encry
pt
and
dec
ryp
t
an
i
m
age,
us
in
g
t
he
valu
es
pro
duced
by
the
m
idd
le
sq
ua
re
ge
ner
at
or,
as
so
ci
at
ed
w
it
h
the
X
OR
s
ymm
et
ric
encry
ption
te
chn
iq
ue.
T
he
X
OR
‘
1
’
di
git
ha
nd
le
s
the
bi
ts,
base
d
on
th
e
operati
on
or
exclusi
ve
bitwise
(
XO
R
)
as
s
how
n
in Figu
re
5.
Figure
5
.
Pr
i
nc
iple o
f
e
ncr
ypti
on
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.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
5615
-
5627
5622
We
us
e
as
a
ke
y
a
bit
string
K
of
g
ive
n
le
ng
t
h
L.
It
is
encr
y
pted
by
pe
rfor
m
ing
the
exclusi
ve
-
or
op
e
rati
on
bit
by
bit
of
t
he
ke
y
K
with
t
he
cl
ear
te
xt,
di
vid
ed
int
o
blo
c
ks
M
of
le
ngth
L
each
.
Re
cal
l
that
the
excl
us
ive
-
or
is
ass
ociat
iv
e,
com
m
utati
v
e
and,
that
it
ha
s
a
ne
utral
el
e
m
ent
0,
a
nd
t
hat
any
c
hain
K
is
it
s
own
in
ve
rse:
K
⊕
K
=
0.
T
hu
s
,
we
can
s
ee
that
the
de
crypti
on
al
gori
thm
is
identic
al
to
the
enc
r
ypti
on
al
gorithm
, w
it
h
the
sam
e k
ey
:
M = C
⊕
K
=
(M
⊕
K)
⊕
K
=
M
⊕
(K
⊕
K
)
=
M
⊕
0
=
M
The
ba
sic
idea
of
this
pr
oce
ss
is
to
perfor
m
an
"exclusi
ve"
or
"
⊕
",
bi
t
by
bit
between
the
key
gen
e
rated
by
a
PRNG,
an
d
th
e
i
m
age
to
be
encr
y
pted,
IO.
This
al
gorithm
is
co
m
plete
ly
sy
m
m
e
tric
al
,
t
hat
is
the
sam
e
op
era
ti
on
is
app
li
ed
again
to
the
en
crypted
im
age,
IC
to
find
the
or
i
gin
al
i
m
age.
The
i
m
ages
us
ed
in
our
a
ppli
cat
ion are as
in Fi
gur
e 6
a
nd
of si
ze
256x25
6.
We
pr
ese
nt
in
the
f
ollo
wing
Fig
ur
es
7
-
14
i
m
ages
of
the
two
im
ages
"E
I:
I
nnoce
nt
Ch
il
dr
en"
a
nd
"FA:
Alge
rian
Wo
m
an"
acco
r
ding
to
dif
fer
e
nt
seeds
to
see
the
influ
e
nce
of
the
siz
e
of
th
e
seed
in
this
ci
ph
e
r
.
No
te
that
the
encr
y
ption
of
t
he
2
im
ages
ha
s
fail
ed
with
seq
uels
that
ha
ve
a
see
d
of
2,
3,
4,
5
or
6
dig
it
s
.
Fr
om
these
resu
lt
s
we
can
co
nclu
de
that
to
qu
a
ntify
an
im
age
it
is
necessary
that
the
seed
is
su
f
fici
ently
wide
(larg
e
d
i
git i
s great
er tha
n 6
).
This c
onfirm
s the c
ond
it
io
n
m
ade
by Vo
n Ne
um
ann
usi
ng
10
-
dig
it
nu
m
bers.
Figure
6
.
IO
-
E
I
–
IO
-
FA
Figure
7
.
IC
-
EI
–
IC
-
F
A wit
h
seed=
26
Figure
8
.
IC
-
EI
–
IC
-
F
A wit
h
seed=
684
Figure
9. IC
-
EI
–
IC
-
F
A wit
h
seed=
2863
Figure
10. IC
-
EI
–
IC
-
F
A wit
h
see
d=
43295
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
-
8
708
Crypto
graphic
adaptati
on
of t
he mid
dle s
quare
g
e
ner
ato
r
(Han
a
Ali
-
P
ac
ha)
5623
Figure
11. IC
-
EI
–
IC
-
F
A wit
h
see
d=
928763
Figure
1
2.
IC
-
EI
–
IC
-
F
A wit
h
see
d=
854726
7
Figure
13. IC
-
EI
–
IC
-
F
A wit
h
see
d=
594126
78
Figure
14. IC
-
EI
–
IC
-
F
A wit
h
see
d=
236589
741
5.7.
Analysis
of
hi
stogr
ams
A
histo
gram
is
a
st
at
ist
ic
a
l
c
urve
in
dicat
ing
the
distrib
utio
n
of
the
pi
xels
of
a
n
i
m
age
accor
ding
to
their
value
.
I
n
our
w
ork,
processe
d
im
age
s
are
gr
ay
scal
e
i
m
ages
wh
ose
pix
el
val
ue
s
var
y
in
the
range
[0,
25
5].
We
ha
ve
dr
a
w
n
an
d
analy
zed
t
he
histo
gr
am
s
of
the
enc
rypte
d
i
m
ages
of
"EI
-
Inn
oce
nt
Chil
dr
e
n"
and
"FA
-
Alge
r
ia
n
Wo
m
en",
the
il
lustrate
d
pl
ots
the
histo
gra
m
s
of
the
enc
rypted
im
ages,
HC
of
t
he
im
ages
EI
a
nd
F
A
re
sp
ect
ively
acc
ordin
g
t
o
the
siz
e
of
the
se
ed.
N
ote
that,
the
historg
ra
m
fo
r
the
plai
ntext
is
denoted
as
HP
.
It
th
us
em
erges
f
ro
m
the
prece
di
ng
res
ults
that
the
histogram
s
of
Fig
ur
es
21
-
23
a
r
e
unif
or
m
ly
distrib
uted
wit
h
res
pect
to
th
e
histogram
s
of
the
or
i
gin
al
im
ages
from
Fi
gure
15
with
r
especti
ve
enc
r
ypte
d
i
m
ages
fr
om
Figures
16
-
20
(
there
is
a
le
ak
age
of
inf
o
rm
at
ion
ab
out
th
e
pix
el
distri
buti
on
of
t
he
im
ages)
.
This
m
akes
cryptanaly
sis
increasin
gly
diff
i
cult
becau
se
th
e
encr
ypte
d
i
m
ages
pro
vid
e
no
el
e
m
ent,
bas
ed
on
the exploit
at
io
n of t
he hist
ogr
a
m
, to
d
e
sig
n
a
stat
ist
ic
al
att
a
ck on t
he p
rop
os
e
d
im
age encrypti
on tech
ni
qu
e
.
Figure
15
.
HP
-
EI
–
HP
–
FA
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.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
5615
-
5627
5624
Figure
16. HC
-
EI
–
HC
-
F
A w
it
h
seed=
26
Figure
17. HC
-
EI
–
HC
-
F
A
a
ve
c
seed=
684
Figure
18. HC
-
EI
–
HC
-
F
A w
it
h
seed=
2863
Evaluation Warning : The document was created with Spire.PDF for Python.