TELKOM
NIKA
, Vol. 11, No. 10, Octobe
r 2013, pp. 5
617 ~ 5
626
ISSN: 2302-4
046
5617
Re
cei
v
ed Ap
ril 24, 2013; Revi
sed
Jun
e
24, 2013; Accepted July 1
1
,
2013
Routing Algorithm for DTN Ba
sed on Congestion
Control
Song Ningni
ng*
1
, Liu Qian
1
, Zhou Xian
w
e
i
1
, Xue Peipei
1
1
Universit
y
of Scienc
e an
d T
e
chn
o
lo
g
y
Bei
j
i
ng, CHINA
30
Xue
y
u
a
n
R
oad, Ha
idi
an D
i
strict, Beijin
g 1
000
83 PR C
h
in
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: 6221
29
4@1
6
3
.com*
1
, 5266
6
75@
163.com
1
,
x
w
zhouli@sina.com
1
,
dia
g
life
@
12
6.c
o
m
1
A
b
st
ra
ct
Different fro
m
Internet netw
o
rks, DT
N ha
s it
s ow
n uni
que c
haracter
i
stics, like inte
rmitten
t
conn
ectivity, low
data trans
fer rate, high
laten
cy a
nd
li
mited stor
ag
e and
nod
e resourc
e
s. Rou
t
in
g
techno
lo
gy is alw
a
ys the ke
y to DT
N research. In th
is p
aper, a ro
uting
protocol th
at is PNCMOP wa
s
p
r
o
p
o
s
e
d
.
It ma
ke
s som
e
im
pro
v
em
en
t o
f
Ep
i
dem
i
c
ro
u
t
ing
p
r
o
t
o
c
o
l
, a
nd a
l
so
to
de
cre
a
se
th
e
e
n
d
-
to
-end
avera
ge
de
lay
and
i
m
pr
ove
d
e
livery
rati
o, e
s
peci
a
lly
w
i
th
c
ong
estio
n
co
ntrol. Si
mu
lati
on
results s
how
th
at
PNCMOP achi
eved a g
o
o
d
p
e
rformanc
e.
Ke
y
w
ords
:
DT
N, congesti
on
control,
routi
n
g
protocol
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Delay tole
ran
t
Netwo
r
k
(DTN) i
s
a
new Ne
twork [1] [2], these net
works tend t
o
have
intermittent
con
n
e
c
tivity, low d
a
ta tran
sfer
rate,
high late
ncy, limited storag
e an
d
node
resou
r
ces, an
d the frequ
en
t chang
e of n
e
twork to
p
o
lo
gy stru
cture
etc. Com
m
on
DTN n
e
two
r
ks
have:
A. Remote Areas of Netwo
r
k T
r
an
smi
ssi
on
Network infra
s
tru
c
ture in d
e
veloping
co
untrie
s
or
re
mote are
a
s a
r
e u
s
ually no
t perfect
and cann
ot access the Internet, u
s
e th
e oppo
rt
unity network te
chnolo
g
y to provide non
-re
a
l-
time, but the
price is
ch
ea
p, relatively a
v
ailabl
e net
work services [
3
] [4]. DakNet
[5] develope
d at
MIT, deploye
d
in the rem
o
te area
s of
India to
provi
de Intern
et service
s
to th
e oppo
rtunity to
netwo
rk.
Da
kNet incl
ude:
and Kio
sk d
e
vice
s depl
o
y
ed in the village, MAP (mobil
e
acce
ss
points) eq
uip
m
ent and Int
e
rnet d
eploy
ment in the
t
o
wn AP d
e
vices o
n
pu
blic t
r
an
spo
r
t vehi
cle
s
,
interface co
mmuni
cation
betwe
en the
s
e device
s
u
s
i
ng Wi-Fi. Th
e villagers throug
h PDA a
n
d
Kiosk d
e
vice
s to exchan
g
e
data; between the
ru
ral
and urb
an bus after the
Kiosk ne
ar the
equipm
ent,
MAP and Ki
osk d
e
vice
to exchan
ge
data b
u
s to
town, MAP
conne
cted
to
the
Internet to
u
p
load
or do
wnlo
ad d
a
ta
via the AP.
A simila
r
system also p
r
o
v
ide informat
ion
servi
c
e
s
to the nomad
s of northe
r
n Finl
and, t
he Saami Netwo
r
k
Con
n
e
c
tivity
[6] and Berkel
ey
and Intel Re
search Institut
e are jointly d
e
velope
d Tier proje
c
t [7].
B. Military Ad
-Hoc
Networks [2]
Such
networks may b
e
deploye
d
in
a ho
stile e
n
vironme
n
t, d
ue to n
ode
mobility,
unpredi
ctable
environ
ment
al facto
r
s
or delibe
r
ate human
i
n
terf
eren
ce may
cau
s
e network
fragme
n
tation
. In addition
, when a
hi
gh-p
r
io
rity busin
ess, lo
w-pri
o
rity bu
si
ness n
eed
s to
comp
ete fo
r ban
dwi
d
th
and th
e hi
g
h
-p
riority b
u
s
ine
s
s, which will
ma
ke
the me
ssa
ge i
s
forwa
r
d
ed to
experie
nce
large
r
qu
e
u
ing del
ay. Mean
while, f
o
r the relia
ble and
saf
e
ty
con
s
id
eratio
n
s
, su
ch sy
ste
m
s re
quire stringent
protect
i
on mea
s
u
r
e
s
underlyin
g in
frastructu
re.
C. Vehicle
Ne
twork
Equippe
d wit
h
sho
r
t-ran
g
e
wirele
ss in
terface,
the i
n
crea
se in t
he vehicl
e, a
vehicle
traveling o
n
t
he road
due t
o
the spee
d, the den
si
ty u
nevenn
ess i
s
formed
a ve
hicle
opp
ortu
nity
netwo
rks. Su
ch
netwo
rks
have a
hug
e
potential fo
r
se
cu
rit
y
app
licat
ion
s
in
T
r
af
f
i
c A
cci
de
nt,
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 10, Octobe
r 2013 : 561
7 –
5626
5618
traffic dete
c
tion, traffic conge
stion fo
recast
a
ddition, the opp
ortunitie
s
for commu
nication
vehicle
s
an
d road
sid
e
access point
s to provi
de Inte
rn
et access an
d comm
ercial
application
s
.
D. Wildlife Tracking
In a wi
de ran
ge of
wildlife
tracking
ap
plication
s
, DTN more
advant
age
s than
tra
d
itional
static sen
s
or
netwo
rk m
e
sh stru
ct
ure. Zebra
Net [8] is a de
sign by
Princeto
n Un
iversity, used
to
track the Afri
can
sava
nna
h ze
bra
network sy
stem
. SWIM (Sha
re
d Wi
rele
ss In
fostation M
o
d
e
l)
[9] is a m
onit
o
ring
o
c
ean
whal
e un
derwater op
port
unity to network.
The
sp
e
c
ial T
ag e
m
b
edde
d
in the whal
e’s periodi
c colle
ction of monit
o
ring d
a
ta.
E. Terres
t
rial
Mobile Networks
[2]
In many ca
se
s, due to th
e
confli
ct in the
m
obility of the nod
e or
a radio fre
que
ncy (RF
)
,
these
net
works may b
e
co
me unthi
nkab
le sepa
rated.
Difficult to
fo
rm a
n
e
nd-to
-end
(E2E
) p
a
th
and the frag
mentation of t
he net
wo
rk
can predi
ct. For exampl
e, a
comm
uter b
u
s
can be
use
d
as
a tool for
messag
e sto
r
e an
d forward, be
ca
use it has o
n
l
y a limited numbe
r of
RF
comm
uni
cati
on ability is l
i
mited
comm
unication
ran
ge. Kind of
bus travel
from one
place to
anothe
r pl
ace
,
it can
be i
n
the vicinity of t
he
client
(clie
nts)
and
it will
be d
e
stin
ed f
o
r the
lo
catio
n
of the remote
machin
e (re
m
ote c
lient
s)
messag
e exchang
e se
rvice.
2. Architectu
r
e OF DT
N
Figure 1, Int
e
rnet
archite
c
ture
in
cludi
n
g
the
a
ppli
c
a
t
ion layer,
ne
twork laye
r, t
r
an
spo
r
t
layer, data li
nk laye
r a
n
d
physi
cal lay
e
r. DT
N
ar
ch
itecture, i
n
order to
solve
the DT
N in t
h
e
pre
s
en
ce
of h
i
gh del
ay, intermittent
con
nectio
n
,
low
data tra
n
smission
rate, th
e
storage
an
d
the
node
re
sou
r
ce co
nstraine
d
,
on the ba
si
s of the In
tern
et architectu
re and
betwee
n
the ap
plication
layer and th
e transpo
rt layer the ne
w proto
c
ol l
a
yer insert an end
-to-e
n
d
data-o
r
ient
ed
messag
es-b
u
ndled laye
r, also kno
w
n a the bun
dle
lay
e
r of (Bun
dle
layer) [10]. T
he Bundle lay
e
r
to form a
net
work
cove
rin
g
the u
s
e
of
store
-
a
nd-fo
rward me
ch
an
ism to
overcome n
e
two
r
k the
interrupt pro
b
lems [1
1]. The Bun
d
le
layer p
r
oto
c
o
l
with a spe
c
ific a
r
ea
of the unde
rlyi
ng
agre
e
me
nt
with
ea
ch oth
e
r with different
p
r
oto
c
ol
stack, p
r
ovidi
ng a
similar se
rvice to t
h
e
gateway con
necte
d to different LA
N bo
unda
ry so
th
at you can
communi
cate
betwe
en diffe
rent
types of networks. The Bu
ndle layer [12
]
Key
feat
ure
s
inclu
de: dat
a transmissio
n, in acco
rda
n
ce
with the "storage-ca
rry
-Fo
r
ward" intermi
ttent
conne
cti
on between
pro
c
e
ssi
ng n
ode
s, can ta
ke
advantag
e of predi
ctabl
e or rando
m co
nn
ection
s.
Figure 1. Internet and
DT
N archite
c
tu
re
comp
ari
s
o
n
3. Routing I
N
DT
N
Routin
g is a
pro
c
e
ss th
at packet
s
a
r
e tra
n
smitte
d from the
sou
r
ce no
de
to the
destin
a
tion n
ode. Routing
proto
c
ol
s g
enerally
incl
ude: the
est
ablishment
o
f
the network
topology, the maintena
nce of these three
parts
of the n
e
twork topol
o
g
y and routin
g algorith
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Routin
g Algorithm
for DTN Based o
n
Co
nge
stion Con
t
rol (Song
Nin
gning
)
5619
A. Routing Fe
ature
s
in DT
N
In the DTN, d
ue to the limited radi
o freq
uen
cy covera
ge and di
strib
u
tion of a wid
e
rang
e
of mobile
no
des, limite
d
e
nergy
re
sou
r
ce
s, high
inte
rfere
n
ce o
r
chann
el imp
a
irments l
a
cks t
h
e
necessa
ry co
nne
ctivity betwee
n
no
de
s, netwo
rk inte
rmittent, DT
N routin
g de
si
gn ne
ed to
b
e
con
s
id
ere
d
the followin
g
problem
s:
(1)
routin
g p
o
licy, DT
N covers
a vari
e
t
y of network
link
con
n
e
c
tivity is relatively poor,
so i
t
is
necessa
ry according to the
different sce
nar
io
s, usi
ng
different ro
uting strat
egie
s
;
(2) the
ca
pa
ci
ty of the
conn
ection, th
e
ca
pacity
si
ze
cl
ose
rel
a
tion
ship to th
e a
m
ount of
data
can
be exch
ang
e
d
betwe
en th
e two nod
es;
(3) the
node
cache
spa
c
e and its m
anag
ement p
r
obl
e
m
s, in
orde
r to deal
with the lon
ger
netwo
rk inte
rruption n
ode
s need to buf
fer packet
s
, interme
d
iate node
s nee
d enou
gh buffe
r
spa
c
e to sto
r
e messag
es
waiting to be
sent;
(4) the p
r
o
c
e
ssi
ng ca
pabili
ty, DTN intended to be
co
nne
cted not throu
gh the traditional net
work
proto
c
ol com
m
unication d
e
vice,
the
s
e device
s
a
r
e n
o
t
only
small in
si
ze and u
s
ually
limite
d
pro
c
e
ssi
ng capability;
(5) the e
n
e
r
g
y
problem, DTN no
des m
a
y sometime
s be du
e to the small
size, limited carry
energy, is the lack of infra
s
tructu
re an
d an unfri
e
ndly environ
ment, resulting no
d
e
s cann
ot be
timely cha
r
ge
sup
p
leme
nta
r
y ene
rgy no
de of
data p
a
c
ket tran
smi
s
sion a
nd
storage, a seri
es
of operation
s
and eve
r
ywh
e
re
requi
re
s
energy, thus t
he complexit
y
of the algori
t
hm is sm
all,
the fewe
r n
u
mbe
r
of by
tes sent rou
t
ing stra
tegy
can
be eff
e
ctive in
red
u
cin
g
en
ergy
con
s
um
ption.
B. Routing G
oal in DT
N
DTN
ha
s its unique
ch
aracteri
stic
s, a
nd ro
uting g
oal is diffe
re
nt from the t
r
adition
al
netwo
rks. T
h
e traditio
nal
routing
is th
e
sho
r
te
st pat
h or the mini
mum nu
mbe
r
of hop
s.
DT
N
routing i
s
m
o
re co
ncern
ed
about the
su
cce
ssful
deliv
e
r
y ratio of p
a
ckets. T
hat is t
he ro
uting g
o
a
l
in DTN i
s
to maximize the
packet su
cce
ssfully pa
sse
d
to the desti
nation no
de.
C. Epidemi
c
Routin
g Algorithm
Epidemi
c
rou
t
ing algorith
m
, which is
essent
ially a
flooding ro
u
t
ing proto
c
ol,
it is the
easi
e
st ro
utin
g algorith
m
b
a
se
d on re
pli
c
ation
strateg
y
.
This routing
mech
ani
sm, whe
never two node
s
to re
ach e
a
ch oth
e
r ca
n be the
area of
comm
uni
cati
on, the two nodes will
swap each ot
her messages, copy
the
data
packet
s
to each
other. When
a new n
ode
become
s
re
a
c
ha
ble du
e to
mobility or other rea
s
on
s, and then
copi
es
the extra backup. Ea
ch net
work no
de ca
rrying thei
r m
e
ssag
es forwarde
d to all node
s en
cou
n
ter.
If sufficient network cache
and netwo
rk
resou
r
ces su
ch as b
and
wi
dth, the spre
ad of the
routing
proto
c
ol
s to g
uara
n
tee
minim
u
m pa
cket deli
v
ery delay, a
nd to find th
e
sho
r
te
st pat
h to
rea
c
h the de
stination nod
e.
4. PNCMOP
Priority ba
se
d No
de
Ca
che
Man
age
ment Optim
a
l Path alg
o
rithm
(PNCMOP) i
s
desi
gne
d ba
sed o
n
the
Epidemi
c
ro
u
t
ing algo
ri
th
m. The Epid
emic al
gorith
m
has
no p
a
cket
arrival
notification, the
r
ef
ore,
wh
en
a
pa
cket a
rri
ves at
the
destin
a
tion
n
ode, the
p
a
cket
remai
n
ing
co
pies
will co
ntinue to "co
n
ta
ct - infe
ctio
n" exist and diff
use b
e
twe
en
the node
s, it will
cau
s
e con
g
e
s
tion
for stori
ng
re
stri
cted node
s.
Thre
e Stages of PNCMOP:
(1)
copy forwardin
g
stag
e; based
on
cop
y
Bundle distributed strateg
y
;
(2)
con
g
e
s
tio
n
control stag
e; priority-b
ased nod
e ca
ch
e manag
eme
n
t;
(3) optimal
p
a
th computat
ion
stage,
ba
sed
on
the
measure of t
he
wei
ghte
d
averag
e of t
he
routing m
e
ch
anism. Next, this thre
e stag
e workflo
w
wi
ll be introdu
ced:
A. Copy Forwardin
g
Stage
In this pa
per,
the ea
sie
s
t sou
r
ce no
de
sen
d
s
repli
c
a
way mad
e
i
m
prove
d
. In orde
r to
establi
s
h
ro
ute, sou
r
ce n
o
de b
r
oa
dcast
few
copi
es
o
f
a ro
ute me
ssag
e in th
e n
e
twork
until a
n
y
copy m
e
ssa
g
e
re
ach de
sti
nation n
ode.
Traditio
nal bl
oodin
g
alg
o
rit
h
ms
ca
use hi
gh ove
r
hea
d
for
the netwo
rk,
so the copy forw
arding
sta
ge must be i
m
prove
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 10, Octobe
r 2013 : 561
7 –
5626
5620
Definition 1:
_
i
A
VD
e
l
a
y
is the avera
g
e
conn
ectio
n
d
e
lay for node
_/
ii
A
V
D
e
l
ay
t
c
on
(1)
Whe
r
e:
t
is unit time;
i
con
is th
e total numb
e
r of nei
ghb
o
r
s
who
co
nne
ct with n
ode
i
in
t
.
Definition 2:
0
de
l
a
y
is a
co
nsta
nt con
n
e
c
tio
n
delay, sy
stem can
cha
nge its val
u
e
according to
overhe
ad of the network.
The co
py number
NU
M
shoul
d be pro
per to the netwo
rk
becau
se ca
che, band
widt
h
and othe
r re
source a
r
e limi
t
ed in DTN n
e
twork.
i
NU
M
n
(2)
Whe
r
e:
0
0
0(
_
)
1(
_
)
i
i
i
if
A
V
D
e
la
y
d
e
l
a
y
n
if
AV
D
e
lay
d
e
l
ay
The sou
r
ce
n
ode sen
d
s
NU
M
copie
s
to the
neigh
bor no
d
e
s
wh
ose
_
i
A
VD
e
l
a
y
is
bigge
r than
0
del
a
y
.
Becau
s
e
of that it use
s
floodin
g
me
ch
anism
to
cau
s
e the hi
gh routing expe
n
s
e
s
, we
increa
se
a p
a
th field
set i
n
the
root m
e
ssag
e pa
cket to save th
e no
de ID th
at the pa
cket
ever
arrive
d, for e
x
ample:
{
,
...}
jk
se
t
I
D
I
D
. First, When
the
p
a
cket a
rrive
s
a no
de
i
, nod
e
i
add
it
s
own ID to th
e set an
d forward
NU
M
copie
s
of this message to its nei
ghbo
rs
exce
p
t
those
neigh
bors wh
ose no
de ID has already
in the se
t.
This alg
o
rith
m can re
du
ce the numbe
r of
packet
s
in the netwo
rk d
u
ring
the root di
scovery procedure.
B. Conge
stio
n Control Stage
In this pape
r, we con
s
ide
r
a Bundle p
r
io
rity order:
(1) p
a
cket
s whose all de
stination no
de
s are
nei
ghb
or
node
s are prefer
entially transmitted to;
(2) pa
ckets that are ve
ry far a
w
ay i
n
th
e net
work
will
not b
e
tra
n
smitted, are
a
ssi
gne
d a
hig
her
priority.
Epidemi
c
ca
che ma
nag
e
m
ent use FI
FO (First
In Firs
t Out) strategy to ens
ure the
fairne
ss of th
e pa
cket
ca
che, but
bad
manag
eme
n
t nod
e
con
g
e
s
tion. In thi
s
p
aper,
u
s
e
pri
o
rity
based di
scarding strategy to node cach
e manag
eme
n
t:
Without affe
ct
ing the
overal
l netwo
rk deli
v
ery ratio, n
o
des
ca
n di
scard
a me
ssag
e und
er
the following conditions:
(1) a
copy ha
s bee
n se
nt to the destin
a
tion nod
e;
(2) the
r
e i
s
n
o
t eno
ugh
ba
ndwi
d
th
routi
ng in
the
enti
r
e life
cy
cle
o
f
b
e
twe
e
n
the n
ode
an
d
the
destin
a
tion n
ode of ;
(3) even if t
he node di
scard
som
e
copies of
,the others
are still li
kely to be
successf
ull
y
delivere
d
.
In orde
r to estimate whet
her pa
cket
s that
satisfy Con
d
ition a., use the con
f
irmation
messag
e tha
t
sent b
a
ck f
r
om th
e de
sti
nation a
nd transmitted
to
all nod
es in t
he net
wo
rk; f
o
r
estimating
whether th
e m
e
ssag
e satisf
y the con
d
itio
n
b., usin
g th
e prio
rity poli
c
y; Con
d
ition
c. is
the mo
st difficult to
estima
te. This
arti
cl
e u
s
e
s
t
he ho
p
count
a
s
a wea
k
pre
d
ict
o
r,
giving
pri
o
rity
to discardi
ng
the packet th
at has be
en tran
smi
tted fa
rther, be
ca
use these m
e
ssage
s are m
o
st
likely to have been
su
ccessfully received.
In this
pap
er, the cache
manag
eme
n
t stra
t
egy im
mediately d
e
l
e
te the m
e
ssage
ha
s
been
su
ccessfully received
replicas
(lo
w
est pri
o
ri
ty), a
nd then di
sca
r
d lower p
r
io
ri
ty packet
s
ha
s
rea
c
he
d the thre
shol
d num
ber of hop
s, then drop pa
ckets u
nde
r bu
t bigger ho
ps.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Routin
g Algorithm
for DTN Based o
n
Co
nge
stion Con
t
rol (Song
Nin
gning
)
5621
C. Optimal Path Comp
utation Stage
This
articl
e i
s
con
c
e
r
n
ed i
n
the figu
re
(,
)
GV
E
, the optim
al p
a
th ch
osen f
o
r a
given
sou
r
ce no
de
sV
to the de
stina
t
ion nod
e
dV
.
T
here a
r
e
sev
e
r
a
l met
r
i
cs
we
must
con
s
id
er
whe
n
cho
s
e
a optimal
pa
th su
ch
as h
ops, mi
nimu
m ene
rgy of
link,
waiting
delay. In
DTN,
Waiting
delay
is inversely proportio
nal to possibility of
link
connectivi
ty that is easy to get, so
we
can cal
c
ulate path
co
nne
cti
v
ity
degree in
stead of waiting delay.
(1) h
o
p
s
The numb
e
r
of hops in the route disco
v
ery
process,
when you want
to send the packet
node
se
nd
s
a ro
ute requ
est, the
req
u
e
st me
ssag
e
tran
smitted
throug
h multi
p
le ho
ps to t
h
e
destin
a
tion n
ode; the
de
st
ination n
ode
sen
d
s a
re
sp
onse
when
th
e re
sp
on
se
messag
e a
rri
ves
at the sou
r
ce
node, the sou
r
ce
n
ode extract se
ction p
a
ths ho
ps.
(2) the p
o
ssib
ility of path conne
ctivity
Definition 3:
i
path
D
is path conn
e
c
tivity degree
of
i
pa
t
h
Definition 4:
(,
)
ij
P
is en
cou
n
ter p
r
oba
bility of node
i
(
iV
) and n
ode
j
(
jV
).
The possibilit
y of path
connectivity consi
s
ts
of
every link
along the path,
and link
con
n
e
c
tivity p
o
ssibility is e
qual to
(,
)
ij
P
.
Definition 5:
ne
w
t
is the time poi
nt when n
ode
s meet at this time;
old
t
is the time point
whe
n
nod
es
meet at last time;
olde
r
t
is the time point wh
en
node
s meet a
t
the time before la
st
time.
(,
)
(
,
)
ne
w
o
ld
ij
ij
old
o
lde
r
tt
PP
tt
(3)
Definition
6:
/
ij
ij
tt
n
is
used to
measure the
averag
e time
of no
de
i
an
d no
de
j
meet. Whe
r
e:
t
is
unit time,
ij
n
is the total nu
mber of no
de
i
conn
ect with
node
j
in
t
Path conn
ecti
vity degree
i
path
D
is affected by
(,
)
ij
P
and
ij
t
:
1
(,
1
)
1
i
d
path
x
x
xs
ij
DP
t
(4)
Therefore, th
e path of least wa
it for the overhe
ad is
conne
ctiv
ity to
the most likel
y path.
(3)
Nod
e
Mini
mum Rem
a
in
ing Energy
i
e
R
B
i
e
R
B
can
be
obtain
ed in th
e routi
ng di
scovery
pr
o
c
e
ss.
Only
be p
o
ssibl
e
whe
n
the val
ue
is hig
her th
a
n
the en
ergy thre
shold
to
sele
ct
the p
a
th. This
allo
ws th
e no
de
s in the
network
energy co
nsumption a
s
evenly dist
ri
b
u
ted a
s
po
ssible, to avoid
the re
sult o
f
the excessi
v
e
con
s
um
ption
of energy be
caus
e som
e
n
ode
s overb
u
rdene
d.
(4) Multi
-
metric wei
ghted a
v
erage
routin
g
Think
abo
ut hop
s, path conne
ctivity degre
e
an
d no
de minimu
m remai
n
ing e
n
e
rgy, and
with the
weig
hted ave
r
a
g
e
metho
d
fo
r
cal
c
ulatio
n of
link
co
st fun
c
tion
F
, the li
n
ear pla
nning
model is
sho
w
n bel
ow.
0
0
0
1
mi
n
0
0
Fh
E
D
hh
EE
DD
(5)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 10, Octobe
r 2013 : 561
7 –
5626
5622
Whe
r
e
is the weig
ht of adjusting h
ops;
is the weig
ht of adjusting remai
n
ing
energy;
is th
e weight of
a
d
justin
g the
p
a
th conn
ectiv
i
ty degre
e
. We set
00
0
,,
hE
D
acco
rdi
n
g
to situation of
the network.
Whe
n
co
nsi
d
ering the o
p
timal path sel
e
ction bi
as p
r
oble
m
, set the three p
a
rameters
,
and
.When
is larg
e, the optimal path selectio
n re
sul
t
s are bia
s
e
d
in favor of the path
of the smaller number of h
ops; When
is large, the optimal path sel
e
ction results are biased in
favor of the remaini
ng
minimum en
ergy path;
Whe
n
is la
rge, the mo
st excelle
nt path
sele
ction results are m
o
re
con
c
e
r
ne
d ab
out the path conne
ctivity degree.
5. Simulations
The a
r
ticl
e
use
s
NS2 t
o
sim
u
late
PNCM
OP. Analysi
s
the
followin
g
th
re
e ro
uting
perfo
rman
ce indicators:
pa
ckets su
ccessfully
de
livery
ratio, en
d-to
-end
average
delivery del
a
y
and no
de en
e
r
gy con
s
u
m
pt
ion.
A. packets su
ccessfully del
ivery ratio;
1
00%
nu
mb
er
o
f
s
u
ccess
f
ul
l
y
d
e
l
i
ver
ed
p
a
cket
s
de
liv
e
r
y
r
a
tio
to
t
a
l
n
u
m
b
e
r
o
f
p
a
cket
s
(6)
(1) the
rel
a
tionship
between
nod
e
ra
te of move
m
ent an
d th
e
delivery
ratio
of the
packet
s
.
The n
u
mbe
r
of the mo
bi
le nod
es i
s
40, the maxi
mum rate of
movement
of node
s
respe
c
tively lm/s, 3m/s, 6
m
/s and
9m/s, the resi
den
ce
time is 0, th
e simul
a
tion t
i
me is 5
0
0
s
, the
data stream i
s
a
CBR
stre
am, 10 com
m
unicati
on
conne
ction
s
, send two pa
ckets pe
r seco
nd.
The sim
u
latio
n
results a
r
e
as follo
ws Fig
u
re 2.
Figure 2. Nod
e
mobility rate and pa
cket
s delivery ratio.
In Figu
re
2,
Nod
e
s in
cre
a
s
e i
n
th
e
rate
of move
men
t, messag
e d
e
livered
successfully
redu
ce
d. Wh
en the node
mobility rate is low, the tw
o proto
c
ol pa
ckets succe
s
sfully pass ra
te is
less, alon
g
wi
th the no
de
mobility rate
of increa
se
th
e Epidemi
c
p
a
cket d
e
livery rate fa
ster t
han
PNCM
OP fast after both the decli
ne rate
is basi
c
ally the sam
e
.
(2) T
he rel
a
tionship bet
we
en numb
e
rs o
f
node and th
e delivery rati
o of the packets.
The
nod
e m
o
bility rate i
s
3
m
/s, the
num
bers
of no
de
s
re
spe
c
tivel
y
20, 4
0
, 60
and
80,
the re
siden
ce
time is 0, the simulatio
n
time
is 50
0s,
and the data
strea
m
is a
CBR stre
am, 1
0
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Routin
g Algorithm
for DTN Based o
n
Co
nge
stion Con
t
rol (Song
Nin
gning
)
5623
comm
uni
cati
on co
nne
ctio
ns, se
nd
s two packet
s
p
e
r se
co
nd. T
he simul
a
tio
n
results a
r
e
a
s
follows
Figure 3.
Figure 3. The
numbe
rs of n
ode
an
d pa
ckets delivery ratio.
In Figure 3:
(1)
Wh
en
a little numb
e
rs o
f
node, p
a
cke
t
s deliv
e
r
y rat
i
o wa
s
slightl
y
lowe
r, be
ca
use th
e
little number
of node
s, the less opp
ortun
i
ty to
contact betwe
en the
node
s,
pa
cke
t
s can
not arri
ve
at the destina
tion node on t
i
me, while th
e node cache
is limited, when the sto
r
a
ge spa
c
e i
s
full,
discards the
packet
s
in ca
che;
(2)
When th
e num
bers
of node
in
crease, pa
ckets delive
r
y ratio incre
a
se
s, the
approp
riate n
ode
s, packet
s
delivery ratio is high
er.
(3) In
cre
a
si
n
g
the numbe
r of nodes, pa
ckets
delive
r
y ratio would d
e
crea
se. The
reason
is that th
e n
u
mbe
r
s of n
ode in
crea
se
s, the
net
wo
rk
may o
c
cu
r cong
estio
n
, re
sulting
in
the
packet
s
delivery ratio de
creased. But Copying
and f
o
rwarding st
rategy
was u
s
ed in PNCM
OP,
saving th
e m
e
ssag
e copy
in the net
work, and j
o
ine
d
the nod
e ca
che
mana
ge
ment conge
st
ion
control sche
me, packet
s
delivery ratio
of P
NCMOP i
s
high
er than
that of Epidemic.
Therefore, a
n
appropri
a
te increa
se in th
e num
b
e
r of
packet replication, help to
improve
the packet su
ccessfully del
ivery ratio.
B. end-to-e
nd
average d
e
li
very delay
(1)
Th
e relati
onship betwe
en
m
o
ving rate
of node
and end
-to-e
nd
ave
r
ag
e delivery
delay.
The n
u
mbe
r
of the mo
bi
le nod
es i
s
40, the maxi
mum rate of
movement
of node
s
respe
c
tively lm/s, 3m/s, 6
m
/s and
9m/s, the resi
den
ce
time is 0, th
e simul
a
tion t
i
me is 5
0
0
s
, the
data stream i
s
a
CBR
stre
am, 10 com
m
unicati
on
conne
ction
s
, send two pa
ckets pe
r seco
nd.
The sim
u
latio
n
results a
r
e
as follo
ws Fig
u
re 4.
In Figu
re
4,
whe
n
the
certain no
de
s,
with the
rate
of g
r
o
w
th of
the
node
m
o
ve, the
averag
e dela
y
increa
se
s.
PNCM
OP ag
reem
en
t, the average d
e
l
a
y less tha
n
the Epidemi
c
delay. Whe
n
the rate of m
o
vement of the node
s i
s
small, the dela
y
of both the rate of ch
ang
e is
slo
w
. Whe
n
node mo
bility rate grad
u
a
lly incre
a
se
d the lead to establi
s
h t
he co
nne
ctio
n is
bro
k
en, the d
e
lay betwe
en
the two have
a more
sub
s
tantial gro
w
th.
PNCMOP a
g
r
eem
ent of the
gro
w
th rate i
s
sl
ower tha
n
the growth
rate
of the
Epidemi
c
pro
t
ocol. PNCM
OP agreeme
n
t
Bundle
prio
rity routing
me
cha
n
ism, it i
s
possibl
e to
cho
o
se a
bet
ter pe
rform
a
n
c
e n
ode
a
s
the
next hop is re
latively small, so the delay.
(2) T
he rel
a
tionship bet
we
en numb
e
rs o
f
node and e
n
d
-to-end ave
r
age delive
r
y delay.
The
nod
e m
o
bility rate i
s
3
m
/s, the
num
bers
of no
de
s
re
spe
c
tivel
y
20, 4
0
, 60
and
80,
the re
siden
ce
time is 0, the simulatio
n
time
is 50
0s,
and the data
strea
m
is a
CBR stre
am, 1
0
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 10, Octobe
r 2013 : 561
7 –
5626
5624
comm
uni
cati
on co
nne
ctio
ns, se
nd
s two packet
s
p
e
r se
co
nd. T
he simul
a
tio
n
results a
r
e
a
s
follows
Figure 5.
Figure 4. Nod
e
mobility rate and averag
e delivery del
ay
Figure 5. The
numbe
rs of n
ode an
d average delive
r
y delay
In Figure 5,
messag
es Fo
r Epide
m
ic
a
g
re
e
m
ent, when a
sm
all
numbe
r of n
ode
s, the
numbe
r of
co
pies th
e a
ppropriate
an
d condu
cive
to t
he tra
n
smi
s
si
on of the
me
ssage,
so th
e
delay i
s
sma
ller;
When
a
n
in
cre
a
se in
the n
u
mb
er of no
de
s, th
e in
cre
a
se in
the n
u
mb
er of
informatio
n b
r
oug
ht a co
p
y
of the messag
e
exch
an
ge between
node
s certai
n
extent, affected
the delivery
o
f
messag
es, l
eadin
g
to an
increa
se
in t
r
ansmi
ssion
d
e
lay. For P
N
CMOP p
r
oto
c
ol,
whe
n
fewer
n
ode
s, by in
creasi
ng the
n
u
mbe
r
of
cop
i
es of th
e p
a
c
kets to
ma
ke up fo
r n
e
twork
spa
r
se in
sufficient, to en
su
re that the
sm
aller
the t
r
an
smissi
on d
e
lay
;
when th
e nu
mber
of nod
e
s
increa
se
s, th
e co
ntrol
pa
ckets
and
the
numbe
r of
co
pies, to
avoid
the del
ay ca
use
d
du
e to t
h
e
large n
u
mb
er of redund
ant
.
C. Nod
e
average re
mainin
g energy
The PNCMO
P
agree
ment
to incre
a
se the num
b
e
r of
packet re
plication algo
rith
ms are
relatively co
mplex, so
en
ergy
con
s
um
ption than
Ep
idemic. But P
NCM
OP p
r
ot
ocol t
o
en
su
re the
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Routin
g Algorithm
for DTN Based o
n
Co
nge
stion Con
t
rol (Song
Nin
gning
)
5625
ca
se of a hi
gher tran
smi
ssi
on rate of
the
packets and a smal
ler tran
smi
s
sion delay th
an
Epidemi
c
si
g
n
ificantly red
u
ce th
e
con
s
umpt
io
n e
n
e
rgy of the
netwo
rk no
d
e
, the me
ssage
delivery, tran
smissio
n
dela
y
and equal
i
z
ation of energ
y
consumptio
n.
Figure 6. Nod
e
remai
n
ing e
nergy comp
arison b
e
twe
e
n
two proto
c
ol
s
6. Conclusio
n
In this p
ape
r, On the
basis of the E
p
i
demic,
de
sig
n
ba
sed
con
gestio
n
control DT
N
routing
the a
g
ree
m
ent P
N
CMOP
and t
he
simulati
o
n
analysi
s
. PNCMOP
algo
rithm workflow
is
divided into three
pha
se
s,
the sele
ction
phase
ba
se
d on co
py Bundle di
stribut
ed stag
e, ba
sed
on the pri
o
rit
y
Bundle nod
e ca
che m
a
n
ageme
n
t co
n
gestio
n
co
ntrol pha
se an
d
the optimal path
based o
n
mu
lti-the metri
c
weig
hted ave
r
age
ro
ut
ing mech
ani
sm. Nod
e
in
the
netwo
rk
sto
r
a
ge
and rational u
s
e of en
ergy i
s
consi
d
e
r
ed
in the wh
ol
e
pro
c
e
ss, to try to prolong t
he su
rvival time
of the tra
n
sfe
r
no
de
s. The
simulatio
n
P
NCM
OP r
outi
ng p
r
oto
c
ol p
a
ckets succe
ssfully p
a
ss
rate,
the avera
ge
end-to
-e
nd d
e
lay and
nod
e re
sidu
al en
er
gy "thre
e
p
e
rform
a
n
c
e i
ndicators an
g
l
e of
PNCM
OP an
d Epidemi
c
ro
uting proto
c
ol
s do an
alysi
s
and compa
r
i
s
on.
Ho
wever, du
e to time con
s
traint
s, this
study is still
not deep e
n
o
ugh, sum
m
arized the
sho
r
tco
m
ing
s
and are
a
s fo
r furthe
r im
provement of the followin
g
aspect
s
:
1)
Due
to the
limitations of
the
simulatio
n
conditi
o
n
s,
not en
oug
h t
o
refle
c
t all
th
e characte
ri
stics
of the DTN
n
e
twork, h
o
w t
o
de
sign
an
a
pprox
im
ation model of
net
work simul
a
tion
sce
n
a
r
io
s
of real DT
N p
endin
g
furthe
r study;
2) Algo
rithm
packet di
stri
b
u
tion qu
antity and
distri
buti
on way, determining m
u
lti-metric
wei
ght
ed
averag
e ro
uting mechani
sm
,
and
and other pa
ram
e
ters of weight
ed
3)
Routing
se
curity
i
s
su
es nee
d
to be
fu
rther
in
-depth
study
formul
a, th
e calculation
of
prob
ability in path metri
c
, the dete
r
mina
tion of
energ
y
thresh
old a
nd othe
r issu
es a
r
e yet to
be furthe
r re
search.
Referen
ces
[1] Fall
K.
A
d
e
la
y-tolera
nt n
e
tw
ork architect
u
re for
ch
all
e
nge
d i
n
tern
ets
[. Proceed
in
gs
of th
e 2
0
0
3
confere
n
ce on Appl
icatio
ns,
techn
o
lo
gi
es,
a
r
chitec
tures, a
nd pr
otocols f
o
r computer c
o
mmunicati
ons.
ACM. 2003; 2
7
-
34.
[2]
Xi
umei
F
,
Z
h
i
gua
ng
S, Ba
n
x
i
a
n
Z
.
Del
a
y
T
o
ler
ant Net
w
ork arc
h
itectur
e
a
n
d
its ke
y
techno
lo
gies
.
Electron
ics
. 20
08; 1(1): 16
1-1
70.
[3]
He Ni
ng
hui,
Li
Hon
g
she
ng,
Gao Jin
g
. Ene
r
g
y
s
a
vi
ng R
o
u
t
ing Al
gorithm
Based
on C
l
u
s
ter in W
S
N
.
T
E
LKOMNIKA Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2013; 1
1
(2): 8
39-8
47.
[4]
Liu H
u
i. A N
o
vel QoS R
outi
ng Al
gorithm
i
n
W
i
reless M
e
sh N
e
t
w
orks.
T
E
LKOMNIKA Indon
esi
a
n
Journ
a
l of Elec
trical Eng
i
ne
eri
n
g
. 201
3; 11(3)
: 1652-1
6
6
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 10, Octobe
r 2013 : 561
7 –
5626
5626
[5]
Pentla
nd A, F
l
etcher R, Hass
on A. Dakn
et:
Rethi
n
kin
g
con
nectivit
y
in
dev
elo
p
in
g nati
ons
.
Compute
r
.
200
4; 37(1): 78
-83.
[6]
Doria
A, Ude
n
M, Pande
y
D.
Provid
ing c
o
n
nectivit
y
to th
e
saami
noma
d
i
c
commun
i
t
y
.
Generati
ons
,
200
9; 1(2): 3.
[7]
Bre
w
er E, et al.
T
i
er project
. 2006;
http://tier.cs.berkeley.edu/
wiki/index.php?title=Hom
e
[8]
Juan
g P, Oki
H, W
ang Y, et al.
Energy-effi
cient co
mp
utin
g for w
ildlife trackin
g
: Desig
n
tradeoffs and
early ex
peri
enc
es w
i
th Z
ebraN
et
. ACM Sigpla
n
Notices. AC
M. 2002; 37(
10
): 96-107.
[9]
Small T
,
Haas Z J.
T
he shar
ed w
i
rel
e
ss i
n
fostation
mod
e
l
:
a new
a
d
h
o
c
netw
o
rkin
g p
a
rad
i
g
m
(o
r
w
here there is
a w
hale, ther
e is a w
a
y).
Procee
din
g
s of the 4th ACM
i
n
ternati
o
n
a
l s
y
mposi
u
m on
Mobil
e
ad h
o
c
net
w
o
rk
ing & c
o
mputi
ng. AC
M. 2003; 23
3-2
44.
[10]
Scott K L, Burleig
h
S.
Bundl
e protoco
l
specifi
c
ation
. 20
07.
[11]
Forrest W. Dela
y
-
T
o
ler
ant Ne
t
w
orks (DT
N
s).
AT
utorial v1.1
.
2003.
[12]
W
e
i Z
.
Resear
ch and s
i
mul
a
tion of d
e
la
y to
lera
nt net
w
o
rk
(Master'
s
deg
r
ee thesis). C
hen
g W
ang
,
Instructor, Shangh
ai: Sha
ngh
ai Jia
o
tong U
n
iversit
y
, 2
0
0
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.