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.
8
,
No.
6
,
D
ece
m
ber
201
8
, pp.
5144
~
51
52
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
8
i
6
.
pp
5144
-
51
52
5144
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Revealin
g AES E
ncryptio
n Device
Key on
328P
Mic
roco
ntroll
ers
with Differenti
al Po
wer An
alysis
Septafi
an
s
yah
D
w
i Pu
tra
1
, Ad
ang Suw
andi
A
hm
ad
2
, S
arwon
o Sutik
no
3
,
Yu
su
f Kurni
aw
an
4
,
Arwi
n
Datum
aya W
ahyudi
Sum
ar
i
5
1
Mana
gement
of
Inform
at
ic
s
,
Pol
it
ekni
k
Nege
r
i L
ampung,
Indone
sia
2
,3,4
,5
School
of
El
e
ct
ri
ca
l
Eng
in
ee
ring
and
Infor
m
at
ic
s, Ins
ti
tut T
eknol
ogi
Bandu
ng,
Indone
si
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Feb
3
, 2
01
8
Re
vised
Ju
l
19
,
201
8
Accepte
d
Aug 2
, 2
01
8
Thi
s
rese
arc
h
de
m
onstrat
es
the
rev
ealing
of
an
a
dvanc
ed
en
cr
y
pt
ion
standa
rd
(AES)
enc
r
y
pt
i
on
devi
c
e
k
e
y
.
The
enc
r
y
p
ti
o
n
devi
c
e
is
ap
pli
ed
to
an
ATMEGA
328P
m
ic
roc
ontrolle
r.
The
sa
id
m
ic
roc
on
trol
l
er
i
s
a
devi
c
e
comm
only
used
in
the
interne
t
of
thi
ngs
(Io
T).
W
e
m
ea
sur
ed
power
consum
pti
on
when
the
en
cr
y
pti
o
n
proc
ess
is
t
aking
pla
c
e.
Th
e
m
essage
sen
t
to
the
en
cr
y
pti
o
n
devi
c
e
is
ran
d
om
l
y
gen
erate
d
,
but
the
ke
y
us
ed
has
a
f
ixe
d
val
ue
.
The
nove
l
t
y
of
thi
s
rese
arch
is
the
cre
a
ti
on
of
a
s
y
stematic
a
nd
opti
m
al
ci
rcu
it
in
ca
rr
y
i
ng
the
diff
ere
nt
ia
l
power
ana
l
ysis
or
difference
of
m
e
an
s
(DP
A/DoM
)
te
chni
que,
so
the
technique
c
an
be
a
ppli
ed
in
ke
y
rev
ea
l
ing
on
a
m
ic
roc
ontroller
devi
c
e
b
y
using
500
trace
s
in
12
0
sec
onds.
Ke
yw
or
d:
AES
DPA
SCA
Encr
y
ption Att
ack
Secu
re
Dev
ic
e
IoT
Copyright
©
201
8
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
:
Ad
a
ng
S
uwan
di
Ah
m
ad,
CAIRG
-
Re
sea
rch G
rou
p
,
School
of Elec
tric
al
Engineer
ing
a
nd
Inform
at
ic
s,
Insti
tut T
eknolo
gi Ba
ndung
,
Jl. G
a
nes
ha No. 10,
Le
bak
Sil
iwan
gi,
C
oblo
ng, L
b. Sil
iwa
ngi, C
ob
l
ong, K
ota
Ba
nd
ung,
4013
2
, I
ndonesi
a
.
Em
a
il
: adan
gSahm
ad@
ya
hoo.com
1.
INTROD
U
CTION
The
inc
reasin
g
nu
m
ber
of
c
om
plex
syst
e
m
s
in
com
pu
te
r
a
pp
li
cat
io
ns
inc
reases
the
nee
d
for
a
sec
ure
data
exc
ha
ng
e
inside
t
hem
.
Th
os
e
a
pp
li
cat
ion
s
us
e
t
he
i
nter
net
as
the
m
edia
for
the
pr
i
vate
exc
ha
nge
of
inf
or
m
at
ion
or
data.
C
ryptol
og
y
is
a
sci
en
ce
ab
ou
t
data
safety
.
T
he
use
of
the
crypt
ogra
phic
ap
plica
ti
on
is
increasin
g
ov
e
r
the
ye
ar.
Thi
s
increasin
g
nu
m
ber
is
fo
ll
owed
by
the
nee
d
for
data
m
ov
em
ent
in
the
internet,
a
m
on
g
m
ob
il
e
syst
e
m
s,
anda
m
on
g
t
he
I
nt
ern
et
of
Thi
ng
s
(IoT
).
T
he
i
m
pr
ov
em
en
t
of
crypt
ogr
aph
ic
functi
onsis
no
t
lim
i
te
d
to
t
he
co
nf
i
den
ti
al
it
y
and
the
c
on
cealm
ent
of
i
nfo
rm
ation
by
un
a
utho
rized
pa
rtie
s.
Crypto
gr
a
phic
functi
ons
a
dva
nce
it
sel
f
i
n
be
com
ing
the
instr
um
ent
for
ver
ify
in
g
th
e
a
uth
e
ntici
ty
,
integrity
,
and
dig
it
al
sig
natu
re
of
a
dat
a
or
inf
or
m
at
ion
.
The
a
dv
a
nc
e
m
ent
in
cryptogra
ph
ic
f
un
ct
ion
s
is
consi
de
red
as
an
i
nteresti
ng
researc
h
obj
ect
w
hethe
r
from
the
a
ppli
cat
ion
si
de,
t
he
st
re
ng
t
h,
or
t
he
te
chn
i
qu
e
of
at
ta
ckin
g
the said
cr
ypt
ogra
phic
appli
cat
ion
on h
a
r
dw
a
re
or
softwa
re.
In
the
past,
th
e
at
ta
ck
on
a
crypto
gr
a
phic
syst
e
m
is
viewed
on
ly
the
oret
ic
al
ly
.
A
con
ve
ntio
nal
cryptanaly
st
ge
ner
al
ly
us
es
li
near
,
diff
e
r
entia
l,
an
d
br
ute
f
or
ce
te
c
hn
i
qu
e
to
ana
ly
ze
and
obta
in
the
vu
l
ner
a
bi
li
ti
es
of
a
n
enc
ryptio
n
al
gorithm
[1
]
,
[2
]
.
T
he
crypt
analy
st
rep
re
se
nts
the
crypt
ographic
al
gorith
m
as
a m
a
them
a
ti
cal
obj
ect
.
The
at
ta
ckin
g
te
chn
iq
ue
of
c
rypto
gr
a
phy
on
ha
rdware
or
e
m
bed
de
d
pla
tfor
m
is
a
ver
y
interest
ing
top
ic
an
d
ver
y
i
m
po
rtant
m
a
tt
er
to
be
research
e
d
especia
ll
y
in
this
m
o
dern
day.
Acc
ordin
g
to
sta
ti
sti
cs
in
2018,
the
num
ber
of
I
oT
de
vi
ces
or
pervasi
ve
hard
war
e
c
om
pu
ti
ng
is
proj
ect
e
d
to
hit
as
hig
h
as
75.
4
bill
ion
dev
ic
es
i
n
2025.
B
ut,
the
re
a
re
so
m
e
po
te
nt
ia
l
err
ors
an
d
threat
m
od
el
s
on
th
os
e
I
oT
syst
e
m
s.
The
m
ai
n
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N:
20
88
-
8708
Reveali
ng AE
S Encry
ptio
n De
vi
ce Ke
y on
328P
Mi
cr
oc
on
tr
ollers w
it
h Dif
fe
rential P
ower
…
(
Ad
ang S
.
A
.
)
5145
pro
blem
with
the
e
rror
m
od
el
is
that
I
oT
is
a
ne
w
obj
e
ct
so
t
hat
the
s
ec
ur
it
y
of
s
uc
h
obj
ect
s
is
not
m
uc
h
consi
der
e
d
i
n
the
desi
gn
ph
a
se.
I
oT
pr
oducts
that
a
r
e
avail
able
now
,
s
uch
as
a
m
ic
ro
proces
so
r
or
m
ic
ro
co
ntro
ll
e
r
with
em
be
dded
s
of
t
war
e,
are
ver
y
vuln
erab
le
t
o
bec
om
e
an
at
ta
ck
su
r
face.
I
oT
is
ve
ry
dep
e
ndent
on
t
he
dev
el
opm
ent
of
a
m
ic
ro
co
ntr
oller
un
it
(
MC
U)
te
c
hnol
og
y.
MC
U
with
lo
w
powe
r
a
nd
co
st
will
be
m
assivel
y
us
e
d
e
spe
ci
al
ly
as
the
m
ai
n
com
po
ne
nt
in
I
oT
de
vices.
T
hese
M
CUs
c
on
ta
i
n
a
flas
h
m
e
m
or
y
that
c
arr
ie
s
a
relat
ively
-
siz
ed
pro
gra
m
(at
le
ast
64
kB).
But,
the
s
hortfall
of
the u
sa
ge
of
MC
U
is
that
there
is
no
s
pe
ci
al
per
ip
her
al
su
c
h
as
tr
ue
ra
ndom
nu
m
ber
gen
e
rato
r
(TR
NG
s
)
or
crypt
ogra
phic
co
proc
essors
that
is
nee
ded
t
o
im
pr
ov
e
t
he
secur
it
y.
By
th
at
reason,
we
s
ee
a
nee
d
of
a
n
exam
inati
on
on
the
vulne
rab
i
li
ti
es
and the
points
of att
acks
of th
e MC
U dev
ic
e
s which
will
be
us
e
d
on the
Io
T d
e
vices.
Howe
ver,
in
f
act
,
wh
e
n
a
crypto
gr
a
phic
al
gorithm
i
m
pl
e
m
ented
in
ha
rdwar
e
will
pr
oduce
hi
gh
perform
ance,
it
can
be
m
ass
pro
du
ce
d
an
d
l
ow
c
os
tl
y
[1]
,
[2
]
,
[
3]
.
Com
m
on
fo
rm
s
of
crypto
gr
a
phic
dev
ic
es
are
un
i
ver
sal
s
erial
bus
(USB)
to
ke
ns
,
sm
art
car
d
[
4]
,
c
hips,
fiel
d
pr
ogra
m
m
able
gate
arr
ay
F
PGA
[5
]
,
[
6]
,
and
m
ic
ro
-
co
nt
ro
ll
ers
[7]
.
Th
ese
crypt
ogra
phic
de
vices
ha
ve
sm
al
l
di
m
e
ns
io
ns
a
nd
lo
w
powe
r
c
on
s
um
pt
ion
.
Howe
ver,
unnoti
ced
by
IC
c
ryptanaly
st
an
d
IC
desig
ner
s
,
crypto
grap
hi
c
dev
ic
es
a
re
m
or
e
vu
lne
rabl
e
an
d
easi
ly
accessi
ble b
y ph
ysi
cal
at
ta
ck
te
chn
iq
ue
s
[8]
.
So,
the u
nde
rly
ing
ass
um
ption
o
f
cl
a
ssica
l crypta
na
ly
sis is
no lo
ng
e
r p
os
si
ble to be a
da
pt
ed.
The
existe
nce
of
a
side
c
ha
nnel
at
ta
ck
(S
CA
)
at
ta
cks
on
th
e
secur
it
y
of
c
r
yp
togra
ph
ic
de
vices
nee
ds
in
-
dep
t
h
rese
ar
ch.
So
m
e
con
c
rete
cha
racteri
s
ti
c
le
aks
are
oc
currin
g.
So
m
e
side
c
hannel
inf
or
m
at
ion
,
s
uch
as
tim
e
[9]
[10]
,
so
und
[11]
,
el
ect
ro
m
agn
et
ic
fiel
ds
[12]
and
powe
r
co
ns
um
ption
[13]
can
be
us
ed
by
an
adv
e
rsa
ry
to
obta
in
t
he
m
ast
erk
ey
st
or
e
d
i
n
the
de
vice.
These
le
a
ks
ar
e
unav
oid
a
ble,
an
d
it
is
easy
for
a
n
at
ta
cker
to
m
e
asur
e
the
valu
e
of
su
c
h
side
inform
ation
su
ch
as
a
pro
be
and
a
high
-
frequ
e
ncy
os
ci
ll
os
c
op
e
.
Side
c
hannel
a
naly
sis
is
an
in
novative
ne
w
r
esearch
area
a
nd
ve
r
y
dif
fer
e
nt
f
ro
m
the
cl
assic
al
cryptana
ly
sis
appr
oach.
The
refor
e
,
now,
it
no
t
on
ly
f
ocuses
on
the
sec
ur
it
y
of
crypt
ogra
phic
al
go
rit
hm
s
bu
t
al
so
on
th
e
secur
it
y o
f
t
he whole
syst
em
t
hat is a c
rypt
ographic
dev
ic
e t
hat im
ple
m
ents
a
cry
ptogra
ph
i
c
al
gorithm
[13]
.
This att
ackin
g t
echn
iq
ue has s
how
n
the
over
al
l resu
lt
of si
m
ple p
ow
er
ana
ly
sis
(S
PA) a
nd D
P
A
[14]
.
The
res
ult
pro
du
ce
d
f
ro
m
DP
A
re
vealed
c
orrect
48
bit
from
64
bit
of
the
overall
key
(7
5%
).
T
he
S
PA
an
d
DPA
ha
ve
pr
oved
that
bo
t
h
of
them
can
re
cov
e
r
75%
of
the
secret
key
and
the
rest
c
ou
l
d
be
ob
ta
i
ne
d
by
us
in
g
brutef
orce
.
T
he
sec
ond
te
ch
nique
ha
s
bee
n
pro
pose
d
in
m
any
arti
cl
es
that
is
by
us
in
g
t
he
c
orre
la
ti
on
factor
bet
wee
n
the
trace
s
a
nd
t
he
ha
m
m
i
ng
weig
ht
of
the
proces
sed
data
[
15
]
,
[16
]
.
In
s
om
e
previo
us
researc
h
,
s
ubke
y
and
a
sec
ret
key
of
AE
S
a
nd
data
enc
ryp
ti
on
sta
ndar
d
(
DES)
ha
ve
bee
n
rec
overe
d
by
us
in
g
a q
uite l
arg
e
num
ber
o
f
trace
s
[17
–
19]
. P
re
vious DPA
at
ta
ckin
g
patte
r
n
use
d
a large qua
ntit
y of
traces (
>100
0
traces)
to
get
75%
correct
bi
t
of
the
m
ast
er
key.
There
ar
e
so
m
e
i
m
pr
ove
m
ents
fr
om
t
he
previ
ou
s
at
t
ackin
g
m
od
el
by
cal
culat
ing
t
he
c
orrelat
ion
c
oe
ff
i
ci
ent
of
the
tra
ce
an
d
the
ha
m
m
ing
weig
ht
of
the
proce
ss
ed
data.
Howe
ver,
w
he
n
cal
culat
ing
t
he
co
rr
el
at
io
n
coeffic
ie
nt,
the
at
ta
ck
m
us
t
hav
e
the
capa
bili
ty
to
fu
ll
y
con
trol
th
e
value
of the p
l
ai
ntext that is to
b
e enc
rypted
b
y t
he
cryptographic
de
vice
[2
0], [2
1]
.
The
DPA f
lo
wis
is
sho
w
n
in
F
igure
1.
Figure
1. D
PA
Flow
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
87
08
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
201
8
:
5144
-
5152
5146
Un
li
ke
tra
diti
onal
cryptanaly
s
is,
SCA
ta
rg
et
s
ph
ysi
cal
crypt
ogra
ph
ic
syst
em
i
m
ple
m
entation
.
P
ower
analy
sis
at
ta
cks
are
one
ty
pe
of
SC
A
that
e
xp
l
oit
power
i
nfor
m
at
ion
ch
ang
e
s.
P
ow
e
r
analy
sis
at
ta
cks
can
be
la
un
c
hed
with
si
m
ple
equ
i
pm
ent
and
at
ta
c
ks
in
a
sho
rt
tim
e.
Po
we
r
a
na
ly
sis
is
a
po
t
ent
an
d
us
ef
ul
at
ta
c
k
against
the
act
ual
i
m
ple
m
entat
ion
of
the
c
r
yptogra
ph
ic
al
gorithm
on
the
hardw
a
re.
Fro
m
the
var
io
us
so
urce
s
of
t
he
side
c
ha
nn
el
i
nfor
m
at
i
on
m
entione
d
earli
er,
s
uch
as
tim
e
m
easur
e
m
ent,
el
ect
ro
m
agn
et
ic
ra
diati
on,
er
ror
m
essage;
inf
or
m
at
ion
de
rive
d
from
power
c
on
s
um
ption
m
ay
be
th
e
m
os
t
dif
ficult
m
at
ter
to
be
co
ntr
olled
b
y
the
crypto
gr
a
phic
desig
ner.
All
cal
culat
ions
per
f
orm
ed
by
encr
ypti
on
de
vices
operate
on
ze
ro
an
d
one
log
ic
gates
.
T
he
pro
cess
of
c
om
pu
ti
ng
e
ncr
y
ptio
n
a
nd
dec
ryption
will
le
ad
t
o
c
hanges
i
n
powe
r
f
or
m
and
m
or
e
sp
eci
fical
ly
th
e
log
ic
gate.
Atta
cker
enc
ry
ption
de
vices
can
m
on
it
or
powe
r
dif
fer
e
nc
es
an
d
get
us
e
fu
l
si
de
channel
in
for
m
at
ion
in
key
sp
ace
sea
rch
es
.
(
D
PA
)
,
i
ntrodu
ce
d
by
K
oc
her
et
al
.
is
a
sta
ti
sti
cal
app
ro
ach
t
o
m
on
it
or
su
c
h p
ow
e
r
si
gn
al
s
.
Sp
eci
fic
DPA
at
ta
ck
fo
rm
s
a
gainst
enc
ryption
dev
ic
es
r
unni
ng
DES
al
gorithm
s
are
c
on
ta
ine
d
i
n
their
DPA
stu
dy
[19]
.
Ba
se
d
on
th
e
res
ults
of
the
st
ud
y,
i
t
was
co
nf
irm
ed
that
D
PA
at
ta
cks
are
ver
y
po
te
nt
and
ca
n
e
ven
be
us
e
d
to
m
on
it
or
t
he
dif
fer
e
nce
of
eve
ry
s
ing
le
bit
of
tra
ns
ist
ors
in
e
nc
ryptio
n
de
vice
s.
The
pap
e
r
will
s
how
how
this
at
ta
ck
can
be
us
e
d
against
the
A
ES
encr
y
ptio
n
al
gorithm
and
wh
at
fact
or
s
ca
us
e
the
vu
l
ner
a
bili
ty
to
occ
ur.
The
pur
pose
of
this
resea
rch
is
to
in
vestigat
e
SCA
an
d
t
o
dev
el
op
a
DPA
base
d
at
ta
ck
on
a
n
MC
U
ta
rg
et
that
a
pp
l
ie
s
AES
-
12
8.
I
n
this
a
rti
cl
e,
r
eaders
will
be
introd
uced
t
o
t
he
idea
of
SC
A
in
sea
rc
hing
for
the
key.
Nex
t,
we
will
intr
oduc
e
the
powe
r
a
naly
sis
te
chn
i
qu
e
with
D
PA
-
D
oM
(
dif
fer
e
nce
of
m
eans)
on
a
n
MC
U.
T
he
le
ast
sign
i
ficant
bit
(L
SB)
m
od
el
of
inte
rm
ediat
e
val
ue
i
s
intr
oduce
d
a
s
a
p
arti
al
m
e
ans
i
n
execu
ti
ng
D
P
A.
Finall
y,
we
m
ade
so
m
e
c
on
cl
us
io
ns
ab
out
therse
fin
dings
an
d
com
m
e
nts
ab
ou
t
so
m
e
fu
tu
re
worksbase
d
on
the
resea
rc
h
r
esults.
O
ne
t
opic
reg
a
rd
i
ng
t
he
fu
t
ur
e
w
orks
is
about
fi
nd
i
ng
t
he
best
ap
proac
h
in m
i
ti
gating D
PA
att
ack
on
a
n
MCU
.
2.
RESEA
R
CH MET
HO
D
This
sect
ion
in
tro
du
ces
the
pe
rfor
m
ed
and
la
b
set
up
f
or
D
PA
at
ta
cks
in
this
researc
h.
Figure
2
a
s
sh
ow
n
the
re
se
arch
ste
ps
us
e
d
in
this
pa
per
.
The
resea
rc
h
ste
ps
use
d
in
t
his
pap
e
r
was
do
ne
by
set
ti
ng
t
hem
in
a
syst
e
m
at
ic
m
eansf
or
rev
ie
wing
the
SCA
te
chn
i
qu
e
on
an
e
ncr
y
ption
dev
ic
e.
T
he
te
st
will
m
ake
a
D
P
A
at
ta
ck
on
an
AES
e
ncr
y
ptio
n
de
vice
by
usi
ng
a
la
borat
or
y
-
te
sti
ng
e
nvir
on
m
ent.
Th
e
en
vironm
ent
it
sel
f
is
sh
ow
n
in
F
i
gur
e 3
.
Figure
2. Re
se
arch M
et
hod
The
e
xpect
ed
m
ai
n
resu
lt
of
this
resea
rch
is
to
re
co
ver
the
secret
key
aft
er
the
enc
ryption
pro
cess
is
finish
e
d
.
M
or
e
com
pr
ehe
ns
iv
el
y,
the
desig
n
ste
p
ofenc
ryp
ti
on
de
vice
wi
ll
pr
od
uce
an
AES128
dev
ic
e
that
runs
on
an
ATMEG
A
328P
m
ic
ro
con
t
rol
le
r.
Af
te
r
the
e
ncr
ypti
on
de
vice
is
ob
ta
ined
,
the
m
et
ho
d
of
m
easur
em
ent
is
desi
gn
e
d
by
m
aking
a
ci
rc
ui
t
with
a
series
resist
or
on
the
V
ground
.
T
he
ne
xt
ste
p
is
to
ge
ner
at
e
the
cl
ock
by
send
i
ng
ci
phe
rtext
to
the
encr
y
ption
dev
ic
e
previo
us
ly
ob
ta
i
ned.
The
la
st
pa
rt
of
the
s
ec
ond
te
st
is t
o
do
a
DPA
at
ta
ck
so t
he
c
har
act
erist
ic
s a
nd the m
od
el
c
an be
ob
ta
in
ed
and b
e
fur
t
her
analy
zed
.
Table
1
li
sts
the
m
ai
n
com
po
ne
nts
a
nd
set
ti
ng
f
or
the
te
sti
ng
.
To
r
un
and
sim
ulate
t
his
at
ta
ck
te
chn
iq
ue
we
buil
d
the
de
vice
unde
r
te
st
(
D
UT).
T
he
syst
em
’s
d
esi
gn
a
rc
hitec
ture
is
s
ho
wn
i
n
F
i
gure
3.
This
DU
T
en
vir
on
m
entisc
on
sist
e
d
of
at
le
ast
three
c
onnecte
d
c
om
po
ne
nts:
AES
c
rypt
o
process
or,
a
pe
rsonal
com
pu
te
r
(P
C
)
,
and
dig
it
al
sam
pl
ing
os
ci
ll
osc
ope
(DSO
).
The
crypt
o
pro
cesso
r
is
the
D
UT
f
ro
m
wh
ic
h
side
channel
in
f
or
m
at
ion
would
be
ha
rv
est
e
d
by
the
DSO,
c
reati
ng
a
traces
cu
r
ve.
The
PC
c
ollec
ts
the
traces
a
nd
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N:
20
88
-
8708
Reveali
ng AE
S Encry
ptio
n De
vi
ce Ke
y on
328P
Mi
cr
oc
on
tr
ollers w
it
h Dif
fe
rential P
ower
…
(
Ad
ang S
.
A
.
)
5147
perform
s
sta
t
is
ti
cal
analy
ses
t
o
find
the
key
by
m
od
el
ing
t
he
traces
cu
rve
us
in
g
key
gu
esses.
The
DUT
and
the osci
ll
os
co
pe
co
m
m
un
ic
at
e u
si
ng U
S
B
and RS
232.
Table
1.
L
ab
S
et
up
f
or
DPA
-
DoM
Variable
Descripti
o
n
Alg
o
rith
m
an
d
len
g
th
of
the
k
ey
Sa
m
p
le f
requ
en
cy
AES
-
1
2
8
bit
1
Gsa
m
p
l
e/s
FPGA a
rchitectu
re
Xilin
x
Ar
tix
-
7
Tr
ig
g
er
sig
n
al
Header pin
with S
MA
co
n
n
ecto
rs
Sh
u
n
t r
esis
to
r
5
0
0
m
Oh
m
-
Sta
ck
p
o
le
VCC
-
Extern
al
5
Volt
-
2A
Secret
k
e
y
0
0
11
22
3
3
44
55
6
6
77
88
9
9
AA
BB
CC DD
EE
FF
PC
–
sa
m
p
lin
g
Intel i5
with 8
G R
AM
The
dif
fere
ntia
l
po
we
r
analy
sis
at
ta
ck
was
us
ed
in
this
s
tud
y
is
a
diff
e
ren
ce
of
m
eans
appro
ac
h
(DoM)
was
pr
opos
e
d
by
Kocher
at
19
99
[
19
]
.
H
oweve
r,
the
Ko
c
he
r
publica
ti
on
im
plem
ents
the
at
ta
c
k
on
a
DES
al
gorith
m
that
is
locat
ed
in
a
n
F
P
GA.
I
n
t
his
r
esearch
,
we
f
ocused
on
a
DPA
at
ta
c
k
on
a
n
A
ES
encr
y
ption
de
vi
ce.
Sp
eci
fical
ly
,
the
dev
ic
e
i
s
an
AE
S
enc
r
ypti
on
syst
em
i
m
ple
m
ented
in
a
m
ic
ro
co
ntr
oller
dev
ic
e.
T
he ba
sic
idea of
DPA is to
m
ake one hyp
oth
esi
s
on
e
b
y
one
bit of
t
he whole
ke
y bit
s.
Ne
xt, we sele
ct
a
functi
on
known
as
functi
on
sel
ect
ion
.
T
he
sel
ect
ion
func
ti
on
gets
the
i
nput
value
of
th
e
key
gu
e
ss
k
g
,
wh
e
re
k
g
= (
k
g1
,
k
g2
,
...
..
k
g255
).
This
te
chn
i
qu
e
will
div
ide
sever
al
cu
rv
es
of
traces
(en
cry
pt
ion
de
vice
m
e
asur
em
ent
resul
t)
into
two
set
s
S
δ
≡
m
i
(
t
) |
L
k
0
,
i
= 0
and
S
ψ
≡
m
i
(
t
)
|
L
k
0
,
I
=1
,
a
nd
the
t
races
are
a
dju
ste
d
to
the
le
aka
ge
values
of
hypotheti
cal
ke
ys
L
k
0
,
i
.
T
he
a
dv
e
rsa
rywil
l
focus
on
one
LS
B
bit
(
a
le
ast
s
ign
ific
a
nt
bit)
L
k
0
,
i
=
∈
{
0,1}
t
hen
the
ou
t
pu
t
of
the
bit
determ
ines
w
her
e
the
po
sit
ion
of
the
t
races
c
urve
is
place
d
.
Lay
in
g
the
traces
done
by
lookin
g
at
if
L
SB
=
0
will
be
placed
on
set
0
(
L
k
0
,
i
=
0)o
r
oth
e
rw
ise
set
1
(
L
k
0
,
i
=
1)
[
18]
,
[
22
]
.
F
ur
the
r
m
or
e,
bo
t
h
set
s
of
tra
ces
are
m
utu
al
ly
red
uce
d
or
t
o
obta
in
the
diff
e
ren
ce
bet
we
en
the
two
cu
r
ves
cal
culat
ed.
In
the
correct
hy
po
t
he
ti
cal
key
that
is
k
0
=
k
,
the
n
the
tr
ue
pr
e
dicti
ve
value
will
be
se
pa
rated
f
ro
m
the
s
ha
pe
of
t
he
whole
c
urve.
At
the
te
ch
nic
al
end
of
this
a
tt
ack
pro
du
c
es
a
pea
k
on
t
he
diff
e
re
ntial
trace
curve
a
nd
t
he
po
i
nt
in
ti
m
e
wh
en
the
ta
r
geted
operati
on
is
cal
c
ulate
d
∆
k
0
(
t
).
I
n
oth
e
r
hypoth
eses
the
k
0
,
k
c
urve
t
races
will
look
flat
.
The
c
orre
ct
k
g
values
ca
n
be
ide
ntifie
d
by
lo
ok
i
ng
at
the
diff
e
re
nt
pe
aks
on
the
cu
r
ves
of
the
ta
ble
(
Do
M
traces).
F
or
m
ally, Do
M
cal
cul
at
ion
s a
re indic
at
ed
by:
∆
k
0
(
t
)
=
S
δ
−
S
ψ
(1)
Figure
3. Setu
p
a
rc
hitec
ture fo
r
at
ta
ck
a
gain
st ATME
GA3
28P
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
87
08
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
201
8
:
5144
-
5152
5148
In
this
pa
per
,
DPA
at
ta
ck
te
chn
i
qu
e
s
us
e
t
he
DPA
AES1
28
bit
at
ta
ck
.
The
num
ber
of
traces
data
is
500
cu
rv
e
trac
es
with
in
form
at
ion
a
bout
t
he
plainte
xt
a
nd
the
res
ulti
ng
ci
ph
e
r
-
te
xt.
T
he
sta
ges
i
n
ca
rr
yi
ng
ou
t
the
DPA
at
ta
ck
on
this
re
por
t
are
to
ob
se
r
ve
th
e
first
rou
nd
of
the
AE
S
-
128
enc
ryptio
n
protoc
ol.
The
DP
A
at
ta
cks
ta
r
get
is o
utbreak
fro
m
Add
R
oundK
e
y and S
ub
By
te
s on AES
roun
d op
e
rati
ons.
Al
go
rit
hm
1.
DPA
Do
M
f
or
1s
t su
bkey
k
1
Inpu
t
:
N pairs
traces
with
pla
intext an
d
k
g
=
key guess
Out
p
ut
: R
ec
overe
d
key f
o
k
1
1:
for
k
g
=
0 to
25
5
do
2:
f
or
i =
1
to
N
do
3:
Matrix
data(
i,k
g)
←
LSB of
(
S
box (Ci
[
0]
⊕
k
g
)
);
4:
end for
5:
f
or
eac
h
sa
m
ple p
oin
t
p
=
1,
2, . , M
of pow
e
r
trace
do
6:
Do
M
p,
k
g
←
(
S
δ
≡
m
i
(
t
)
|
L
k
0
,
i
=
1)
–
(
S
ψ
≡
m
i
(
t
) |
L
k
0
,
i
=
0)
;
7:
end for
8:
Do
Mt
race
,
k
g
← {DoM
1,
k
g
, DoM
2,
k
g ....
DoM
M,
k
g
};
9:
end for
10
:
k
1
[
0]
← absm
ax
| D
oMtrace,
k
g
|
11
:
e
nd
In
eac
h
t
race
i
,
I
i
,
a
16byte
is
an
inte
rm
ediate
sta
te
of
t
he
ou
t
pu
t
value
of
the
ci
ph
e
r
a
fter
S
ub
By
te
s
op
e
rati
on
is
pe
rfor
m
ed
in
the
first
r
ound.
T
he
n,
as
m
any
as
n
byte
s
of
eac
h
sta
te
∈
{0.
.
.15}
is
de
no
te
d
by
I
i
,
n
.
The
key
val
ue
us
e
d
in
the
firs
t
ro
un
d
is
deno
te
d
as
K,
and
t
he
value
of
n
of
each
byte
will
be
de
no
te
d
by
Kn
.
The
plainte
xt
us
e
d
on
eac
h
trace
is
den
ote
d
as
Xi
,
n
.
Th
e
refor
e
,
the
m
ath
em
atical
m
od
el
of
the
first
rou
nd
ob
s
er
vation ca
n be:
I
i
,
n
=
S
[
x
i
,
n
⊕
K
n
]
(2)
D
=
L
SB
(I
i
,
n
)
(
3)
Fr
om
the
equ
at
ion
,
the
value
of
X
i
,
n
is
a
known
va
riable:
on
e
byte
of
pla
in
-
te
xt.
K
n
is
the
secret
key
const
ants.
The
S
var
ia
ble
is
th
e
default
value
of
the
AE
S
s
ubsti
tuti
on
ta
ble
.
Wh
il
e
the
outpu
t
of
the
S
functi
on
is
I
(
i
,
n
)
a
n
unknow
n
va
riable
whose
value
de
pends
on
a
1
-
byte
secret
key
and
a
known
var
ia
ble
su
c
h
a
s
plain
-
te
xt
,
u
se
of
t
ough
te
sts
ca
n
crack
AE
S
pas
swor
ds
easi
ly
if
f
ound
the
va
lue
of
a
key
gu
e
ss
is
c
orrec
t
.
More
sp
eci
fical
ly
K
n
is
an
8
-
bit
value,
so
at
le
ast
256
te
sts
will
be
per
f
or
m
ed
to
con
fi
rm
the
cor
rect
K
n
value.
In
the
16
byte
s
of
K
n
that
m
ake
up
al
l
AES
-
12
8
ke
ys
can
be
f
ound
on
ly
by
sp
l
it
ti
ng
fo
r
eac
h
byte
separ
at
el
y.
The
flo
wch
a
rt DPA
-
D
OM
on A
E
S
as
s
how
n
in
F
i
gure
4.
Figure
4. Flo
w
char
t
DPA
-
D
O
M on A
ES
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N:
20
88
-
8708
Reveali
ng AE
S Encry
ptio
n De
vi
ce Ke
y on
328P
Mi
cr
oc
on
tr
ollers w
it
h Dif
fe
rential P
ower
…
(
Ad
ang S
.
A
.
)
5149
3.
RESU
LT
S
AND DI
SCUS
S
ION
MC
Us
vu
l
ner
a
bili
ty
with
D
P
A
at
ta
ck
is
prov
e
d
i
n
t
his
work.
In
ou
r
t
est
,
we
ha
ve
s
ucceede
d
in
desig
ning a
n
at
ta
cking
e
nvir
onm
ent that ru
ns on
a
n ATME
GA3
28P m
ic
ro
con
t
ro
ll
eras s
how
n
in
F
ig
ur
e
5.
T
he
DPA
te
ch
niqu
e
nee
ds
a
n
i
nput
of
s
om
e
po
we
r
trac
e
a
nd
public
data
su
c
h
as
t
he
ci
ph
e
rtext
to
ca
rr
y
the
recoveri
ng
ke
y
al
go
rithm
.
The
re
searc
he
rs
m
ade
a
co
rr
el
at
ion
bet
w
een
secret
ke
y,
public
data
,
an
d
m
easur
em
ent
t
races
to
recove
r
the
secret
key.
The
cal
culat
ion
is
m
ade
fo
r
ever
y
key
gues
s
.
If
the
key
gu
ess
is
correct
, it wil
l
be
s
how
n
in
the
for
m
of
a
grap
h
sim
il
ar to
the
one in
F
ig
ure
6.
Figure
5. AE
S
DU
T
a
nd Tr
ac
es
DPA
is
a
pr
act
ic
al
way
of
te
sti
ng
w
heth
er
th
e
value
of
the
Kn
gu
e
ssin
g
pr
ocess
is
cl
os
er
to
the
truth.
The
K
n
ca
nd
i
da
te
is
us
e
d
with
the
e
qu
at
io
n
above
t
o
obta
in
the
value
I(
i,
n)
f
or
each
X
(i
,n)
trace
.
A
sel
ect
ion
functi
on
can
be
m
ade
based
on
the
pr
ocess
of
cal
culat
ing
the
value
I(
i,
n).
I
n
this
stud
y
,
the
on
e
-
bit
va
lue
of
I(
i,
n)
(LSB
)
is
us
e
d
as
the
ou
t
pu
t
of
t
he
sel
e
ct
ion
f
unct
ion.
Each
trace
is
searchi
ng
t
he
value
of
one
bi
t
LSB
then
di
vi
de
d
i
nto
t
wo
gr
oups
base
d
on
t
he
outp
ut
of
t
he
sel
ect
ion
funct
ion
.
T
he
a
ver
a
ge
diff
e
re
nce
be
tween
each
s
ubset
is
then
obse
rv
e
d
.
If
the
outp
ut
value
of
t
he
S
-
bo
xes
est
im
ated
by
the
sel
e
ct
ion
functi
on
has
a
sm
a
ll
co
rr
el
at
i
on
value
f
or
t
he
traces, the DPA
t
ech
nique
will
sh
ow
the sp
ikes
of
the tra
ces aver
a
ge
in
dicat
in
g
that
the
gu
e
ss
value
of
Kn
is
true.
F
or
e
ve
ry
wro
ng
K
n,
the
predict
io
n
of
I
(i,n)
values
w
ould
not
be
relat
ed
to
the d
at
a
bei
ng
processe
d
by t
he
ta
r
get
dev
ic
e.
Figure
6
sho
ws
the
distrib
utio
n
of
key
gu
e
ssing
for
the
t
he16
th
sta
te
.
It
is
vi
sible
that
the
key
guesse
s
ind
e
x
num
ber
61
has
a
sig
nif
ic
ant
diff
e
re
nc
e
in
data
distri
bu
ti
on
c
om
par
ed
to
key
gu
es
s
ind
e
x
num
ber
50
to
60.
The
la
r
ge
diff
e
ren
ce
is
assum
ed
as
the
correct
key
gu
e
ss.
T
he
te
st
resu
lt
shows
that
the
at
ta
c
k
ha
s
su
ccee
ded
in rec
ov
e
rin
g
the w
hole
12
8
-
bit
key
(10
0%
key
recovery)
.
Th
e
at
ta
cking
sim
ulati
on
te
st
is
done
by
us
in
g
1,0
50
tra
ces
and
ta
kes
16
m
inu
te
s
of
e
xecu
ti
on.
T
he
key
can
be
dir
ect
ly
reco
ver
e
d
beca
us
e
of
th
e
AES
al
gorithm
vu
ln
erab
il
it
y
in
init
ia
l
AddRou
ndKeyo
per
at
io
n,
wh
ic
h
is,
basical
ly
,
an
XO
R
op
e
rati
on
of
pl
ai
ntext
and
m
ast
erk
ey
.
The
resu
lt
pr
oduce
d
the
ke
y
us
ed,
a
nd
th
e
key
guesses
from
the
si
m
ul
at
ion
co
rr
e
spo
nd
t
he
seq
uen
ce
of th
e sim
ulate
d
sta
te
s (
43
126 2
1 22 40
174 2
10
166 2
71 24
7 2
1 136 9
207 7
9 6
0).
Figure
6. Gl
obal
Su
ccess
Rat
e A
E
S12
8
f
or s
ta
te
nu
m
ber
16
Inde
x of
key
50 to
61
Re
velead
Key
50
51
52
53
54
55
56
57
58
59
60
61
Wro
ng
Key
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
87
08
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
201
8
:
5144
-
5152
5150
The
sam
e analy
sis can
be re
pe
at
ed
f
or
all
16
-
byte
sta
te
s (
n
= 0
,...,
15)
i
n o
btainin
g
al
l o
f 128
-
bit AE
S
ci
ph
e
r
keys
f
rom
the
encr
ypti
on
de
vice.
The
value
of
the
s
uccess
of
this
at
ta
ck
is
sh
ow
n
in
Fig
ure
7.
The
x
-
axis
s
hows
t
he nu
m
ber
of trac
es an
d
t
he
y
-
a
xi
s ind
ic
at
es t
he
k
ey
ob
ta
ine
d.
(a)
(b)
Figure
7. The
c
om
par
is
on of t
race a
ver
a
ges f
or f
al
se
a
nd
c
orrect
sta
te
gu
e
s
ses
Figures
7.a
an
d
7.b
s
how
a
ver
y
sig
nifican
t
diff
ere
nce
be
tween
the
rig
ht
gu
ess
(
blu
e
l
ine)
an
d
the
wrong
key
guess
(
red
li
ne
)
against
powe
r
consum
ption
.
The
gr
a
ph
in
bo
t
h
im
ages
m
easur
es
the
value
of
Do
M
(
Y
-
a
xis)
with
t
race
num
ber
(X
-
axis
).
The
co
rr
ect
ke
y
gu
e
sses
ha
ve
the
highest
l
evel
of
t
race
gr
a
ph
diff
e
re
nce
wh
e
n
c
om
par
ed
t
o
the
a
ver
a
ge
tra
ce
grap
h.
T
he
te
st
res
ults
s
how
n
in
Ta
ble
2
re
pr
ese
nt
th
e
at
ta
ck
su
ccee
ds
in
an
overall
12
8
-
bi
t
key
(
100%
ke
y
acqu
isi
ti
on)
.
Test
in
g
of
at
t
ack
sim
ulati
on
us
e
d
500
trac
es
an
d
it
too
k
120
sec
onds
.
T
he
key
can
be
directl
y
obta
ine
d
beca
us
e
of
the
wea
kn
e
ss
of
the
A
ES
al
gorit
hm
fo
un
d
in
the
init
ia
l
Ad
dR
oundKey
ope
rati
on
wh
ic
h
is
basical
ly
the
XO
R
plainte
xt
op
erati
on
ag
ai
ns
t
the
m
aster
key
directl
y.
Table
2.
T
he
r
e
su
lt
of
a
DPA
at
ta
ck
No
Variable
Testin
g
Res
u
lts
1.
The n
u
m
b
e
r
o
f
tr
a
ces n
eeded
500
2.
Execu
tio
n
ti
m
e
1
2
0
seco
n
d
s
3.
A nu
m
b
e
r
o
f
k
ey
b
i
ts g
ain
ed
1
2
8
b
its
4.
So
m
e
m
iss
in
g
key
b
its
0
b
its
4.
CONCL
UDI
NG RE
MAR
KS
The
im
ple
m
entat
ion
of
a
n
AE
S
enc
ryptio
n
s
yst
e
m
in
anM
CU
has
a
hi
gh
vu
l
ner
a
bili
ty
to
th
e
m
ast
e
r
key
of
the
AE
S
enc
ryptio
n
dev
ic
e
it
sel
f.
A
D
PA
at
ta
ck
is
a
sta
ti
sti
ca
l
at
ta
ck
base
d
on
the
powe
r
us
a
ge
analy
sis
re
qu
i
r
ed
by
the
enc
r
ypti
on
de
vice.
W
e
al
rea
dy
ge
t
the
m
ai
n
prob
le
m
of
t
his
powe
r
-
base
d
a
tt
ack.
Af
te
r
analy
sin
g
the D
P
A
on
t
he
DU
T
,
we
obta
ined
an
at
ta
ck
s
urface o
n
the A
ES
e
nc
ryp
ti
on
d
evice.
Th
e
m
a
in
su
sce
ptibil
it
y
of
AES1
28
li
e
s
in
a
predict
a
ble
powe
r
est
im
at
ion
value
after
the
S
ubB
yt
es
fu
nctio
n
i
n
each
rou
nd.
W
e
ha
ve
su
ccee
ded
de
m
on
strat
ing
it
by
us
i
ng
a
m
in
i
m
u
m
trace
and
ti
m
ena
m
e
l
y,
on
ly
500
trace
s
a
nd
it
too
k
120
se
conds
.
T
he
m
ai
n
vulne
ra
bili
ty
of
AES
t
o
DPA
is
in
t
he
first
r
ound
of
at
ta
ck
(af
te
r
S
BOX
op
e
rati
on).
Th
e
vu
l
ner
a
bili
ty
is
wh
e
n
the
XO
R
of
plaint
ext
an
d
the
m
ast
er
key
f
ollo
wed
by
a
no
n
-
li
near
su
bst
it
ution
of
SubBy
te
s
fun
c
ti
on
(
SBO
X).
The
at
ta
cke
r
can
easi
ly
m
ake
the
em
pir
ic
al
com
pu
ta
tio
n
i
n
determ
ining
th
e
interm
ediat
e
value.
This
value
is
essenti
al
in
rec
ov
e
rin
g
the
secret
key.
We
belie
ve
tha
t
this
vu
l
ner
a
bili
ty
can be s
olv
e
dby hidi
ng and m
askin
g
th
e inter
m
ediat
e v
al
ue.
A
s
F
i
gure
8
s
hows
, we c
ou
l
d ob
ta
in
the
interm
ediat
e
value
m
od
el
le
d
as
ham
ing
weig
ht
(
8
bit)
f
or
e
ve
ry
m
essa
ge
. I
n
f
uture
re
search
, w
e
will
try
to
rand
om
iz
e
the
interm
ediat
e
v
al
ue
by
us
in
g
inform
ation
f
u
sion
[
23
–
31]
a
nd
c
on
st
an
t
we
ight
enco
din
g
[
27
]
,
[32]
appro
ac
h.
Tho
se
ap
proa
ches
co
uld
dec
ei
ve
at
ta
cker
s
in
reco
ve
rin
g
A
ES
encr
y
ption
dev
ic
e
’s
secret
key.
We
pro
po
se
d
that
m
et
ho
d
cal
le
d
as
c
ogniti
ve
-
ma
skin
g
.
C
ogniti
ve
-
m
asking
is
bu
il
t
on
the
m
os
t
sig
nificant
powe
r
usa
ge
analy
sis
aff
ect
ing
a
sin
gle
m
essage
enc
r
ypti
on
process
.
Co
gn
it
ive
-
m
askin
g
is
a
c
on
ce
pt
dev
el
op
e
d
fro
m
inform
ation
f
us
i
on.
T
he
c
oncept of
in
for
m
at
ion
f
us
io
n
i
s
to
c
om
bin
e
two
inf
orm
ation
q
uic
kly
and accu
ratel
y t
o
get t
he best
value o
f
m
asking
decisi
on
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N:
20
88
-
8708
Reveali
ng AE
S Encry
ptio
n De
vi
ce Ke
y on
328P
Mi
cr
oc
on
tr
ollers w
it
h Dif
fe
rential P
ower
…
(
Ad
ang S
.
A
.
)
5151
Figure
8. Ham
m
ing
w
ei
gh
t
di
stribu
ti
on
f
or
pl
ai
ntext num
ber
148, 14
9,
a
nd
150
ACKN
OWLE
DGE
MENTS
The
fi
rst
aut
hor
ack
nowle
dg
e
s
suppo
rt
from
the
I
ndonesi
a
En
dowm
ent
Fu
nd
for
E
du
cat
ion
(LPDP
-
Lem
bag
a
Pengel
ola
Da
na
P
end
i
dik
a
n)
sc
hola
rs
hip
,
Mi
ni
stry
of
Fina
nc
e,
The
Re
publ
ic
of
Indonesi
a.
The
auth
or
s
grate
f
ul
ly
kn
owle
dge
the ano
nym
ou
s r
e
viewers f
or t
heir val
ua
ble c
omm
ents.
REFERE
NCE
S
[1]
S.D.
Putra,
A.S.
Ahm
ad,
and
S.
Sutikno,
“
Design
of
an
AES
D
evi
c
e
as
Device
Under
Te
st
in
a
DP
A
Atta
ck”
,
in
Inte
rnational
Jo
urnal
of
N
et
wor
k
Se
curit
y
,
20
18.
[2]
S.S.
Chawla
an
d
N.
Goel,
“
FPGA
implementa
ti
on
of
an
8
-
bit
AES
arc
hit
ectu
re:
A
roll
ed
an
d
m
aske
d
S
-
Bo
x
appr
oac
h
”
,
in
20
15
Annual IEEE India
Conf
ere
nc
e
(
INDICO
N
)
,
2015,
pp
.
1
–
6.
[3]
H.
Chen,
Y.
C
hen,
and
D
.
H.
Summ
erv
il
le,
“
A
surve
y
on
th
e
application
of
FP
GAs
for
net
work
infra
structur
e
sec
urity
”
,
Comm
unic
ati
ons
Surve
ys
&
Tutor
ial
s,
IEE
E
,
vol
.
13
,
no
.
4
,
pp
.
541
–
561
,
2011.
[4]
T.
S.
Mess
erg
es,
E.
Dabbish,
R
.
H.
Sloan,
and
o
the
rs,
“
Exa
m
ining
sm
art
-
ca
rd
sec
uri
t
y
under
th
e
threat
of
pow
er
ana
l
y
sis
at
t
ac
ks”
,
Computers,
IE
EE
Tr
ansacti
ons
on
,
vol. 51
,
no.
5,
pp
.
541
–
552
,
2002.
[5]
A.
Ariva
zha
gan
and
o
the
rs,
“
RTL
Mode
ll
ing
f
or
the
Ciph
er
Blc
ok
Ch
ai
ning
Mode
(Cbc)
f
or
Data
Secur
i
t
y
”
,
Indone
sian J
our
nal
of
Elec
tric
al
Engi
ne
ering
and
Computer
Sc
ie
n
ce
,
vol
.
8
,
no
.
3
,
2017.
[6]
S.
Oukili
and
S.
Bri,
“
High
throughpu
t
FP
G
A
I
mpl
ementation
of
Data
Enc
r
y
p
ti
on
Standa
rd
with
ti
m
e
var
ia
ble
sub
-
ke
y
s”
,
Inte
rnat
io
nal
Journal
of
Elec
tri
cal
and
Co
mputer
Engi
n
ee
r
ing
,
vo
l. 6, no. 1, p. 298, 2016.
[7]
P.
Sara
vana
n
,
N.
Raj
adur
ai,
and
P.
Kalpa
na,
“
Pow
er
ana
l
y
sis
attac
k
on
8051
m
ic
roc
ontr
o
ll
ers”
,
in
Computati
on
al
Inte
lligen
ce and Com
puti
ng
R
ese
arch
(
ICCIC)
,
2014
IEEE
In
te
rn
ati
onal
Con
fe
ren
ce
on
,
2014
,
pp
.
1
–
4.
[8]
F.
Koeune
and
F.X.
Standaert
,
“
A
tut
oria
l
on
p
h
y
sic
al
se
cur
i
t
y
and
side
-
cha
nne
l
atta
cks,
Found
at
ions
of
Se
cur
i
t
y
Anal
y
sis
and
D
e
sig
n
III:
FO
SA
D 2004/
2005
tut
or
ia
l
lectur
es
”
,
Springer
-
Verl
ag, Berlin, He
id
el
ber
g
,
2005.
[9]
D.R.
Ran
i
and
S
.
Venka
t
eswarl
u,
“
Secur
ity
ag
ai
ns
t
Ti
m
ing
Ana
l
y
s
is At
ta
ck
”
,
Inte
r
nati
onal Journal
of
E
le
c
tric
al
an
d
Computer
Engi
n
ee
ring
(
IJE
C
E)
,
vol.
5
,
no
.
4
,
p
.
7
59,
20
15
.
[10]
P.
C.
Koche
r
,
“
Tim
ing
at
tacks on
implementa
t
ions o
f
Diffie
-
Hel
lman,
RS
A,
DS
S,
and
othe
r
s
y
st
ems
”
,
in
Adv
anc
es
i
n
Cryptol
ogy
—
CR
Y
PTO’96
,
1996
,
pp.
104
–
113
.
[11]
G.
Dee
p
a,
G
.
S
ri
T
ej
a
,
and
S.
Venka
te
sw
arl
u
,
“
An
Overvi
ew
of
Acoustic
Sid
e
-
Channel
Att
ack
”
,
In
te
rnationa
l
Journal
of
Computer
Sc
ie
nc
e
&
Comm
unic
ati
on
Net
works
,
vo
l. 3, no. 1, p. 15, 201
3.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
87
08
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
201
8
:
5144
-
5152
5152
[12]
M.
Masoum
i
and
M.H.
Reza
y
ati,
“
Novel
appr
oac
h
to
p
r
ote
c
t
adv
anced
enc
r
y
p
ti
on
st
anda
rd
al
gori
th
m
implementa
t
ion
aga
inst
diff
ere
nt
ia
l
elec
tromagne
ti
c
and
power
an
aly
s
is”
,
Informat
ion
Forensic
s
and
Sec
urity,
IEE
E
Tr
ansacti
ons on
,
vol. 10, no. 2, p
p.
256
–
265
,
201
5.
[13]
P.
Koche
r
,
R.
Le
e
,
G.
McGra
w,
A.
R
aghuna
t
han,
and
S.
Modera
tor
-
R
avi,
“
Secur
ity
as
a
n
ew
dimension
i
n
embedde
d
s
y
s
tem
design”
,
in
Proce
ed
ings o
f
the
41st
a
nnual
Desi
gn
Aut
omat
ion Confe
renc
e
,
200
4,
pp
.
753
–
760
.
[14]
L.
Goubin
and
J.
Pata
rin
,
“
DES
and
diffe
ren
tial
power
anal
y
s
is
the
“
dupli
ca
t
i
on
?
m
et
hod
”
,
in
Cryptographic
Har
dware
and
E
mbedde
d
Syst
em
s
,
1999,
pp.
158
–
172.
[15]
E.
Bri
er,
C.
Cl
av
ie
r,
and
F.
Ol
ivier,
“
Corr
elati
on
power
anal
y
sis
with
a
l
ea
k
age
m
odel
”
,
in
Crypt
ographic
Har
dware
and
Embe
dded
S
yste
ms
-
CHES
20
04
,
Springer
,
20
04,
pp
.
16
–
29
.
[16]
H.
Li,
K.
W
u,
B
.
Peng,
Y
.
Zh
ang,
X.
Zh
eng,
and
F.
Yu,
“
Enha
nc
ed
cor
re
la
t
ion
pow
er
an
aly
sis a
ttac
k
on
sm
art
ca
rd”
,
in
Y
oung
Co
mpu
te
r Sc
ie
nt
ists,
20
08.
IC
Y
CS
2008
.
The
9th
Int
ernat
ional
Conf
ere
nc
e
for
,
2008,
pp.
2143
–
2148.
[17]
E.
Biha
m
and
A.
Sham
ir,
“
Diffe
ren
t
ia
l
cr
y
pt
a
na
l
y
s
is
of
DES
-
li
ke
cr
y
p
tos
y
st
e
m
s
”
,
in
Adv
anc
es
in
Cryptol
og
y
-
CR
Y
PTO
,
1991,
vol.
90
,
pp
.
2
–
21
.
[18]
S.
Guill
e
y
and
R.
Pac
al
e
t,
“
SoC
s
sec
urity
:
a
w
ar
ag
ai
nst
side
-
c
hanne
ls”
,
in
Ann
ale
s
des
t
él
é
com
municat
ions
,
20
04,
vol.
59
,
no
.
7
–
8
,
pp.
998
–
1009
.
[19]
P.
Koche
r,
J
.
Jaf
fe,
and
B.
Ju
n,
“
Diffe
ren
t
ia
l
power
anal
y
s
is
”
,
in
Adv
anc
es
in
Cry
ptol
ogy
—
CR
Y
P
TO
’99
,
1999,
pp
.
388
–
397.
[20]
S.D.
Putra,
A.S.
Ahm
ad,
and
S.
Sutikno,
“
Po
wer
ana
l
y
sis
at
t
ac
k
on
implementation
of
DES”
,
in
Information
Technol
ogy
Syst
ems and
Innov
a
t
ion
(
ICITSI)
,
2016
Inte
rnat
ional
Confe
renc
e
on
,
2016,
pp
.
1
–
6.
[21]
Y.
Souis
si
,
S.
Guill
e
y
,
S.
Bh
asi
n,
and
J.
L.
Dan
ger
,
“
Com
m
on
f
ramework
to
ev
al
ua
te
m
oder
n
e
m
bedde
d
s
y
stem
s
aga
inst
sid
e
-
channel
a
tt
a
cks”
,
in
Technol
ogi
es
for
Hom
el
and
Secur
it
y
(
HST)
,
2011
IEE
E
In
te
rnat
ional
Confe
ren
c
e
on
,
2011
,
pp
.
86
–
91.
[22]
B.
Gier
lichs,
E.
De
Mulder,
B.
Prene
el
,
and
I.
Verba
uwhede
,
“
Empiric
al
compari
son
of
si
de
cha
nne
l
ana
l
y
si
s
disti
nguishers
on
DES
in
har
dw
are
”
,
in
Circuit
Theory
and
Desi
gn,
2009.
ECCTD
2009.
European
Confe
renc
e
o
n
,
2009,
pp
.
391
–
3
94.
[23]
A.
S.
Ahm
ad
and
K.
O.
Bac
hri,
“
Cognit
ive
art
if
icial
intel
li
gen
ce
m
et
hod
for
m
e
as
uring
tra
nsform
er
per
form
anc
e
”
,
in
2016
Fut
ure
Tec
hnologi
es
Con
fer
enc
e
(
FTC)
,
2016,
pp
.
67
–
73
.
[24]
K.O.
Bac
hri
,
A.
D.W.
Sum
ari
,
B.
A.
Soedja
rno
,
a
nd
A.S.
Ahm
ad,
“
The
implementa
ti
on
of
A3S
informati
on
fusion
al
gorit
hm
for
in
te
rpre
ti
ng
Diss
olve
d
Gas
Anal
y
sis
(
DGA)
base
d
on
Doerne
nburg
Rat
io
”
,
i
n
2017
Inte
rnation
al
Symposium on
E
le
c
tronic
s and
S
mar
t
Dev
ices (
ISESD)
,
2017,
pp
.
335
–
340.
[25]
L.
Goeir
m
ant
o
,
R.
Mengko,
an
d
T.
L.
Ra
ja
b
,
“
Dire
ction
of
vent
ric
l
e
cont
r
ac
t
i
on
base
d
on
pr
ec
ordi
al
lead
E
CG
signal
”
,
in
2016
4th
Int
ernati
onal
Confe
ren
ce on Cybe
r an
d
IT
Se
rvic
e
Manage
me
nt
,
2016
,
pp
.
1
–
3.
[26]
S.D.
Putra
,
A
.
S
.
Ahm
ad,
and
S.
Sutikno,
“
DP
A
-
count
ermea
sure
with
knowl
edge
growing
sy
stem”
,
in
201
6
Inte
rnational
Sy
mpos
ium
on
El
e
ct
ronics
and
Smar
t
Dev
i
ce
s (
ISE
SD)
,
2016,
pp.
1
6
–
20.
[27]
S.D.
Putra,
M.
Yudhipra
wira
,
Y
.
K
urnia
wan,
S.
Sutikno,
and
A.
S.
Ahm
ad,
“
Secur
ity
anal
y
s
is
of
BC3
al
gorit
hm
f
or
diffe
ren
ti
a
l
power
anal
y
sis
a
tt
a
c
k”
,
in
2017
Int
e
rnational
Sympo
sium
on
El
ec
tro
nic
s
and
Smar
t
Dev
ices
(
ISESD
)
,
2017,
pp
.
341
–
3
45.
[28]
C.
O.
Sere
at
i
,
A.
D.W.
Sum
ari
,
T
.
Adiono,
and
A.
S.
Ahm
ad,
“
Cognit
ive
artifi
ci
a
l
i
nte
lligen
ce
(CAI
)
software
bas
ed
on
knowledge
growing
s
y
stem
(KG
S)
for
diagnos
ing
hea
r
t
b
loc
k
and
arr
y
t
h
m
ia
”
,
in
2017
6th
Int
ernati
ona
l
Confe
renc
e
on
E
le
c
tric
al
Engi
n
e
ering
and
In
formatic
s (
ICEE
I)
,
2017,
pp
.
1
–
5.
[29]
C.
O.
Ser
eati,
A.D.W.
Sum
ari
,
T.
Adiono,
a
nd
A.S.
Ahm
a
d,
“
Im
ple
m
ent
at
ion
Know
le
dge
Grow
ing
Sy
st
em
Algorit
hm
using
VH
DL”
,
in
201
6
Inte
rnat
ional
Symposium
on
El
e
ct
ronics
and
Smar
t
Dev
ices
(
ISESD)
,
2016,
p
p.
7
–
10.
[30]
A.D.W.
Sum
ari
,
A.S.
Ahm
ad,
A.I.
W
ur
y
anda
ri
,
and
J.
Sem
biring,
“
Brai
n
-
inspi
red
Know
le
dge
Grow
ing
-
Sy
st
e
m
:
Towa
rds
A
True
Cognit
ive
Agen
t”
,
In
te
rnationa
l
Journal
of
Com
pute
r
Scienc
e
&
Arti
ficial
Intelligence
(
IJCSA
I)
,
vol.
2
,
no
.
1
,
pp
.
26
–
36,
2012
.
[31]
H.R.
A.
Ta
lompo
,
A.S.
Ahm
ad,
Y.S.
Gondokar
y
o
no,
and
S.
Sutik
no,
“
NA
IDS
des
ign
using
ChiMI
C
-
KGS”
,
in
201
7
Inte
rnational
Sy
mpos
ium
on
El
e
ct
ronics
and
Smar
t
Dev
i
ce
s (
ISE
SD)
,
2017,
pp.
3
46
–
351.
[32]
S.
D.
Putra,
A
.
S.
Ahm
ad,
S.
Sutikno,
and
Y
.
Kurniawa
n,
“
Atta
ck
ing
AES
-
Masking
Enc
r
ypti
on
Devi
ce
w
it
h
Corre
lation
Po
wer
Anal
y
s
is”
,
in
Inte
rnat
ional
Journal
of
Co
mm
unic
ati
on
N
et
works
and
Inf
orm
ati
on
Sec
uri
ty
(
IJCNIS)
,
2018,
pp.
397
–
402
.
Evaluation Warning : The document was created with Spire.PDF for Python.