Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
1
3
,
No.
2
,
Febr
uar
y
201
9
, pp.
649
~
656
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
3
.i
2
.pp
649
-
656
649
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
A compr
ehensi
ve review
of hone
y en
cryption
sc
hem
e
Ab
io
dun
Esth
er Om
ola
r
a
1
, Ama
n
Jantan
2
, O
lud
are
Isaa
c A
bi
od
u
n
3
1
,2
School
of
Co
m
pute
r
Scie
n
ce
s
,
Univer
si
ti Sain
s Mal
a
y
sia
,
Pen
a
ng,
Mal
a
y
s
ia
3
Depa
rtment of
Com
pute
r
Scie
n
ce
,
Bingh
am Unive
rsit
y
,
Karu, Ni
ger
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
J
ul
1
2
, 2
018
Re
vised
N
ov
27
, 2
018
Accepte
d
Dec
2
7
, 201
8
W
e
pre
sent
a
co
m
pre
hensive
sur
ve
y
of
the
Hone
y
Enc
r
y
p
ti
on
(H
E)
sche
m
e
.
Hone
y
En
cr
y
pt
i
on
is
an
enc
r
y
pt
ion
sche
m
e
tha
t
provide
s
resil
i
e
nce
ag
ai
nst
brute
-
forc
e
at
t
a
ck
b
y
serv
ing
up
pla
usibl
e
-
loo
king
but
fak
e
pla
intext
for
eve
r
y
inv
al
id
ke
y
used
b
y
an
in
t
rude
r
to
de
cr
y
pt
a
m
essage
.
Our
goal
is
to
furnish
rese
archers
with
the
fra
m
ework
of
the
sche
m
e
not
just
for
implementa
t
ion
purpose
but
to
i
dent
if
y
the
gaps
in
th
e
sch
eme
and
answer
the
open
quest
i
ons
tha
t
remai
n
unanswere
d
b
y
the
sm
al
l
set
of
rese
arc
h
ca
rri
ed
out
sin
ce
it
s
incepti
on
.
W
e
ide
n
ti
fi
ed
two
m
aj
or
open
ar
eas
which
ar
e
the
difficulty
of
cre
a
ti
ng
sem
ant
i
ca
l
l
y
and
cont
ex
tua
lly
pl
ausibl
e
-
l
ooking
and
convi
nc
ing
deco
y
m
essage
th
a
t
is
good
enou
gh
to
fool
the
at
t
ac
ker
int
o
bel
i
evi
ng
he
h
as
the
origi
n
al
m
es
sage
.
Second
l
y
,
t
y
po
prob
le
m
;
where
a
f
ak
e
pla
intext
appear
s
val
id
to
a
le
g
it
imate
user
wh
en
he
m
ista
ken
l
y
enters
a
wrong
ke
y
.
Our
findi
ngs
consolidate
th
e
nee
d
for
furthe
r
rese
ar
ch
as
stat
e
-
of
-
the
-
ar
t
rese
ar
ch
fai
ls
to
produc
e
convi
nc
ing
dec
o
y
s
tha
t
ar
e
good
en
ough
to
kee
p
the a
t
tacke
r
from
ac
qu
iri
ng
t
he
m
essage
.
Ke
yw
or
d
s
:
Brute
-
force
Deco
ys
Distrib
ution t
ra
ns
f
or
m
ing
e
ncode
r
(
DTE
)
Honey e
ncr
y
ption
Passwor
d
-
base
d
e
ncr
y
ption
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
:
Am
an
Jantan
,
School
of Com
pu
te
r
Scie
nces,
Un
i
ver
sit
i Sai
ns M
al
ay
sia
, P
e
nang.
Em
a
il
:
a
m
an@usm
.
m
y
,
styleest2011@
gm
ail.co
m
1.
INTROD
U
CTION
T
he
cra
ft
of
decep
ti
on
is
ind
is
pe
ns
able
i
n
the
eve
nt
of
co
nfronti
ng
an
enem
y.
It
enab
le
s
a
n
env
i
ronm
ent
wh
e
re
a
n
ad
ve
rsar
y
is
tra
pped
into
ta
king
act
ion
s
th
at
c
on
s
um
e/
wastes
his
res
ources
[1
-
3].
Em
plo
yi
ng
de
cepti
on
an
d
de
coy
te
ch
niques
in
netw
ork
sy
stem
s
help
to
detect
,
trace,
m
on
it
or
an
d
de
te
r
the
act
ivit
ie
s
of
a
n
a
dversa
ry
[
4
-
6].
It
is
sta
ge
d
t
o
m
ake
the
adv
er
sary
’
s
li
fe
dif
fic
ult
w
he
re
a
false
rea
li
ty
is
pro
j
ect
ed
as
a
reali
ty
to
hi
m
.
Indee
d,
Sir
S
un Tzu
e
nca
psula
te
d
the
art
of
decep
ti
on
in
a
per
s
picu
ous
se
ntence
wh
e
n
he
sai
d,
"
Th
e
ar
t
of
w
ar
te
ac
hes
us
not
on
the
li
kel
i
hood
of
the
e
ne
my
’
s
no
t
c
om
ing
,
bu
t
on
ou
r
ow
n
r
ea
din
ess
to
r
e
cei
ve
hi
m;
not
on
the
c
hance
of
his
not
attac
ki
ng
,
bu
t
r
ath
e
r
on
the
fact
th
at
we
have
made
our
po
sit
io
n u
na
s
s
ailab
le
,
"
[
7].
W
e
trace
histo
rical
exam
ples
of
the
us
e
of
deco
ys
to
194
3
w
he
n
the
Br
it
ish
found
the
corpse
of
a
ho
m
el
ess
m
an
an
d
wen
t
t
hroug
h
e
xtra
ordinar
y
le
ng
t
h
to
fa
br
ic
at
e
his
death
a
nd
c
rea
te
d
a
pr
i
or
bu
t
fak
e
existi
ng
perso
nal
li
fe
for
him
to
deceive
t
he
Ge
rm
ans
.
His
reali
sti
c
but
fa
ke
pe
rs
ona
include
d
him
been
a
captai
n
in
the
m
ilit
ary
,
ha
ving
a
fathe
r
w
hom
he
sen
ds
/rec
ei
ves
le
tt
ers
f
r
om
and
a
fianc
ee'
s
le
tt
er
an
d
photo.
Also
,
the
B
riti
sh
plante
d
s
ome
fak
e
pa
pe
rs
on
him
ind
ic
at
ing
a
false
locat
i
on
f
or
an
Alli
ed
at
ta
ck.
O
f
c
ourse
,
the
Ger
m
ans
f
ound
t
he
dea
d
m
an
’
s
body
an
d
t
he
le
tt
ers
on
him
. T
hey
rea
d
t
he
le
tt
ers
a
nd
belie
ve
d
eve
r
yt
hin
g
on
it
based
on
the
‘
sup
po
se
d
’
evide
nce
f
ound
on
him
.
Su
bs
equ
e
ntly
,
they
div
e
rted
their
a
tt
ention
a
nd
m
i
li
ta
r
y
warfare
to
s
ome
ot
her
re
gion.
U
nkno
wn
to
them
,
the
y
we
r
e
co
nned
a
nd
t
he
Alli
ed
tro
op
la
nded
.
For
a
long
tim
e,
the
Germ
an
m
il
i
ta
ry
con
ti
nue
d
to
t
hi
nk
that
the
re
was
a
div
e
rsion
e
ven
a
fter
t
he
Alli
e
d
tr
oop
la
nd
e
d.
History
has
it
that th
is
B
riti
sh
facade
of th
e
hum
an
deco
y
sa
ved ove
r
f
ort
y (40) t
housa
nd
Alli
ed
li
ves
[
8]
.
In
rece
nt
ti
m
es,
dec
oy/hon
ey
syst
e
m
s
su
ch
as
ho
neypo
ts
,
honey
t
ok
e
ns,
honey
acco
unts,
honeyw
ords
w
hich
portray
fa
lse
res
ource
ha
ve
bee
n
de
ploy
ed
i
n
va
rio
us
netw
ork
syst
em
s
to
detect
,
obser
ve
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
2
,
Fe
bru
ary
201
9
:
649
–
656
650
and
th
war
t
at
ta
cks
from
cy
ber
crim
inals
[9
-
10]
.
H
one
y
en
crypti
on
(H
E
)
pro
posed
by
Ju
el
s
a
nd
Ri
ste
np
a
rt
[
11
-
12
]
is
struc
ture
d
unde
r
the
dec
oy
-
fr
am
ewor
k.
It
is
an
encr
ypti
on
sc
hem
e
that
yield
s
valid
-
looking
but
fak
e
m
essage
upon
decr
y
ptio
n
with
a
wro
ng
key
th
us
a
n
adv
e
rsa
ry
gai
ns
no
in
f
or
m
at
i
on
ab
out
the
ori
gin
al
m
essage.
The
a
dve
nt
of
hi
gh
s
pee
d
a
nd
s
uperc
har
ge
d
par
al
le
l
a
nd
distri
bu
te
d
sy
stem
s
(su
c
h
a
s
G
ra
ph
ic
al
Pr
oc
essin
g
Unit
,
Fiel
d
Progr
a
m
m
able
Gate
Arrays)
pa
ved
the
way
f
or
ga
therin
g,
a
naly
zi
ng
a
nd
proce
ssin
g
la
r
ge
c
hunks
of
data
of
te
n
re
f
err
e
d
to
a
s
bi
g
data
[
13
]
.
H
owever
,
t
his
hu
ge
advancem
ent
pu
t
c
rypto
syst
em
s
a
t
a
disa
dv
a
ntage
as
at
ta
cke
rs
le
ver
a
ge
on
the
high
c
om
pu
ta
ti
on
al
powe
r
of
this
syst
em
s
to
carry
ou
t
brute
-
fo
rce
at
ta
cks
[
14
]
.
A
pr
e
dom
inant
netw
ork
at
ta
ck
that
oft
en
j
e
opar
dizes
c
om
pu
te
rs
c
onnecte
d
to
a
net
work
is
the
brute
f
or
ce
at
ta
ck.
E
ven
th
ough
c
onve
ntion
al
e
ncr
y
ptio
n
sc
hem
es
c
on
ti
nu
e
to
gua
ran
te
e
sec
ur
it
y
by
increasin
g
the
siz
e
of
t
he
key
or
com
pu
ta
ti
onal
in
feasibil
it
y
of
sea
rc
hing
for
t
he
key
,
m
os
t
c
rypto
syst
em
s
fail
to w
it
hst
an
d
cr
ypta
naly
sis at
ta
ck
s
pecifica
ll
y t
he
br
ute
-
fo
r
ce at
ta
ck
[
15]
.
Honey
enc
ryp
ti
on
wa
s
pro
po
s
ed
as
a
c
ounterm
easur
e
to
brute
-
f
or
c
e
at
ta
ck
on
c
onve
ntion
al
encr
y
ption
sc
hem
es
sp
eci
fica
ll
y
fo
r
m
in
-
en
tro
py
syst
em
s
li
ke
pass
wor
ds
[
1
1
-
12]
.
It
was
ob
se
r
ved
f
ro
m
stud
ie
s
of
pe
rsi
ste
nt
data
br
ea
ch
t
hat
us
e
rs
c
ho
s
e
sim
ple,
w
eak
a
nd
predic
ta
ble
pa
sswo
rds
w
hic
h
m
akes
them
su
sce
ptible
to
brute
f
or
ce
at
ta
ck
[
16
-
1
7
]
.
H
on
ey
e
ncr
y
ptio
n
act
s
as
a
sup
plem
entary
encr
ypti
on
to
f
ort
ify
the
conve
ntion
al
P
assw
ord
-
Ba
sed
En
c
ry
ptio
n (P
BE) sc
hem
e
.
Honey
enc
ryp
ti
on
a
ddresses
the
flaw
s
of
pass
wor
d
-
bas
ed
enc
ryptio
n
schem
es
and
is
currently
e
m
plo
ye
d
in
s
ecur
i
ng
m
os
t
password
-
base
d
syst
e
m
in
the
form
of
ho
ney
words.
H
ow
e
ve
r
,
ho
ney
e
ncry
ption
has
not
bee
n
e
m
plo
ye
d
in
m
os
t
syst
e
m
s
su
c
h
as
,
it
s
a
pp
li
c
at
ion
for
e
ncr
y
pting
hum
an
w
ritt
en
doc
um
ents
li
ke
e
-
m
ails
,
et
c
.
T
he
c
halle
nge
i
s
how
do
we
create
co
ntext
ually
and
sem
antic
al
ly
cor
re
ct
deco
y
-
m
ess
age
that
can
act
ually
f
ool
a
n
at
ta
cke
r
?
Ot
her
pr
ob
le
m
s
l
ike
ty
po
-
safety
ha
ve
rem
ai
ned
una
ddres
se
d.
F
or
insta
nc
e,
if
a
le
gitim
at
e
receiver
m
ist
aken
ly
enters
a
wro
ng
pa
sswo
rd
.
Give
n
th
at
dec
oy
syst
em
can
com
plete
ly
add
ress
brute
f
or
ce
at
ta
ck
wh
ic
h
sta
nd
ard
e
nc
ryptio
n
schem
es
are
susc
eptible
to
,
th
en
the
re
is
nee
d
to
f
os
te
r
rese
arch
in
this
area
.
M
or
e
ov
er
,
t
he
cu
rr
e
nt
ad
van
ce
m
ade
on
quant
um
co
m
pu
te
rs
pro
pels
us
to
s
earch
for
qu
a
nt
um
-
safe
c
ryptosyst
e
m
s.
Since
al
l our
e
ncr
y
ptio
n
schem
es
are
ex
cl
us
ively
ba
se
d
on
Ma
them
atical
prob
le
m
s
wh
ic
h
are
est
a
blishe
d
base
d
on
the
dif
fic
ult
y
of
s
ol
vin
g
discrete
l
og
a
rithm
and
nu
m
ber
facto
ri
zat
ion
pro
blem
[
18
]
,
this
pu
ts
us
at
an
‘
e
ncr
ypti
on
l
ess
’
sta
te
as
s
oon
as
qu
a
ntu
m
com
pu
te
rs
s
ol
ve
al
l
the
un
de
rly
ing
Ma
them
at
ics
us
e
d
t
o
sec
ure
our
m
od
ern
c
r
yptosyst
em
s.
The
HE
sc
hem
e
is
not
dev
ise
d
un
der
th
e
c
ompu
ta
ti
onal
dif
ficult
y
of
breaki
ng
t
hem
al
on
e
but
the
real
tric
ker
y
t
hat
cry
pt
ography
was
m
eant
to
be
buil
t
upon
[
10
-
12]
.
Con
se
quently
,
ho
ney
e
ncr
y
ption
if
pro
pe
rly
desi
gned
a
nd
im
ple
m
ent
ed,
will
be
a
go
od
s
upple
m
entary
encr
y
ption
sc
hem
e
for
t
he
quant
u
m
era.
T
her
e
fore,
the
t
ar
get
of
t
his
pap
e
r
is
to
ca
pture
a
nd
pr
e
s
ent
a
synopsis
of
t
he
cu
rr
e
nt
sta
te
of
H
on
ey
enc
ryptio
n
sc
hem
e
to
ide
ntify
t
he
ga
ps
i
n
t
he
schem
e
to
en
able
it
s
op
ti
m
iz
ation
fo
r
real
-
li
fe ap
pli
cat
ion
s.
Our
pa
per
is
or
ga
nized
as
fo
l
lows
.
Fir
st,
w
e
set
the
sta
ge
by
prese
nting
a
detai
le
d
bac
kgr
ound
of
honey
e
nc
ryption
sc
hem
e.
Al
so
,
we
pr
es
ent
a
n
up
-
to
-
date
re
view
of
the
li
te
ratur
e
in
HE
.
F
ur
t
her
m
or
e
,
w
e
discuss
t
he
c
r
it
eria
of
hone
y
enc
ryptio
n,
issue
s
a
nd
c
halle
ng
e
s.
W
e
co
nclu
de
with
our
propos
i
ti
on
s,
su
ggest
io
ns
an
d
ide
ntific
at
ion o
f prom
isi
ng
areas
for fut
ur
e
researc
h.
2.
BACKG
ROU
ND OF
R
ESE
ARCH
W
e
giv
e
a
bri
ef
ove
rv
ie
w
of
the
co
nventi
onal
enc
ryptio
n
to
en
sure
a
ba
sic
unde
rstan
ding
of
how
Honey
e
ncr
y
ption
ev
olv
e
d.
I
n
a
co
nventi
onal
pass
w
ord
-
ba
sed
e
nc
ryptio
n,
a
n
a
dv
e
rsa
ry
perform
ing
a
brute
-
force
at
ta
ck
to
ob
ta
in
th
e
key
us
ed
for
enc
r
ypti
ng
a
m
essa
ge
gets
gibberi
sh
(
non
-
un
if
orm
distribu
ti
on)
or
a
n
error
sym
bo
l
a
s
the
ex
pecte
d
ou
t
pu
t
w
he
n
he
trie
s
a
wrong
key
.
T
his
ou
t
put
is
a
po
i
nter
and
disti
nguishe
r
t
hat
the
key
he
is
try
ing
is
inc
orr
ect
and
he
co
nt
inu
es
his
sea
r
ch
ti
ll
he
gets
a
plausib
le
-
lookin
g
m
essage
wh
ic
h
m
ay
be
the
plainte
xt.
D
ur
i
ng
his
at
ta
ck
,
he
quic
kly
discard
s
the
m
es
sage
w
hen
the
distri
bu
ti
on
is
no
n
-
un
i
form
. T
his
giv
es
him
m
or
e
tim
e
to
con
ti
nu
e
his
searc
h,
his
pro
ba
bili
ty
of
r
eco
ver
i
ng
the
m
essage/plai
ntext
is
high.
Fi
gu
r
e
1
sho
ws
a
detai
le
d
ex
pla
nation
of
how
a
conve
ntional
encr
ypti
on
schem
e
sp
eci
fical
ly
Passwor
d
-
based
e
ncr
y
ption re
sp
on
ds
t
o
a
bru
te
f
orce at
ta
ck
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A com
pr
ehe
ns
i
ve revi
ew
o
f
ho
ney e
ncry
ption sche
me
(
A
biodun
Esther
O
m
ola
r
a
)
651
Figure
1
.
A
n
il
lustrati
on of t
he
sett
ing
of the
conv
e
ntio
nal e
ncr
y
ption sche
m
e in a
br
ute
-
f
or
ce
att
ack
The
se
nder
en
crypts
his
m
es
sage
us
in
g
a
ke
y
and
a
ci
ph
e
r
.
He
sen
ds
t
he
ci
ph
e
rtext
t
o
t
he
receive
r
.
The
receive
r
de
crypts
the
ci
pherte
xt
us
i
ng
a
decr
y
ption
al
gorithm
and
th
e
sam
e
key
us
ed
by
the
se
nder
.
A
n
at
ta
cker
that
in
te
rcep
ts
the
ci
ph
e
rtext
m
ay
t
ry
to
rec
ov
e
r
t
he
m
essage
by
rand
om
l
y
gu
e
ssing
t
he
key
.
In
t
he
conve
ntion
al
s
et
ti
ng
scena
rio
,
the
at
ta
cke
r
can
im
m
ediat
e
ly
te
ll
fr
om
the
suppose
plain
te
xt
that
the
key
he
su
ppli
ed
is
i
n
c
orrect
because
of
the
non
-
unif
or
m
distrib
utio
n
of
th
e
plainte
xt.
We
de
scrib
e
an
at
ta
ck
m
od
el
of
the HE sc
hem
e
briefly
.
Fo
r
an
e
ncr
y
ption
C=
e
nc(
M,
K)
of
m
essage
M.
If
K
an
d
M
are
draw
n
f
ro
m
a
known
distrib
ution.
The
ta
r
ge
t
of
a
n
a
dv
e
rsa
ry
is
to
rec
over
t
he
m
essage
M.
H
e
trie
s
to
dec
ode
C
us
in
g
dif
f
eren
t
keys.
For
eve
r
y
key
he
trie
s
,
he
gets
M1
,...,
Mn.
F
or
a
m
i
nim
u
m
entro
py
distribu
ti
on
li
ke
pass
wor
ds,
M
is
guara
nteed
to
app
ea
r
on
his
li
st.
This
is
pos
sible
becau
se
us
ers
c
hoos
e
si
m
ple
password
s
that
can
be
e
asi
ly
gu
es
sed
.
Also
,
at
ta
cker
s
are
awar
e
of
ho
w
us
er
s
ch
oose
their
pa
sswo
rd
s
(fr
om
pr
evio
us
ly
releas
ed
de
ta
il
s
of
le
aked
pass
words
on
t
he
i
nter
net).
T
her
e
fore,
the
s
ecur
it
y
her
e
de
pends
on
the
pro
ba
bili
ty
of
th
e
ad
ve
rsar
y
be
en
a
ble
to
pick
the
m
e
ssage
M
f
r
om
al
l
n
possible
m
essages
sho
ul
d
one
of
the
ke
ys
he
trie
d
wa
s
correct
.
In
t
h
e
even
t
th
a
t
a
n
adv
e
rs
ar
y
cor
re
ctly
g
uesses
th
e
key,
he
is
s
till
s
tu
c
k
with
a
spoof
data
and
ca
nn
ot
asce
rtain
w
hich
is
th
e
co
rr
ec
t
m
e
ssage
espe
ci
all
y
wh
en
he
has
no
idea
of
th
e
t
ar
ge
t
me
ssag
e.
H
e
wins
o
nly
if
he
can
de
te
r
min
e
th
e
mes
sage
fr
om
th
e
list
of
me
ssa
ges
he
r
ecov
er
ed
du
ri
ng
his
a
tt
ack
.
F
igur
e
2
sho
ws
a
det
ailed
e
xpl
ana
tion
of how
hon
ey
encryptio
n sch
em
e
re
sponds
to a b
rute forc
e at
ta
ck
.
Figure
2
.
A
n
il
lustrati
on of t
he
sett
ing
of the
H
on
ey
e
nc
ryp
ti
on
sc
hem
e
in a bru
te
-
f
or
ce
at
ta
ck
Distrib
ution
T
r
ansfo
rm
ing
E
nc
od
e
r
:
H
on
ey
encr
y
ption
work
s
with
a
co
m
po
nen
t
re
ferr
ed
to
as
the
distrib
ution
t
ra
ns
f
or
m
ing
e
nc
od
e
r
(
DTE
).
A
DTE
is
a
pai
r
of
al
gorithm
DTE=
(e
ncode,
decode
)
that
ta
kes
M
as
an
i
nput
a
nd
retu
rn
s
a
valu
e
in
S
as
ou
t
put.
Dec
od
e
ta
ke
s
as
in
put
a
value
S
an
d
retu
r
ns
a
n
ou
t
pu
t
m
essag
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
2
,
Fe
bru
ary
201
9
:
649
–
656
652
M.
H
on
ey
e
ncry
ption
i
nvolv
e
s
a
DT
E
-
and
-
then
-
encr
ypt
process.
This
m
eans
a
se
nder
a
pp
li
es
the
DT
E
to
th
e
or
i
gin
al
m
essage
he
inte
nd
s
e
ncodin
g
a
nd
th
en
us
es
a
ny
c
onve
ntio
nal
e
nc
ryptio
n
sc
hem
e
as
the
sec
ond
la
ye
r
of
enc
ryptio
n.
The
DTE
m
odel
s
al
l
possible
m
essage
relat
ive
to
the
ori
gi
nal
m
essage
and
m
aps
them
to
a
seed
sp
ace
s
uc
h
t
hat
a
ny
key
su
ppli
ed
w
he
n
decr
y
pting
a
m
essage
pro
duce
s
a
relat
ive,
but
fa
ke
m
essage
from
the
or
igi
nal
m
e
ssage
an
d
t
his
m
akes
it
dif
fic
ult
f
or
the
ad
ve
rsar
y
t
o
determ
ine
if
he
ha
s
re
cov
e
re
d
the
or
i
gin
al
m
essage
or
not.
The
DTE
re
pr
ese
nts
th
e
m
od
el
of
the
m
essage.
A
go
od
DT
E
is
de
sign
e
d
to
m
od
el
the
m
essage
distri
bu
ti
on
well
suc
h
that
if
a
see
d
is
sel
ect
ed
unif
or
m
ly
a
t
ran
do
m
and
a
pp
li
ed
to
it
,
the
m
e
ssage
is
recovere
d.
Fi
gure
3
a
a
nd
3b
giv
e
a
desc
ript
ion
of
t
he
DTE
f
or
en
co
ding
a
nd
it
s
rev
e
rse
s
,
DT
D
f
or
decodin
g
a
m
essage
.
Figure
3
a
.
Fr
a
m
ewo
r
k of D
T
E for
Ho
ney E
ncodin
g
Figure
3
b
.
Fr
a
m
ewo
r
k of D
T
D
f
or
Honey
De
co
ding
The
al
gorithm
o
f
the e
nc
od
e
a
nd d
ec
ode
proc
ess of
HE
is
as
fo
ll
ows
[
12
]
:
HEnc
(K,
M)
S
¬$ enc
ode
(M)
R ¬$
{0,
1}
n
S
’
¬ H
(R,
K)
C ¬
S
’
⊕
S
retur
n
(R,
C)
HDec (K,
(R,
C))
S
’
¬ H
(R,
K)
S
¬ C
⊕
S
’
M ¬ decod
e(S)
retur
n
M
H
r
ep
rese
nts th
e
crypt
ographi
c h
as
h functi
on
, K
re
pr
e
sents
a
key
, M
re
pr
e
sents the
m
ess
age, S
represe
nt
s the
seed,
R
re
pr
e
se
nts a
rand
om
st
rin
g,
C
re
pr
ese
nts the
ci
pherte
xt,
a
nd
¬$
re
presents
un
i
form
ra
ndom
assignm
ent
.
2.
1
.
Criteria
fo
r
a G
ood En
coder
In
order
to
c
re
at
e
a
plausi
ble
/c
onvin
ci
ng
de
coy
th
at
do
es
no
t
ex
pose
the
ori
gi
nal
m
essage
t
o
the
adv
e
rsa
ry
, it i
s
fun
dam
ental
to
kee
p
in
m
ind
two
crit
eria
:
a)
Indisti
nguis
ha
bili
ty
:
Deco
y
m
essages
m
us
t
be
dif
ficult
to
disti
nguis
h
f
ro
m
real
m
ess
ages
.
S
uccess
f
ul
decep
ti
on
of
a
n
a
dversa
ry
li
e
s
on
the
ho
ney (d
ec
oy)
m
essa
ge.
Au
t
om
at
ed
too
ls
an
d
hum
a
ns
sho
uld
not b
e
able
to
te
ll
de
coy
m
essage
f
ro
m
true
m
essage
.
Dec
oys
m
us
t
be
dr
a
w
n
f
r
om
a
pr
ob
abili
ty
distribut
ion
ov
e
r
po
s
sible
m
essages
sim
i
l
ar
to
the
distri
bu
ti
on
of
t
he
plainte
xt.
M
odel
ing
hum
an
la
ngua
ge
re
quir
es
honey
m
essage
s
that
a
ppear
a
s
they
a
re
us
e
d
in
real
w
or
l
d.
A
good
D
TE
m
us
t
pr
od
uce
convinci
ng
dec
oy
m
essage
s.
The
pro
ba
bili
ty
of
te
ll
ing
deco
y
from
true
m
es
sage
m
us
t
be
m
ini
m
a
l
to
su
c
cessf
ully
deceive
the att
acker
.
b)
Confide
ntial
it
y:
A
good
DTE
m
us
t
m
od
el
t
he
hum
an
la
nguag
e
a
nd
at
t
he
sam
e
tim
e
hid
e
t
he
st
ru
ct
ural
inf
or
m
at
ion
of
the
ori
gin
al
te
xt.
H
um
an
la
ngua
ge
s
uc
h
as
e
-
m
ails,
hu
m
an
-
ge
ner
at
ed
do
cum
ents
requir
es
a
co
ns
ide
ra
ble
co
ntext
-
an
d
-
con
te
nt
rele
van
t
inf
or
m
at
ion
.
An
enc
oder
th
at
does
no
t
re
ve
al
the
str
uctu
r
al
sk
el
et
on
of
t
he
unde
rly
ing
m
essage/plai
n
te
xt
pro
vid
es
b
et
te
r
sec
ur
it
y
.
3.
T
OOLS F
OR
BUILDI
NG E
NCODE
RS
Seve
ral
sta
ti
stical
too
ls
em
pl
oyed
f
or
co
nst
ru
ct
in
g
the
e
ncode
rs
to
m
od
el
natu
ral
la
ngua
ge
ar
e
discusse
d
in t
hi
s secti
on
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A com
pr
ehe
ns
i
ve revi
ew
o
f
ho
ney e
ncry
ption sche
me
(
A
biodun
Esther
O
m
ola
r
a
)
653
3.1
.
Pr
obabil
istic
Langu
ag
e M
od
el
s
The
la
ngua
ge
m
od
el
is
the
pr
ob
a
bili
ty
distribu
ti
on
ov
e
r
se
qu
e
nces
of
word
s
a
nd
it
is
use
d
to
pre
dict
the
ne
xt
wor
d
or
ne
xt
se
nten
ce.
I
n
t
his
m
od
el
,
a
t
raine
d
n
-
gr
am
gen
e
ra
ti
on
m
od
el
pr
ovide
s
pla
us
ible
le
tt
er
seq
uen
ces
wh
i
ch
will
loo
k
ve
ry
si
m
i
la
r
to
words.
Key
in
this
case
sh
ould
set
up
the
init
ia
l
sta
te
o
f
the
seq
uen
ce
w
hic
h
would
le
ad
t
o
a
dif
fer
e
nt
c
om
bin
at
ion
of
the
le
tt
ers.
It
is
us
e
d
t
o
assig
n
a
pro
bab
il
it
y
to
a
sentence
. Give
n
a
series
of wor
ds
K
1,
K
2,
K
3.
.
.....
Kt
. It
assi
gn
s
a
li
kelih
oo
d
;
(
)
=
(
1
,
2
,
3
,
…
,
)
(
1)
3
.
2
.
Ma
r
kov
Model
s
A
stoc
hastic
m
od
el
us
e
d
to
r
epr
ese
nt
ra
ndom
ly
chan
ging
syst
e
m
s
wh
ere
fu
tu
re
sta
te
s
dep
e
nd
on
current
sta
te
s
and
do
no
t
de
pend
on
pa
st
even
ts
t
hat
occ
urred
be
fore
it
.
Pr
e
dicta
ble
e
ven
ts
t
hat
occ
ur
ov
e
r
tim
e
are
fash
ion
e
d
us
in
g
Ma
rko
v
m
od
el
s.
Ma
rkov
m
od
el
is
the
Ma
rk
ov
chain
w
hic
h
m
od
el
s
the
sta
te
of
a
syst
e
m
w
it
h
a s
tochasti
c
var
ia
ble that c
ha
ng
e
s over t
i
m
e.
3
.
3
.
Gr
amm
ar
Model
or Code
-
b
ook
Meth
od
Gen
e
rati
on m
e
thod
wh
e
re s
y
nt
act
ic
al
trees an
d dict
io
nar
ie
s
are use
d
to
g
e
ner
at
e
p
la
us
i
bl
e text o
r
t
he
cod
e
-
book
wh
e
re a
n
e
xisti
ng book wit
h
te
xt a
nd int
er
vals is
us
e
d
to
en
c
ode
the c
om
bin
at
ion o
f
t
he word
s.
4.
PROP
OS
E
D MET
HO
DS
T
O I
MPROV
E
THE
HO
NEY
ENCR
YPTI
ON SC
HEME
This
sect
io
n
de
scribes
pro
po
sal
s
from
sever
al
resea
rch
e
r
s
in
the
past.
The
pro
posal
sect
ion
is
div
ide
d
i
nto
t
wo
gro
ups.
Th
e
first
par
t
is
fo
c
us
e
d
on
t
he
enc
od
e
r
of
t
he
DTE
.
T
he
s
econd
par
t
des
cribes
pro
po
sal
s
on ty
po pr
ob
le
m
s in
HE
.
T
able
1 d
epict
s a
d
escri
ption o
f
m
od
el
s g
i
ven b
y
dif
f
eren
t a
uthor
s
.
4.1
.
T
ypo
-
Pr
ob
le
ms in
H
oney e
ncry
pt
i
on
In
t
he HE sc
he
m
e, a u
ser
is s
upplied
with a
seem
ing
ly
r
eal
bu
t
false te
xt
f
or
e
ve
ry k
ey
he
supp
li
es.
A
sign
ific
a
nt
dra
wb
ac
k
with
the
HE
sc
hem
e
is
ty
po
safety
w
he
n
a
le
giti
m
a
te
us
er
m
ist
aken
ly
enters
a
n
in
c
orrec
t
key.
Ty
po
e
rro
r
is
a
se
rio
us
pro
blem
in
HE
if
not
a
dequatel
y
prov
i
ded
f
or
.
This
sect
io
n
pro
vid
es
a
re
vie
w
of
researc
h
pro
po
sed
un
der
ty
po
sec
ur
it
y
in
honey
e
ncr
y
ption
sc
hem
e
.
T
a
ble
2
dep
ic
ts
a
d
escri
ption
of
ty
po
-
base
d
sec
ur
it
y
so
luti
on
pro
vide
d
by d
i
f
fe
ren
t
auth
or
s
.
5.
ISSU
E
S
AND CH
ALLE
NGE
S
As
discusse
d
in
the
intr
oduct
ion
s
ect
ion,
H
E
pro
vid
e
s
sec
ur
it
y
bey
ond
brute
-
f
or
ce
bound;
i
n
set
ti
ngs
wh
e
re
m
ini
m
um
entro
py
key
s
are
us
e
d
to
s
ecur
e
m
essage
s.
H
oweve
r
,
th
ere
are
ope
n
ar
eas
in
HE
wh
i
ch
need
to
be
r
esea
rc
he
d
. I
n
this
sect
io
n,
we b
riefly
desc
ribe som
e of the
problem
s o
f
H
E
.
a.
H
oney
e
ncr
y
ption
is
dif
ficult
to
ap
ply
in
a
s
et
ti
ng
w
he
re
t
he
plainte
xt
is
la
r
ge
or
the
di
stribu
ti
on
of
th
e
m
essage
is
un
known
.
A
la
r
ge
plainte
xt
requires
a
substan
ti
al
con
te
nt
to
be
us
ed
to
c
onstruct
the
DTE
s
o
that
fak
e
-
te
xts
that
looks
li
ke
the
ori
gi
nal
te
xts
can
be
us
e
d
as
the
dec
oy
m
essage.
T
he
con
te
nts
of
t
he
deco
y
m
essage
al
so
need
to
ha
ve
a
go
od
co
ntext
ual
m
eanin
g
relat
ive
to
the
or
igi
na
l
m
essage.
T
he
auth
or
s
ac
knowle
dged
t
his
di
f
ficult
y
in
thei
r
pa
per
w
he
n
they
pointe
d
ou
t
that
“...
Th
e
k
ey
cha
ll
e
ng
es
of
honey
enc
rypt
ion
sc
hem
e
are
develo
pm
e
nt
of
ap
pro
pr
i
ate
insta
nces
of
a
new
ty
pe
of
ra
ndomize
d
mess
ag
e
e
ncodin
g
sc
he
me
c
alled
a
distri
buti
on
-
tra
ns
f
ormin
g
e
nc
od
e
r
(
DT
E
)
”
[11
-
13]
.
This
pro
bl
e
m
was
al
s
o
st
rengthe
ned
wh
e
n
Ju
el
s
[
9]
poi
nted
out
the
c
om
plexity
of
encodin
g
ho
ne
y
do
c
um
ents,
for
instance,
e
-
m
a
il
s
req
uire
ge
ne
rati
ng
fak
e
but
sem
antic
all
y
and
co
nte
xtu
al
ly
reali
sti
c
natu
ral
la
ngua
ge
m
essage.
b.
Hav
i
ng
a
c
on
te
xt
-
sensiti
ve
de
coy p
r
oduce
s
good
insta
nces
of
the m
essage
b
ut if
not
co
rrec
tl
y
con
struct
ed
will
rev
eal
the
structu
re
of
t
he
or
i
gin
al
m
es
sage
giv
in
g
th
e
adversa
ry
an
uppe
r
ha
nd/hi
gh
pro
ba
bili
ty
of
recoveri
ng the
m
essage
usi
ng
chosen
-
ci
ph
ert
ext
at
ta
ck (CC
A)
.
c.
HE
is
ta
il
or
e
d
t
o
w
ork
i
n
lo
w
-
entr
op
y
set
ti
ng
s
li
ke
pass
w
ords,
RS
A
k
ey
s,
PI
N
S
a
nd
C
redi
t
card
n
um
bers.
Exten
ding
it
to
suppo
rt
oth
e
r
set
ti
ng
s
a
nd
file
ty
pe
re
qu
i
res
an
e
xtensi
ve
desig
n
of
t
he
DTE
t
o
m
eet
t
he
crit
eria
disc
us
s
ed
in
sect
io
n
2.1
.
d.
HE
is
fas
hione
d
t
o
pr
oduce
f
ake
but
valid
-
l
ooking
te
xt
f
or
e
ver
y
key
s
upplied
by
any
one.
T
he
refor
e
,
a
le
gitim
at
e
us
er
that
m
ade
a
typ
o
e
rror
w
hile
try
ing
to
retrie
ve
an
e
ncr
y
pted
m
essage
will
recover
a
valid
-
lookin
g
but
fa
ke
m
essage
and
he has
no
way
of kn
ow
i
ng thi
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
2
,
Fe
bru
ary
201
9
:
649
–
656
654
T
able
1
.
Descr
i
ption o
f
m
o
del
s g
i
ven b
y
dif
f
eren
t a
uthor
s
Au
th
o
rs
Descripti
o
n
Ju
stif
icatio
n
Flaws
Jo
et
al.
[
1
9
]
The
au
th
o
rs
p
rop
o
sed
th
e
statis
tical
co
d
e
sch
e
m
e
wh
ich
is
a
u
n
i
f
icatio
n
o
f
th
e
stru
ctu
ral
co
d
e
sch
e
m
e
an
d
h
o
n
ey
en
cry
p
tio
n
sch
e
m
e.
In
th
is
p
rop
o
sal,
th
e
HE
p
rov
id
es
th
e
se
m
an
tic
f
eatu
re
o
f
lan
g
u
a
g
e
wh
ile
th
e
stru
ctu
ral
co
d
e
sch
e
m
e
p
rov
id
es
th
e
syntactic
f
eat
u
res
o
f
n
atu
ral
lan
g
u
ag
e.
The
p
erfo
r
m
an
ce
was
ev
alu
ated
to
f
in
d
h
o
w
m
an
y
ti
m
es
it
was
requ
ired
to
g
en
e
rate
m
e
an
in
g
f
u
l
f
alse
tex
t
f
ro
m
th
e
o
rigin
al
tex
t
in
th
e
co
rpu
s.
The
p
rob
ab
ility
wa
s
0
.38
wh
ich
m
ea
n
s
at
least
f
o
u
r
repetitio
n
s
are
req
u
ired
to
g
en
erate
co
n
v
in
cin
g
deco
y
m
e
ss
ag
e.
This
ap
p
roach
p
rov
id
es
p
lau
sib
le
f
alse
tex
t
relative
to
th
e orig
in
al text as
d
ecoys.
It
d
o
es
n
o
t
support
o
th
er
d
ata
-
f
o
r
m
at
b
u
t
g
en
erate
s
d
ecoys
o
n
ly
f
o
r
sh
o
rt
len
g
th
m
ess
ag
e
.
The
a
m
b
ig
u
ity
b
etween
th
e
o
rigin
al
p
lain
tex
t
an
d
th
e
f
alse
tex
t
is
m
u
ch
,
an
d
th
e
ad
v
ersar
y
m
a
y
u
se
th
is
to
f
ig
u
re
o
u
t
th
e
d
if
f
erence
b
etween
th
e
f
alse
tex
t
an
d
o
rigin
al
tex
t
wh
en
h
e
h
as
a
little
k
n
o
wled
g
e
o
f
th
e
target
m
ess
ag
e
Ch
atterjee
et
al.
[
2
0
]
This
p
rop
o
sal
d
e
scrib
es
h
o
w
to
b
u
ild
a
Natu
ral
L
an
g
u
ag
e
Enco
d
er
(NL
E)
called
No
Crack
u
sin
g
ex
istin
g
pas
swo
rd
m
o
d
els.
The
p
erfo
r
m
an
ce
was
ev
alu
ated
b
y
m
e
asu
ring
th
e
ti
m
e
requ
ired
to
recov
er
a
p
arti
cu
lar
v
au
lt
an
d
th
e
ti
m
e
to
ad
d
a
p
ass
wo
rd to
a vau
lt
.
The
an
aly
sis
sh
o
wed
th
at,
th
e
s
m
aller
th
e
v
au
lt,
th
e
f
aster
th
e
recov
ery.
Lar
g
e
v
au
lts
requ
ire
a
lo
n
g
ti
m
e
to
gen
er
ate a
d
eco
y
vau
lt.
This
p
rop
o
sal
p
e
rf
o
r
m
s
b
est
wh
en
the v
au
lt is sm
a
ll.
A
sin
g
le
p
ass
w
o
rd
can
b
e
recov
ered qu
ick
ly
.
This
ap
p
roach
crea
tes
realistic
d
ecoy
v
au
lts
o
n
t
h
e
f
ly
d
u
ring
b
rute
-
f
o
rce
attacks
.
This
ap
p
roach
d
o
es
n
o
t
su
p
p
o
rt
large
v
au
lt
as
it
is
v
ery
slo
w.
The
syste
m
reflect
s
h
u
m
an
lan
g
u
ag
e
b
u
t
was
b
o
rne
in
th
e
co
n
tex
t
o
f
p
ass
wo
rd
secu
rity
an
d
d
o
es
n
o
t
su
p
p
o
rt
h
u
m
a
n
-
written
d
o
cu
m
en
ts
lik
e
e
m
a
ils
o
r
lo
n
g
m
ess
ag
es.
Go
lla
et
al
.
[
2
1
]
This
p
rop
o
sal
u
sed
Ku
llb
ack
–
Leibler
(KL
)
d
iv
ergen
ce
to
p
rov
e
th
at
th
e
ap
p
roach
b
y
[
2
0
]
d
eg
rades
secu
rity
.
The
p
rop
o
sal
also
p
o
in
ted
o
u
t
th
at
all
f
ix
ed
NLE
is
s
u
scep
tib
le
to
t
h
e
KL
d
iv
ergen
ce
attack
an
d
p
rop
o
se
d
th
e adap
tiv
e NL
E
.
The
ad
ap
tiv
e
NLE
was
co
n
stru
cted
u
sin
g
Mar
k
o
v
m
o
d
el
.
It
was
ev
alu
ated
u
sing
KL
d
iv
ergen
ce
attack,
an
d
th
e
an
a
ly
sis
sh
o
wed
th
at
th
e
real
v
au
lt
rank
s
a
m
o
n
g
4
0
.12
%
o
f
the
m
o
st lik
el
y
v
au
lt
.
This
p
rop
o
sal
i
m
p
rov
es
p
revio
u
s
m
eth
o
d
s
o
f
secu
ring
th
e
v
au
lts
f
o
r
p
ass
wo
rd
secu
rity
.
The
ad
ap
tiv
e
NLE
in
crea
ses
th
e
m
ess
ag
e
sp
ac
e,
allo
win
g
m
o
re
in
stan
ces
o
f
o
n
lin
e
g
u
ess
in
g
o
f
th
e orig
in
al vau
lt.
On
e
o
f
th
e
m
ajo
r
d
rawback
s
o
f
ad
o
p
tin
g
th
e
fi
x
ed
o
r
static
NLE
is
in
tersection
attack.
There
is
n
o
ad
eq
u
ate
clo
su
re
o
f
p
rov
id
in
g
m
a
x
i
m
a
l
secu
rit
y
.
Beu
n
ardeau
et al
.
[
2
2
]
This
p
rop
o
sal
co
n
ten
d
s
th
at
t
h
e
p
rop
o
sed
m
eth
o
d
b
y
[
20
-
21
]
wo
rks
reason
ab
ly
well
to
secu
re
sh
o
rt
p
ass
wo
rds
b
u
t
f
ail
to
m
o
d
el
n
atu
ral
lan
g
u
ag
e
as
u
se
d
in
real
-
wo
rl
d
scen
arios
su
ch
as
e
-
m
a
ils
an
d
written d
o
cu
m
en
ts.
This
p
rop
o
sal
ex
p
l
ain
s
th
at
co
n
tex
t
-
relevan
t
in
f
o
r
m
ati
o
n
is
requ
ired
to
m
o
d
el
h
u
m
an
lan
g
u
ag
e
to
p
ro
d
u
ce
co
n
v
in
cin
g
d
ecoy
m
ess
ag
es
th
at
f
o
o
l h
u
m
an
and
au
to
m
ated
too
ls
.
The
co
rpu
s
Qu
o
tatio
n
DTE
is
p
rop
o
sed
..
Gra
m
m
a
r
m
o
d
el
o
f
lan
g
u
ag
e
is
u
sed
t
o
b
u
il
d
th
e
DTE
an
d
u
sers
are req
u
i
red
to
o
n
ly
q
u
o
te
f
ro
m
a
k
n
o
wn
pu
b
l
ic do
cu
m
en
t.
This
p
rop
o
sal
sugg
ests
h
o
w
to
e
x
ten
d
th
e
sch
em
e
to
allo
w
en
co
d
in
g
h
u
m
a
n
-
w
ritten
tex
ts.
Qu
o
tin
g
f
ro
m
a
p
u
b
lic
d
o
cu
m
en
t
restr
ict
u
sers
to
th
e
v
o
cabu
lary
o
f
th
e
d
o
cu
m
en
t do
m
ain
.
Fix
ed
co
d
eb
o
o
k
is
n
o
t
ab
le
to
p
rov
id
e
all
th
e
requ
ired
co
m
b
in
atio
n
s o
f
w
o
rds
.
Fo
r
in
stan
ce,
it
is
u
n
lik
ely
th
at
a
u
ser
will
b
e
ab
le
to
en
co
d
e
a
tex
t
f
ro
m
a
co
m
p
u
ter
scien
ce
d
o
m
ain
u
sin
g
a
co
d
e
-
b
o
o
k
f
ro
m
a
f
lo
wer
-
b
ased
d
o
m
ain
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A com
pr
ehe
ns
i
ve revi
ew
o
f
ho
ney e
ncry
ption sche
me
(
A
biodun
Esther
O
m
ola
r
a
)
655
T
able
2
.
Descr
i
ption o
f
typo
-
ba
sed
s
olu
ti
on
gi
ven
by
dif
fe
re
nt
aut
hors
Au
th
o
rs
Descripti
o
n
Po
sitiv
es an
d
Negativ
es
Ch
atterjee
et al.
[
2
3
]
This
p
rop
o
sal
p
resents
a
ty
p
o
-
to
lerant
ch
ecker
wh
ich
wo
rks
relativel
y
well
with
th
e
ex
istin
g
p
ass
wo
rd
au
th
en
ticatio
n
sy
ste
m
.
This
p
rop
o
sal
p
o
i
n
ted
o
u
t
th
at
at
le
ast
a
m
in
u
te
wo
u
ld
h
av
e
b
een
sav
ed
for
2
0
%
o
f
t
h
e
u
sers
i
f
t
h
e
ty
p
o
-
to
lerant
sche
m
e
is ad
o
p
ted
.
This
ap
p
roach
i
s
su
itab
le
f
o
r
ex
istin
g
p
ass
wo
rd
-
b
ased
au
th
en
ticatio
n
syste
m
as
it
ap
p
lies
caps
lo
ck
co
rr
ecto
r,
f
irst
case
f
lip
co
rr
ecto
r
an
d
also
ex
tra
ch
ar
acter
a
t
th
e
en
d
co
rr
ecto
r
to
i
m
p
rov
e
u
sab
ility
b
u
t
th
is
ap
p
roach
is
n
o
t
su
itab
le
to
h
an
d
le
th
e
ty
p
o
p
rob
le
m
i
n
the HE
sch
e
m
e.
Ch
o
i
et
al.
[
2
4
]
The
sch
e
m
e
p
rov
id
ed
two
ty
p
es
o
f
ty
p
o
-
saf
ety
b
o
th
in
an
o
ff
lin
e
an
d
o
n
lin
e
settin
g
to
h
an
d
le
d
if
f
erent
ty
p
o
p
rob
le
m
s
wh
ile
still
retainin
g
m
ess
ag
e
recov
er
y
in
a ty
p
ical
HE
sch
e
m
e
.
The
Ty
p
e
A
p
roto
co
l
is
easier
to
i
m
p
le
m
en
t
as
it
requ
ires
o
n
ly
a
serv
er
b
u
t
th
e
m
ajo
r
d
rawback
is
t
h
at
th
e
size
o
f
th
e
k
ey
is
s
m
all
an
d
also
th
ere
is
th
e
u
n
certa
in
ty
o
f
d
etectin
g
t
y
p
o
s
in
so
m
e
settin
g
s.
Ty
p
e
B
is
an
i
m
p
r
o
v
e
m
en
t
o
v
er
T
y
p
e
A
as
a
u
ser
can
easily
n
o
tice
ty
p
o
s
if
h
e
re
m
e
m
b
ers
h
is
p
in
,
h
o
wev
er,
a
k
e
y
p
rob
le
m
h
ere
is th
at us
er
ha
s to
r
e
m
e
m
b
e
r
th
e
p
in
to v
erify hi
s
m
ess
ag
e.
Ch
atterjee
et al.
[
25
]
This
p
rop
o
sal
p
resents
a
p
erso
n
alized
ty
p
o
toleran
t pas
s
wo
rd ch
ecki
n
g
.
This
resear
ch
p
rop
o
ses
a
si
m
p
l
e
b
lack
listin
g
p
roced
u
re
in
wh
ich
a
s
m
all
set
o
f
risk
y
ty
p
o
s
is
p
roh
ib
ited
f
ro
m
b
ein
g
ad
m
is
sib
le
in
to
t
h
e
ty
p
o
cache.
This
resea
rch
is
a
n
i
m
p
rov
e
m
en
t
o
v
er
ex
istin
g
ty
p
o
-
to
lerant
p
ass
wo
rd
sch
e
m
e
s
b
u
t
is
n
o
t
d
esi
g
n
ed
to
wo
rk
o
n
ty
p
o
s
co
m
m
itt
ed
on
a
d
e
co
y
s
y
ste
m
.
How
e
v
er,
it
can
b
e
m
o
d
if
ied
f
o
r
Ho
n
ey
encr
y
p
tio
n
.
6.
CONCL
US
I
O
NS
A
ND FUT
UR
E
CHALL
ENGES
In
this pap
e
r,
i
ssu
es,
c
halle
ng
es
and
d
et
ai
le
d
li
te
ratur
e
rev
ie
w
of
the
H
one
y
Encr
ypti
on
(
HE)
sc
hem
e
is
pro
vid
e
d
.
T
he
ai
m
is
to
furn
is
h
c
urre
nt
/a
sp
iring
re
se
arch
e
rs
a
nd
pract
it
ion
ers
wi
th
a
c
om
pr
eh
ensive
ov
e
r
view
of
th
e
sta
te
-
of
-
art
r
esearch
i
n
the
schem
e.
F
ro
m
the
survey
of
var
i
ou
s
pro
pos
al
s,
we
co
nclu
de
that
the
cu
rr
e
nt
te
c
hn
i
qu
e
s
use
d
in
produci
ng
de
coy
m
essage
do
no
t
m
od
el
hum
an
la
ngua
ge
entirel
y
an
d
s
o
fail
to
pro
du
ce
d
ec
oy
s that are
acce
pt
able an
d
c
onvi
ncin
g
to
lu
re the
att
acker
a
wa
y from
the
genuine
re
s
ource.
Pr
ese
ntly
,
hone
y
encr
ypti
on
schem
e
has
be
en
im
ple
m
ente
d
f
or
c
re
dit
car
d
num
ber
s,
pa
sswor
ds
a
nd
RSA
pin
s
.
T
he
re
is
an
ur
ge
nt
nee
d
for
H
E
to
be
a
dap
t
ed
f
or
ot
her
s
et
ti
ng
s
s
uch
a
s
dec
oys
f
or
hu
m
a
n
gen
e
rated
m
es
sage
s
uc
h
as
e
-
m
ails,
co
nv
i
ncin
g
dec
oys
to
co
nfr
on
t
ea
ves
droppin
g
a
tt
ack
duri
ng
onli
ne
chatt
ing
,
et
c
.
This
stu
dy
has
raised
m
any
qu
est
io
ns
in
nee
d
of
f
ur
t
her
in
vestigat
io
n.
T
he
refor
e
,
we
pro
po
s
e
f
ur
the
r
re
searc
h
in the f
ollow
i
ng a
reas:
a.
Natu
ral
Langu
age
Processi
ng
in
H
on
ey
E
ncr
y
ption
:
H
ow
do
we
capt
ur
e
the
em
pirical
pr
operti
es
of
la
nguag
e
?
H
ow
do
we
m
odel
the
hu
m
an
l
angua
ge
it
sel
f
as
an
ef
fecti
ve
to
ol
f
or
desig
ning
c
onvi
ncin
g
deco
y
s
?
b.
How d
o we
honey enc
rypt
(pr
oduce
dec
oys)
without
rev
eal
i
ng the
struct
ure o
f
the
m
essa
ge
?
c.
How
do
we
ge
ner
at
e
dec
oy
/
honey
m
essages
that
fool
m
achines
a
nd
hum
an
from
reali
zi
ng
real
m
essages
from
d
ecoy m
e
ssages
?
d.
How
do
we
ha
nd
le
m
auling
and
prev
ent
a
dv
e
rsa
ries
fro
m
le
arn
ing
pa
r
ti
al
inform
at
io
n
of
the
or
i
gina
l
m
essage fro
m
the
deco
y
duri
ng a
n
e
xh
a
us
ti
ve
k
ey
-
sea
rc
h
?
e.
How
do
we
a
ddress
ty
po
pro
blem
s
in
the
H.
E
schem
e
?
T
hi
s
prob
le
m
requires
im
m
ediat
e
at
te
ntion
a
nd
extensi
ve
re
sea
rch eve
n bef
or
e any im
ple
m
e
ntati
on
of the
HE
sc
hem
e.
REF
E
RE
NCE
S
[1]
Om
ola
ra
AE,
Ja
nta
n
A,
Abiodu
n
OI,
Pos
ton
H
E.
A
Nove
l
App
roa
ch
for
th
e
A
dapt
a
ti
on
of
Hone
y
En
cr
y
pti
on
to
Support
Natur
al
La
nguag
e
Mes
sage
.
In
Procee
dings
of
th
e
In
te
rnational
Mul
ti
Confe
ren
ce
of
Engi
ne
ers
and
Computer
Scient
ists
2018
(Vol.
1
).
[2]
Diss
o
JP
,
Jone
s
K,
Bai
ley
S.
A
pla
usible
soluti
on
to
SC
ADA
sec
urity
hone
y
pot
s
y
s
te
m
s.
In
Broadband
a
nd
Wirel
ess
Computing,
Comm
unication
and
Appl
ications
(BW
CCA
),
2013
Ei
ghth
Inte
rna
ti
ona
l
Confer
ence
on
2013
Oct
28
(pp
.
443
-
448).
IE
EE.
[3]
Bringe
r
ML
,
C
hel
m
ec
ki
CA,
F
uji
noki
H
.
A
su
rve
y
:
R
ec
en
t
ad
vanc
es
and
fu
tu
re
tr
ends
in
ho
ne
y
po
t
r
ese
ar
ch.
Inte
rnational
Jo
urnal
of
Comput
er
Net
work
and
I
nformation
Se
cu
rity
.
2012
Sep
1;
4(10):63.
[4]
Dhanji
PK
,
Singh
SK
.
A
ss
aul
t
D
iscove
r
y
and
Lo
ca
l
iz
ing
Advers
ar
y
in
Remote
Networks.
Indone
sian
Jo
urnal
of
El
e
ct
rica
l
Eng
in
ee
ring a
nd
Computer
Sc
ie
nc
e
.
2
018
Jan
1;9(1
):
8
1
-
84
[5]
Zha
ng
D.
Inc
on
sistency
:
the
go
od,
the
bad,
and
the
ugl
y
.
In
Inf
orm
ati
on
Re
use
&
I
nte
gration
,
2009.
IRI'
09
.
IEEE
Inte
rna
ti
ona
l
Co
nfe
ren
c
e
on
200
9
Aug 10
(pp
.
18
2
-
187).
I
EE
E
.
[6]
Pawlic
k
J,
Colb
ert
E
,
Zhu
Q.
A
Gam
e
-
The
ore
ti
c
Ta
xonom
y
and
Surve
y
of
Def
en
sive
Dec
ep
ti
on
f
or
C
y
be
rsec
uri
t
y
and
Priva
c
y
.
arX
iv
pr
epr
int a
rXiv
:1712.
05441.
20
17
Dec
14.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
2
,
Fe
bru
ary
201
9
:
649
–
656
656
[7]
S.
Tz
u
.
"
The
art
of
war.
"
In
Strat
egi
c
Studies
,
pp
.
63
-
91.
Rout
le
dg
e,
2008
.
[8]
Montagu,
E. T
h
e
Man
W
ho
Nev
e
r
W
as,
J.
B.
Li
pp
inc
ott Com
pan
y
,
Phila
d
el
phi
a, PA (1954).
[9]
Juels
A.
A
bod
yguar
d
of
l
ie
s:
th
e
use
of
hon
e
y
obje
c
ts
in
infor
m
at
ion
sec
uri
t
y
.
In
Proceedi
ngs
of
th
e
19th
AC
M
sympos
ium
on
A
cc
ess c
on
trol
mo
del
s and te
chnologies
2014
Jun
2
5
(pp. 1
-
4)
.
ACM
.
[10]
Juels
A,
Rive
st
RL
.
Honey
words
:
Mak
ing
pas
swor
d
-
crac
ki
ng
det
e
ct
abl
e
.
In
Proce
ed
ings
of
the
2013
ACM
SIG
SA
C
conf
ere
nce
on
Com
puter &
comm
unic
ations sec
uri
t
y
20
13
Nov 4
(pp
.
14
5
-
160).
ACM
.
[11]
Juels
A,
Riste
n
par
t
T
.
Hone
y
Enc
r
y
pt
ion:
En
c
r
y
pt
ion
be
y
ond
the
brut
e
-
forc
e
bar
rie
r
.
IE
EE
S
e
curit
y
&
Priv
acy
.
2014
Jul;12(4):
5
9
-
62.
[12]
Juels
A,
Riste
n
par
t
T
.
Hone
y
enc
r
y
p
ti
on:
Se
cur
ity
b
e
y
ond
t
he
brut
e
-
forc
e
bound.
In
Ann
ual
Inte
rnat
ion
a
l
Confe
renc
e
on
the
Theory
and
Appl
ic
a
ti
ons
of
Cryptographic
Techni
qu
es
2014
Ma
y
11
(
pp.
293
-
310).
Springer
,
Berl
in
,
He
ide
lb
e
rg.
[13]
Kw
ok
SH
,
La
m
EY.
Eff
ective
uses
of
FP
GAs
for
brute
-
for
ce
a
tt
a
ck
on
rc4
ci
p
her
s.
IE
EE
Tr
ansacti
ons
on
Ve
r
y
Lar
ge
-
Scal
e
In
tegr
ati
on
(
VLSI)
S
yste
ms
.
2008
Aug;16(8):1096
-
10
0.
[14]
Couture
N,
Kent
KB.
The
eff
ec
t
i
vene
ss
of
brute
forc
e
attac
ks
on
RC4.
In
Comm
u
nic
ati
on
Ne
tworks
and
Serv
ic
es
Re
search
,
2004.
Proce
edi
ngs.
Se
cond
Annual
Co
nfe
ren
c
e
on
200
4
Ma
y
19
(pp. 3
33
-
336).
I
EE
E
.
[15]
Naja
fab
adi
MM
,
Khos
hgofta
ar
T
M,
Kem
p
C,
Se
l
i
y
a
N,
Zu
ec
h
R.
Mac
hine
learni
n
g
for
de
tecti
ng
b
rute
for
ce
atta
ck
s
at
the
net
work
l
eve
l
.
In
Bi
oinf
or
matic
s
and
Bi
oe
ngine
ering
(BIB
E),
2014
IEE
E
I
nte
rna
ti
ona
l
Confer
ence
on
2014
Nov 10
(pp. 379
-
385).
I
EE
E
.
[16]
Bonnea
u
J.
The
scie
nce
of
guessing:
ana
l
y
z
ing
an
anon
y
m
i
ze
d
cor
pus
of
70
m
i
ll
ion
passw
ords.
In
Sec
urity
an
d
Priv
ac
y
(SP
),
20
12
IEEE
S
y
m
po
sium
on
2012
Ma
y
20
(pp
.
5
38
-
5
52).
IE
EE.
[17]
Florenc
io
D,
H
erl
e
y
C
.
A
la
rg
e
-
sca
l
e
stud
y
of
web
passw
ord
habi
ts.
In
Proceedi
ngs
of
the
1
6th
int
ernat
iona
l
conf
ere
n
ce on
W
orld W
ide We
b
2
007
Ma
y
8
(pp.
657
-
666).
ACM
.
[18]
Gam
ido
HV
,
Sis
on
A
M,
Medina
RP
.
Modifie
d
AES
for
Te
xt
and
Im
age
Enc
r
y
p
ti
on
.
Indone
sian
Journal
of
El
e
ct
rica
l
Eng
in
ee
ring a
nd
Computer
Sc
ie
nc
e
.
2
018
Ma
y
27;11(
3).
[19]
Jo
HJ
,
Yoon
J
W.
A
new
count
er
m
ea
sure
aga
inst
brute
-
forc
e
at
t
a
cks
tha
t
use
hig
h
per
form
anc
e
c
om
pute
rs
for
big
da
ta a
n
aly
s
is.
Int
ernati
onal
Journ
al
of
Distribute
d
Sensor Ne
twork
s
.
2015
Jun 1
;11
(6):406915.
[20]
Chat
terje
e
R
,
B
onnea
u
J,
Juels
A,
Rist
enpa
r
t
T.
Cr
ac
ki
ng
-
res
ista
nt
p
assw
ord
vaults
using
n
at
ura
l
l
angua
g
e
enc
oder
s.
In
Secur
it
y
and
Priv
a
c
y
(SP
),
2015
IE
E
E
S
y
m
posiu
m
o
n
2015
Ma
y
17
(
pp.
481
-
498)
.
I
E
EE
.
[21]
Golla
M,
Beusc
her
B,
Dürm
uth
M.
On
the
sec
urity
of
cr
ac
k
in
g
-
resista
nt
passw
ord
vaul
ts.
In
Proce
ed
ings
of
the
2016
ACM
SIGS
AC
Conf
ere
nce
on
Computer
an
d
Comm
unic
ati
o
ns Se
curity
2016
Oct
24
(pp. 123
0
-
1241).
ACM
.
[22]
Beuna
rde
au
M, F
err
adi
H,
Géra
u
d
R,
Nac
cache
D
.
Hone
y
En
cr
y
pt
ion
for
L
angua
g
e.
In
Inte
rnat
ion
al
Confe
ren
ce
o
n
Cryptol
ogy
in
M
alay
sia
2016
De
c
1
(pp
.
127
-
144
).
Springer
,
Ch
a
m
.
[23]
Chat
terje
e
R,
At
ha
y
le
A,
Akhawe
D,
Juels
A,
R
iste
npar
t
T
.
pAS
SW
ORD
tY
PO
S
and
how
t
o
cor
r
e
ct
th
em
sec
ure
l
y.
In
Sec
uri
ty and Privacy
(SP
),
20
16
IEEE
S
y
m
po
sium
on
2016
Ma
y
22
(pp
.
7
99
-
8
18).
IE
EE.
[24]
Choi
H,
Nam
H,
Hur
J.
Pas
sw
or
d
t
y
pos
resilience
in
hone
y
enc
r
ypti
on.
In
Information
Net
workin
g
(
IC
OIN)
,
201
7
Inte
rna
ti
ona
l
Co
nfe
ren
c
e
on
201
7
Jan
11
(pp. 59
3
-
598).
I
EE
E
.
[25]
Chat
terje
e
R,
W
oodage
J,
Pnue
li
Y,
Chowdhur
y
A,
Riste
npar
t
T
.
The
TypTop
Syste
m:
Pe
rs
onali
z
ed
Typo
-
tolerant
Passwor
d
Chec
ki
ng
.
In
Proc
ee
d
ings
of
th
e
201
7
ACM
SIG
SAC
Confer
en
ce
o
n
Com
pute
r
and
Com
m
unic
at
io
ns
Secur
ity
2017:
3
29
-
346
Evaluation Warning : The document was created with Spire.PDF for Python.