Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
12
,
No.
3
,
Decem
ber
201
8
, p
p.
941
~
949
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
2
.i
3
.pp
941
-
949
941
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
A Compreh
ensive Re
view
of Appl
ications
of D
on’t Care
Bit
Fill
ing T
ec
hn
iqu
es for
Test Powe
r Reducti
on
in Digi
tal VLSI
Systems
Sa
n
joy Mitr
a
1
,
De
bapras
ad
Da
s
2
1
Depa
rtment of
Com
pute
r
Scie
n
ce
and Engi
ne
ering,
Tr
ipura Insti
t
ute
of
T
ec
hnolo
g
y
,
Agart
ala, Ind
ia
2
Depa
rtment of
El
e
ct
roni
cs
and
Com
m
unic
at
ion Engi
ne
eri
ng,
TS
SO
T,
As
sam
Univer
sit
y
,
Sil
cha
r
,
India
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Feb
21
,
2018
Re
vised Jul
10,
2018
Accepte
d
Aug
2
9
, 201
8
Mass
ive
power
consum
pti
on
during
VLSI
te
s
ti
ng
is
a
ser
iou
s
thre
at
to
rel
i
abi
l
ity
con
c
ern
s
of
ubiquitous
sili
con
in
dustr
y
.
Man
y
low
-
power
m
et
hodologi
es
a
re
found
in
th
e
te
chn
ic
a
l
l
it
er
ature
to
addr
ess
t
his
i
ss
ue
of
te
st
m
ode
high
power
consum
pt
ion
and
don’t
care
bit
(X)
fil
li
ng
appr
oac
h
es
are
one
of
the
m
in
thi
s
fra
te
rn
ity
.
The
se
don’t
c
are
(X)
bit
filling
te
chn
ique
s
have
dra
wn
the
signifi
c
ant
attention
of
industr
y
and
ac
ad
emia
fo
r
it
s
highe
r
compati
bilit
y
wi
th
exi
sting
desig
n
flow
as
nei
th
er
m
odifi
ca
t
ion
of
the
CUT
is
req
uire
d
nor
they
n
ee
d
to
rer
un
the
ti
m
e
-
consum
ing
ATPG
proc
ess.
Th
is
pape
r
pre
sen
ts
a
comprehe
nsive
rev
ie
w
of
the
ap
pli
c
at
ions
of
do
n’t
c
ar
e
bit
fil
li
ng
te
chn
iqu
es
for
m
it
iga
ti
n
g
prime
two
conc
er
ns
of
d
y
n
amic
power
dissipat
ion
n
amel
y
shif
t
power
a
nd
ca
ptur
e
powe
r,
oc
cur
red
dur
in
g
full
sca
n
te
sting
.
Ke
yw
or
ds:
Ca
ptu
re
po
wer
DF
T
Don’
t
care
(X)
bits
Sh
ift
powe
r
WT
A
X
-
fill
ing
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
:
Sanjo
y M
it
ra,
Trip
ur
a
Insti
tute o
f
Tec
hnolog
y,
Nar
si
ng
a
r
h,
T
r
ipura,
India.
Em
a
il
:
m
ai
l.s
m
it
ra@g
m
ai
l.com
1.
INTROD
U
CTION
The
te
sti
ng
ph
eno
m
enon
in
s
il
ic
on
industry
is
known
for
it
s
ov
e
rr
at
e
d
po
wer
c
onsu
m
ption
a
nd
this
rem
ai
ns
un
al
te
red
in
ca
se
of
scan
te
sti
ng.
This
te
st
m
ode
high
po
wer
consum
ption
i
s
po
te
nt
en
ough
for
dam
aging
the
chip
or
eve
n
m
ay
beco
m
e
a
reason
f
or
power
-
in
du
ce
d
te
st
yi
el
d
loss.
As
a
res
ult,
sign
i
ficant
researc
h
at
te
ntion
was
fo
c
us
e
d
towa
rds
fin
d
ing
s
uitable
low
-
powe
r
te
sti
ng
strat
egies.
T
he
two
pri
m
e
v
ariant
s
of
te
st
po
wer
duri
ng
sca
n
te
st
are
sh
i
ft
powe
r
(i.e.
,
the
pow
er
co
nsum
ed
du
ri
ng
t
he
scan
sh
ifti
ng
ope
rati
ons
)
and capt
ur
e
po
wer (i.e.
, th
e
pow
e
r
c
onsu
m
ed durin
g
t
he
re
sp
onse
capt
ur
e
cyc
le
).
In
c
rease
i
n
s
wi
tc
hin
g
act
ivit
y
occurre
d
duri
ng
t
he
s
hiftin
g
of
te
st
data
int
o
t
he
sca
n
c
hai
n
is
view
ed
as a
m
ajo
r
sho
r
tc
omi
ng
of sca
n
-
base
d
te
st schem
e. Redu
ct
ion
in
t
he
num
ber
o
f
sca
n
cel
l’
s sign
al
transi
ti
on
s is
a fam
il
ia
r
app
r
oach
to red
uce
the p
owe
r
di
ssi
pation durin
g
s
can
-
base
d
te
sti
ng
a
nd
m
ay
b
e
cl
assifi
ed
into thr
ee
cat
egories:
a)
Diff
e
re
nce
between
the
ad
j
ac
ent
sca
n
cel
ls’
values
,
wh
il
e
t
he
te
st
sti
m
ulu
s
is
loa
ded
int
o
the
sca
n
-
chai
n,
is t
erm
ed
as th
e scan
-
in t
ran
si
ti
on
s.
b)
Diff
e
re
nce am
i
d
the
test
sti
m
u
li
an
d
it
s
res
pons
e
of the
sam
e scan
cell
s is term
ed
as capt
ure tra
ns
it
ion
s
c)
Wh
il
e
un
l
oad
i
ng
the
res
ponse
s
in
sca
n
-
out
m
od
e,
the
diff
e
ren
ce
bet
ween
the
ad
j
ace
nt
sc
an
cel
ls’
value
s
are
known
as s
can
-
out t
ransi
ti
on
s
Scan
-
sh
i
ft
-
po
w
er
is
relat
e
d
to
first
a
nd
th
ir
d
cat
egory
a
nd
needs
to
be
m
i
nim
iz
ed
in
ord
er
to
a
pply
higher
sh
i
ft
f
r
equ
e
ncy
for
r
edu
ci
ng
the
te
sti
ng
ti
m
e
and
cost.
The
ca
pture
-
powe
r
r
el
at
es
to
the
s
econd
cat
egory.
T
he
te
st
m
od
e
capt
ur
e
-
po
wer
s
ho
uld
be
ke
pt
sm
al
le
r
than
peak
-
pow
er
th
res
hold
of
CUT
t
o
avo
i
d
the
crop
ping
up
of
ICs
da
m
aging
e
xcess
ive
heat.
Mo
r
eov
e
r,
higher
switc
hing
act
ivit
y
or
iginate
s
fr
om
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.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
9
4
1
–
9
4
9
942
si
m
ultaneou
s
t
est
ing
of
m
ult
i
ple
cor
es
a
nd
t
est
com
pacti
on
.
Adversel
y,
hot
sp
ots
m
ay
b
e
dev
el
ope
d
in
creased
from
excessive
switc
hing
act
i
vity
durin
g
sca
n
s
hift
that
m
ay
dam
age
the
s
il
ic
on
,
the
bo
ndin
g
wires,
an
d
e
ven
the p
ac
ka
ge
1
.
1.
B
ackgro
und
Power
-
awa
re
strat
egy
f
or
fi
ll
ing
of
don’t
care(
X)
bits
of
the
te
st
c
ube
duri
ng
sca
n
te
sti
ng
a
re
gen
e
rall
y
m
ade
to
be
ex
plo
it
ed
f
or
ei
the
r
s
hift
powe
r
re
duct
ion
or
ca
pt
ur
e
powe
r
reducti
on
or
e
ve
n
bo
th
of
them
.
Du
ring
s
can
te
sti
ng
,
do
n’
t
care(
X
)
bits
can
be
eff
ect
i
vely
rep
la
ced
in
orde
r
to
lowe
r
dow
n
the
to
gg
li
ng
act
ivit
y
up
t
o
the
desire
d
le
ve
l.
Ef
fici
ent
fill
ing
of
don’
t
c
are
bits
m
ay
l
ead
t
o
si
gn
ific
ant
re
duct
io
n
i
n
te
st
powe
r
c
on
s
umpti
on.
Well
know
n
X
-
fill
ing
te
ch
niques
s
uch
as
0
-
fill
,
1
-
fill
,
m
ini
m
um
tra
ns
it
ion
-
fi
ll
and
adj
ace
nt
-
fill
ha
ve
s
how
n
a
no
ta
ble
re
du
ct
io
n
in
a
ver
a
ge/pe
ak
power
du
ring
scan
s
hiftin
g
or
duri
ng
la
unch
to
captu
re.
T
he
f
undam
ental
m
erit
of
lo
w
-
power
don’t
care
bit
fill
ing
te
c
hniq
ue
is
that
it
can
neithe
r
re
su
lt
s
i
n
area
ov
e
rh
e
ad
nor de
gr
a
des p
erfor
m
ance.
1
.
2
.
Pr
ob
le
m
The
iss
ue
of
e
xc
essive
te
st
po
wer
has
t
urne
d
out
to
be
m
or
e
te
rr
ible
with
t
he
boundless
ut
il
iz
ation
of
at
-
sp
ee
d
sca
n
t
est
ing
,
w
hich
i
s
cu
rr
e
ntly
ob
l
igatory
for
perform
ance
ver
if
ic
at
ion
.
Re
gardless
of
t
he
te
sti
ng
strat
egy
util
iz
ed,
the
iss
ue
of
excessive
po
wer
am
id
te
stin
g
ca
n
be
pa
rt
into
tw
o
s
ub
pro
blem
s
i)
Ex
cessi
ve
powe
r
duri
ng
s
hift
cy
cl
e
ii
)
Excessive
powe
r
duri
ng
Lau
nc
h
-
To
-
Ca
pture
(LTC)
cy
cl
e.
T
his
is
well
re
porte
d
that
switc
hing
ta
sk
s
of
a
ci
rcu
it
dep
e
nd
a
bl
y
pr
om
pt
heat
dissem
inati
on.
On
the
off
c
han
ce
that
the
ci
rcu
it
tem
per
at
ur
e
is
too
hi
gh
no
t
withstan
ding
a
m
id
a
bri
ef
te
rm
of
on
-
li
ne
te
st
sessi
on
it
m
igh
t
ha
ve
th
e
accom
pan
yi
ng
ci
rcu
it
r
el
at
ed
issues suc
h
as
c
hip
dam
age,
lo
wer reli
abili
ty
, lo
w
te
st t
hroug
hput a
nd
yi
el
d l
os
s.
1
.
3
.
Analy
tical co
nt
ri
b
ut
io
n
This
pa
per
int
rod
uces
a
com
pr
ehen
sive
s
urvey
on
the
us
es
of
don'
t
care
bit
fill
ing
m
et
ho
ds
f
or
m
od
erati
ng
pri
m
e
two
c
oncer
ns
of
dy
nam
ic
powe
r
dissipat
ion
t
o
be
s
peci
fic
sh
i
ft
po
wer
an
d
ca
pture
powe
r,
happe
ned
am
id
fu
ll
scan
te
stin
g.
In
this
pa
pe
r,
we
ha
ve
id
entifi
ed
an
d
an
al
yz
ed
the
m
erit
s
and
shortco
m
ing
s
of
va
rio
us
te
st
powe
r
re
duct
ion
te
ch
niques
base
d
on
don’
t
care
bit
fill
ing
an
d
th
e
va
riants
of
te
st
powe
r
cov
e
re
d
in
our
analy
sis
include
sh
ift
powe
r
and
capt
ur
e
powe
r.
Be
sides
these
two
vari
ants
of
te
st
po
we
r
,
there
are
so
m
e
te
st
power
r
edu
ct
io
n
te
c
hniqu
es
w
hich
r
edu
ce
both
s
hi
ft
an
d
capt
ur
e
powe
r
an
d
w
e
hav
e
identifie
d
their
m
erit
s
and
shortc
om
ing
s
al
so
.
T
he
m
os
t
sign
i
ficant
as
pec
t
of
t
his
a
naly
ti
cal
work
li
es
in
th
e
fact
that
we
hav
e
dr
a
w
n
a
relat
ive
p
erfo
rm
a
nce
com
par
iso
n
am
on
g
these
don’t
care
bit
fill
ing
te
chn
iq
ue
s
an
d
this
com
par
iso
n
is
ba
sed
on
t
he
%
of
te
st
po
wer
re
du
ct
io
n
on
t
he
giv
e
n
be
nch
m
ark
ci
rc
uit
corres
ponding
t
o
a
par
ti
cula
r
fill
in
g
te
ch
nique.
2.
TE
ST POWE
R AT A
GL
A
NC
E
2
.
1
.
Va
ri
ants
of Tes
t
P
ower
At
-
s
peed
te
sts
are
widely
rec
ognized
by
the
industry
as
it
ai
ds
to
unc
ov
e
r
tim
ing
-
spe
ed
-
relat
ed
a
nd
even
un
-
m
od
el
le
d
de
fects
of
t
he
CU
Ts
.
At
-
s
peed
te
sts
ty
pical
ly
entai
l
lon
ger
l
ow
-
f
re
qu
e
ncy
sh
i
ft
phase
an
d
a
sh
ort
ca
ptu
re
phase.
Test
data
are
us
ua
ll
y
lo
aded
into
scan
chains
at
a
lo
wer
fr
e
quency
in
ord
er
t
o
m
ai
ntain
CUT’s
po
wer
const
raint a
nd
al
so
to re
du
ce
sh
ift
power.
Although,
h
i
gh test
co
st m
igh
t h
ave to be
inc
urred
f
or
this
ty
pe
of
st
rategy.
Be
st
possible
re
duct
ion
i
n
the
CU
T’s
s
hift
-
powe
r
diss
ipati
on
f
aci
li
ta
te
s
hig
her
sh
if
t
fr
e
qu
e
ncy
to
pro
vid
e
te
st
pa
r
al
le
li
s
m
yi
e
lding
reduce
d
te
sti
ng
tim
e.
In
va
li
dation
of
te
st
resu
lt
s
m
ay
arise
due
to
IR
-
dr
op
a
nd/or
gro
und
bounce
eff
ect
s
trig
ger
e
d
by
exces
sive
at
-
sp
ee
d
tr
ansiti
on
s
durin
g
the
captu
re
phase
.
H
ence
, c
on
ta
i
nm
ent o
f
capt
ure
-
po
wer
dissip
at
ion
unde
r
the
CUT
’s peak
pow
e
r
c
onstrai
nt
b
ecom
es inevi
ta
ble.
2
.
2
.
Met
ri
cs
of Po
w
er Es
ti
mat
i
on
The
m
ajo
r
fr
ac
ti
on
of
powe
r
is
dissipated
in
a
CM
OS
ci
rcui
t
wh
en
switc
hi
ng
el
em
ents
s
witc
h
from
log
ic
0
to
1
or
vice
ve
rsa.
S
witc
hing
el
em
ents
will
m
ake
transiti
ons
on
l
y
in
the
e
ven
t
of
ei
the
r
the
pri
m
ary
inputs
al
te
r
value
or
the
scan cell
s
chan
ge
va
lues.
Sa
nk
a
rali
ng
am
et
al
.
(2
00
0)
[
1]
sh
owe
d
that
the
total
po
we
r
consum
ption
i
n
scan
-
based
t
est
ing
is
not
only
based
on
the
num
ber
of
transiti
ons
in
te
st
set
bu
t
al
so
on
t
he
relat
ive
posit
io
n
of
wh
e
re
the
transiti
on
occ
ur
s
.
O
ne
c
omm
on
m
et
ric
weigh
te
d
t
ransi
ti
o
ns
m
et
ric
(W
TM)
is
us
e
d
to
est
i
m
a
te
the
te
st
po
w
er.
Th
e
m
e
tric
is
strongly
associat
ed
with
th
e
switc
hing
act
ivit
y
in
the
interna
l
nodes
of
C
UT
durin
g
sca
n
-
s
hift
operati
on.
It
was
e
xp
e
rim
ental
ly
sh
own
by
Sanka
rali
nga
m
et
al
.
(2
00
0)
[1
]
that
highe
r
va
lue
of
WTM
corres
pondin
g
to
sca
n
vect
or
s
res
ults
in
m
or
e
powe
r
dissipati
on
in
C
UT.
The WT
M
for t
he
sca
n
-
in
test
stim
uli
can b
e
com
pu
te
d by
:
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 Co
mp
r
ehe
nsi
ve Revi
ew
o
f
Ap
plicati
ons
of Don’t Ca
re Bi
t Fi
ll
ing
Tec
hniqu
es
for Test
…
(
Sanjo
y
Mi
tra
)
943
1
1
1
,
,
)
)(
(
l
i
i
j
i
j
j
t
t
i
l
W
T
M
(1)
Wh
e
re
the
s
ca
n
-
c
hain
le
ng
t
h
is
denoted
by
and
a
sca
n
ve
ct
or
wit
h
sca
nn
e
d
in
befo
re
and
s
o
on
.
S
can
in
aver
a
ge
s
hift
powe
r,
an
d pea
k p
ow
e
r for a
te
st set
can
be
e
s
tim
a
te
d
as
fo
ll
ow
s
n
t
t
i
l
P
m
j
l
i
i
j
i
j
a
v
e
1
1
1
1
,
,
)
(
)
(
(2)
1
1
1
,
,
)
,...,
3
,
2
,
1
(
)
(
)
(
m
a
x
l
i
i
j
i
j
n
j
p
e
a
k
t
t
i
l
p
(3)
Eq
uations
2
an
d
3
s
how
that
reducin
g
the
te
st
vector
s
’
tra
nsi
ti
on
an
d
the
weig
ht
(l
-
i)
are
the
key
facto
r
s
for
reducin
g
t
he
a
ver
a
ge
a
nd
pea
k
powe
r.
The
s
a
m
e
equ
at
io
ns
can
be
us
e
d
to
est
i
m
at
e
al
so
the
a
ver
a
ge
a
nd
pea
k
-
powe
rs
in
sca
n
-
out m
od
e.
We
co
ns
ide
r
the
num
ber
of
transiti
ons
in
s
can
cel
ls
fo
r
e
ach
scan
-
c
hain
to
com
pu
te
t
he
capt
ur
e
-
powe
r.
This
is
becau
se
of
th
e
li
near
relat
io
ns
hi
p
exists
be
tween
ca
ptu
r
e
transiti
on
s
on
the
scan
-
c
ha
in
an
d
peak
-
power
of
the
ci
rc
uit.
S
o,
it
is
at
tem
pted
to
m
ini
m
iz
e
t
he
Ham
m
ing
di
sta
nce
bet
wee
n
te
st
sti
m
uli
and
it
s
respo
ns
e
on
ea
ch
scan
cel
l.
T
his
will
reduce
the
peak
-
pow
er
of
t
he
ci
rcu
i
t
in
te
st
m
od
e.
Fil
li
ng
the
ent
ire
or
la
rg
er
num
ber
of
uns
pecifie
d
bits
in
the
te
st
set
to
reduce
the
pea
k
-
powe
r
m
a
y
aff
ect
the
com
pr
essio
n
eff
ic
ie
ncy an
d m
ay
incr
ease t
he
total
pow
e
r, i
.e.
scan
-
in a
nd scan
-
ou
t t
ra
nsi
ti
on
s. It i
s r
equ
i
red
t
o
m
ai
nt
ai
n
the
captu
re
-
powe
r
within
t
he
ci
rc
uit’s
peak
-
pow
er
li
m
it
fo
r
pr
op
e
r
operati
on
.
Fil
li
ng
one
unspe
ci
fied
bit
in
the
te
st
stim
ul
i
m
a
y
aff
ect
m
any
un
s
pecifie
d
bit
s
in
the
te
st
resp
onse
wh
ic
h
m
ay
cause
captu
re
-
powe
r
vio
la
ti
on
s.
So
,
it
is
ne
cess
ary
to
est
im
a
te
the
im
pact
of
fill
ing
of
eac
h
un
s
pecifie
d
bit
in
the
te
st
set
.
The
lo
gic
va
lu
es
are
assigne
d
to
t
he
un
s
pecifie
d
bit(s)
base
d
on
it
s
i
m
p
act
on
ca
pture
-
po
wer.
The
im
pact
of
fill
ing
of
on
e
un
s
pecifie
d bit
w
it
h
t
he
lo
gic
value v
(i.e.
0
or 1)
for
the
n
t
h
sca
n
cel
l
of
m
th test
v
ect
or
can be c
om
pu
t
ed
as
n
a
l
l
f
o
r
n
a
l
l
f
o
r
i
m
p
a
c
t
n
m
S
n
m
R
n
m
S
n
m
R
v
n
m
C
)
,
(
)
,
(
)
,
(
)
,
(
)
,
,
(
Wh
e
re a
nd
ar
e
log
ic
values
of
the test
sti
m
ulu
s and
res
pons
e
of sam
e scan
c
el
ls resp
ect
ivel
y
.
3.
SHIFT P
OW
ER R
E
D
UC
T
ION
T
HROU
GH
DON’T
C
AR
E B
IT
FIL
LING
In[2
]
the
idea
i
s
to
us
e
a
duri
ng
w
hich
no
n
-
rand
om
filling
is
us
e
d
to
assign
val
ues
to
don’t
care
bit
s
(X
s
)
of
eac
h
te
st
patte
rn
of
a d
et
erm
inist
ic
t
est
set
seq
uen
c
e
Ex
per
im
e
nts
perform
ed
on
I
SCAS
’89
a
nd
I
TC’9
9
ben
c
hm
ark
cir
cuits ha
ve bee
n done t
o
est
im
at
e the r
e
duct
ion i
n pea
k pow
er
ob
ta
ine
d d
uri
ng
TC.
In
[
3]
the
te
st
patte
rn
s
are
ex
tract
ed
f
r
om
the
AT
PG
co
ns
i
der
i
ng
a
po
st
-
ATPG
X
-
fill
ing
.
SR
-
a
war
e
X
-
fill
ing
is
c
arr
ie
d
out
on
ATPG
ext
ra
ct
ed
te
st
patte
rn
s
perf
or
m
ed
base
d
on
pr
e
-
com
pu
te
d
t
r
ansiti
on
pro
bab
il
it
y
(T
P)
.
Scan
chai
ns
sti
tc
hin
g
was
done
by
us
in
g
m
ini
m
u
m
WH
D
(
Weig
hted
Ham
m
ing
Dis
ta
nce)
searchi
ng.
Sca
n
cel
l
value
sim
il
arity
in
bot
h
sca
n
i
n
a
nd
s
can
out
di
recti
on
is
im
pr
oved
with
the
ai
d
of
patte
rn
si
m
ulati
on
and
the
TP
est
i
m
a
ti
on
.
T
he
reas
on
is
as
it
atte
m
pts
to
X
-
bits
assignm
ent
to
sa
m
e
bits
and
al
i
ke
respo
ns
es a
re
ga
there
d
f
or the
ref
e
rr
e
d
sca
n
st
it
ching
m
et
hod.
In
[4
]
s
ub
sta
nt
ia
l
red
uctio
n
i
n
sh
i
ft
powe
r
durin
g
sca
n
te
sti
ng
is
achie
ve
d
th
rou
gh
e
nsuri
ng
t
hat
the
captu
re
switc
hi
ng
li
m
it
is
wit
hin
the
pre
-
de
f
ined
th
reshold
.
Def
ect
co
ver
a
ge
is
al
so
in
cre
ased
as
com
par
ed
to
oth
e
r
X
fill
ing
heurist
ic
s
w
hile
keep
i
ng
te
st
patte
rn
c
ount
pe
rsiste
nt.
Hen
c
e
the
ap
proac
h
can
be
tre
at
ed
as
the
balance
d
way
t
o
deal
with
po
wer
e
ff
ic
ie
ncy
and
de
fect
co
ve
rag
e
sim
ultaneousl
y.
Fil
l
adjace
nt
(FA)
te
c
hn
i
qu
e
has
got
the
ne
gative
i
m
pact
on
the
peak
ca
pture
powe
r
li
m
it
a
ti
on
an
d
to
res
olv
e
this
issue
Pr
e
ferre
d
Fil
l
(P
F)
te
chn
iq
ue
f
or
s
pecifyi
ng
as
m
any
Xs
as
nece
ssary
in
orde
r
t
o
li
m
it
the
pea
k
ca
pture
unde
r
the
pow
er
bu
dg
et
.
This
is
done
i
n
a
ste
pwise
[
5,
6]
fas
hion
a
nd
co
n
cu
rr
e
ntly
with
the
ap
plica
ti
on
of
MFA/
MFA+P
te
c
hn
i
qu
e
i
n
order
to
m
ini
m
iz
e
the
nu
m
ber
of
Xs
s
pecifie
d
acco
rd
i
ng
to
PF.
As
P
in
cre
ases,
the
de
fect
cov
e
rag
e
of
th
e
te
st
vecto
rs
incre
as
es
bu
t
they
co
ns
um
e
m
or
e
sh
ift
powe
r.
In
order
t
o
evalu
at
e
the
scan
-
i
n
swi
tc
hing
act
ivit
y
of
ever
y
te
st
cub
e
T
gen
erate
d
usi
ng
each
of
th
ese
fill
ing
s,
the
y
us
ed
the
nor
m
al
iz
ed
weight
ed
switc
hing
act
ivit
y
(N
-
WSA)
[7
]
.
This
m
e
tric
cou
nts
the
num
ber
of
tra
ns
it
ions
in
su
ccessi
ve
scan
cel
ls,
ta
kin
g
al
so
into
a
ccount
their
relat
ive
po
sit
io
ns
,
an
d
norm
al
iz
es
this
value
by
di
vid
in
g
it
by
t
he
upper
bo
und
of
the
volum
e
of
switc
hing
flip
-
flo
ps
.
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.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
9
4
1
–
9
4
9
944
In
[
8]
the
pea
k
sh
ift
po
wer
m
ini
m
iz
at
ion
pro
blem
is
fo
rm
ula
te
d
as
a
m
od
ifie
d
form
of
inter
val
colo
ur
i
ng
pro
bl
e
m
al
so
known
as
B
ottl
en
eck
Col
ourin
g
Pr
oble
m
.
Here
it
is
exp
la
in
ed
ho
w
pea
k
toggle
m
ini
m
iz
at
ion
m
ay
be
viewed
as
a
case
of
Bott
le
neck
Colo
ur
i
ng
P
robl
e
m
.
A
te
st
vec
tor
orde
rin
g
al
gorithm
m
entioned
as
i
nterleave
d
te
st
vecto
r
order
i
ng
(
I
-
Orde
rin
g)
ha
ving
r
unni
ng
ti
m
e
O
(log
(n))
is
i
ncor
porated
her
e
.
It
is
ob
s
erv
e
d
that
the
app
r
oac
h
is
eff
ect
ive
in
re
du
ci
ng
peak
t
oggles
w
hen
DP
Fil
l
is
app
li
ed
in
com
bin
at
ion
w
it
h
I
-
orde
rin
g
and
pe
rcen
ta
ge
i
m
pr
ovem
ent
consi
ste
ntly
increases
with
in
crease
in
ci
rcui
t
siz
e.
In
Ta
ble
1
va
rio
us
as
pects
includi
ng
m
e
rits
and
dem
erit
s
of
s
hift
powe
r
re
duct
io
n
m
et
ho
dolo
gies
are
po
i
nted o
ut
.
Table
1.
X
-
Fil
li
ng
Tech
niques
for
Sh
i
ft P
ow
e
r
Re
duct
io
n
4.
X
-
FILL
IN
G
FOR
CAPT
U
RE PO
WER
RED
UC
TI
ON
In
[
9]
a
te
st
c
ube
is
processe
d
base
d
on
it
s
case
ty
pe
a
nd
four
num
ber
s
of
X
-
cases
ar
e
f
or
m
ulate
d
base
d
on
the
c
om
bin
at
ion
of
wh
et
her
PPI
a
nd
PP
O
are
wi
th
X
bits
or
wi
thout
X
bits.
I
n
res
pect
of
four
t
h
X
-
case
of
a
te
st
cub
e
,
dep
e
nding
on
c
orres
pondin
g
bit
-
pair
value
s
(i.e
.
‘
0’,’1
’
an
d
‘
X’)
of
PP
I
a
nd
PP
O
ty
pe
cl
assifi
cat
ion
(
ty
pe
A
-
D)
is
c
on
si
der
e
d.
As
for
exam
ple,
in
case
of
Ty
pe
-
B
or
Ty
pe
-
C
bit
-
pair
s
Ca
se
-
2
or
Ca
se
-
3
process
ing
is
ca
rr
ie
d
out
an
d
f
or
Ty
pe
-
D
bit
-
pai
rs
X
-
fill
ing
was
done
th
rou
gh
as
sign
m
ent
justi
ficat
io
n.
Hen
ce
,
de
pend
ing
upon
the
a
pp
li
cabil
it
y
of
an
X
-
cas
e
a
n
d
it
s
ty
pe,
justi
f
ic
at
ion
,
as
sig
nm
ent
an
d
assi
gnm
ent
j
ust
ific
at
ion
ar
e
app
li
ed
to
c
onduct
X
-
Fil
li
ng
in
ord
er
to
ge
t
fu
ll
y
sp
eci
fied
te
st
vector
wh
ic
h
is
ex
pec
te
d
to
yi
el
d
reduce
d
c
aptu
re
power.
Re
du
ct
io
n
of
s
witc
hing
act
ivi
ty
sp
eci
fical
ly
for
sin
gle
ca
pt
u
re
pulse
m
ay
resu
lt
in
ina
de
qu
acy
i
n
IR
-
dro
p
re
du
ct
io
n
fo
r
la
un
c
h
-
off
-
capt
ur
e
cl
oc
ki
ng
base
d
at
-
s
pe
ed
scan
te
sti
ng
w
hich
a
pp
li
e
s
two
ca
ptu
re
pu
lse
s
and
this
pro
ble
m
is
addresse
d
in
[10]
.
T
he
m
et
ho
d
i
n
[
10]
ge
ner
at
es
‘C
ool’
te
st
vecto
rs
capa
ble
of
re
du
ci
ng
captu
re
s
witc
hi
ng
act
ivit
y
irr
especti
ve
of
a
ny
fa
ult
co
verage,
ti
m
ing
,
and
a
rea
over
he
ad
im
pact.
Her
e
al
s
o
ei
gh
t
num
ber
s
of
X
-
ty
pe
s
are
def
i
ned
base
d
on
t
he
patte
r
n
i
n
w
hich
X
ty
pe
s
occ
ur
.
Ba
la
nced
re
du
ct
io
n
of
first
and
seco
nd
ca
pture
tra
ns
it
io
ns
a
re
ac
hi
ev
ed
by
dy
nam
i
cal
ly
op
ti
ng
a
ta
rg
et
ca
ptur
e.
Th
e
ta
r
get
captu
re
sel
ect
ion
is
m
ade
ba
sed
on
the
he
ur
ist
ic
total
est
i
m
at
ed
captu
re
tran
sit
ion
act
ivit
y
(TECTA
).
TE
CTA
i
s
der
i
ved
from
e
xisti
ng
ca
pture
transiti
ons
(E
CTs)
an
d
po
te
ntial
captur
e
tr
ansiti
on
s
(P
CT
s)
.
Ca
pt
ur
e
s
w
it
chin
g
act
ivit
y
is
eval
uated
th
rou
gh
the
weig
hted
s
um
of
captur
e
transiti
on
s
at
al
l
FFs
wh
ic
h
is
def
ine
d
he
re
as
a
m
et
rics
nam
ed
weig
hte
d
s
witc
hing
act
ivit
y
at
FF
(F
F
-
WSA).
La
rg
e
r
t
he
val
ue
of
justi
f
ic
at
ion
easi
nes
s
JE(
s)
resu
lt
s i
n
easi
e
r
the
justi
ficat
ion o
f
a
lo
gic val
ue
on
s.
The
OJ
Fil
l
ap
proac
h
[
11]
focuses
on
ad
diti
on
al
fa
ult
detect
ion
an
d
co
m
pact
ion
aspe
ct
s
instea
d
of
lookin
g
f
or
onl
y
W
S
A
re
duct
ion
a
nd
pro
du
c
es
‘sile
nt’
te
st
cub
e
s
le
avin
g
un
s
pecifie
d
bit
s
sp
are
d
f
or
de
te
ct
ing
add
it
io
nal
fa
ults.
SA
T
-
based
j
ust
ifie
d
ta
r
ge
t
extracti
on
pr
ocess
is
ca
rr
ie
d
out
to
sel
ect
i
vely
sp
eci
fy
X
-
val
ues
for prese
r
ving
the li
kelih
ood
t
o detec
t ad
diti
on
al
fa
ults by
ATPG.
Mi
ni
m
iz
ation
of
the
total
we
igh
te
d
s
witc
hi
ng
act
ivit
y
cropp
e
d
up
in
the
CUT
is
do
ne
th
r
ough
the
form
ulati
on
of
ILP
base
d
z
ero
-
one
li
nea
r
pro
blem
(ZO
LP)
in
[1
2].
ZOLP
e
quat
io
ns
are
f
or
m
ulate
d
to
m
ini
m
iz
e
te
st
m
od
e
switc
hing
act
ivit
y
by
fi
ll
ing
X
bits
of
par
ti
al
ly
sp
eci
f
ie
d
te
st
patte
rns.
T
ransf
or
m
ation
of
the
Boo
le
a
n
f
unct
ion
al
it
y
of
CUT
int
o
L
P
equ
at
io
ns
with
an
opti
m
iz
at
io
n
f
un
ct
io
n
was
done
in
the
fir
st
ph
ase
to ach
ie
ve
t
he a
bove. A
fter
w
ard
s
, speci
fied
bits val
ues
a
re
assigne
d
t
o
in
put va
riables i
n ZOLP
m
od
el
SA
T
–
Fil
l
[
13
]
assum
es
that
t
he
tran
sit
ion
pro
ba
bili
ty
of
a
flip
-
flo
p
cause
tho
se
of
ot
her
flip
-
fl
op
is
diff
e
re
nt
as
f
or
al
l
flip
-
fl
ops
a
nd
su
c
h
transiti
on
pro
bab
il
it
y
is
est
i
m
at
ed
fo
r
a
n
ind
i
vidual
flip
-
flo
p.
SA
T
s
olv
e
r
is
us
e
d
f
or
sea
rchi
ng
s
uch
value
assignm
ents
not
causi
ng
tra
nsi
ti
on
s.
Assum
ing
un
e
qual
effe
ct
s
of
powe
r
dissi
pat
ion
f
or
transiti
on
s
for
eac
h
fl
ip
-
fl
op,
a
stron
g
ef
fect
on
po
wer
dissipati
on
at
la
un
c
h
-
on
captur
e
Fillin
g
T
echn
iq
u
e
Key
Heu
ristics
Test
Cu
b
e
Si
m
u
latio
n
Ben
ch
m
a
rk
Clai
m
ed
Mer
its
Sh
o
rtco
m
in
g
s
No
n
r
an
d
o
m
Heu
ristic bas
ed
X
-
Fillin
g
[
2
]
MT
-
Fill, 0
-
Fill, 1
-
Fill
Tetr
a
MAX
ISCAS’89
and
IT
C’9
9
Up
to 8
9
% p
eak p
o
wer
redu
ctio
n
,
No
DFT
co
st
Ai
m
s to
m
in
i
m
i
ze
o
n
ly
p
eak p
o
wer
SR awar
e
X
-
Fillin
g
and
scan
stitch
in
g
[
3
]
Scan
chain
r
eo
rder
in
g
,
W
HD
bas
ed
scan
stitch
in
g
,
TP
esti
m
atio
n
Tetr
a
MAX
ISCAS’89
and
IT
C’9
9
Sig
n
if
ican
t sh
i
f
t power
redu
ctio
n
(
u
p
to 6
4
%)
irr
esp
ectiv
e of
sca
n
chain
size.
Increase in
co
m
p
u
tatio
n
ti
m
e and
test
patter
n
cou
n
t
Def
ect Awa
re
X
-
Fillin
g
[
4
]
MFA,
MFA+
P whe
re
‘P
’
is
u
ser def
in
ed
para
m
eter,
N
-
W
SA
A
TPG
ISCAS’89
and
IWLS’05
Hig
h
er
p
o
ten
tial to d
etect un
m
o
d
eled
def
ects,
peak
captu
re
p
o
wer
co
n
strain
t
W
ith
the in
cr
ease i
n
def
ect
co
v
erage,
sh
if
t power
also
in
crea
ses
.
DP
-
Fill[
8
]
Bo
ttlen
eck co
lo
ring
p
rob
le
m
,
I
-
o
rderin
g
Tetr
a
MAX
IT
C’9
9
Sig
n
if
ican
t dro
p
d
o
wn
in
p
eak
to
g
g
le by
ap
p
ly
in
g
I
-
o
rdering
and
DP
-
Fill jo
in
tly
.
Peak
inp
u
t tog
g
le
in
crea
ses
with n
u
m
b
er
o
f
iteration
s
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 Co
mp
r
ehe
nsi
ve Revi
ew
o
f
Ap
plicati
ons
of Don’t Ca
re Bi
t Fi
ll
ing
Tec
hniqu
es
for Test
…
(
Sanjo
y
Mi
tra
)
945
cy
cl
e
m
akes
se
le
ct
ion
orde
r
of
the
flip
-
flo
p
high
an
d
this
orde
r
is
set
tl
ing
on
by
app
ly
in
g
the
correla
ti
on
a
m
id
the tra
ns
it
ion o
f
a
flip
-
flo
p
a
nd
powe
r di
ssip
at
ion
at la
un
c
h
-
on ca
pture cy
cl
e.
In
the
sta
te
se
nsi
ti
ve
X
–
Fil
li
ng
ap
proac
h
[14
]
,
a
pr
oba
bili
ty
-
base
d
WSA
m
od
el
fo
r
ca
ptu
re
op
e
rati
on
was
f
or
m
ulate
d
an
d
aim
ed
to r
edu
ce tra
ns
it
ion
s
of
both
sca
nn
e
d
flip
-
fl
op
s
an
d
inter
nal gat
es. X
-
bits in PI
s
are
sta
t
e
dep
e
nd
e
nt
and
owin
g
to
this
fill
ing
in
di
viduall
y
would
com
pr
om
ise
t
he
fill
ing
ef
fec
ti
ven
ess
a
nd
he
nce
a
sta
te
-
base
d
m
e
thod
was
f
or
m
ulate
d
in
[
14
]
wh
ic
h
warra
nt
that
the
total
c
aptu
re
switc
hi
ng
act
ivit
y
(TCSA
)
of
any test
cube is
to be
kep
t a
s
m
ini
m
u
m
as
po
ssible.
5.
SHIFT
AND CAPT
U
RE P
OWER
R
E
D
UC
TI
ON
The
he
ur
ist
ic
m
ini
m
u
m
set
of
pr
im
ary
inputs
(MS
PI)
is
a
ppli
ed
AB
-
Fil
l
[
15
]
a
ppr
oac
h
t
o
c
on
t
ro
l
t
he
nu
m
ber
of
tran
sit
ion
al
FFs
in
or
de
r
to
lo
we
r
dow
n
capt
ure
power.
The
f
il
li
ng
process
is
cl
assifi
ed
into
fou
r
diff
e
re
nt
ty
pes
(A
-
E)
of
fill
in
g.
A
back
t
racki
ng
ta
ble
com
pu
te
s
cl
arit
y
of
correla
ti
on
am
ong
sca
n
-
in
c
ubes
a
nd
scan
-
out
bits
and
s
pecifies
th
e
value
of
eac
h
sh
i
ft
-
in
c
ub
e
to
assign
t
he
appr
opriat
e
li
ne.
The
s
pecifie
d
val
ue
can
e
ff
ect
uall
y
con
t
ro
l
t
he
captu
re
s
witc
hi
ng
act
ivit
y
f
or
eac
h
flip
-
fl
op.
Mi
nim
iz
at
i
on
of
the
Ha
m
m
ing
distance am
id PP
I
a
nd P
PO i
n
eac
h
te
st
pair
is the
pr
im
e obj
ect
iv
e
beh
i
nd this
X
-
fill
ing
[
15
]
a
ppr
oach.
BA
fill
[16]
was
pu
t
f
orw
ard
e
d
t
o
re
du
ce
pea
k
ca
pture
s
witc
hing
act
ivit
y
by
m
anag
i
ng
the
relat
ion
s
hip
be
tween
t
he
ca
ptu
re
a
nd
sti
m
ul
us
values
of
the
c
om
bin
at
ori
al
log
ic
with
out
nee
ding
a
ny
extra
DF
T.
Weig
hted
tra
ns
it
io
n
c
ount
(
WTC)
m
e
tric
was
us
e
d
t
o
c
om
pu
te
sh
i
f
t
switc
hing
act
ivit
y
fo
r
sca
n
-
i
n
a
nd
scan
-
out
data.
Bi
twise
XO
R
op
e
rati
on
w
a
s
app
li
ed
to
i
nput
te
st
stim
uli
a
nd
res
pons
e
ve
ct
or
of
eac
h
sc
an
cel
l
and th
us
t
he
e
xt
ent of ca
pture
act
ivit
y was es
tim
a
te
d.
The
im
pact
of
an
X
-
bit
on
a
CUT’s
s
hift
-
a
nd
ca
ptu
re
-
power
(m
entione
d
as
S
-
im
pact
and
C
-
im
pact)
is
m
od
el
le
d
in
[17]
and
a
pp
li
ed
th
ese
to
ste
er
the
X
-
fill
ing
(m
entioned
as
S
-
fill
and
C
-
fill
).
In
it
ia
ll
y,
X
-
fill
ing
was
done
f
or
s
hift
-
powe
r
re
duct
ion
(te
rm
ed
S
-
fill
ing
)
s
ubje
ct
to
the
co
nst
raint
that
ca
pt
ur
e
po
wer
la
i
d
within
the
CUT
’s
pea
k
powe
r
li
m
it
.
U
pon
vio
la
ti
on
of
the
c
onstr
ai
nt
X
-
fill
ing
aim
ing
to
ca
pt
ur
e
po
wer
re
duct
ion
(n
am
ed
C
-
fill
ing)
was
pe
rform
ed.
A
fter
c
ap
ture
powe
r
reducti
on,
rem
ai
ni
ng
X
-
bits
are
f
il
le
d
up
wit
h
S
-
fill
in
g
and
the
f
ur
t
he
r
vio
la
ti
on
of
captu
re
powe
r
if
a
ny
was
c
heck
e
d
once
a
gain
an
d
s
uc
h
ste
ps
it
erate
a
m
on
g
th
em
sel
ves.
Sh
ift
-
I
n
Transi
ti
on
Prob
a
bili
ty
(S
ITP)
an
d
Sh
if
t
-
O
ut
Tran
sit
io
n
Prob
a
bili
ty
(S
OTP)
are
est
im
at
ed
for
fill
ing
a
ny
X
-
bit
an
d
su
m
of
S
ITP
a
nd
S
OTP
is
te
rm
ed
as
Sh
ift
T
rans
it
ion
Pro
bab
il
it
y
(S
TP).
T
he
va
lue
of
STP
agai
ns
t
‘0’
an
d
‘1’
deter
m
ines
the
f
il
lin
g
bin
a
ry
valu
e
of
the
ne
xt
X
bit.
Like
wise,
Ca
ptu
re
T
ra
ns
it
io
n
Pr
oba
bili
ty
(
CTP)
play
s a
de
ci
sive role i
n X
-
fill
ing
.
It
was
at
te
m
pt
ed
in
[
18]
to
cl
us
te
r
t
he
sca
n
f
li
p
-
flo
ps
with
com
m
on
su
cce
sso
r
s
into
sin
gle
scan
c
hai
n
so
that
the
s
pe
ci
fied
bits
p
e
r
patte
r
n
was
distrib
uted
over
m
ini
m
al
sc
an
chai
ns
an
d
on
th
e
basis
of
thi
s
arr
a
ng
em
ent
it
becam
e
po
ssible
that
al
l
the
bits
to
so
m
e
scan
chain
s
in
a
vecto
r
f
ound
as
don’t
care(
X).
Segm
ent
-
base
d
X
-
fill
ing
was
pro
po
se
d
f
or
s
uch
sca
n
chai
ns
in
order
to
r
edu
ce
te
st
po
wer
an
d
al
so
t
o
kee
p
def
ect
c
overa
ge
consist
ent
.
CSP
Fil
li
ng
[
19
]
be
gin
s
wi
th
the
a
ppli
cat
ion
of
the
MTC
-
fill
ing
t
echn
i
qu
e
to
f
il
l
the
entire
un
s
pecifie
d
bit
s in
the test
cu
be.
Capt
ur
e
-
tra
ns
it
ion
s
of
eac
h
patte
r
n
are c
om
pu
te
d
after
fill
ing
all
X
-
bit
s w
it
h
a
log
ic
val
ue
of
‘0’
or
‘
1’.
T
he
captu
re
po
we
r
thre
shold
is
check
e
d
a
gain
st
each
patte
r
n
and
i
f
any
vio
la
ti
ng
patte
rn
is
f
ound
the
n
t
hat
in
div
id
ual
patte
r
n
i
s
sub
j
ect
ed
t
o
f
ur
t
her
ca
pture
powe
r
re
duct
io
n
in
orde
r
t
o
br
ing
it
within
the
thr
esh
old
.
Ps
eu
do
pri
m
ary
inp
uts
of
the
ca
pt
ur
e
vio
la
te
d
t
est
cub
e
a
re
s
ubj
ect
ed
to
m
i
nim
u
m
captu
re
tra
ns
it
ion
schem
e
and
the
MTC
-
fill
ing.
T
his
pract
ic
e
seq
uen
ce
is
rep
eat
e
d
f
or
e
ver
y
vio
la
te
d
patte
r
n
with
a
vie
w
to
m
ake
captu
re
transiti
ons
ag
ai
ns
t
in
div
id
ua
l
patte
rn
w
it
hin
th
e
ca
pture
t
hr
es
hold
lim
it
.
D
on’t
care
bits
in
the
te
st
patte
rn
ar
e
fill
ed
wit
h
th
e
value
of
the
m
os
t
near
sp
ec
ifie
d
bit
on
the
le
ft
side
for
re
du
ci
ng
transiti
on
bet
w
een
co
ns
ec
utiv
e
bits
res
ulti
ng
in
dro
p
dow
n
of
s
hift
power.
The
s
hift
power
al
s
o
de
pe
nd
s
on
the posit
io
n of
transiti
on a
nd
weig
hted
t
ran
si
ti
on
m
et
ric (
WTM)
us
ed
to
c
om
pu
te
sh
ift
-
in
transiti
ons.
A
fa
n
-
ou
t
awa
r
e
m
od
ifie
d
ad
ja
cent
X
-
fill
ing
te
chn
iq
ue
is
pr
opos
e
d
in
[20]
to
reduce
sh
i
ft
and
ca
pture
powe
r.
Be
side
s
the
aver
ag
e
and
peak
te
st
po
wer
re
duct
ion,
tim
e
co
m
plexity
and
nu
m
ber
of
it
erati
ons
ar
e
al
so
reduce
d
in
this
m
et
ho
d.
The
m
et
ho
d
pr
im
aril
y
e
m
ph
asi
zes
on
reduci
ng
c
aptu
re
po
wer
hen
ce
res
ulti
ng
in
th
e
sli
gh
tl
y
hig
he
r
con
s
um
ption
of
s
hift
powe
r
as
co
m
par
ed
to
oth
e
r
ad
j
ace
nt
fill
ing
te
chn
iqu
es
.
The
ord
ers
of
fill
ing
X
-
bits
wer
e
determ
ined
by
fa
n
-
outs
of
the
l
og
ic
cel
ls.
It
was
ass
um
ed
that
scan
cel
l
with
X
-
bit
hav
i
ng
la
rg
er
fa
n
-
out
corres
pondin
g
to
gi
ven
te
st
ve
ct
or
bears
higher
im
pact
on
the
pea
k
powe
r
a
nd
t
hese
cel
ls
are
identifie
d ac
co
r
di
ng
ly
.
In
the
[
21]
com
bin
at
ion
of
0
fill
ing
and
1
f
il
li
ng
are
app
li
ed
in
the
te
st
set
to
red
uce
te
st
po
we
r.
T
he
peak
switc
hi
ng acti
vity
ag
ai
nst
each
te
st patt
ern
is
recor
der
for
0
-
fill
and
1
-
fill
. Th
e
sort o
f
fill
ing
w
hich y
ie
lds
le
sser
pea
k
sw
it
ching
act
ivit
y
was
picked.
The
opti
m
a
l
f
eat
ur
es
of
0
-
fi
ll
and
1
-
fill
are
m
ixed
fo
r
droppi
ng
dow
n
a
ver
a
ge shift s
witc
hing
and a
ver
a
ge
c
aptu
re swit
chin
g.
In Table
2
a
nd
3 vari
ou
s
as
pe
ct
s inclu
ding
m
erit
s an
d de
m
erit
s o
f
s
hift
and capt
ur
e
po
wer re
du
ct
io
n
m
et
ho
dolo
gies
are
po
i
nted o
ut.
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.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
9
4
1
–
9
4
9
946
Table
2
.
X
-
Fil
li
ng
Tech
niques
for
Ca
pture P
ower
Red
uction
Table
3.
X
-
Fil
li
ng
Tech
niques
for
Sh
i
ft and
Ca
ptu
re
P
ow
e
r
Red
uction
Metho
d
Key
Heu
ristics
Ben
ch
m
a
rk
Clai
m
ed
Mer
its
Sh
o
rtco
m
in
g
s
LCP
-
X
Fillin
g
[
9
]
Ju
stif
icatio
n
,
Ass
ig
n
m
en
t,
Ass
ig
n
m
en
t
-
Ju
stif
icatio
n
ISCAS’89
In co
m
p
ariso
n
wit
h
r
an
d
o
m
X
-
f
illin
g
,
th
e average an
d
m
ax
i
m
u
m
n
u
m
b
er
of
no
d
e tr
an
sitio
n
s are
redu
ced b
y
49
.3%
an
d
13
.3%
resp
ectiv
ely
It
is co
n
cerned
wi
t
h
on
ly
captu
re
p
o
wer
redu
ctio
n
Do
u
b
le
Cap
tu
re(
DC)
-
X
Fillin
g
[
1
0
]
FF
-
W
SA
Metr
ic,
TE
CT
A,
Ju
stif
icatio
n
easin
e
ss
(JE)
ISCAS’89
Sig
n
if
ican
t
IR
-
d
rop
r
ed
u
ctio
n
with
o
u
t any
circuit
m
o
d
if
icatio
n
an
d
ov
erhead
,
f
au
l
t cov
erage los
s
Increased
pro
cess
i
n
g
ti
m
e du
e to
m
u
lti
pas
s p
roced
u
re,
Ju
stif
icatio
n
pro
ce
ss
con
su
m
es
m
u
ch
r
u
n
n
in
g
ti
m
e
Op
t
-
Ju
stif
icatio
n
(OJ)
-
Fill[
1
1
]
Ju
stif
icatio
n
tar
g
et,
Un
satis
f
y
-
ab
le
ju
stif
icatio
n
tar
g
et,
p
o
st ATPG
stag
e
ISCAS’89
Ab
ility
to
detect
a
d
d
itio
n
al f
au
lts,
silen
t test cub
es ar
e pro
d
u
ced b
y
switch
in
g
activity
redu
ctio
n
th
rou
g
h
di
v
ersif
ied
assig
n
m
en
t of
X
-
b
its, ov
er
p
atte
r
n
cou
n
t r
esu
ltin
g
in
test cos
t r
ed
u
cti
o
n
P
eak s
witch
in
g
act
iv
ity
redu
ctio
n
is ins
ig
n
i
f
ican
t in
seco
n
d
captu
re
cy
c
le,
n
o
any
ass
u
red p
eak redu
ctio
n
IL
P bas
ed
do
n
’t
care
f
illin
g
[
1
2
]
Zer
o
-
o
n
e linear
p
rob
le
m
(Z
O
LP)
eq
u
atio
n
s
ISCAS’89
Ab
o
v
e 64
% an
d
4
0
% peak
W
S
A
redu
ctio
n
in LOC
sch
e
m
e and
L
OS
sch
e
m
e
r
esp
ectiv
el
y
Red
u
ctio
n
in
capt
u
re
p
o
wer
is
co
m
p
a
red with
ran
d
o
m
f
illin
g
o
n
ly
SAT
-
Fi
ll[
1
3
]
SAT
so
lv
ers aided
id
en
tif
icatio
n
o
f
ass
ig
n
m
en
ts n
o
t
trigg
ering
tr
an
sitio
n
s
ISCAS’89
an
d
I
TC’99
Valu
e
ass
ig
n
m
en
ts
is
g
u
id
ed
b
y
th
e
p
o
ss
ib
ility
o
f
th
e
trans
itio
n
s
o
f
o
th
er
f
lip
-
f
lo
p
s,
Prob
ab
ility
o
f
trans
itio
n
o
f
in
d
iv
id
u
al
f
li
p
-
f
lo
p
affectin
g
th
o
se
o
f
o
th
er
f
li
p
-
f
lo
p
s
at
LOC
c
y
cle
was tak
en
into
acc
o
u
n
t.
Increased
co
m
p
u
ta
tio
n
al ti
m
e
an
d
this
app
roach
is li
m
ited
to
LOC
m
o
d
e
on
ly
State
-
Sen
sitiv
e
X
-
Fillin
g
[
1
4
]
Cap
tu
re
s
witch
in
g
activ
ity
(CS
A),
T
o
t
al
Cap
tu
re
switch
in
g
activ
ity
(
TCSA)
,
p
rob
ab
ility
b
ased
W
SA
m
o
d
el
ISCAS’89
X
-
b
it assig
n
m
en
t i
s d
o
n
e thro
u
g
h
state
-
b
ased
r
ig
id
pr
o
b
ab
ilities.
Co
m
p
u
tin
g
ti
m
e
ef
f
icien
t app
roach
.
Sig
n
if
ican
t average an
d
peak
capt
u
re
p
o
wer red
u
ctio
n
.
State chan
g
es h
as
g
o
t the
p
rob
ab
ility
o
f
gettin
g
inf
l
u
en
ced
b
y
no
n
deter
m
in
is
m
X
-
Stat
-
Fill[
2
5
]
Test vecto
r
reorder
in
g
f
o
r
scan
bas
ed
architectu
res
realize
d
w
ith
tog
g
le
-
m
a
sk
in
g
f
lip
-
f
lo
p
s
IT
C
’99
7
.4%
r
ed
u
ctio
n
of
av
erage peak
capt
u
re
switch
in
g
Increa
sed
co
m
p
u
ta
tio
n
al cos
t
Gree
d
y
app
roach
m
a
k
es it
su
b
o
p
tim
al f
o
r
p
ea
k
tog
g
le
redu
ctio
n
Metho
d
Key
Heu
ristics
Test Cu
b
e
Ben
ch
m
a
rk
Mer
its
Sh
o
rtco
m
in
g
s
Ad
jacent
Back
trackin
g
Fill[
1
5
]
MPSI
,
E
T
y
p
e
Fil
lin
g
PI
Relax
atio
n
Tetr
a
M
A
X
ISCAS’89
Red
u
ced cap
tu
re
switch
in
g
activ
ity
and
sh
if
t
–
in
tr
an
sitio
n
s
Increased
r
u
n
ti
m
e
Bo
u
n
d
e
d
Ad
jacent
Fill[
1
6
]
W
TC
m
et
ric,
bitwi
se
XOR
Tetr
a
M
A
X
ISCAS and
Ind
u
strial cir
cu
its
Peak
captu
re
p
o
we
r
redu
ced u
p
to
98
.86
% an
d
95
.
2
6
%
redu
ctio
n
in S
I
an
d
SO
switch
in
g
r
esp
ectiv
ely
,
n
o
p
attern in
f
latio
n
Pattern r
ejectio
n
i
-
Fill[
1
7
]
C
-
i
m
p
act,
S
-
i
m
p
ac
t
SIT
P,
SOT
P,ST
P,
CTP
MI
N
TE
S
T
ISCAS’89
and
IT
C’9
9
Red
u
ctio
n
in th
erm
a
l st
ress,
in
crea
sed
test par
al
lelis
m
and
sh
if
t f
requ
en
cy
Increased
r
u
n
ti
m
e,
Hig
h
peak
sh
if
t
p
o
wer
Seg
m
en
t
b
ased
X
-
Fillin
g
[
1
8
]
Seg
m
en
t f
illin
g
ATL
A
NTA
ISCAS’89
and
IT
C’9
9
p
ersisten
t,
d
ef
ect c
o
v
erage
sig
n
if
ican
t r
ed
u
cti
o
n
in cap
tu
re
p
o
wer
an
d
test ti
m
e
Inv
o
lv
e
m
en
t of
DFT
co
st
CSP Fill[
1
9
]
MTC
-
f
illin
g
techn
iq
u
e,
Cap
tu
re
p
o
wer
th
r
esh
o
ld
,
Mini
m
u
m
capt
u
re
trans
itio
n
sch
e
m
e,
W
T
M
--
-
--
ISCAS’89
Av
erage sh
if
t a
n
d
captu
re
p
o
wer
is redu
ced b
y
4
0
%
an
d
11
%
resp
ectiv
ely
wh
en
co
m
p
a
red to
0
-
Fillin
g
The % of
r
ed
u
ctio
n
h
ig
h
lig
h
ted
is
m
e
asu
red again
st
b
as
ic 0
-
f
illin
g
Fan
-
o
u
t
awar
e
m
o
d
if
ied
ad
jacent
X
-
f
illin
g
[
2
0
]
Scan
cel
l con
tain
in
g
X
-
b
it with
m
ax
i
m
u
m
FAN
-
OUT
ATL
A
NTA
ISCAS’89
Up
to 4
0
% an
d
20% r
ed
u
ctio
n
in
capt
u
re
&
sh
if
t power
when
co
m
p
a
red to
adjac
en
cy
and
L
CP
f
illin
g
r
esp
ectiv
ely
,
Red
u
ced
si
m
u
l
atio
n
ti
m
e
.
Increased
sh
if
t
p
o
wer
d
iss
ip
atio
n
Co
m
b
in
ed
0
an
d
1
-
Fill[
2
1
]
Peak
switch
in
g
activ
ity
Tetr
a
M
A
X
ISCAS‘89
Up
to
73
.7%
&
69
.
8
3
%
redu
ctio
n
in
av
erage sh
if
t and
capt
u
re
trans
itio
n
s resp
ectiv
ely
w.r
.t
.
r
an
d
o
m
f
ill a
p
p
roach
Po
wer
r
ed
u
ctio
n
largely
d
ep
en
d
s o
n
X
-
b
its co
u
n
t and
in
ternal d
esig
n
of
th
e cir
cu
it.
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 Co
mp
r
ehe
nsi
ve Revi
ew
o
f
Ap
plicati
ons
of Don’t Ca
re Bi
t Fi
ll
ing
Tec
hniqu
es
for Test
…
(
Sanjo
y
Mi
tra
)
947
6.
COMP
AR
I
S
ON OF TE
ST
POWE
R RE
DUCTIO
N O
N
BE
N
CHM
ARK
CIRC
UI
TS
In
t
his
sect
io
n
we
c
om
par
e
th
e
perce
nta
ge
powe
r
reducti
on
am
on
g
var
i
ous
X
-
Fil
li
ng
m
et
hods
.
He
re
we
com
par
e
th
e
per
ce
ntage
of
aver
a
ge
s
hift
power
,
as
sho
wn
in
Fi
gures
1
-
8
,
peak
pow
er
an
d
captu
re
powe
r
reducti
on
on
s
el
ect
ed
be
nchm
ark
ci
rc
uits.Var
i
our
X
-
fill
ing
m
et
ho
ds
ar
e
m
entione
d
in
com
par
ison
w
it
h
thei
r
abbre
via
ti
on
s
.
FA
,
MF
A,
MF
A+10
an
d
MF
A+20
[
4]
re
pr
e
sents
fill
adj
ac
ent,
m
od
ifie
d
f
il
l
adj
acent,
m
od
i
fied
fill
adjace
nt
te
chn
iq
ue
with
an
oth
er
10%
of
X
bits
ar
e
sp
eci
fie
d
to
r
edu
ce
capt
ur
e
powe
r
a
nd
m
od
i
fied
fill
adjace
nt
te
chn
iq
ue
with
ano
t
her
20%
of
X
bits
are
s
pecified
t
o
re
duce
capt
ur
e
power
resp
ect
i
vely
.
SB
and
SB
-
MC
fill
[18]
re
pr
ese
nt
segem
ent
bas
ed
fill
an
d
se
gm
ent
based
f
il
l
with
m
ulti
ple
chain.
B
-
fil
l
represents
ba
la
nced
fill
[2
2].
MTR
Fil
l
[2
3]
repres
ents
m
ini
m
u
m
transiti
on
ra
ndom
X
-
fill
ing
.
i
-
fill
[1
7]
is
the
i
m
pact
-
or
ie
nted
X
fill
ing
m
et
ho
d
for
sh
i
ft
an
d
captu
re
powe
r
r
edu
ct
io
n.
B
A
Fil
l
sign
ifie
s
the
bo
unde
d
ad
j
acent
fill
[1
6]
and
4m
fill
ing
schem
e
is
descr
i
bed
in
[
24
]
.
X
fill
ing
f
or
ca
pture
an
d
s
hift
powe
r
re
du
ct
ion
re
pr
e
sen
te
d
as
CSP f
il
l [
19]
.
Figure
1. Perce
ntage o
f
P
ower
Red
uction i
n
s
13207
Figure
2. Perce
ntage o
f Av
e
ra
ge
P
ower
Red
uc
ti
on
in
s1
58
50
Figure
3. Perce
ntage o
f Av
e
ra
ge
P
ower
Red
uc
ti
on
in
s3
84
17
Figure
4. %
of
Av
e
P
ow
e
r
Re
du
ct
io
n
i
n
s
38
584
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.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
9
4
1
–
9
4
9
948
Figure
5.
%
of
Peak Po
we
r
Re
du
ct
io
n
i
n
s
13
207
Figure
6
.
%
of
Peak Po
we
r
Re
du
ct
io
n
i
n
s
15
850
Figure
7
.
Be
nc
hm
ark
w
ise
%
of Ave.
Captu
r
e Powe
r
Re
du
ct
io
n
Figure
8
.
Be
nc
hm
ark
w
ise
%
of Ave.
Captu
r
e Powe
r
Re
du
ct
io
n
7.
CONCL
US
I
O
N
Lo
w
po
wer
V
LSI
desig
ns
[
26
,
27]
an
d
lo
w
powe
r
te
sti
ng
strat
egies
are
ga
ining
sig
nific
ant
at
te
ntion
of
this
ub
i
qu
it
ou
sly
gro
wing
sil
ic
on
industr
y.
W
e
ha
ve
an
al
yz
ed
sta
te
of
the
art
captu
r
e
power,
sh
ift
powe
r
and
sh
i
ft
an
d
captu
re
po
wer
reducti
on
m
eth
od
ologies
ba
s
ed
on
fill
ing
of
uns
pe
ci
fied
bits
in
the
te
st
cu
be.
This
analy
ti
cal
rev
ie
w
w
ork
fo
c
us
es
on
m
e
rits
and
sho
rt
com
ing
s
of
va
rio
us
don’t
care
fill
ing
m
et
h
ods
app
li
ed
to
re
du
ce
te
st
po
wer
i.
e.
captu
re
pow
er,
sh
ift
po
wer
etc.
Finall
y
we
pu
t
up
a
com
par
iso
n
of
te
st
powe
r
reducti
on
po
te
ntial
of
resp
ect
ive
fill
ing
m
e
t
hodolo
gies
with
the
par
am
et
er
of
te
st
power
red
ucti
on
pe
rc
entag
e
on
sel
ect
e
d
be
nch
m
ark
ci
rc
uits.
This
com
par
iso
n
with
res
pe
ct
to
te
st
po
w
er
re
du
ct
io
n
pe
rcen
ta
ge
on
sel
ect
ed
ben
c
hm
ark
s
pro
vid
es
us
t
he
us
ef
ul
c
omparati
ve
i
nfo
r
m
at
ion
reg
a
r
di
ng
the
pe
rform
ance
of
te
st
powe
r
reducin
g
m
et
ho
dolo
gies
a
nd
m
a
y
serv
e
a
s
a
f
ru
it
f
ul
in
put
to
f
ur
the
r
r
esearch
an
d
de
velo
pm
ent
wo
r
ks
i
n
this fiel
d.
REFERE
NCE
S
[1]
Ranga
na
tha
n
Sa
nkar
alingam,
Ra
m
a
Rao
Orugan
ti
and
Nur
A
.
T
ouba,
“Stat
ic
co
mpacti
on
t
ec
hni
ques
to
con
trol
scan
vect
or pow
er
dissipati
on
,
”
Proc.
18th
IE
EE
VLSI T
est
S
y
m
p
osium
,
2000,
pp
.
35
-
40
.
[2]
Nabil
Bade
r
ed
dine
,
Patr
ic
k
Gira
rd,
Serge
Pravoss
oudovit
ch,
Christi
an
L
andr
ault,
Arna
ud
Vira
ze
l
an
d
W
under
li
ch
,
H
an
s
-
Joac
him,
“
Mini
mizing
Pe
a
k
Po
wer
Consum
pti
on
during
Scan
T
esti
ng:
Test
Pat
t
ern
Modif
i
cat
io
n
wit
h
X Fi
l
li
ng
H
euristi
cs,
”
Proc.
IEE
E
DTIS’06:
Des &
Te
st of
In
te
gra
te
d
S
y
s
in N
ano
sca
le T
e
ch
,
Sep
2006,
Tunis
(Tuni
sia)
,
pp
.
35
9
-
364
.
[3]
Sung
y
oul
Seo
,
Yong
Le
e
,
H
y
eo
ncha
n
Li
m
,
Jooh
wan
Lee
,
Hongb
om
Yoo,
Yojoung
Kim
and
Sungho
Kang,
“
Sca
n
Chain
Re
ord
ering
-
aware
X
-
F
il
l
ing
and
St
it
ch
in
g
for
Scan
Sh
if
t
Powe
r
R
educ
t
io
n,
”.
Proc.
24th
I
EE
E
A
TS,
2015
,
pp.
1
-
6
.
[4]
S.Ba
latsouk
a,
V
Te
nentes.,
X.K
avousia
nos
and
K.Cha
kra
bar
t
y
,
“D
ef
ec
t
Aware
X
-
Fi
lling
for
L
ow
-
Powe
r
Scan
Testing,
“
Proc
Design,
Autom
a
ti
on
&
T
est in
E
urope
Conf
er
ence
&
Exhi
b
it
ion
(
DA
TE
2010)
,
pp
873
–
878.
[5]
Remersaro,
Xijiang
Li
n,
Sudha
kar
M
Redd
y
,
I
rit
h
Pom
era
nz
a
nd
Janusz
Raj
ski
,
“Scan
Base
d
Tests
wit
h
Lo
w
Swit
chi
ng
A
ct
i
vi
t
y”
Proc.
IEEE
D
esig
n
&
T
est
of
Com
pute
rs,
vol
.
24(3),
pp
.
268
–
275,
Ma
y
2007.
[6]
Remersaro
Remersaro
,
Xij
ia
ng
Li
n,
Sudhak
ar
M
Redd
y
and
Ir
i
th
Pom
era
nz,
“P
refe
rr
ed
Fi
l
l:
A
Scal
abl
e
Me
thod
to
Re
duc
e
Captur
e Powe
r for
S
can Based
Designs.”
Proc.
IEEE
In
ternat
ion
al
Te
st
C
onfe
ren
c
e2
006,
pp.
1
-
10.
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 Co
mp
r
ehe
nsi
ve Revi
ew
o
f
Ap
plicati
ons
of Don’t Ca
re Bi
t Fi
ll
ing
Tec
hniqu
es
for Test
…
(
Sanjo
y
Mi
tra
)
949
[7]
Grze
gorz
Mrugalski,
Janusz
Raj
ski,
Dari
usz
Cz
y
sz
and
Jer
z
y
T
y
sze
r
,
”N
ew
Test
Data
Dec
ompr
essor
for
L
ow
Powe
r
App
li
ca
tions
,
”
Proc.
44
th A
CM/IEE
E
Desi
gn
Autom
at
ion Confere
nc
e
200
7,
pp
.
539
–
544
.
[8]
“
DP
-
fil
l:
A
D
y
n
amic
Program
m
ing
appr
oa
ch
to
X
-
fil
li
ng
fo
r
m
i
nimizi
ng
p
ea
k
t
est
power
in
sca
n
te
sts,
”
Design
,
Autom
at
ion
&
T
est
in
Europ
e
Co
nfe
ren
c
e & Exhi
bit
ion (D
ATE)
,
2015,
pp
.
836
-
8
41
.
[9]
Xiaoqi
ng
W
en,
Y.Yamashita
,
S
.
Kaji
har
a,
La
un
g
-
Te
rng
W
ang,
K.
K.Sal
uj
a
and
K.
Kinoshita,
“O
n
Low
-
Captur
e
-
Powe
r Test
Ge
n
erati
on
for Sc
an
Testing,
”
Proc.
23rd
IE
E
E
VLSI
Te
st
S
y
m
posiu
m
(VTS'
05)
,
200
5,
pp
265
-
270
.
[10]
Xiaoqi
ng
W
en,
Y.Yamashita
,
S.
Kaji
har
a,
L
aun
g
-
Te
rng
W
ang,
K.
K.Sal
uja
an
d
K.
Kinoshita
,
”
Low
Capture
Swit
chi
ng
A
ct
i
vi
t
y
Test
Gen
erati
o
n
for
R
educing
I
R
-
Dr
op
in
A
t
-
S
p
ee
d
S
can
Test
ing
,
”
Journal
of
E
lectr
oni
c
T
esti
ng
,
Volum
e
24(4)
A
ugust
2008,
pp
.
379
–
391
.
[11]
Stepha
n
Egg
ersgl
,
“Pea
k
Capt
ure
Powe
r
Re
d
uct
ion
for
Com
pact
Test
Set
s
Us
ing
Opt
-
Justi
f
ic
ati
on
-
Fi
l
l,
”
Proc.
22nd
As
ia
n
T
est S
y
m
posium
2013,
pp
.
31
-
36
.
[12]
Rohini
Gulve
an
d
Vire
ndra
Sing
h,
“IL
P
Based
Don’t
Care
Bi
ts
Fi
lling
Techn
iq
ue
for
Re
du
ci
ng
Capture
Powe
r,
”
Proc.
IE
EE E
ast
-
W
est
Design
&
Te
st S
y
m
posium
(EWD
TS),
pp
.
1
–
4.
[13]
Masa
y
oshi
Yos
himura,
Yos
hi
yasu
Ta
k
aha
shi
,
Hiroshi
Yam
az
a
ki
,
Toshinori
H
osokawa,
“A
D
on’t
Care
F
i
llin
g
Me
thod to Redu
ce
Cap
ture Pow
er
based
on
Corr
el
ati
on
of FF
Tr
a
nsiti
ons,”
Proc.
I
EEE
24th
AT
S
2015,
pp
.
13
-
18
.
[14]
Jing
-
Li
ng
Yang
and
Qiang
Xu,
“
Stat
e
-
Sensiti
v
e
X
-
Fill
ing
Sch
eme
for
Scan
C
apt
ure
Pow
er
R
educ
t
ion
,
”
IE
E
E
Tr
ansacti
ons on Com
pute
r
-
Ai
ded
Design
of
Int
egrate
d
C
ircui
ts an
d
Syste
ms
,
vol
2
7(7),
pp
1338
-
1
343,
2008
.
[15]
Tsung
-
Ta
ng
Ch
en,
W
ei
-
Li
n
L
i,
Po
-
Han
W
u
and
Jiann
-
Ch
y
i
Rau
,
“A
New
Sch
em
e
of
R
educ
ing
S
hif
t
and
Captur
e
Powe
r Us
ing
th
e
X
-
F
il
l
ing
M
et
ho
dology
,
”
Proc
.
As
ia
n
Te
st
S
y
m
posium
,
2009,
p
p.
105
-
1
10
.
[16]
Ans
hum
an
Chan
dra
and
Rohit
,
K
apur
“
Bounde
d
Adj
ac
ent
Fi
l
l
fo
r
Low
Capture
Powe
r
Scan
Testing,
”
Proc.
26t
h
IEE
E
VLSI T
est
S
y
m
posium
(VTS 2008)
pp
.
13
1
–
138
.
[17]
Jia
Li,
Qiang
Xu
,
Yu
Hu
and
Xia
owei
Li,
“
X
-
Fill
ing
for
Sim
ult
an
eous
Shift
-
and
Capt
ure
-
Pow
er
Reduc
ti
on
in
At
-
Speed
Scan
-
Bas
ed
Te
st
ing,
”
I
E
EE
Tr
ansacti
on
s
on
Ve
ry
Lar
ge
Scale
Inte
gra
ti
on
(
VLSI)
Sys
te
ms
,
vol.
18(7)
,
pp.
1081
-
1092,
2
010.
[18]
Z.
Chen
,
J.
Feng
,
D.
Xiang
and
B.
Yin,
“
Scan
c
hai
n
conf
igu
ra
tion
base
d
X
-
filli
ng
for
low
powe
r
and
high
qu
al
i
t
y
te
sting
,
” IE
T Co
mputer
Digit.
Te
ch.
,
vol
.
4(1)
,
pp
.
1
–
13
,
2010
.
[19]
S
Sivana
ntha
m
,
K
Sara
thkumar,
Jinc
y
P
Manue
l,
P
S
Mall
ic
k
and
J
Raj
a
Paul
Pe
rinba
m
,
“C
SP
-
F
il
li
ng:
A
New
X
-
fill
ing
Techni
qu
e
to
Re
duce
Ca
ptu
re
and
Shif
t
Powe
r
in
Test
Appl
ic
a
ti
ons,”
Proc.
Inte
rna
ti
o
nal
S
y
m
posium
on
El
e
ct
roni
c
S
y
s
tem
Design
(ISED) 2012, pp. 135
–
139
.
[20]
S.
Sivana
ntha
m
,
V.
Sandee
p,
P.
S.
Mall
ic
k
and
J.
Raj
a
Paul
Perinba
m
,
“A
Nove
l
Approach
fo
r
Simult
aneous
Re
duc
ti
on
of
Sh
if
t
and
C
ap
ture
Powe
r
for
Scan
based
Testing,”
Proc.
IEEE
In
te
rna
ti
ona
l
Conf
ere
nc
e
on
Signa
l
Proce
ss
ing,
Com
m
unic
at
ion
,
Co
m
puti
ng
and
N
etw
orking
Techno
logi
es,
2011,
pp.
418
-
423
.
[21]
V
Sinduja
,
Ragha
v
Siddhart
h
a
nd
Anita
J
P,
“
Ef
fici
ent
Don’t
-
Care
Fi
ll
ing
Me
thod
to
Ac
hieve
Re
duction
in
Test
Powe
r,”
Proc.
Inte
rna
ti
ona
l
Confer
ence
on
Adv
in
Com
puti
ng,
Com
m
unic
at
ions
and
Inform
at
ic
s
(ICACCI),
2015,
pp.
478
–
482
.
[22]
Qing
Zha
o,
Xi
ao
Le
Cui
and
Chu
ng
Le
n
L
ee,
“An
Impr
ov
ed
X
-
Fill
ing
Strate
gy
Bas
ed
on
the
Mu
lt
i
l
aye
r
Data
Copy
Sche
me
for
Test
Data
and
Powe
r
Re
du
ct
ion
for
S
oC,
”
Proc.
11th
IEE
E
Inte
rn
at
ion
al
Confe
ren
c
e
o
n
Solid
-
State
an
d
Inte
gra
te
d
Ci
rcu
i
t
T
ec
hnolog
y
(I
CS
ICT),
2012
,
p
p.
1
-
3
.
[23]
Song
D
ongSup,
Jin
-
Ho
Ahn,
Ki
m
Ta
ej
in
and
Kang
Ungho
,
“
M
TR
-
Fill
:
A
Sim
ula
t
ed
Annea
li
n
g
-
Based
X
-
Fill
i
ng
Te
chn
ique
to
Re
duce
T
est
Pow
er
for
sca
n
Based
Designs,”
IEI
C
E
Tr
ans
on
Info
&
Sys
,
vol
91(4)
pp.
1197
-
1200
,
April
2008
.
[24]
Hai
y
ing
Yuan,
Kun Guo,
Xun
S
un
and
Zi
jian
Ju,
“
A
Pow
er
Eff
ic
ie
nt
Te
s
t
Dat
a
Com
pre
ss
ion
Meth
od
for
SoC
u
sing
Alte
rna
ti
ng
St
at
i
stic
al Run
-
L
ength
Coding,”
J El
e
ct
ron Test,
vol
3
2(1),
pp
59
–
68,
Feb
2016
.
[25]
A.S.T
rina
dh
,
S.
Potluri
,
S.
Balac
handr
an,
C
.
S.
Babu
and
V.
Kam
akot
i,
“
X
Stat
:
X
-
Fil
li
ng
A
lgori
thm
for
Pe
ak
Capt
ure
Pow
er Red
uct
ion
in
Sc
an
T
ests,
”
Journ
al
of
L
ow
Powe
r
Elec
troni
cs,
vo
l
10(1)
,
pp.
107
-
1
15,
Mar
ch
2001
.
[26]
L
F
Rahman,
M
ohamm
ad
F.
B.
Am
ir,
M
B
I
Re
az
,
M.
Marufuz
z
aman,
and
H
Hu
sain,
“
Advanc
es
on
Low
Pow
er
Designs
for
SR
AM
Cel
l,
”
TEL
KOMNIKA
Indone
sian
Journal
of
El
e
ct
ri
ca
l
En
gine
ering
,
Vol.
12(8),
pp.
6063
-
6082,
Augus
t
2
014
K
B
Ra
y
,
S
K.
Manda
l
and
B.
S.
Patro,
“
Low
Pow
er
F
GS
RAM
Cel
l
Us
ing
Slee
p
y
a
nd
LE
CTOR
T
ec
hn
ique
”
Indon
esia
n
Journal
of
Ele
ct
rical
Engi
n
ee
r
ing
and
Comput
er
Sci
en
ce,
Vol
.
4(2),
pp
.
333
~
340,
Novem
ber
2
016
.
[27]
K
B
Ray
,
S
K.
Manda
l
and
B.
S.
Patro,
“
Low
Pow
er
FG
SR
A
M
Cel
l
Us
ing
S
le
ep
y
and
LE
C
TOR
Te
chni
q
u
e
”
Indone
sian
J
our
nal
of
Elec
tric
al
Engi
ne
ering
and
Computer
Sc
ie
n
ce
,
Vol.
4(2), pp. 333
~ 340
,
Nov
ember
2016.
Evaluation Warning : The document was created with Spire.PDF for Python.