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.
4456
~
44
66
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v8
i
6
.
pp
4456
-
44
66
4456
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Cachin
g on Name
d Data
Netw
ork:
a Survey
and Futu
re
Re
search
Le
an
n
a Vidy
a Y
ov
i
ta, Nan
a Ra
c
hmana
S
yamb
as
School
of El
ec
tr
i
ca
l
Engi
n
ee
r
ing and
Inform
at
i
cs,
Bandung
Insti
tute
of
Technol
og
y
,
Indone
sia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ma
r
20
, 201
8
Re
vised
Ju
l
16
,
201
8
Accepte
d
J
ul
30
, 2
01
8
The
IP
-
base
d
s
y
stem
c
ause
i
nef
ficien
t
con
tent
de
li
ver
y
pro
ce
ss
.
Th
is
ine
ffi
ci
en
c
y
wa
s
at
te
m
pte
d
to
be
solved
with
the
Conte
nt
Distribut
ion
Network.
A
rep
l
ic
a
serve
r
is
locate
d
in
a
par
ticu
la
r
locati
on
,
usually
on
the
edge
router
tha
t
is
cl
osest
to
the
user.
The
user’s
req
uest
will
b
e
serve
d
from
tha
t
rep
l
ica
serve
r.
How
eve
r,
c
a
chi
ng
on
Content
Distribut
ion
Network
is
infl
exi
b
le.
Thi
s
sy
st
em
is
diffi
cul
t
to
support
m
obil
ity
and
co
ndit
ions
of
d
y
nami
c
content
demand
from
consum
ers.
W
e
n
ee
d
to
shift
th
e
par
adi
gm
to
cont
en
t
-
ce
n
tric.
In
Nam
ed
Data
Network,
da
ta
c
an
be
place
d
on
the
content
store
on
rout
ers
tha
t
are
cl
osest
to
the
consum
er.
Cac
h
ing
on
N
amed
Dat
a
Network
m
ust
b
e
abl
e
to
store
c
onte
nt
d
y
namic
a
lly
.
It
should
be
sele
ctiv
e
l
y
sele
c
t
content
th
at
is
eligible
to
be
sto
red
or
dele
te
d
from
the
con
te
nt
storag
e
base
d
on
ce
rtai
n
conside
rations,
e.
g
.
the
popul
ari
t
y
of
content
in
the
loc
a
l
are
a
.
Thi
s
surve
y
p
ape
r
exp
la
ins
the
developm
en
t
of
cachi
ng
te
c
hnique
s
on
Nam
ed
Data
Network
tha
t
ar
e
cl
assifie
d
int
o
m
ai
n
point
s.
The
brie
f
expl
an
at
ion
of
a
dvant
ag
es
and
di
sadva
nta
g
es
are
pre
sente
d
to
m
a
ke
it
ea
s
y
to
under
stand.
Fin
al
l
y
,
proposed
t
he
open
challenge
rel
a
te
d
to
the
cachi
ng
m
ec
hani
sm
to i
m
prove
ND
N pe
rform
anc
e
.
Ke
yw
or
d:
Ca
chin
g place
m
ent
Ca
chin
g po
li
cy
Con
te
nt selec
ti
on
Con
te
nt sto
rage
Nam
ed
data ne
twork
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
:
Lean
na Vidya
Yovita
,
School
of Elec
tric
al
Engineer
ing
a
nd
Inform
at
ic
s,
Ba
ndung I
ns
ti
tute o
f
Tec
hnol
og
y,
10
Ga
nes
ha
R
oa
d,
Lb
. S
il
iwa
ngi, C
ob
l
ong, Ba
ndung 4
0132,
W
est
Ja
va, I
ndonesi
a.
Em
a
il
:
le
ann
a@te
lko
m
un
ive
rsity
.ac.id
1.
INTROD
U
CTION
The
e
ra
of
te
le
com
m
un
ic
at
ion
s
beg
a
n
i
n
1876,
wh
e
re
a
ne
twork
was
buil
t
that
ena
bled
2
par
ti
es
t
o
transm
it
their
vo
ic
e
an
d
com
m
un
ic
at
e.Th
e
In
te
r
net
be
gan
in
1969
s
,
f
unded
by
Adva
nc
ed
Re
searc
h
Pr
oject
Ag
e
ncy
[
1]
.
U
sing
I
nter
net
P
ro
t
oco
l
(
IP)
as
an
ad
dr
ess
,
th
e
request
from
the
u
ser
will
be
f
orwarde
d
to
the
serv
e
r,
thr
ough
oth
e
r
node
s
w
it
hin
the
net
w
ork.
The
re
plies
to
the
re
quest
will
be
sent
to
the
us
er
thr
ough
a
par
ti
cula
r
path
that
ha
s
been
form
ed
by
r
ou
ti
ng
process
in
the
netw
ork.
I
f
a
ny
us
e
r
requests
the
sam
e
data,
then
the
p
acket
will
be
sent
again
f
r
om
the
serv
e
r
to
the
use
r.
T
his
cause
s
ineff
ic
ie
nt
pa
cket
delive
ry
proces
s
because
the
pa
cket is al
ways
sent from
a ser
ver
t
hat is far
f
ro
m
the u
ser
. T
o
s
olv
e this
prob
le
m
, th
e co
nc
ept of
Con
te
nt
Distri
bu
ti
on
Netw
ork
[
2]
was
pro
po
s
ed
.
A
re
plica
ser
ver
is
c
r
eat
ed
co
ntains
al
l
the
data
a
s
in
the
m
ai
n
serv
e
r,
pl
aced
at
a
fixe
d
locat
io
n,
cl
ose
r
to
t
he
use
r
.
So
that,
the
request
f
or
cer
ta
in
co
ntent
w
il
l
be
red
irect
e
d
t
o
th
e re
plica
server
and it i
s
no n
e
ed
to
b
e
serve
d by a
n
o
ri
gin
s
erv
e
r
that i
s
f
a
r
ther
a
way.
The
re
plica
serv
e
r
is
update
dp
e
rio
dical
ly
or
w
he
n
any
con
te
nt
chang
es
on
it
s
or
ig
inal
serv
e
r
.
Howe
ver,
this
syst
e
m
will
be
dif
ficult
to
su
pport
m
ob
il
it
y
and
dyna
m
ic
chan
ging
con
te
nt
reques
t
fr
om
consum
ers.
Wh
en
the
c
onsu
m
er
away
fr
om
the
rep
li
ca
serv
e
r,
it
le
ads
to
the
possibil
it
y
that
a
con
sum
er
can
no
lo
nger
be
s
erv
e
d
ef
fici
ent
ly
by
the
rep
li
ca
serv
e
r.
Co
nt
ent
Distrib
ution
Netw
ork
th
at
is
still
based
on
In
te
r
net
P
r
oto
c
ol
(
IP)
ca
us
es
the
re
quest
process
from
the
us
e
r
is
al
ways
ad
dr
e
ssed
to
a
pa
rtic
ular
s
erv
e
r
.
Con
se
quently
,
anaddit
iona
l
process
is
sti
ll
ne
eded
t
om
app
ing
t
he
inte
nd
e
d
I
P
with
t
he
s
erv
e
r
posit
io
n
that
is
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
Cachi
ng on N
amed
Data
Ne
tw
or
k:
a S
ur
vey
and
…
(
Lean
na Vi
dya Y
ovita
)
4457
cl
os
est
to
the
use
r.
Actuall
y,
f
ro
m
the
beg
i
nning
,
t
he
f
oc
us
of
t
he
us
e
r
re
quest
is
the
co
nt
ent
(c
on
te
nt
-
ba
sed),
bu
t i
n
the
pre
vio
us
syst
em
,
the r
e
qu
est
is
addr
essed
t
o
a ce
rta
in serve
r node
with a ce
rtai
n I
P (host
-
ba
sed
).
In
2009, Ja
co
bson et
al.
pro
pose
a c
on
te
nt
-
ba
sed netw
ork p
arad
i
gm
[3]
. Th
is co
nce
pt h
as
b
ee
n
raise
d
a
few
ye
ars
ea
rlie
r
in
it
s
res
earch
pro
j
ect
s
and
it
is
nam
e
d
Co
ntent
-
Ce
nt
ric
Networ
king
(CCN
)
ori
gi
nally
dev
el
op
e
d
at
X
erox'
s
Palo
Alto
Re
search
Ce
nter.
It
is
curre
ntly
dev
el
opin
g
into
Nam
ed
Data
Netw
ork
(NDN)
init
ia
te
d
by
the
NS
F
-
Fun
ded
Fu
ture
I
nter
ne
t
Ar
chite
ct
ure
Pr
oject
[4]
.
This
co
ncep
t
rep
la
ces
the
'
wh
e
re'
par
a
dig
m
to
th
e
co
ncep
t
of
'
w
hat'
,
wh
e
re
the
consum
er
re
quest
is
no
lo
ng
er
ad
dresse
d
to
a
sp
eci
fic
node
bu
t
it
is
intende
d
f
o
r
a
certai
n
con
t
ent
[3]
,
[
5]
-
[7]
.
This
pa
rad
i
gm
causes
the
respon
se
to
t
he
con
te
nt
re
que
sts
no
t
on
ly
ser
ved
by
a
par
ti
cular
s
erv
e
r
but
al
so
can
be
se
rv
e
d
by
the
near
e
st
dev
ic
e
w
hic
h
stores
the
r
eq
ue
ste
d
data.
To
s
u
ppor
t
this
co
nce
pt,
the
N
D
N
router
nodes
are
eq
uipped
with
co
ntent
stora
ge
to
sto
re
th
e
data
[
3]
-
[
6]
,
[8]
.
The
c
oncept
of
cachi
ng
on
N
a
m
ed
Data
Net
work
is
diff
e
re
nt
f
ro
m
cachin
g
in
the
pr
evi
ous
syst
em
.
Each
N
D
N
no
de
has
a
co
nte
nt
stora
ge
to
hold
data
.
Dif
fe
ren
t
with
the
previ
ou
s
netw
ork,
no
de
m
ob
il
i
ty
will
be
s
upporte
d
be
cause
the
co
nt
ent
store
ca
n
be
ta
il
or
e
d
to
t
he
use
r'
s
dem
a
nd
patte
r
n
f
or
the
c
on
te
nt.
Ch
ange
s
in
us
e
r
posit
io
ns
caus
e
th
e
r
ou
te
r
has
to
re
-
cust
om
iz
e
the
con
te
nts
in
th
e
con
te
nt
store
accor
ding
to
t
he
us
e
r
requests
in
the
local
area.
Ca
chein
NDN
is
m
or
e
dynam
ic
.
The
N
DN
a
rc
hitec
ture
s
upports
fle
xib
le
ne
twor
k
topolo
gies,
w
he
re
wireless
node
s
can
enter
and
exit
the
area.
O
ne
ti
m
e,
a
no
de
ca
n
be
a
pr
od
uce
r
and
so
m
et
i
m
es
i
t
t
urns
int
o
the
c
onsu
m
er
.
I
n
the
NDN
node
can
be
em
bed
de
d
with
va
rio
us
c
ache
r
ules,
incl
ud
i
ng
to
determ
ine
wh
ic
h
co
ntent
will
be
sel
ect
ed
an
d
delet
ed
from
con
te
nt
store
[
5]
,
[
9]
,
[1
0]
,
the
sel
ec
ti
on
of
places
w
he
re
a
con
te
nt
will
be
cac
hed
[
10]
-
[
13]
,
an
d
c
ache
po
li
cy
th
at
i
m
ply
the
m
od
el
of
co
operati
on
betwee
n
no
des
to
determ
inethe
cachin
g
de
ci
sion
[14]
-
[17]
.
Re
la
te
d
to
node
m
ob
il
it
y,
sever
al
te
chn
i
qu
e
s
have
been
st
ud
ie
d
to
m
a
intai
n
the
per
f
orm
ance
of
the
syst
em
even
thou
gh
the
node
s
m
ov
e
in
and
ou
t
of
cov
e
ra
ge
[
18]
-
[21]
.
This
s
urve
y
pa
per
ex
plains
th
e
de
velo
pm
ent
of
cac
hing
te
c
hn
i
qu
e
s
on
N
a
m
ed
Data
Net
work
wh
ic
h
is
an
i
m
po
rtant
basis
for
unde
rstan
ding
the
la
te
st
ND
N
cachin
g
te
chn
i
qu
e
s
in
de
velo
ping
bette
r
fu
t
ur
e
te
chn
iq
ues
for
enh
a
ncin
g
N
DN
perf
or
m
ance
as
an
ef
fic
ie
nt
forw
a
r
d
c
omm
un
ic
at
ion
so
luti
on.
T
he
br
ie
f
exp
la
natio
n
of
adv
a
ntage
s
an
d
disa
dvanta
ge
s
are
prese
nted
to
m
ake
it
eas
y
to
unde
rstan
d.
Fi
nally
,
pro
po
s
ed
the ope
n
c
halle
ng
e
r
el
at
ed
to
t
he
cac
hing m
echan
ism
to
im
pro
ve ND
N pe
rfor
m
ance
.
The
rem
ai
nd
er
of
this
pa
pe
r
is
or
ga
nize
d
as
fo
ll
ow.
I
n
se
ct
ion
2
descr
i
bed
t
he
sta
te
of
the
a
rt
of
cachin
g
on
th
e
Nam
ed
Data
Netw
ork
.
T
he
cachi
ng
te
ch
niques
a
re
gro
up
e
d
i
nto
cac
he
placem
ent,
cache
con
te
nt
sel
ect
ion,
an
d
cac
he
po
li
cy
desi
gn.
In
t
his
sect
iona
lso
ex
plaine
d
the
ad
va
ntage
s
an
d
draw
bac
ks
of
each grou
p
of
cachin
g
te
ch
niq
ue
s.
The
te
ch
nique to s
up
port
m
ob
il
it
y are
exp
la
ine
d
to
o.
In
secti
on 3
d
e
scrib
e
d
the
c
halle
ng
e
and
op
e
n
iss
ue
s
relat
ed
to
the
cac
hing
i
n
Nam
ed
Data
Netw
ork.
A
nd
finall
y,
sect
i
on
4
put
forw
a
r
d
the
conclusi
on of thi
s p
a
per.
2.
RELATE
D
W
ORK
So
m
e
su
rv
ey
pap
e
rs
on
cachin
g
ha
ve
bee
n
done
be
f
or
e
.
Paper
[
22]
em
ph
asi
zes
discuss
io
n
on
te
chn
iq
ues
of
cache
re
place
m
ent
fo
r
we
b
serv
ic
es
i
n
the
IP
-
base
d
syst
em
.
Paper
[23]
e
m
ph
asi
zes
dis
cussi
on
on
se
veral
in
-
ne
twork
cac
hing
m
echan
is
m
s
in
Inform
at
ion
-
Ce
ntric
Net
w
orkin
g,
2014
and
ea
rlie
r.
Pa
pe
r
[5
]
discusse
s
in
for
m
at
ion
-
cent
ric
m
ob
il
e
caching,
incl
ud
i
ng
cachin
g
in
cel
lular,
ad
hoc,
a
nd
a
hy
br
i
d
network.
This
pap
e
r
e
xpla
ins
dif
f
ere
nt
cache
l
ocati
on
for
eac
h
sc
hem
e
and
so
m
e
cache
m
echan
is
m
s.
Discus
sion
on
cachin
g
rep
la
c
e
m
ent
m
echan
is
m
on
ly
,
pr
es
ented
by
pa
pe
r
[
22]
an
d
[
24]
,
for
t
he
m
ob
il
e
node
.
D
ue
t
o
th
e
auth
or
'
s
knowle
dg
e
,
so
far
th
ere
has
been
no
pa
per
s
urvey
that
discuss
es
t
he
la
te
st
cac
hing
te
ch
nique
s
and
m
app
in
g
the
te
chn
i
qu
e
in
gro
up
s
base
d
on
the
ba
sic
te
ch
niq
ue
.
T
his
sche
m
e
m
akes
it
ea
sy
to
unde
rsta
nd
t
he
basic te
ch
niqu
es of the
cachi
ng m
echan
ism
s.
This
s
urvey
pa
per
f
ocu
se
s
on
cachin
g
te
c
hniqu
es
,
incl
ud
i
ng
the
rece
nt
st
u
dies
.
T
o
m
ake
it
cl
ear
to
exp
la
in
, in
this
p
ape
r
the cach
ing
m
echan
ism
is
div
ided
int
o 4 g
ene
ral g
r
ou
p,
they are cac
he
placem
ent, cache
con
te
nt
sel
ect
ion,
cache
poli
cy
design
a
nd
ca
chin
g
f
or
m
ob
i
le
nodes.
T
he
e
xp
la
nation
be
gi
ns
with
the
ca
chin
g
diff
e
re
nces
in
Nam
ed
Data
N
et
work
with
it
s
pr
e
vious
netw
ork,
t
he
ur
gen
c
y
of
cac
hing
discuss
io
n
on
Na
m
ed
Data
Net
work
to
im
pr
ov
e
it
s
pe
rfor
m
ance,
an
d
the
n
m
a
pp
i
ng
the
cac
hing
te
ch
nique
s
base
d
on
the
basi
c
m
echan
ism
.
T
he
ai
m
is
to
facil
it
at
e
the
read
er
to
know
the
basis
of
the
de
velo
pm
e
nt
of
the
se
cachin
g
te
chn
iq
ues
.
Th
e
adv
a
ntages
a
nd
disad
va
ntag
es
of
each
gro
up
of
te
ch
niqu
es
are
pr
ese
nte
d
with
a
su
cci
nc
t
and
fo
c
us
to
m
ake
it
easy
to
unde
rstan
d.
T
his
pa
per
c
on
cl
ud
es
with
an
ex
plan
at
ion
of
t
he
propose
d
resea
rc
h
on
cachin
g
on
N
DN
t
hat
is
sti
l
l
op
e
n,
s
o
it
can
co
ntin
ue
to
be
de
velo
pe
d
to
exam
ine
t
he
best
te
c
hn
i
qu
e
s
to
su
pp
or
t
N
DN.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4456
-
4466
4458
3.
STATE
OF
T
HE A
RT O
F
CACHI
NG O
N NAME
D
DATA
NETW
ORK
3.1
.
C
ompone
nt
of Named
Data Ne
twor
k
R
ou
te
r
Nam
ed
Data
N
et
work
s
hifts
t
he
'
wh
ere'
to
'
wh
at
'
par
a
digm
.
The
us
er
se
nd
s
his
re
quest
for
a
co
ntent
to
the
net
work,
an
d
t
he
n
the
netw
ork
w
ouldd
et
e
rm
ines
w
ho
is
the
m
os
t
eff
ic
ie
nt
no
de
that
ca
n
se
rve
this
request.
So,
t
he
us
e
r
does
no
t
nee
d
t
o
know
wh
e
re
is
the
c
onte
nt
se
r
ver
.
T
his
par
a
dig
m
causes
t
he
rep
ly
f
or
a
request
no
t
al
ways
be
do
ne
by
the
se
r
ver
,
bu
t
a
ny
node
t
hat
is
in
t
he
ne
twork
.
T
he
N
DN
arc
hitec
tur
e
causes
data com
m
un
ic
at
ion
processes
m
or
e eff
ic
ie
nt
and n
et
wor
k
lo
ads wil
l be si
gnific
antly
r
e
du
ced.
The
N
D
N
N
o
de
co
ns
ist
s
of
3
com
po
ne
nts,
nam
ely
Con
te
nt
Stor
a
ge
(C
S)
,
Pe
ndin
g
Int
erest
Table
(P
I
T)
an
d
F
orwardin
g
I
nform
at
ion
Ba
sed
(
FI
B)
[
3]
,
[
25]
.
Wh
e
n
c
on
s
ume
r
B
wan
ts
a
da
ta
fr
om
the
produce
r
,
the
co
nsum
er
will
send
a
re
quest
f
or
ce
rtai
n
con
te
nt
us
i
ng
the
I
nterest
Pa
c
ket.
T
he
N
D
N
r
oute
r
that
r
ecei
ves
the
re
qu
est
from
the
con
s
um
er
will
check
w
hethe
r
the
con
te
nt
is
in
it
s
CS.
If
th
ere
is,
the
rou
te
r
will
i
m
m
ediat
el
y
s
end
the
re
ques
te
d
data
to
the
con
s
um
er.
If
the
data
is
no
t
in
CS,
then
the
router
c
hec
ks
the
Pendin
g
I
nte
re
st
Table
to
see
if
the
co
ntent
has
be
en
reque
ste
d
an
d
ha
s
not
bee
n
re
plied
with
m
at
ching
data
pack
et
s
.
I
f
in
the
P
IT
th
ere
is
su
c
h
in
form
ation
,
the
n
the
i
nfor
m
at
ion
wil
l
be
up
dated
by
add
i
ng
i
nfo
r
m
at
ion
that
co
nsum
er
B
al
so
request
ed
the
sam
e
data.
The
i
nfo
rm
at
ion
on
this
P
IT
m
akes
a
r
ev
erse
path
f
or
s
end
i
ng
data
to
the
co
ns
um
er.
If
in
t
he
P
IT
the
re
i
s
no
data
requ
est
co
ntent
t
ha
t
is
the
sam
e
as
co
nsum
er
B,
the
n
checks
are
m
a
de
on
F
orwa
rdi
ng
I
nfo
rm
at
io
n
Ba
sed
(FIB).
The
intere
st
pa
cket
will
be
f
orwarde
d
t
o
th
e
dat
a
pro
vid
er
node
accor
ding
to
th
e
inform
at
ion
i
n
FI
B.
If
F
IB
does
not
store
the
co
ntent
pr
ovide
r'
s
no
de
da
ta
,
the
interest
p
a
cket
will
b
e
discar
de
d by the
N
DN r
ou
te
r
. T
his
process
desc
ribe
d
in
Fig
ure
1.
St
a
r
t
T
h
e
r
e
i
s
a
c
o
n
t
e
n
t
t
h
a
t
i
s
r
e
q
u
e
s
t
e
d
i
n
C
o
n
t
e
n
t
S
t
o
r
a
g
e
S
e
n
d
t
h
e
c
o
n
t
e
n
t
t
o
t
h
e
c
o
n
s
u
m
e
r
C
h
e
c
k
i
n
t
h
e
P
I
T
T
h
e
r
e
i
s
a
n
i
n
f
o
r
m
a
t
i
o
a
n
a
b
o
u
t
t
h
e
c
o
n
t
e
n
t
t
h
a
t
i
s
r
e
q
u
e
s
t
e
d
Up
d
a
t
e
t
h
e
P
I
T
C
h
e
c
k
i
n
t
h
e
FI
B
T
h
e
r
e
i
s
a
n
i
n
f
o
r
m
a
t
i
o
a
n
a
b
o
u
t
t
h
e
c
o
n
t
e
n
t
t
h
a
t
i
s
r
e
q
u
e
s
t
e
d
F
o
r
w
a
r
d
p
a
c
ke
t
i
n
t
e
r
e
s
t
D
r
o
p
t
h
e
I
n
t
e
r
e
s
t
P
a
c
ke
t
E
n
d
No
Y
e
s
No
Y
e
s
Y
e
s
No
Figure
1. Proce
sses that
occ
ur
on each
ND
N r
ou
te
r wh
e
n
it
re
cei
ves
the
Int
erest Pac
ket
[
25]
3.2. C
on
te
nt S
to
r
age
Con
te
nt
Stora
ge
(CS)
is
on
e
of
the
im
po
rtant
com
po
ne
nts
in
the
N
DN
r
ou
t
er
no
de.
CS
is
essenti
al
to
al
low
the
data
to
be
sto
re
d
in
N
D
N
r
ou
te
r
nodes
s
o
that
if
the
co
nsu
m
er
request
f
or
a
co
ntent,
it
is
no
t
necessa
rily
serv
ed
by
the
certai
n
serv
e
r,
bu
t
can
be
ser
ve
d
by
a
ro
ute
r
no
de
that
has
the
con
te
nt
in
it
s
CS.
CS
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
Cachi
ng on N
amed
Data
Ne
tw
or
k:
a S
ur
vey
and
…
(
Lean
na Vi
dya Y
ovita
)
4459
is
on
e
of
the
li
m
it
ed
resour
ce
s
on
N
D
N
r
ou
te
rs.
T
her
e
fore
it
sh
ould
be
ut
il
iz
ed
as
eff
ic
ie
ntly
as
possib
le
in
order t
o
im
pr
ove
NDN per
f
orm
ance.
The
siz
e
of
th
e
co
n
te
nt
st
or
e
aff
ect
s
t
he
del
ay
and
num
ber
of
ho
ps
t
hat
pa
ckets
m
us
t
ta
ke
to
go
t
o
consum
er
[
26
]
.
T
his
co
ndit
ion
af
fects
the
overall
netw
ork
l
oad
due
to
the
ci
rcu
la
t
ion
of
data
in
th
e
netw
ork
[
4]
,
[
8]
.
CS
al
so
pe
rfor
m
s
diff
ere
nt
eff
ect
with
the
var
i
ou
s
ca
che
poli
cy
i
m
plem
ented
in
t
he
node
[27]
.
I
n
this
pa
per,
the
cac
hin
g
st
rategies
a
re
cl
assifi
ed
as
cache
placem
ent,
cache
c
on
te
nt
sel
ect
ion
,
cache
po
li
cy
desig
n
a
nd
cac
hing
for
the
m
ob
il
e
no
de
.Each
group
de
scribe
d,
inclu
ding
it
s
adv
ant
ages
an
d
draw
backs
in secti
ons
3.3 t
o
3.6
.
3.3. C
ache Pl
ac
ement
Ca
che
placem
ent
fo
c
us
es
on
determ
ining
w
hich
node
s
will
store
a
data
pac
ket.
I
n
th
e
P
ubli
sh
er/S
ubs
criber
net
work,
it
has
bee
n
propose
d
a
m
et
ho
d
t
o
ch
oose
a
node
to
sto
re
pa
ckets
ba
sed
on
local
con
te
nt
po
pu
la
rity
and
co
nten
t
stora
ge
ca
pac
it
y
per
node
[
28]
.
I
n
N
DN
ne
tworks
,
pack
et
s
are
i
niti
al
ly
placed
on ev
e
ry
node i
n
the
netw
ork
so
t
hat the c
on
su
m
er can
dire
ct
ly
access the
con
te
nt to
t
he
c
losest
node
.
Pape
r
[
11
]
pr
opose
s
a
pac
ket
data
flo
od
i
ng
m
echan
ism
,
wh
ere
data
pac
ke
ts
are
store
d
i
n
al
l
nodes
that
are
in
the
best
path
but
lim
i
te
d
to
the
m
axi
m
u
m
nu
m
ber
of
hops
f
or
the
s
pr
ea
d
of
th
e
pa
cket.
In
pa
pe
r
[12]
the
pac
ka
ge
is
de
plo
ye
d
to
be
sto
red
in
netw
ork
node
s
bu
t
sti
ll
m
ai
ntained
to
ens
ur
e
t
her
e
a
re
no
redu
nd
a
nt
pac
kets,
t
o
sa
ve
r
eso
ur
ces
,
us
in
g
bloom
filt
er.
The
la
ck
of
bl
oo
m
fil
te
r
te
chn
i
qu
e
relat
edt
o
f
al
se
po
sit
ive
pro
ble
m
s
cor
recte
d
by
A.
Hiday
at
et
al
[13]
.
In
t
his
te
chn
i
que,
bloom
fil
t
er
is
com
bin
in
g
with
seq
uen
ti
al
sear
ch
al
go
rithm
.
Pape
r
[
29
]
pr
opos
es
a
m
ec
han
ism
that
c
om
bin
es
the
te
chn
i
qu
e
of
pa
cket
insertio
n
and
packet
delet
ion
by
ad
ding
a
Ca
chin
g
Co
ntributi
on
par
am
et
er
in
the
interest
pac
ket.
Th
e
node
will
decide
w
he
ther
it
will
cache
the
data
pac
ket
or
not.
If
t
he
da
ta
pack
et
ca
nnot
be
cac
he
d
on
the
ce
rtai
n
node,
the
n
it
will
be
forw
a
r
ded
t
o
a
no
t
her
node
.
A
trai
l
m
echan
is
m
is
bu
il
t
to
st
or
es
inf
orm
ati
on
ab
out
the
pa
th
to
t
he
ne
xt
node
t
hat
can
store
the
co
ntent.In
pap
e
r
[
10
]
,
nodes
that
oft
en
get
a
con
te
nt
r
e
qu
e
st
from
c
on
s
um
er
hav
e
a
hig
h
con
t
rib
ution
va
lue.
A
node
will
store
a
c
on
te
nt
that
ha
s
hi
gh
co
ntri
buti
on
value
if
stora
ge
capa
ci
ty
is
avail
able.
Pa
pe
r
[
9]
pro
poses
the
m
ov
em
ent
of
data
to
wa
rd
the
ed
ge
router
cl
os
est
to
t
he
consum
er
for
e
very
sp
eci
fic
c
on
te
nt
re
quest
.
T
he
cache
place
m
ent
can
be
r
esum
e
into
3,
i.e.
functi
on
base
d,
div
e
rsity
and
flo
od
i
ng
as
s
how
n
i
n
Fi
g
ure
2.
Com
par
iso
n
of
the
t
hr
ee
te
chn
i
ques,
i
nclu
ding
t
he
t
echn
ic
al
fo
c
us,
th
e
adv
a
ntage
s,
a
nd
dr
a
w
back
s
desc
ribe
d
in
Ta
ble 1.
Table
1.
C
om
par
iso
n of
Ca
ch
e Place
m
ent Techn
i
qu
e
s
Clas
sif
icatio
n
Main f
o
cu
s
Ad
v
an
tag
es
Drawback
s
Fu
n
ctio
n
-
b
ased
[
2
8
]
[
1
0
]
[
2
9
]
Fo
cu
s o
n
sp
ec
if
ic
p
ara
m
et
ers to
b
e ac
h
iev
ed
.
i.e.
a
m
in
i
m
u
m
d
elay
,
th
e
nu
m
b
e
r
o
f
hops
,
etc.
-
Sp
ecif
ically
to
m
a
x
i
m
i
ze
th
e ac
h
iev
e
m
en
t of
certa
in
para
m
et
ers
-
The alg
o
rith
m
is
m
o
re
co
m
p
lex
if
th
ere
ar
e
m
an
y
pa
r
a
m
ete
rs.
Div
ersity
[
9
]
[
1
2
]
E
m
p
h
asize th
e sp
r
ea
d
of
co
n
ten
t on
the n
etwo
rk b
y
av
o
id
in
g
t
o
keep
th
e sa
m
e
p
ackets
in th
e netwo
rk. to sav
e
reso
u
rces on
the n
etwo
rk
-
More eff
icien
tly
st
o
re
co
n
ten
t becau
se o
n
ly
d
if
f
erent con
ten
ts t
h
at ar
e
sto
red b
y
dif
f
erent no
d
es
-
Lar
g
er
average d
el
ay
-
Hav
e to co
o
p
erate
with
o
th
er
n
o
d
es in
i
m
p
le
m
en
tin
g
con
ten
t
sto
rage r
u
les
-
The p
rocess
in
g
ti
m
e
m
a
y
inc
rease
as it inv
o
lv
es k
n
o
wled
g
e f
ro
m
o
th
er
n
o
d
es
Flo
o
d
in
g
[
3
]
[
1
1
]
E
m
p
h
asize to
sto
re
con
ten
t as
m
u
ch
as po
ss
ib
le on
the
n
etwo
rk to
ens
u
re
th
e ea
se an
d
sp
eed o
f
con
s
u
m
e
r
s in
accessin
g
data
-
Users can a
ccess
p
ackag
es with
m
i
n
i
m
al
d
elay
-
Req
u
ests
f
o
r
co
n
te
n
t can
b
e served
by
m
u
lti
p
le
n
o
d
es
-
Netwo
rk lo
ad
s in
cr
ease becau
se
n
o
d
es sto
re
th
e sam
e
con
ten
t.
Figure
2. Cl
assifi
cat
ion
of cac
he plac
em
ent t
echn
i
qu
e
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4456
-
4466
4460
3.4
.
C
ache C
onten
t
Sele
c
tio
n
The
cache
c
on
te
nt
sel
ect
ion
te
chn
iq
ues
fo
c
us
on
dete
rm
i
ning
w
hich
c
onte
nt
will
be
cached
a
nd
wh
ic
h
co
ntent
sh
ould
be
re
m
ov
ed
from
t
he
cache.
S
ome
of
the
co
nte
nt
sel
ect
ion
te
chn
i
qu
e
s
to
cache
are
Ca
chin
g
E
ver
y
thing
Eve
ryw
he
re
(CEE
)
[3]
,
[23]
,
wh
e
re
e
ach
node
st
or
e
s
al
l
of
t
he
dat
a
from
the
pro
du
ce
r
and
it
m
eans
no
c
on
te
nt
sel
ec
ti
on
a
nd
P
rob
(
p)
[
3]
,
[
5]
,
[
23]
wh
e
re
data
is
cache
d
with
pro
ba
bili
ty
p
an
d
no
t
cached
with
prob
a
bili
ty
1
-
p.
As
a
re
su
lt
,
data
packet
s
that
are
cache
d
by
on
e
r
ou
te
r
m
ay
be
di
ff
e
ren
t
from
the
oth
e
r
r
ou
te
rs.
Pape
r
[
30
]
pro
poses
the
c
on
ce
pt
that
e
ver
y
r
oute
r
c
ache
the
data
with
the
pro
bab
il
it
y
determ
ined
by the
num
ber
of h
op b
et
wee
n
pro
ducer
a
nd
th
e
router
.Select
ion
of
co
nte
nt
to
be
cache
d
ba
sed
on
the
predict
io
n
that
the
co
ntent
will
be
reques
te
d
by
the
loca
l
con
s
um
er
p
ropo
s
ed
by
pa
per
[31]
.
Re
la
te
d
to
t
he
cache
c
on
te
nt
sel
ect
ion
,
c
onte
nt
cent
ric
net
work
pe
rfo
rm
a
nce
is
al
s
o
a
ffec
te
d
by
C
S
r
eplace
m
ent
ru
le
s
a
nd
us
er
locali
zat
io
n
[32]
.
T
he
cac
he
c
on
te
nt sele
ct
ion
ca
n be
re
su
m
e as in
Ta
bl
e 2
.
Table
2.
C
om
par
iso
n of
Ca
ch
e Co
ntent S
el
e
ct
ion
Tec
hniq
ue
s (In
se
rtio
n
a
nd E
victi
on
)
Clas
sif
icatio
n
Main f
o
cu
s
ad
v
an
tag
es
Drawback
s
Po
p
u
larity
[
2
9
]
[
8
]
[36
]
Fo
cu
s o
n
pack
et
selectio
n
bas
ed
on
the
n
u
m
b
er
of
r
eq
u
ests
f
o
r
th
e pack
et.
Its alr
e
ad
y
a
cco
m
m
o
d
ated
the
selectio
n
of
con
ten
t bas
ed
on
co
n
su
m
er
interest
less
po
p
u
lar
co
n
ten
t can b
e o
m
itted
,
wh
ile it
is
still n
eeded
or r
eq
u
ested
by
so
m
e
co
n
su
m
ers
Prob
ab
ility
[
3
]
[
5
]
[
2
3
]
[
3
7
]
E
m
p
h
asize th
e s
ele
ctio
n
o
f
p
ackag
es/co
n
te
n
t
w
ith
a
ce
rtain
p
rob
ab
ility.
A
p
ac
k
et
can b
e ca
ch
ed
with
a
certa
in
pro
b
ab
ility.
More fair
in
deter
m
in
in
g
the
p
ackag
e to b
e ca
ch
ed
or deleted
.
-
So
m
e no
d
es
m
a
y
s
to
re
th
e sa
m
e con
t
en
t
-
Need sp
ecif
ic strat
eg
ies to
deter
m
in
e
th
e
p
rob
ab
ility
Predic
-
tio
n
b
ased
[
3
4
]
[
3
1
]
Tigh
ten
th
e selecti
o
n
of
p
ackag
es b
ased
on
p
redictio
n
s wh
eth
er
th
e
selectio
n
of
con
ten
t will
p
rov
id
e the target
v
alu
e
set.
-
Av
o
id
sto
ring
u
n
n
e
cess
ary
co
n
ten
t
-
Acco
m
m
o
d
ate
the
f
u
tu
re
n
eeds
of
the u
ser
-
The p
redictio
n
m
a
y
be in
co
r
rect
if
th
e
co
n
d
itio
n
o
f
the n
e
two
rk o
r
u
ser chan
g
es.
-
Internal calcu
latio
n
of
the rou
ter
is
m
o
re
co
m
p
lex
.
Anothe
r
te
ch
ni
qu
e
relat
ed
t
o
the
cache
c
on
te
nt
sel
ect
ion
is
Pr
edict
io
n
-
base
d
cachi
ng
[
33]
.
The
con
te
nt
will
be
decide
d
t
o
be
cached
by
r
ou
t
er
based
on
t
he
num
ber
of
requests
.
In
this
s
chem
e,
it
is
add
ed
a
new
ta
ble
in
t
he
r
oute
r,
nam
ed
the
Pe
nd
i
ng
S
pecies
I
nte
rest
Table
(
P
S
IT
)
.
T
his
ta
ble
stores
the
li
st
of
the
m
os
t
req
ueste
d
co
ntent
based
on
data
i
n
the
PI
T
.
S
uppose
there
is
c
onte
nt
that
is
regula
rly
requeste
d
by
the
consum
er
every
Mon
day,
bu
t
there
is
al
so
non
-
re
gula
r
co
nt
ent,
for
exam
ple,
the
co
nte
nt
s
of
the
World
Cu
p
even
t.
Af
te
r
th
at
,
Dynam
ic
C
ache
A
dju
s
tm
e
nt
al
gorithm
is
us
ed
to
decid
e
a
pack
a
ge
th
at
will
be
cached
or
no
t
based
on
it
s
wastage
valu
e.
A
co
ntent
w
il
l
be
viewed
i
n
siz
e.
I
f
the
CS
is
sti
ll
su
ff
ic
ie
nt,
the
pac
ka
ge
is
store
d.
I
f
the
C
S
is
fu
ll
then
t
he
pac
ket
in
th
e
CS
will
be
sel
ect
ran
dom
ly
a
nd
the
n
com
par
e
d
it
with
the
ne
w
data
pack
et
s
.
I
f
they
are
bo
t
h
sam
e,
the
value
of
the
hit
par
am
et
er
wi
ll
increase.
Re
-
te
sti
ng
is
done
by
com
par
ing
the
hit
par
am
et
er
with the am
ou
nt o
f data
that has
bee
n
s
or
te
d.
If the h
it
v
al
ue
is hig
her, th
e p
acke
t
is
gi
ve
n
al
l
ocati
on
i
n
t
he
buffer,
a
nd
oth
e
rw
i
se,
the
co
nte
nt
is
not
al
locat
ed
in
CS
.
S
el
ect
ion
of
a
c
on
te
nt
ca
n
al
so
be
cal
cul
at
ed
base
d
on
local
popula
r
it
y
and
ho
p
c
ount
re
duct
ion
gain
t
hat
ca
n
be
gi
ven
by
the
pack
et
[
29
]
.
Anothe
r
co
nte
nt
sel
ect
ion
te
chn
i
qu
e
is
Ma
x
-
Gain
I
n
-
netw
ork
Ca
chi
ng
(MAGIC)
[
34]
.
The
pro
pose
d
m
et
ho
d
ai
m
s
t
o
reduce
ba
nd
width
c
on
s
umpti
on
a
nd
co
nsi
der
c
onte
nt
popula
rity
as
w
el
l
as
ho
p
re
duct
ion.
Wh
e
n
recei
ving
the
interest
pa
cket,
each
router
will
cal
culat
e
the
Local
Gain
an
d
c
ompare
it
with
th
e
valu
e
store
d
on
t
he
Ma
xG
ai
n
fiel
d.
If
t
he
local
ca
che
of
t
he
r
oute
r
gai
n
is
gr
eat
er
tha
n
the
Ma
xG
ai
n
value,
t
hen
the
router
will
upda
te
the
Ma
xG
a
in
val
ue
in
the
interest
pac
ket.
This
Ma
xGai
n
value
will
be
cop
ie
d
on
a
ddit
ion
al
fiel
ds
in
t
he
da
ta
pack
et
.
Al
ong
t
he
pack
et
delivery
path
,
if
the
L
ocal
Gain
va
lue
is
the
s
a
m
e
as
the
Ma
xG
ai
n
value
i
n
th
e
data p
ac
ket, it
will
b
e cac
he
d
in
the
data p
a
cket.
If
a
data
pac
ke
t
enters
the
r
oute
r
node
a
nd
the
r
oute
r
didn
’t
ha
ve
it
i
n
it
s
Co
ntent
St
or
a
ge,
t
he
n
the
node
will
che
ck
it
s
Co
ntent
Stor
a
ge
c
ondi
ti
on
.
If
it
is
f
ull,
the
n
it
wil
l
be
sel
ect
ed
wh
ic
h
pack
et
will
be
delet
ed
f
ro
m
C
on
te
nt stor
a
ge t
o
pro
vid
e spa
c
e to sto
re th
e ne
w
pac
ket. Tec
hn
i
qu
e
s that ar
e comm
on
ly
u
s
ed
in
the
NDN
syst
e
m
to
sel
ect
wh
ic
h
pac
kets
will
be
delet
e
d
in
CS
is
Least
Re
centl
y
Used
(LRU
)
and
Least
Fr
e
qu
e
ntly
Us
ed
(
LF
U)
[3]
,
[29]
,
[
22]
.
De
ghga
n
et
al
in
the
pa
per
[
2]
pro
posed
an
othe
r
te
ch
nique
to
giv
e
a
tim
er
to
a
package.
The
ti
m
er
is
us
ed
to
determ
ine
ho
w
long
a
pack
et
m
a
y
be
in
the
con
t
ent
stora
ge
befor
e
it
is
finall
y
delet
ed.
Pa
pe
r
[
35
]
pro
po
se
d
the
Re
cent
Usa
ge
Fr
e
qu
e
ncy
(RF
U)
al
gorithm
,
wh
ic
h
dete
rm
i
ned
t
he
popula
rity
of
con
te
nt
within
a
lim
it
ed
time
range.
T
he
lowest
popula
ri
ty
value
will
cau
se
a
co
ntent
to
be
rem
ov
ed fr
om
the content
sto
re.
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
Cachi
ng on N
amed
Data
Ne
tw
or
k:
a S
ur
vey
and
…
(
Lean
na Vi
dya Y
ovita
)
4461
Accor
ding
to
the
pa
per
[
24]
,
the
perform
ance
of
cachi
ng
ca
n
be
im
pr
ov
e
d
by
us
in
g
eff
ic
i
ent
cachin
g
rep
la
cem
ent
m
et
hods
.
In
m
obil
e
network
s
,
this
is
a
c
halle
nge,
because
the
env
i
ronm
ent
i
s
dif
fer
e
nt
fro
m
the
fixe
d
netw
ork
conditi
ons.
Th
e
pa
ram
et
ers
use
d
by
t
he
rep
l
ace
m
ent
r
ule
inclu
de
rece
ncy,
po
pula
rity
,
m
essag
e
siz
e,
cost
t
o
ac
hieve
obj
ect
s
,
and
acce
ss
del
ay
[24]
.
T
he
c
ache
c
on
te
nt
s
el
ect
ion
te
ch
ni
qu
e
s
can
be
re
su
m
e
a
s
in Fig
ure
3
a
nd the c
om
par
iso
n of cac
he
cont
ent sele
ct
ion t
echn
i
qu
e
sas in
Table
2.
Figure
3. Cl
assifi
cat
ion
of cac
he
c
on
te
nt sele
ct
ion
tec
hn
i
qu
e
s
3.5. C
ache Pol
ic
y
De
sign
Ca
che
poli
cy
fo
cu
ses
on
te
ch
niques
ho
w
co
ntent
is
store
d
in
nodes
.
O
ne
of
the
cac
he
poli
cy
relat
ed
te
chn
iq
ues
is Uti
li
ty
-
dr
ive
n
cachin
g
[
8]
.
T
his
te
ch
nique
is
a
util
it
y
-
dr
ive
n
cachi
ng
te
ch
nique
in w
hic
h
a
util
it
y
value
is
li
nk
e
d
to
a
con
te
nt.
U
ti
liti
es
are
a
functi
on
of
a
hit
po
s
sibil
it
y
of
con
te
nt.
Th
e
go
al
is
to
m
axi
m
i
ze
the
total
am
ou
nt of
u
ti
li
ty
co
ntent
in content
sto
r
age.
Pape
r
[
38]
m
o
deled
t
he
cac
he
on
it
s
syst
em
into
2
la
ye
rs
.
The
first
la
ye
r
is
the
in
div
i
dual
cachin
g
i
n
each
node
a
nd
la
ye
r
2
is
the
accum
ulati
on
of
al
l
the
cach
e
on
the
netw
ork.
T
he
stu
dy
analy
zes
how
m
uch
stora
ge
c
on
te
nt
shou
l
d
be
pro
vid
e
d
in
the
sy
stem
to
m
eet
t
he
perform
ance
of
4
ap
plica
ti
on
s
,
i.e.
we
b
t
r
aff
ic
,
file
sh
arin
g,
a
nd
vid
e
o
traf
f
ic
that
are
disti
nguish
e
d
in
to
us
er
-
ge
ne
ra
te
d
con
te
nt
(
UG
C
)
an
d
vi
deo
o
n
dem
and
(V
oD)
.
Assa
ntachai
et
al
[14]
pro
posed
a
hybri
d
c
achin
g
schem
e.
If
a
ny
new
c
on
te
nt
is
reque
ste
d
by
the
consum
er
an
d
no
t
exist
ye
t
on
the
router
no
de,
the
n
the
ne
w
co
ntent
will
be
save
d.
The
con
te
nt
rep
la
ce
m
ent
schem
e
us
ed
is
a
com
bin
at
ion
of
t
he
co
ncep
t
of
a
c
oope
rati
ve
ap
proac
h
an
d
distribu
ti
ve
ap
proac
h.
Cooperati
ve
ca
chin
g
is
a
sche
m
e
in
wh
ic
h
e
ach
node
m
akes
a
rep
la
cem
ent
decisi
on
ba
se
d
on
the
knowl
edg
e
receive
d
f
r
om
oth
e
r
nodes
r
esi
ding
in
t
he
sam
e
reg
ion.
Distrib
utive
cachin
g
is
us
e
d
to
m
ake
de
ci
sion
s
ind
e
pende
ntly
us
in
g
inter
nal
knowle
dge
to
achieve
local
m
axim
u
m
per
f
orm
ance.
In
pa
pe
r
[14]
the
netw
ork
is
div
ide
d
int
o
2
par
ts,
t
hat
is
th
e
norm
al
reg
io
n
(
reg
i
on
on
th
e
edg
e
)
a
nd
t
he
backb
on
e
re
gion
(t
he
re
gion
that
connects
t
he
norm
al
reg
io
ns
)
.
In
the
norm
al
reg
i
on,
if
t
her
e
is
a
cache
hit
i
nterest,
t
he
c
onte
nt
is
m
ov
ed
t
o
th
e
fron
t
of
the
se
qu
e
nce,
a
nd
w
hen
t
he
cache
m
isses
then
th
e
data
at
the
ta
il
of
the
seq
ue
nce
is
rem
ov
e
d.
T
he
backb
on
e
re
gi
on
f
ollo
ws
the
norm
al
reg
io
n
patte
r
n,
only
the
backb
on
e
node
s
work
wit
h
oth
e
r
nod
es
in
the
sam
e
reg
ion
to
deci
de
to
cac
he
.
Co
operati
ve
cachin
g
poli
cy
desig
n
is
al
so use
d
in
[
39]
wi
th
area
s d
ivide
d
i
nto
cl
us
te
rs
Pape
r
[
15
]
,
[40]
descr
i
be
d
that
the
m
echan
ism
to
cache
a
con
te
nt
has
a
cru
ci
al
im
p
act
on
th
e
eff
ic
ie
ncy
of
c
on
te
nt
delive
ry
and
util
iz
at
ion
of
CS.
Pape
r
[
9]
pro
poses
the
m
echan
is
m
to
div
i
de
file
s
int
o
sm
a
ll
er
pack
et
s
cal
le
d
c
hunk.
The
am
ou
nt
of
c
hunk
disse
m
inate
d
de
pe
nds
on
the
po
pula
rity
of
t
he
c
onte
nt.
The
num
ber
of
ch
unks
is
dete
rm
ined
by
the
Chun
k
Ma
r
king
W
i
ndow
(C
M
W
)
w
hich
e
xponentia
ll
y
enlar
ges
ever
y
num
ber
o
f
chu
nk
s
s
ucc
essfu
ll
y
deliver
ed
In
[
41
]
C
on
te
nt
-
Ce
nt
ric
net
work
is
im
pl
e
m
ented
u
sin
g
tw
o
ty
pe
s
of
ap
plica
ti
ons.
F
or
eac
h
app
li
cat
io
n,
it
is
create
d
a
se
par
at
e
li
st
a
nd
each
ide
ntifie
d
with
a
uniq
ue
ID.
T
he
C
S
is
sepa
rated
an
d
each
app
li
cat
io
n
ca
n o
nly be
st
or
e
d i
n
it
s ow
n
c
onte
nt stor
e
. T
he st
or
a
ge
c
on
te
nt p
arti
ti
on m
echan
ism
is test
e
d
with
two
m
et
ho
ds
:
sta
ti
c
cache
par
ti
ti
on
in
g
an
d
dynam
ic
cache
par
ti
ti
on
in
g.
I
n
sta
ti
c
par
ti
tio
ni
ng,
the
cac
he
can
on
ly
be
us
e
d
a
s
sp
eci
fied
. W
hile
in
dynam
i
c
cache
pa
rtit
ion
in
g,
u
nuse
d
c
ache
by
a
n
ap
pl
ic
at
ion
can
be
sh
are
d
with
oth
e
r
a
ppli
cat
ion
s.
Ca
c
he
with
sp
li
tt
ing
te
chn
i
qu
e
al
so
pro
posed
in
[
42
]
.
Th
e
co
nte
nt
sto
rag
e
is
div
ide
d
into tw
o part,
one
par
t
for
a
po
pu
la
r
c
on
te
nt a
nd the
oth
e
r for less
popula
r
c
on
te
nt.
Pa
pe
r
[
43
]
s
plit
the c
onte
nt
stora
ge
i
nto
t
hree
r
egi
on
s
.
T
he
data
is
cat
e
gories
as
a
sel
f
-
data
,
fr
ie
nd
s
data,
a
nd
stra
nger
data.
Pa
pe
r
[
16]
m
or
e
sp
e
ci
fic
on
cachi
ng
m
a
nag
em
ent
in
m
e
m
or
y
wh
e
r
e
m
ulti
pr
ocess
or
is
use
d
wit
h
ce
rtai
n
inte
r
connect
m
echan
ism
s to
re
du
ce
po
wer
us
a
ge.
Ca
chin
g
te
chn
i
qu
e
s
that
coupl
ing
data
cach
e
placem
ent,
rep
la
ce
m
ent,
and
locat
ion
wa
s
pro
posed
by
Xiaoyan
H
u,
e
t
al
.
[29]
.
To
s
et
the
pac
ket
to
be
ca
che
d,
it
is
def
ine
d
a
c
achin
g
val
ue
f
or
eac
h
pack
et
o
that
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4456
-
4466
4462
can
be
cache
d
at
no
de
v.
This
caching
value
inv
ol
ves
m
ult
i
plica
ti
on
of
local
po
pula
rity
value
an
d
ho
p
coun
t
reducti
on
gain
of
t
he
it
em
,
then
div
ide
d
by
cache
s
pace
c
onte
ntio
n
wh
ic
h
is
the
sam
e
va
lue
in
al
l
r
oute
rs.
If
an
interest
goe
s
to
node
v,
a
nd
the
it
em
wasnot
cached
y
et
on
the
node
v
befor
e
,
node
v
will
cal
cul
at
e
the
appr
ox
im
at
e
po
te
ntial
val
ue
of
the
c
achi
ng
con
t
rib
ution
of
the
it
em
.
The
data
will
be
ca
ched
at
node
v
if
th
e
m
axi
m
al
value
of
cachin
g
co
ntributi
on
is
po
sit
ive.
If
the
con
te
nt
stora
ge
is
fu
ll
,
it
will
sel
ect
a
pack
age
wit
h
the
le
ast
co
ntr
ibu
ti
on
cac
hing
value
to
del
et
e.
Re
la
te
d
to
the
cac
hi
ng
locat
ion
dete
rm
ina
ti
on
,
the
cache
locat
ion
com
po
ne
nt
will
m
a
i
ntain
t
he
trai
l
t
o
gu
i
de
t
he
c
onte
nt.
This
trai
l
is
only
create
d
if
the
co
nten
t
is
not
cached
on
t
he
local
node
.
T
he
cache
poli
cy
desi
gn
can
be
res
um
e
as
in
Fig
ure
4.
The
com
par
ison
of
cache
po
li
cy
d
e
sig
n
t
echn
i
qu
e
s as i
n Table
3.
Figure
4. Cl
assifi
cat
ion
of cac
he po
li
cy
d
e
sig
n
Table
3.
C
om
par
iso
n of cac
he
poli
cy
d
esi
gn t
echn
i
qu
e
s
Clas
sif
icatio
n
Main f
o
cu
s
Ad
v
an
tag
es
Drawback
s
Co
o
p
erative
[
9
]
[
2
9
]
[44
]
[
1
7
]
Co
o
rdin
ate and
col
lab
o
rate
with
o
th
er
n
o
d
es in
the n
etwo
rk
to
d
eter
m
in
e
cachin
g
p
o
licy
More eff
icien
t in u
sin
g
reso
u
rces
-
Need ad
d
itio
n
al
m
echan
is
m
s
to b
e
ab
le to
m
o
n
ito
r
an
d
sh
are
in
f
o
r
m
atio
n
betwe
en
no
d
es
-
Fo
r
large net
wo
rks
,
th
is
m
e
ch
an
is
m
can b
e
v
ery
ef
f
icien
t
Ind
ep
en
d
en
t
[
8
]
[
3
1
]
[42
]
[
3
8
]
[
4
5
]
[
4
6
]
Cach
in
g
-
related de
cisio
n
s are
p
erfo
r
m
ed
by
th
e n
o
d
e r
eg
ardless
of
in
f
o
r
m
atio
n
f
ro
m
o
th
er
n
o
d
es
-
There
is n
o
need
f
o
r
ad
d
itio
n
al
m
e
ch
an
is
m
s
f
o
r
m
o
n
ito
ring
and
sh
ar
in
g
inf
o
r
m
atio
n
with
o
th
er
n
o
d
es
-
Can
no
t do
r
eso
u
rce
sh
aring
Hy
b
rid
[
1
4
]
Mer
g
in
g
between
co
o
p
erative and
in
d
ep
en
d
en
t techn
i
q
u
es.
-
C
an
m
o
re
ef
f
icien
tly
ap
p
ly
cer
t
ain
m
e
ch
an
is
m
s
to sp
ecif
ic con
d
itio
n
s
-
Need to
be d
ef
in
ed
abo
u
t the
sp
ecif
ic con
d
itio
n
s
f
o
r
a
m
e
ch
an
is
m
-
Ad
d
the co
m
p
u
tati
o
n
pro
cess
3.6 Cac
hing
fo
r
The
mo
bil
e
Node
Gen
e
rall
y,
cac
hing
te
ch
ni
qu
e
s
f
or
m
ob
il
e
node
s
hav
e
a
ba
sic
idea
for
s
ubscri
bing
a
us
e
r
to
a
c
on
te
nt
pro
du
ce
r
[28]
,
[47]
,
pr
e
fetchi
ng
co
nte
nt
to
oth
e
r
r
oute
r
t
ha
t
will
ha
nd
le
consum
er
[
19
]
,
[
20
]
,
colla
bor
at
e
th
e
data
transm
issio
n
m
od
e
f
or
V
AN
E
T
[
44
]
,
a
nd
m
ob
il
e
node
su
pp
or
t
te
ch
niques
that
co
ns
ide
r
an
en
er
gy
[48]
.
In
t
he
m
ob
il
e
env
i
ronm
ent,
the
pr
ob
le
m
s
are
N
D
N
no
des
al
ways
m
ov
e,
inclu
ding
r
ou
te
rs,
pr
oduce
rs
,
an
d
consum
ers
[
21]
.
The
pro
du
c
e
r
m
ov
em
ent
causes
a
great
er
pro
blem
than
the
m
ov
em
ent
of
t
he
c
onsu
m
er
no
de
or rou
te
r node.
Prob
le
m
-
so
lvi
ng
relat
ed
to p
r
oduce
r
m
ov
em
ents is
pr
ese
nte
d by pa
pe
r
[
28]
.
The
publish
/s
ubscri
be
syst
em
is
the
m
echani
s
m
by
w
hich
t
he
s
ubscri
ber
c
an
receive
m
es
sages
from
the
publishe
r.
This
relat
ionsh
ip
is
governed
by
the
m
anager
so
a
use
r
w
ho
subscri
bes
to
certai
n
con
te
nt
will
al
ways
get
th
e
co
ntent
t
he
y
wan
t
w
he
n
publis
her
ge
ner
at
es
the
c
on
te
nt
[
47]
.
I
n
t
he
pr
e
-
exi
sti
ng
pub/
s
ub
syst
em
,
the
pro
du
ce
r
do
e
s
not
st
or
e
m
essages
that
ha
ve
been
pu
blishe
d
befor
e
.
In
this
case
,
i
f
ne
w
su
bsc
ri
ber
s
j
oi
n
the
syst
em
,
they
cou
l
d
not
get
the
co
nte
nt
that
has
be
en
pu
blished
befor
e
t
hey
ent
er
the
syst
e
m
.
To
so
l
ve
the proble
m
,
[
28
]
pro
po
se
d
stora
ge
m
echan
ism
and
re
plica
ti
on
al
go
rith
m
with
diff
ere
ntiat
ed
con
te
nt
cl
ass.
I
n
this
ne
w
syst
e
m
,
st
or
a
ge
ca
n
c
onver
t
t
he
c
on
te
nt
cl
asses
t
hey
sto
re.
T
he
pro
po
se
d
rep
li
cat
ion
al
gorithm
is
to
sel
ect
M
stora
ge
po
i
nts
f
ro
m
N
po
i
nts
that
are
av
ai
la
ble
in
the
net
work
base
d
on
l
ocali
ty
and
popula
rity
, tar
get r
e
plica
ti
on
degree
of eac
h t
op
ic
, a
nd st
or
a
ge
ca
paci
ty
A
te
chn
i
que
f
or
acc
omm
od
at
ing
co
nsum
er
m
ob
il
it
y
in
wireless
netw
orks
is
Proact
iv
e
Mult
il
evel
Ca
che
Sele
ct
io
n
(
PMC
S
),
pro
po
s
ed
by
pa
pe
r
[
18
]
.
In
t
his
schem
e,
if
the
consum
er
will
switc
h
c
ov
e
rage
or
hand
off,
the
c
on
s
um
er
will
send
a
noti
ficat
ion
a
bout
w
hi
ch
r
ou
te
r
to
go
to.
T
he
c
urr
ently
us
ed
rou
te
r
will
sel
ect
a
su
bs
et
of
nei
ghbori
ng
ro
ute
rs
to
rece
ive
con
te
nt
that
has
been
re
qu
est
ed
by
the
co
ns
um
er
bu
t
has
no
t
ye
t
been
se
nt
to
it
.
Wh
e
n
a
hand
off
occurs
,
the
c
onsu
m
er
will
sto
p
requesti
ng
to
sen
d
data.
Durin
g
this
hand
off
pr
oces
s
al
so
,
the
des
ti
nation
r
oute
r
will
cache
the
data
pac
kets
f
ro
m
the
old
r
oute
r,
w
hich
ha
s
no
t
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
Cachi
ng on N
amed
Data
Ne
tw
or
k:
a S
ur
vey
and
…
(
Lean
na Vi
dya Y
ovita
)
4463
been
recei
ve
d
by
t
he
c
onsum
er
up
t
o
a
c
ertai
n
li
m
i
t.
O
nce
t
he
c
onne
ct
ion
t
o
the
ne
w
router
has
bee
n
est
ablished
,
th
en
the
data
tra
ns
m
issi
on
will
be
se
rved
by
the
ne
w
router
.
A
nother
te
ch
nique
is
pro
posed
by
[19]
to
predict
node
m
ob
il
it
y and p
r
ov
i
de
th
e b
est
pr
e
fetchi
ng no
de.
Pape
r
[
20
]
ex
pl
ai
n
the
m
echa
nism
to
su
pp
ort
pr
od
ucer
m
obil
ity,
su
ch
as
push
t
o
sen
d
the
data,
m
ake
so
m
e
cop
ie
s
of
data,
deter
m
ine
the
co
ntent
placem
ent,
and
re
-
an
nounce
if
they
m
ov
e
t
o
a
n
ot
he
r
area
.
Pape
r
[44]
propose
V
A
NE
T'
s
co
m
m
un
i
c
at
ion
m
od
e
switc
hing,
Ve
hi
cl
e
-
to
-
veh
ic
le
(V2V)
an
d
ve
hicle
-
to
-
infr
a
struct
ur
e
(
V2I),
de
pe
nd
i
ng
on
the
popul
arit
y
of
do
wn
l
oad
e
d
c
on
te
nt.
Mob
il
e
no
de
ha
s
the
lim
it
ed
powe
r,
so
the cac
hing p
r
ocess has to
consi
der
t
he
en
erg
y co
nsum
ption
in
the
node
, d
ue
t
o
gre
en ND
N
as ex
plained
i
n
[49]
.
Pa
per
[
48]
pro
po
se
d
a
n
energy
eff
ic
ie
nt
te
chn
i
qu
es
f
or
MA
NET
.
T
he
net
work
is
div
ide
d
int
o
gr
oups
,
m
anag
ed
by
a
Ma
ste
r
Node
.
Pape
r
[
50
]
pr
opos
e
d
a
te
ch
nique
with
op
tim
a
l
sel
ect
ion
of
cl
us
te
r
he
ad
in
W
i
reless
Senso
r
N
et
w
ork
to
im
pr
ov
e e
ff
ic
ie
ncy.
4.
CHALL
ENG
E AND
OPEN
ISSUES
4.1.
Q
oS
-
b
as
e
d
C
achin
g
In
al
l
cachi
ng
te
chn
iq
ues
,
ei
t
her
cache
place
m
ent,
cache
c
on
te
nt
sel
ect
io
n,
or
cache
po
li
cy
design
that
has
bee
n
dev
el
op
e
d
m
os
tl
y
hav
e
not
c
onside
red
the
diff
e
ren
t
treat
m
e
nt
f
or
dif
fer
e
nt
serv
ic
es
.
I
n
st
ud
ie
s
that
hav
e
bee
n
done,
t
he
dat
a
us
ually
only
diff
e
re
ntiat
ed
base
d
on
co
nt
ent
popula
rity
,
con
te
nt
rece
ntly
,
the
est
i
m
at
ed
ben
e
fits
of
c
on
te
nt
stora
ge,
et
c.
T
her
e
a
re
only
a
few
of
st
ud
ie
s
t
hat
ta
ke
into
a
ccount
the
trea
t
m
ent
diff
e
re
ntiat
ion
base
d
on
serv
i
ce
requirem
ents
or
us
e
r
re
qu
i
rem
ents.
In
fac
t,
diff
e
r
ent
us
e
rs
m
a
y
su
bs
cri
be
to
diff
e
re
nt
pr
i
vileged
se
r
vices. So
fa
r,
n
ot
m
uch
resea
rch
h
as
been
d
one
rela
te
d
to
Q
oS
-
bas
ed
cachi
ng
o
n
NDN.
Pape
r
[
45]
is o
ne of
the
pa
pers that
discuss
this
dis
ti
nction
us
in
g
cl
asses
.
The
c
oncept
f
ol
lows
t
he
Diff
e
ren
ti
at
ed
Ser
vi
ce
(
DiffSer
v)
c
on
ce
pt
t
hat
wa
s
previ
ously
use
d
i
n
the
I
P
netw
ork.
F
urt
he
r
de
velo
pm
ent
is
need
ed
for
cachin
g
m
echan
ism
s
that
ca
n
m
eet
diff
ere
nt
requirem
ents
for
serv
ic
es
an
d
use
rs.
T
hese
t
ech
niques
i
nclu
de
how
t
o
c
hoos
e
co
ntent
a
nd
w
her
e
tocac
he
th
e
m
in
the
netw
ork
.
The deci
sio
n
c
an be take
n
i
nd
epende
ntly
o
r
coope
rati
vely
w
it
h othe
r nod
es in t
he netw
ork
.
4.2. C
achin
g f
or Mo
bil
e No
de
Node
m
ob
il
it
y
m
us
t
be
sup
ported
to
pro
vi
de
the
flexi
bili
t
y
of
the
syst
e
m
.
Gen
e
rall
y,
m
ob
il
it
y
char
act
e
risti
c
i
s
div
ide
d
into
pr
od
ucer
m
ob
il
it
y
dan
consum
er
m
ob
il
ity.
Rou
te
r
m
ob
il
ity
is
si
m
il
ar
to
the
consum
er
m
ob
il
ity.
Consum
er
m
ob
il
it
y
is
na
turall
y
sup
por
te
d
by
N
D
N,
bu
t
not
s
o
with
pro
du
ce
r
m
ob
il
it
y.
So
,
t
he
ar
ea
of
the
pro
ducer
-
m
ob
il
i
ty
su
pport
te
chn
i
qu
e
is
on
e
of
resea
rc
h
op
portu
niti
es.
Seve
ral
te
ch
ni
qu
es
are
prese
nted
r
el
at
ed
to
cach
e
in
the
m
ob
il
e
node
t
o
s
upport
produce
r
m
ob
il
i
ty
.
Fo
r
Exam
ple,
in
th
e
paper
[18]
pre
-
fetchi
ng
c
onte
nt
is
pro
po
se
d.
Thi
s
schem
e
was
done
wh
e
n
t
he
m
ob
il
e
node
m
ov
es
to
t
he
new
cov
e
ra
ge
r
oute
r.
Anothe
r
pro
po
s
ed
m
et
hod
is
to
pre
fetch
a
gro
up
of
c
onte
nt,
not
just
1
c
onte
nt,
w
hi
ch
is
us
ua
ll
y
request
ed
by
the
c
ons
um
er
[31]
.
Pre
-
fetchi
ng
cau
se
s
ad
diti
on
al
ti
m
e
need
e
d
to
m
ov
e
con
te
nt
to
a
new
router.
F
ur
t
her
inv
e
sti
gation
of
ot
her
te
c
hn
iqu
es
relat
ed
to
node
m
ob
il
it
y
su
pp
or
t
for
N
DN
is
re
quired
to
ens
ur
e
un
i
nterrup
te
d
data
c
om
m
un
ic
at
ion
se
ven
if
t
he
us
e
r
s
witc
hes
c
ov
erag
e
by
co
nsi
der
in
g
t
he
e
xpect
ed
delay
, cac
he
lo
ad,
a
nd the
com
plexit
y of
t
he
algorit
hm
that m
us
t be ex
ec
uted.
4.3.
En
erg
y
-
aw
are
C
achin
g
NDN
r
oute
rs
i
n
the
m
ob
il
e
w
irel
ess
netw
ork
will
ha
ve
po
wer
restrict
ions.
Ca
chin
g
te
c
hn
i
qu
e
s
that
consi
der
the
a
vaila
bili
ty
of
powe
r
on
the
node
al
so
nee
d
to
be
e
xplo
red
furthe
r.
T
his
process
m
ay
i
nclu
de
sel
ect
ing
node
s
to
place
co
ntent
base
d
on
po
sit
io
n,
di
sta
nce,
e
nergy
avail
abili
ty
at
the
no
de,
re
so
urce
avail
abili
ty
and
ot
her
im
po
rt
ant
thin
gs
that
sh
oul
d
co
ns
ide
r
proces
s
ef
fici
ency.
Co
ver
i
ng
a
te
chn
iq
ue
th
at
can
reduce
the
nu
m
ber
of
re
plac
e
m
ents
that
oc
cur.
If
a
c
onte
nt
is
too
fr
e
quent
ly
rem
ov
ed
from
the
cache,
it
wil
l
no
t
be
e
ff
ic
ie
nt.
4.4
.
T
yp
e
of
D
ata on C
onten
t
S
to
re
Currentl
y,
the
cached
c
on
t
ent
on
t
he
N
DN
router
ca
n
be
ei
ther
a
file
or
sm
aller,
cal
le
d
chun
k
[
9]
[44]
.
Chunk
-
ba
sed
syst
e
m
s
will
m
ake
the
transm
issi
on
process
m
or
e
eff
ic
ie
nt
because
if
a
ch
unk
is
lost
durin
g
tra
ns
m
issi
on
or
it
is
delet
ed
in
CS,
it
on
ly
need
s
to
be
re
plac
ed
with
a
ne
w
chun
k
with
ou
t
hav
i
ng
to
re
place
the
w
ho
le
file
.
Howe
ve
r
,
the
div
isi
on
of
th
e
file
into
ch
unk
ca
us
es
th
e
us
e
r'
s
qu
e
rie
s
to
be
gen
e
rated
ch
unk
-
based.
T
his
m
eans
that
in
the
c
hunk
-
bas
ed
syst
em
,
the
interest
pac
ke
t
for
a
com
plete
file
m
or
e
than
the
file
-
base
d
syst
e
m
.
Fu
rthe
r
ex
plorat
ion
of
ca
chin
g
proce
dur
es
and
m
echan
is
m
s
reg
ard
i
ng
this
form
o
f data
shou
l
d be e
xp
l
ored.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4456
-
4466
4464
5.
CONCL
US
I
O
N
In
t
his
pa
pe
r,
we
hav
e
ex
plica
te
the
ad
van
ta
ges
of
NDN
netw
ork
arch
it
ect
ure
com
par
ed
t
o
tradit
ion
al
I
P
netw
ork
an
d
Con
te
nt
Distri
bu
ti
on
Netw
ork,
a
nd
e
xcess
cachin
g
on
NDN
c
om
par
ed
to
it
s
pr
e
decess
or
syst
e
m
.
The
de
ve
lop
m
ent
of
va
rio
us
cac
hing
te
chn
i
qu
es
ha
s
bee
n
m
app
ed
o
ut.
I
n
this
pa
per
al
s
o
exp
la
ine
d
t
he
adv
a
ntage
s
a
nd
draw
bac
ks
of
eac
h
gro
up.
Finall
y,
it
has
be
en
s
ugge
ste
d
the
r
esearch
opport
un
it
ie
s
r
el
at
ed
to
cachi
ng
on
N
DN
t
ha
t
can
be
in
ves
ti
gated
in
the
f
uture,
i.e.
cac
hi
ng
m
echan
ism
s
that
involve
diff
e
r
ences
in
QoS
r
eq
uirem
ents
for
data
a
nd
us
ers
,
cachi
ng
that
sup
port
s
m
ob
il
it
y
no
des,
a
nd
cachin
g
th
at
conside
rs
e
nergy.
ACKN
OWLE
DGME
NT
This
w
ork
was
su
pp
or
te
d
by
Tel
ko
m
Un
ive
rsity
and
Mi
ni
stry
of
Re
sea
r
ch
Tec
hnology
and
Highe
r
Ed
ucati
on Rep
ub
li
c
of Indo
ne
sia
.
REFERE
NCE
S
[1]
T.
M.
Chen
,
“
Fr
om
Circ
ui
t
Sw
it
che
d
to
IP
-
bas
ed
Networks
”
,
i
n
Enc
yclope
dia
of
Mult
imedi
a
Technol
ogy
and
Net
working
,
2nd
ed.,
M
.
Pag
an,
Ed.
Id
ea Group
Publishing,
2009
.
[2]
B.
G.
Pal
li
s
and
A.
Vaka
l
i,
“
Insight
and
Perspec
t
ive
s
for
Con
te
nt
Deli
v
er
y
Ne
tworks
”
,
vol
.
49
,
no
.
1,
pp.
101
–
106
,
2006.
[3]
V.
Jac
obson,
D.
K.
Sm
et
te
rs,
N.
H.
Briggs,
J.D.
Thornt
on,
M.F.
Plass,
and
R.
L
.
Bra
y
n
ard
,
“
Net
working
Nam
ed
Conte
nt”
,
in
The
5th
ACM
In
te
rn
ati
onal
Con
fe
ren
ce
on
eme
rging
Net
working
Ex
p
erime
nts
and
Te
chnol
ogi
es
,
200
9,
pp.
1
–
12
.
[4]
L.
V.
Yovita
an
d
N.R.
S
y
ambas,
“
Conte
nt
Sto
rag
e
Eff
e
ct
on
t
he
Nam
ed
Da
ta
Network
Tra
ffi
c
Lo
ad
”
,
in
Th
e
Inte
rnational
Co
nfe
renc
e
on
Telecomm
unic
ati
on
Syste
ms
,
S
erv
ice
s,
and
Appl
i
cat
io
ns 2017
,
2017
.
[5]
H.
Jin,
D.
Xu,
C.
Zha
o,
and
D
.
Li
ang
,
“
Inform
at
ion
-
ce
n
tr
ic
m
obil
e
ca
ch
ing
net
work
fra
m
eworks
and
ca
chi
n
g
opti
m
iz
ation :
a
surve
y
”
,
EURA
S
IP
J
.
W
irel.
Com
mun.
Net
w
.
,
vol
.
33,
pp
.
1
–
32
,
20
17.
[6]
Anjal
i
,
“
Com
ponent
s of
N
amed D
at
a
Ne
tworkin
g”
,
In
t. J.
Inno
v.
Eng.
Te
chnol.
,
v
ol.
7
,
no
.
3
,
pp
.
5
43
–
552,
2016
.
[7]
Y.
Jing,
“
Ev
al
ua
ti
ng
Cac
h
ing
Mec
han
ism
s
In
Fu
ture
Inte
rn
et
Ar
chi
t
ec
tur
es”
,
Ca
m
bridge
,
US
A,
MIT
-
CS
AIL
-
TR
-
2016
-
009,
2016
.
[8]
M.
Dehgha
n,
L
.
Mass
ouli
e
,
D.
Towsle
y
,
D.
Mena
sche
,
and
Y.C.
T
a
y
,
“
A
Util
ity
Opt
imiz
at
ion
A
pproa
ch
to
Network
Cac
h
e Design
”
,
in
I
EEE
INFOCOM 20
16
,
2016
.
[9]
K.
Cho,
M.
Lee,
K.
Park,
T.T.
Kw
on,
and
Y.
Choi,
“
W
AV
E :
Popularit
y
-
bas
ed
and
Co
ll
abor
at
iv
e
In
-
ne
twork
Cac
hing
for
Conte
nt
-
Ori
ent
ed
Networks”
,
in
2012
IEE
E
Conf
ere
nce
on
Computer
Comm
unic
ati
ons
Workshops
,
2012,
pp
.
316
–
3
21.
[10]
D.
De
Voluce
a
u,
L.
C
.
C,
and
F.
Sail
han
,
“
Ene
rg
y
-
awa
r
e
W
eb
Cac
hing
for
Mobile
Te
rm
in
a
ls
”
,
in
The
22
nd
Inte
rnational
Co
nfe
renc
e
on
Dist
ribute
d
Comput
i
ng
Syste
ms
Wor
kshops
,
2002,
pp
.
820
–
825
.
[11]
C.
M.I.
N.
Park
and
R.
A.
Reh
m
an,
“
Packe
t
Flooding
Miti
ga
t
ion
in
CCN
-
base
d
W
ire
le
ss
Multi
m
edi
a
S
en
so
r
Networks
for
S
m
art
Cit
ie
s
Pack
et
Flooding
Miti
gat
ion
in
CCN
-
Based
W
ire
le
ss
Multi
m
edi
a
S
en
sor
Networks
fo
r
Sm
art
Cit
i
es
”
,
I
EE
E
Acce
ss
,
vol
.
5
,
no
.
June
,
pp
.
11054
–
11062,
2
017.
[12]
J.H.
Mun
and
H.
Li
m
,
“
Cac
h
e
Sharing
Us
ing
Blo
om
Filt
er
s
in
Na
m
ed
Data
Netw
o
rking
”
,
J.
Net
w
.
Comput.
Appl.
,
vol.
90
,
pp
.
74
–
8
2,
2017
.
[13]
A.
Hida
y
at,
F.
Bukhari
,
and
H.
Sukoco,
“
Bloom
fil
te
r
implem
ent
a
ti
on
in
cach
e
w
it
h
low
l
eve
l
of
fal
se
posit
iv
e
”
,
Telk
omnika
(
Tele
communic
at
ion
Comput.
Elec
tr
on.
Control
.
,
vol
.
15
,
no
.
4
,
pp
.
1
911
–
191
6,
2017
.
[14]
K.
As
santa
chai,
K.
Kane
ko,
an
d
F.
Te
rao
k
a,
“
A
Hy
br
id
Ca
ch
e
Mana
gement
Scheme
for
Inform
at
ion
Cent
r
i
c
Networki
ng
”
,
in
Instit
ute of
Elec
t
ronics,
Informati
on
and
Comm
unic
ati
on
Eng
ine
er
s
,
2015,
pp.
4
–
9.
[15]
S.
Shail
endr
a
,
S.
Sengott
uve
la
n
,
H.K.
Rat
h,
B.
P
ani
gra
hi
,
and
A.
Sim
ha,
“
Perf
orm
anc
e
Evalua
ti
on
of
Cac
hing
Polic
i
es
in
ND
N
-
an
ICN Arch
itect
ur
e
”
,
in
IE
EE R
eg
ion
10
Conf
ere
nce (
TENCO
N)
,
2016,
pp
.
11
17
–
1121.
[16]
M.H.
Aw
ada
l
la
and
A.
Sad
ek,
“
An
eff
icient
c
ache
orga
n
iz
a
ti
on
f
or
on
-
chi
p
m
ult
i
proc
essor
net
wo
rks
”
,
Int
.
J.
Elec
t
r.
Comput.
Eng
.
,
v
ol.
5
,
no
.
3
,
pp
.
5
03
–
517,
2015
.
[17]
P.
Chara
n,
T.
Us
m
ani
,
R.
Paulus,
and
S.H.
Saee
d
,
“
A Coope
rat
ive
Cac
he
Man
age
m
ent
Scheme
for
I
EE
E802
.
15
.
4
base
d
W
ireless
Sensor Net
works
”
,
Int
.
J. Ele
c
tr.
Comput.
Eng.
,
v
ol.
8
,
no
.
3
,
pp
.
1
701
–
1710,
2018
.
[18]
L.
Rui,
S.
Yang,
and
H.
Huang,
“
A
proa
ct
ive
m
ult
i
-
l
evel
ca
ch
e
sele
c
ti
on
sche
m
e
to
enha
nce
co
nsum
er
m
obil
i
t
y
s
upport
in
Nam
e
d
Data Net
worki
ng
”
,
In
t. J.
Distr
ib.
S
ens.
N
et
wor
ks
,
vol
.
13
,
no
.
1
0,
2017
.
[19]
N.
Abani
and
M.
Gerl
a
,
“
Proac
ti
v
e
Cac
h
i
ng
with
Mobili
t
y
P
red
iction
unde
r
Unce
rtaint
y
in
I
nform
at
ion
-
ce
n
tr
ic
Networks”
,
in
4t
h
ACM
Con
fe
re
nce
on
Informati
on
-
Cent
ric
Net
w
orking
(
ICN
2017)
,
2017,
pp
.
88
–
97.
[20]
M.B.
Le
hm
ann,
M.P.
Barc
e
ll
os,
and
A.
Mauthe
,
“
Providing
Prod
uce
r
Mobilit
y
S
upport
in
ND
N
Th
r
ough
Proac
tive
Data
Replic
at
ion
”
,
in
Ne
twork
Operations
and
Manage
ment
Sym
posium
(
N
OMS
2016)
,
2016,
no.
Nom
s,
pp.
383
–
391.
[21]
A.
Azgin,
R
.
Ravi
ndra
n
,
and
G.
W
ang,
“
M
obil
ity
Stud
y
f
or
Nam
ed
Data
Networki
ng
in
W
ire
le
ss
Acc
e
ss
Networks”
,
in
N
ex
t
-
Gen
erati
on
Net
w
orking
Sym
posium
,
2014,
p
p.
3252
–
3257
.
[22]
S.
Podlipni
g
and
L.
B.
Os
z,
“
A
Surve
y
of
W
eb
C
ac
he
R
eplac
eme
nt
Strategie
s
”
,
A
CM
Comput.
Su
rv.
,
vo
l.
35
,
no.
4,
pp.
374
–
398
,
20
03.
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
Cachi
ng on N
amed
Data
Ne
tw
or
k:
a S
ur
vey
and
…
(
Lean
na Vi
dya Y
ovita
)
4465
[23]
M.
Zha
ng,
H
.
L
uo,
and
H.
Zh
an
g,
“
A
Surve
y
of
Cac
hing
Me
chanism
s
in
Inform
at
ion
ce
ntr
ic
N
e
tworking”
,
I
EEE
Comm
un.
Surv.
TU
TO
RIA
LS
,
vol.
17
,
no
.
3
,
pp
.
1
473
–
1499,
2015
.
[24]
E.
Hattab
and
S.
Kawasm
ah,
“
A
Surve
y
of
Repl
a
ce
m
ent
Polic
i
es
for
Mobile
W
eb
Cac
hing
”
,
in
Inte
rnatio
nal
Confe
renc
e
on
Dev
el
opmen
ts of E
-
Syst
ems E
ngi
nee
ring
,
2015.
[25]
D.
Saxen
a
and
I
.
I.
T. Roorkee, “
Nam
ed
Dat
a
Netw
orking:
A Surve
y
”
,
Comput
.
S
ci.
Rev.
,
vo
l. 19, pp
.
15
–
55
,
2016
.
[26]
H.
Situm
ora
ng,
N.R.
S
y
ambas,
a
nd
T.
Juhana
,
“
The
Eff
e
ct
of
Scal
ing
the
Siz
e
of
Topol
og
y
and
Conte
nt
Stor
ed
on
the
Nam
ed
Da
t
a
Networki
ng
”
,
in
Int
ernati
on
al
Confe
ren
ce
on
Telecomm
unic
ati
on
Syst
ems,
Serv
ic
es
,
and
Appl
ic
a
ti
ons 201
6
,
2016
,
pp
.
16
–
21.
[27]
G.
Zha
ng,
B
.
T
ang,
P.
W
ang,
Y.
W
u,
and
X.
Zha
ng,
“
Perfor
m
anc
e
As
sess
me
nt
of
c
ac
he
str
at
eg
ie
s
in
Cont
e
nt
Cent
ri
c
Network
”
,
In
f. Tec
hno
l. J.
,
vol. 12, no. 23, pp.
7083
–
7089,
2013.
[28]
V.
Sourlas,
P.
Flegka
s,
and
G.
S.
Paschos,
“
St
ora
ge
Planni
ng
and
Repl
i
ca
t
ion
As
sign
m
ent
in
Conte
nt
-
Cent
r
i
c
Publish /
Subs
cribe
Networks
”
,
I
nt.
J. Comput. Tel
e
commun.
Ne
t
w.
,
vo
l. 55, no. 1
8,
pp
.
4021
–
403
2,
2010
.
[29]
X.
Hu,
J.
Gon
g,
G.
Cheng
,
a
nd
C.
Pan
t,
“
E
nhanc
ing
In
-
ne
t
work
Cac
h
ing
b
y
Coup
li
ng
C
ac
he
Pla
ce
m
ent
,
Repl
a
ce
m
ent a
n
d
Locat
ion
”
,
in
I
EE
E
ICC
2015
-
Next
Gen
erati
o
n
Net
work
ing
Sy
mpos
ium
,
2015,
pp.
5672
–
5678
.
[30]
Y.
W
ang,
M.
Xu
,
and
Z.
Feng,
“
Hop
-
base
d
Probabil
ist
ic
Ca
chi
ng
fo
r
Inform
at
ion
-
Cen
tric
Networks
”
,
in
Globec
o
m
2013
-
Next
Gen
e
ration
Net
wor
king Sy
mpos
ium
,
2
013,
pp
.
2102
–
2
107.
[31]
N.
Chauha
n
an
d
L.
K.
Aw
asthi
,
“
Prefe
tc
hi
ng
base
d
Coopera
t
i
ve
Cac
hing
in
Mobile
Adhoc
Networks”
,
in
Inte
rnational
Co
nfe
renc
e
on
Eme
rging Trends i
n
Computer
and
E
le
c
t
ronics
Eng
in
ee
ring
,
2012,
pp.
60
–
64.
[32]
G.
Ross
ini
and
D.
Ross
i,
“
A
div
e
int
o
th
e
c
ac
h
in
g
per
form
anc
e
o
f
Conte
nt
Cen
tric
Networki
ng
”
,
i
n
2012
IE
EE
17
th
Inte
rnational
W
orkshop on
Computer
A
ide
d
Mod
el
ing
and
Desig
n
of
Comm
unic
a
ti
on
Links
and
N
et
works
,
2012
.
[33]
H.
Chao,
W
.
J.
Jian,
H.
Cho,
C.
Tsai,
and
J.
Pan,
“
Prediction
-
Based
Ca
c
he
Adaptation
for
Nam
ed
Data
Networki
ng”
,
J
.
Comput.
,
vo
l. 27
,
no
.
1
,
pp
.
45
–
5
5,
2015
.
[34]
J.
Ren
et
al.
,
“
MA
GIC :
a
Dist
ribut
ed
MA
x
-
Gain
In
-
net
work
Cac
hing
Stra
te
g
y
in
Inform
at
ion
-
Cent
ri
c
Netw
orks
”
,
in
2014
I
EEE
IN
FOCOM Workshop
on
Name
-
Or
ie
nt
ed
Mobi
li
t
y
,
2014,
pp
.
470
–
4
75.
[35]
S.
Kang,
S.
Lee,
and
Y.
Ko,
“
A
Rec
ent
Populari
t
y
B
ase
d
Dy
nami
c
Cac
h
e
Mana
gement
fo
r
Conte
nt
Cen
tr
ic
Networki
ng”
,
in
Fourth
Inte
rnat
ional
Confe
ren
c
e
on
Ubiquit
ous
and
Fut
ure
Networ
ks
(
IC
UFN)
,
2012,
pp.
219
–
224.
[36]
W
.
Zha
ng
,
R.
Fa
n,
F.
Li
u,
an
d
P.
La
i
,
“
Ene
rg
y
-
A
ware
Ca
chi
ng
”
,
i
n
IEE
E
21st
Int
e
rnational
Conf
ere
nce
on
Paral
le
l
and
Distribute
d
Syste
ms
,
2015
,
p
p.
473
–
480
.
[37]
I.
Ps
ara
s,
W
.
K.
Chai
,
and
G.
Pa
vlou,
“
Probabili
stic
In
-
Ne
twork
Cac
h
ing
f
or
In
form
at
ion
-
Cent
r
i
c
Networks
”
,
in
sec
ond
ed
it
ion
o
f
th
e
ICN
workshop on
Informatio
n
-
ce
ntri
c
n
et
working
,
2012,
pp.
5
5
–
60.
[38]
C.
Fricke
r
,
P.
Robert
,
J.
Robe
rts
,
and
N.
Sbihi,
“
Im
pac
t
of
tra
ff
ic
m
ix
on
ca
chi
ng
per
form
anc
e
in
a
cont
en
t
-
c
ent
ri
c
net
work
”
,
in
Wo
rkshop on
Eme
r
ging
Design
Cho
ic
es
in
Name
-
Or
ie
nt
ed
Ne
tworking
,
2012
,
pp
.
31
0
–
315.
[39]
N.
Chand,
R.
C
.
Jos
hi,
and
M.
Misra,
“
Cooper
at
iv
e
Ca
chi
ng
S
tra
t
eg
y
in
Mobi
le
Ad
Hoc
Ne
t
works
Based
on
Cluste
rs
”
,
Wire
l. P
ers.
Comm
un.
,
vol. 43, no. 1, p
p.
41
–
63
,
2007
.
[40]
M.
Am
ade
o
and
A.
Molina
ro,
“
CHANET
:
A
Conte
nt
-
Cen
tric
Archi
t
ec
tur
e
f
or
IEE
E
”
,
in
2
011
Inte
rnation
a
l
Confe
renc
e
on
t
he
Ne
twork
of
th
e
Fu
ture
,
2011,
pp.
122
–
127
.
[41]
G.
Carof
iglio,
V
.
Gehle
n
,
D.
Per
ino,
and
B
.
L
abs,
“
Expe
riment
al
Eva
lu
at
ion
of
M
emor
y
Mana
g
e
m
en
t
in
Conte
n
t
-
Cent
ri
c
Network
ing”
,
in
IEEE
In
t
ernati
onal
Confer
enc
e
on
Comm
unic
ati
ons
(
ICC),
2011
,
pp
.
1
–
6.
[42]
N.
E.
Ma
jd,
S.
Misra,
and
R
.
Toura
ni
,
“
Split
-
Cac
he
:
A
Holi
stic
Ca
chi
ng
Fr
amework
for
Im
prove
d
Network
Perform
anc
e
in
W
ire
le
ss
Ad
Hoc
Networks”
,
in
Global
Comm
unic
ati
ons
Conf
e
renc
e
(
GL
OBECOM)
,
2014,
pp.
137
–
142.
[43]
Y.
W
ang,
“
Hier
arc
hi
ca
l
Coop
er
at
iv
e
Cac
h
ing
in
Mobile
Opport
unisti
c
Soci
al
N
et
works
”
,
in
Glo
bec
om
2014
-
A
d
Hoc
and
S
ensor Ne
tworki
ng
Sym
posium
,
2014,
n
o.
3
,
pp
.
411
–
41
6.
[44]
W
.
Quan,
Y.
Liu,
X.
Ji
ang
,
and
J.
Guan,
“
Int
el
l
ige
nt
popu
la
ri
t
y
-
awa
re
cont
en
t
c
ac
hing
and
re
trieving
in
h
ighw
a
y
vehi
cu
la
r
n
et
wor
ks”
,
EUR
ASIP J.
Wirel.
Comm
un
.
Ne
tw.
,
2016
.
[45]
Y.
Kim
,
Y.
Kim
,
J.
B
i,
and
I
.
Ye
om
,
“
Diffe
ren
tia
te
d
forward
ing
a
nd
cachi
ng
in
n
a
m
ed
-
dat
a
n
et
wor
king”
,
J
.
N
e
tw.
Comput.
App
l.
,
v
ol.
60
,
pp
.
155
–
1
69,
2016
.
[46]
W
.
Feng
and
Y.
Zha
ng,
“
A
Gene
ral
Cac
h
e
Parti
t
i
on
Model
for
M
ult
iple
QoS
Cla
ss
es”
,
in
Inte
rnati
onal
Confe
renc
e
on
Computati
on
al
Intelli
g
ence
fo
r
Mode
ll
ing
,
Co
ntrol
and
Aut
om
ati
on,
and
Inte
rn
ati
onal
Conf
ere
n
ce
on
In
te
l
li
gen
t
Age
nts,
We
b
Tec
hnologi
es
and
In
te
rnet
Comm
erc
e
,
2005
,
pp
.
0
–
5.
[47]
A.
Virgil
lito,
“
Publish/Subs
cri
be
Comm
unic
at
ion
Sy
st
e
m
s:
from
Models
to
Applic
ations
”
,
Univer
sita
d
egli
Studi
d
i
Rom
a
“
La
Sapi
e
nza
”
,
2003
.
[48]
A.
Rao,
P.
Kum
ar,
and
N.
Chau
han,
“
Ene
r
g
y
Ef
fic
i
ent
D
y
n
amic
Group
Cac
hing
in
Mobile
Ad
hoc
Networks
for
Im
proving
Data
Acc
essibil
i
t
y
”
,
in
Inte
rnationa
l
Confe
renc
e
on
Re
c
ent
Tr
ends
In
Information
Technol
og
y
,
2012
,
pp.
372
–
376
.
[49]
C.
Fang,
F.R.
Yu,
S.
Mem
ber
,
T.
Huang,
J.
Li
u,
and
Y.
Li
u,
“
A
Surve
y
of
Gree
n
Inform
at
ion
-
Ce
ntri
c
Networki
n
g :
Resea
rch
Iss
ues
and
Cha
ll
eng
es”
,
IE
EE Comm
un.
Surv.
Tutor
ial
s
,
vol.
17
,
no
.
3
,
pp
.
1455
–
1472
,
20
15.
[50]
A.M.E
.
Ta
m
iz
h
a
rasi,
J.J.S
.
M.E
,
A.K.
Pri
y
a
,
and
R.
Maa
r
li
n,
“
En
erg
y
Aw
are
Heu
risti
c
Appro
ac
h
for
Cluste
r
He
ad
Sele
c
ti
on
in
W
ir
el
ess Sensor Ne
t
works
”
,
Bull.
E
l
ec
tr.
Eng
.
In
formatic
s
,
vo
l. 6, n
o.
1
,
pp
.
70
–
75
,
2017.
Evaluation Warning : The document was created with Spire.PDF for Python.