TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 12, No. 10, Octobe
r 20
14, pp. 7486
~ 749
4
DOI: 10.115
9
1
/telkomni
ka.
v
12i8.558
9
7486
Re
cei
v
ed
Jan
uary 6, 2014;
Re
vised July
25, 2014; Accepted Augu
st
20, 2014
Optimization of Routing Protocol in Wireless Sensor
Networks by Improved Ant Colony and Particle Swarm
Algorithm
Tianshun
Hu
ang*, Xiaoqiang Li, Zhiqiang Zhan
g
, Hongy
un Lia
n
Hen
an Occup
a
t
ion T
e
chnical
Coll
eg
e, Hen
a
n
Z
hengz
ho
u, 450
04
6, Chin
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: huan
gtia
nshu
ned
u@1
63.co
m
A
b
st
r
a
ct
T
h
is pa
per
mai
n
ly d
i
scusses
the a
n
a
l
ysis a
n
d
eva
l
u
a
tion
of
routin
g pr
otoc
ols for W
i
re
les
s
Senso
r
Netw
orks. Particle sw
arm al
g
o
rith
m has be
en prov
en to be very goo
d solve
many g
l
oba
l opti
m
i
z
at
i
o
n
prob
le
ms.
Ant
colo
ny al
gor
ith
m
n
o
t on
ly us
es the
positiv
e
feedb
ack pr
in
ciple, th
e ev
ol
ution
proc
ess
of
spee
din
g
u
p
to
a certai
n exte
nt, but also
ca
n be
i
m
pl
e
m
e
n
t
ed in
par
all
e
l i
n
natur
e a
nd d
i
fferent ind
i
vid
u
a
ls
throug
h conti
n
uous
infor
m
ati
on exc
han
ge
and tra
n
s
m
issi
on.
T
he p
a
p
e
r prese
n
ts opti
m
i
z
at
io
n of ro
uti
n
g
protoco
l
in
w
i
reless s
ensor
n
e
tw
orks base
d
on i
m
prove
d
a
n
t colo
ny a
nd
particl
e sw
arm alg
o
rith
m. F
i
n
a
lly
,
simulati
on
res
u
lts ver
i
fy the
effectiveness
o
f
the i
m
prove
d
al
gorith
m
,
an
d i
m
prove
the
search
for o
p
ti
ma
l
routin
g.
Ke
y
w
ords
:
wireless sensor network, particle swarm
alg
o
rith
m, ant col
ony, routin
g protoc
o
l
Co
p
y
rig
h
t
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Ant colony al
gorithm
is a
bioni
c intelligence algorithms, proces
s i
t
in real life f
r
om the
ant foraging
of inspi
r
ation,
to
use probability selection
mechanism
cont
rol path, also joined
with
the extension of time, pheromon
e volatilization factor. M
any st
udies show that,
ant
col
ony
algorithm has
strong ability to find bet
ter solution
s,
this al
gorithm not only uses the
positive
feedba
ck pri
n
ciple, the ev
olution p
r
o
c
e
ss
of sp
e
edi
ng up to a
certain extent,
but also
ca
n
b
e
impleme
n
ted in
pa
rallel
i
n
nature, different
indi
vidu
al
s throug
h con
t
inuou
s info
rmation ex
cha
nge
and tran
smi
s
sion, can work togethe
r, help to find
a better sol
u
tio
n
. Ant colony algorithm
ca
n be
unde
rsto
od a
s
a kin
d
of sp
ecial reinfo
rcement lea
r
nin
g
algorith
m
.
Particle
swa
r
m algorithm i
s
ea
sy to implement, low computatio
n
a
l co
st and u
s
e
s
less
hard
w
a
r
e
re
source. Pa
rticl
e
swa
r
m al
g
o
rithm
ha
s b
een
prove
n
t
o
be
very g
o
od
solve m
a
n
y
global
optimi
z
ation
p
r
oble
m
s [1]. Of
course, PSO
algorith
m
a
n
d
othe
r
glob
al optimi
z
ati
on
algorithm, is
easy to fall into local
optim
um, conv
erg
e
n
ce
pre
c
i
s
ion
is not hi
gh,
disa
dvantag
e
of
slo
w
late
con
v
ergen
ce, in t
he theo
retical
study
on
alg
o
rithm. The P
S
O algo
rithm
analysi
s
did
n
o
t
mature the
o
ry, few rese
arche
r
s fo
r the
conve
r
ge
nce of the algorit
hm is an
alyzed, and mo
st
of
the re
sea
r
ch
on the struct
ure a
nd pe
rfo
r
man
c
e of
th
e improve
d
al
gorithm a
r
e
studied, incl
ud
ing
the analy
s
is
o
f
param
eters,
topology
stru
cture,
parti
cle
to maintain t
he diversity, fusio
n
alg
o
rith
m
and pe
rform
a
nce
comp
ari
s
ons.
Wirel
e
ss
se
n
s
or net
wo
rk
routing p
r
oto
c
ol fr
om th
e d
r
ive mechani
sm ca
n b
e
divi
ded into
proa
ctive rou
t
ing proto
c
ol,
on-d
e
ma
nd
routing
prot
o
c
ol a
nd hyb
r
i
d
routin
g p
r
o
t
ocol: p
r
oa
ctive
routing p
r
oto
c
ol, all routin
g has b
een f
o
rme
d
before
use; on
-dem
and ro
uting p
r
otocol only whe
n
need
ed to form; hybrid ro
uting proto
c
o
l
is a ki
nd of
the new ro
u
t
ing proto
c
ol
the two idea
s
together.
Due
to the re
sou
r
ce
con
s
trai
ne
d wirele
ss
se
nso
r
no
de
s, a
nd the n
u
mb
er of no
de
s in
a
sen
s
o
r
net
wo
rk, sen
s
o
r
no
des d
o
not h
a
ve enou
gh
spa
c
e to
store a larg
e nu
mber of
routi
ng
table, so in practical appli
c
ation, on-d
e
m
and ro
uti
ng p
r
otocol and h
y
brid routin
g proto
c
ol is m
o
re
popul
ar. Wi
re
less sensor
n
e
twork
routin
g proto
c
ol
s
from the net
wo
rk top
o
l
ogy structu
r
e point of
view can b
e
divided into: flat and hiera
r
chical
routing proto
c
ol
s.
Th
e
pape
r pre
s
ents optimi
z
a
t
io
n
of
routing protocol
in wireless sen
s
o
r
net
wo
rks b
a
s
ed
on imp
r
oved ant col
ony and p
a
rt
icle
swarm al
gorit
hm.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Optim
i
zation of Routing Protocol in
Wire
less
Sensor
Networks b
y
I
m
pr
ove
d
… (T
ianshun
Hua
ng)
7487
2. Rou
t
ing
Protocol in
Wireless S
e
nsor Net
w
o
r
ks Based
o
n
Improv
ed
Particle S
w
a
r
m
Algorithm
Cla
ssifi
cation
of routing protocol
s for
wire
le
ss
sen
s
o
r
networks on
the continu
a
t
ion of
the tra
d
itional
cla
s
sificatio
n
metho
d
of A
d
ho
c
network, a
c
cording
to the
differe
n
t
angle
s
ca
n
be
cla
ssifie
d
differently. Acco
rding
to the
routing
st
rate
gy point of vi
ew, can b
e
d
i
vided into a
c
tive
and p
a
ssive
routing
rout
ing two type
s, acco
rdi
n
g
to the logi
cal
stru
cture
of the net
work
manag
eme
n
t and
routin
g
proto
c
ol
s
ca
n be
divided
in
to flat ro
u
t
ing and
hie
r
archical routi
ng.
Active routin
g: also
calle
d the table d
r
iven ro
uting
(Tabl
e Drive
n
), an a
c
tive
routing
path
by
found a simi
lar strate
gy with
tradition
al
routin
g
protocol,
by p
e
riodi
ca
lly broad
ca
st
routi
ng
informatio
n p
a
cket no
de
s, routing i
n
formation e
xcha
nge, a
c
tive d
i
scovery rout
e, at the sa
me
time, node
s
must mai
n
tai
n
to all n
e
two
r
k
nod
es [2].
Its advantag
e
is that
whe
n
a nod
e n
eed
s to
sen
d
data pa
ckets to the d
e
stinat
io
n no
de, as long a
s
the route ex
ists, the req
u
i
r
ed del
ay is very
small. Sh
ort
c
oming
s
need
co
stly ove
r
h
ead, a
s
far
as
po
ssi
ble t
he
routing
u
pdate foll
owed
cha
nge
s in the cu
rrent topology,
wa
st
e of resources to build
a
nd re
build th
ose a
r
e n
o
t use
d
routing.
Particle
swarm algorithm
is from this m
odel in
enlighte
n
an
d use
d
to solve the
optimizatio
n probl
em. In PSO, each o
p
timization p
r
o
b
lems of pot
ential sol
u
tio
n
s is to search a
bird in
spa
c
e, called th
e
particl
e. All of t
he parti
cle
s
are det
ermin
ed by
a function
b
e
ing
optimize
d
fitness (fitne
ss v
a
lue), e
a
ch p
a
rticle
ha
s a
spe
ed dete
r
m
i
ne the directi
on and
distan
ce
they fly. Then the particle
will follow the optimal
parti
cle current search
in the solution space.
Particle
swa
r
m optimi
z
ati
on al
gorith
m
is in
itialized
to a
g
r
oup
of rand
om
particl
e
(sto
cha
s
tic).
And then through the ite
r
ation to find
the optimal
solution. In ea
ch iteration, the
particl
e i
s
u
p
dated
by followin
g
two
"e
xtreme": the
first is the
op
timal pa
rticle
itself to find t
he
solutio
n
, the
sol
u
tion i
s
calle
d individ
ual extre
m
u
m
point
(pe
r
son
a
l be
st
said its
po
sition).
Another extreme is the o
p
timal popul
ations
cu
rr
e
n
t
ly find solutions, this
sol
u
tion is a gl
obal
extremum by
global be
st, GB said its
p
o
sition
).
Also
can o
n
ly use one a
s
a p
a
rticle
neigh
b
o
r
without the
whole p
opul
ation, so
extre
m
e in all
nei
ghbo
rs is
a l
o
cal
extremu
m
. Particle
swarm
optimizatio
n
allows the in
telligen
ce alg
o
rithm i
s
re
al
ized by
con
s
tantly
on the extreme poi
nt
update.
Information d
i
sseminatio
n
proto
c
ol info
rmation con
s
ultation and
has e
n
e
r
gy adaptive
function
ba
se
d on. T
he b
a
s
ic idea
of th
e protoc
ol is:
any two
nod
e
s
a
r
e to
be n
egotiated
bef
ore
the data transmission. Node will me
tadata (Meta2data) for a descri
ption of the origi
nal data,
metadata in
cl
ude
s som
e
key informatio
n of the
original data. The
neighbo
r no
des a
c
cordin
g to
the metad
a
ta
inform
ation,
de
cide
wh
e
t
her you
ne
e
d
to tra
n
spo
r
t the
raw d
a
ta. Wh
en t
h
e
informatio
n redun
dan
cy i
s
very hi
gh, th
e am
ount
of raw
data i
s
m
u
ch
hig
her th
an the
meta
d
a
ta
con
d
ition
s
, u
s
ing
this info
rmation
ne
go
tiation p
r
oto
c
ol can
greatl
y
red
u
ce the
amo
unt of
d
a
ta
transmissio
n based on it.
Comp
ared to
hiera
r
chical
routing p
r
oto
c
ol an
d singl
e layer routin
g proto
c
ol, h
a
s better
scalability, easier
data fusi
on,
thereby reducing power consumpti
on. Single routing protocol
, due
to the expa
n
s
ion
of the n
e
twork
si
ze,
gateway load
will in
crea
se
, leading
to n
e
twork
delay.
In
orde
r to improve the expansibilit
y of the netwo
rk, p
eople put fo
rward the con
c
ept of network
s
t
ratification [3]. LEACH [5] is
the firs
t hierarchic
al ro
uting proto
c
ol
in
sen
s
or n
e
t
works, routin
g
proto
c
ol p
o
si
tion need to
kno
w
the l
o
catio
n
information of se
nso
r
no
de
s based on th
ese
informatio
n, can
be u
s
ed
to cal
c
ulate
the di
sta
n
ce betwe
en no
d
e
s, en
ergy consumption and
estimated
co
nstru
c
tion
mo
re effici
ent ro
uting p
r
ot
o
c
ol
. As the
se
nsor n
ode
s
are
distrib
u
ted i
n
a
regio
n
, and
n
o
IP addre
s
s
scheme
simil
a
r to this,
so
we
can b
u
ild
efficient ro
uting protocol u
s
ing
locatio
n
information, to pro
l
ong the net
work
lifetime, a
s
is sho
w
n by
Equation (1
).
)
2
sin(
)]
(
)
(
[
2
1
fk
D
k
k
Q
k
x
x
m
k
x
x
x
x
i
j
j
i
j
i
(1)
In the flat rou
t
ing proto
c
ol
s, each
nod
e i
n
t
he net
wo
rk status i
n
the
routing
fun
c
tion the
same, no hie
r
archi
c
al ma
nagem
ent
m
e
ch
ani
sm.
Ad
vantage
s of
plana
r
stru
ct
ure
routin
g i
s
no
spe
c
ial no
de
s in the network, the net
work traffi
c i
s
uniformly distribute
d
in the netwo
rk, the
routing
algori
thm is easy to im
plem
ent. Disadvantage is
scalabilit
y small, to a certai
n extent,
limiting the si
ze of the net
work, the net
work dyna
mic resp
on
se sp
eed.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 748
6
– 7494
7488
In the hiera
r
chical
stru
cture, cluste
r he
a
d
node i
s
not
only re
spon
si
ble for the ju
ri
sdi
c
tion
of the clu
s
te
r
informatio
n collectio
n an
d
fusion
pr
oce
s
sing, fo
rwardi
ng is
also respon
sible fo
r t
he
inter cl
uste
r d
a
ta. The form
ation of ea
ch
clu
s
ter
hi
era
r
chi
c
al routing
prot
o
c
ol
s are
usu
a
lly base
d
on
the rete
ntion
of
e
n
e
r
gy of
se
nsor nod
es and cl
u
s
te
r he
ad
s
clo
s
e
deg
ree,
at th
e same time
i
n
orde
r to
prol
ong the
lifetime of the
wh
ole net
wo
rk,
clu
s
ter
hea
d
node
sel
e
ctio
n nee
d p
e
rio
d
i
c
updatin
g. Th
e advanta
g
e
s
of hie
r
a
r
chical
routi
n
g
is suitable
for wi
rele
ss
sen
s
o
r
net
work
environ
ment,
larg
e-scale,
scalabl
e.
Di
sadvantag
e
is that the cl
uster he
ad n
o
d
e
s
relia
bility and
stability ha
s
great
effect o
n
the p
e
rfo
r
mance
of the
whol
e n
e
two
r
k,
colle
cting
and p
r
o
c
e
s
si
ng
information will also be a lot of
cluster head energy consum
ption.
In the theo
reti
cal
study on
algorith
m
. Th
e PSO
algo
rit
h
m analy
s
is
d
i
d not matu
re
theory,
few
re
sea
r
ch
ers for the
co
nverge
nce of
the al
go
rith
m is an
alyze
d
, and
mo
st
of the
re
sea
r
ch
on
the stru
cture and pe
rform
a
nce of
the im
proved al
go
ri
thm are stu
d
i
ed, inclu
d
ing
the analysi
s
o
f
para
m
eters, t
opolo
g
y stru
cture, pa
rticle
to main
tain
the diversity, algo
rithm a
nd pe
rforman
c
e
comp
ari
s
o
n
[4]. PSO has
simple, e
a
sy
to implement,
few paramet
ers to
set, wi
thout gra
d
ient
informatio
n a
nd oth
e
r
ch
a
r
acte
ri
stics, i
n
a
co
ntinuo
us
nonli
nea
r optimization
pro
b
lem
s
a
n
d
combi
nato
r
ial
optimization
probl
em
s hav
e sho
w
n g
o
o
d
results.
With the
tra
d
itional PSO
only to it
s
histori
c
al
be
st positio
n a
n
d
the
neig
h
b
o
rho
od
histori
c
al
be
st positio
n le
arning i
s
differe
nt, each
p
a
rti
c
le
of CLPSO is rand
om
learni
ng to
th
eir
own o
r
other
particl
es, an
d
each on
e ca
n learn from
different parti
cle; the learni
ng strate
gie
s
so
that each pa
rticle
has m
o
re le
arning
obje
c
ts, c
an
be in the l
a
tent sp
ace flight more, thu
s
con
d
u
c
ive to global
sea
r
ch
. CLPSO velocity updating
formula.
）
（
1
)
,
(
)
1
(
)
(
)
(
t
x
j
p
r
t
wv
t
v
ij
j
f
ij
ij
i
(2)
In the pa
rticl
e
swa
r
m alg
o
rithm, ea
ch
opt
imizatio
n probl
em
s
of potential solu
tions can
be tho
ught of
as a p
o
int i
n
the
sea
r
ch
sp
ace dim
e
nsio
n,
which
we call
"
parti
cle" (Parti
cle).
In
flight particl
e
to a ce
rtain
spe
ed in th
e se
ar
ch
sp
ace, thi
s
to
dynamically adju
s
t the sp
eed
according
to
its o
w
n
flight
expe
rien
ce
and
co
mpani
on flying
exp
e
rien
ce.
All p
a
rticle
s have
a
adapt to the obje
c
tive function to be d
e
termin
ed va
l
ue, and that his be
st po
sition discove
r
e
d
so
far.
Implementati
on of
sea
r
ch
diversificatio
n in th
e
previous velocity
of parti
cle
s
u
nder the
action of (Div
ersifi
cation
), and the reali
z
ation of
ce
ntralize
d
se
arch
pro
c
e
ss in th
e cog
n
itive part
and so
cial p
a
rt
of
the
tra
c
tion unde
r (intensifi
c
at
ion
)
, so
the
bal
ance an
d
re
stri
ct ea
ch
other
betwe
en the
s
e three
dete
r
mines th
e m
a
in alg
o
rithm
can. Pa
rticle
swarm
optim
ization
se
arch is
formed
by su
ch a g
r
o
up o
f
rando
m initializatio
n of the pa
rticle
s
and a p
opul
a
t
ion com
p
o
s
e
d
,
perfo
rmed in
an iterative m
anne
r,
as is
shown by Figu
re 1.
Figure 1. The
Principl
e of Particle Swarm Move
Dire
cted
diffusion p
r
oto
c
ol
is a ro
utin
g
mech
ani
sm
based on th
e que
ry. The
whole
pro
c
e
s
s can
be divide
d int
o
interest diff
usio
n, gr
a
d
ie
nt is e
s
tabli
s
h
ed an
d the p
a
th to strengt
hen
the three
sta
ges. In
the
st
age
of intere
st spr
ead, th
e si
nk no
de t
o
the
se
nsor nod
e
sen
d
s
its
want to obtai
n inform
ation
types or
co
n
t
ent. A ta
sk type, the targ
et area, the
d
a
ta tran
smi
s
sion
rate, time st
amp me
ssag
e paramete
r
of inte
rest.
Each sen
s
or no
de afte
r re
ceiving t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Optim
i
zation of Routing Protocol in
Wire
less
Sensor
Networks b
y
I
m
pr
ove
d
… (T
ianshun
Hua
ng)
7489
informatio
n,
save it
in
CACHE.
Whe
n
the
inform
ation
req
u
ire
m
ents thro
ug
hout the
sen
s
or
netwo
rk, it would esta
blish a gradie
n
t field bet
we
en
sen
s
o
r
nod
e and sin
k
no
d
e
, a gradient
field
is ba
se
d on t
he minimi
zati
on of the
co
st and en
er
gy
adaptive p
r
in
ciple. O
n
ce the colle
ction
of
sen
s
o
r
nod
es to a sink no
de data of intere
st, can a
c
cording to th
e gradi
ent field for the fast
est
path for data
transmissio
n.
HREEM
R is
prop
osed b
a
sed on the
directed
diffu
sio
n
routin
g me
chani
sm, the g
oal is to
improve routi
ng relia
bility by maintainin
g multip
le available path
s
.
The proto
c
o
l
in operatio
n
durin
g the localizatio
n algo
rithm and DD the same so
urce nod
e an
d sink p
o
int optimal path P, at
the same tim
e
in order to
protocol
can
still the
co
nst
r
uction of m
o
re than one do not
want
wi
th P
redu
nda
ncy p
a
th gua
rante
e
norm
a
l ru
n
n
ing of P fa
ilure (eme
rge
n
c
y mea
s
ures
in ord
e
r to av
oid
the occu
rre
n
c
e of the ma
in path failure phen
omen
on and ta
ke
). HREEM
R protocol p
r
op
o
s
ed
disjoi
nt multipath and wi
ndi
ng path two d
i
fferent multip
ath mech
ani
sm.
12
2
2
,,
4
24
lc
c
l
ll
l
(3)
Funda
mentall
y
spe
a
ki
ng, most solutio
n
s
a
r
e ba
sed
on
imp
r
oved
combi
ned
wit
h
othe
r
optimizatio
n
algorith
m
, is t
o
g
r
eatly im
p
r
ove
th
e difficulty of un
derstandi
ng
and
implem
entati
on
of algorithm
complexity, wh
ich ma
ke
s th
e parti
cl
e swarm optimi
z
at
ion algo
rithm
lost so
me core
advantag
es a
ttractive, su
ch as
simpli
city, underst
an
ding the impl
ementation
si
mplicity, this
the
algorith
m
of large a
r
ea a
p
p
licatio
n is somewhat
unf
avorabl
e. Th
erefo
r
e, the improve
d
part
i
cle
swarm al
gori
t
hm as the
starting p
o
int
for impr
ove
d
particl
e swarm alg
o
rith
m is simpl
e
and
effective, it is
necessa
ry an
d meanin
g
ful.
The ba
sic p
a
r
ticle swa
r
m
optimizatio
n algorit
h
m
in the solutio
n
space sea
r
ch,
ever
y
particl
e of fli
ght time
con
s
tant i
s
1,
so
metime
s lea
d
to pa
rticle
s i
n
the
optimal
sol
u
tion
of the
back an
d forth near "o
sci
llation" phen
omeno
n [5]. A
daptive adj
ustment of the time of flight
method to dy
namically adj
ust the time
of flight, with
the increa
se
of the it
erative algeb
ra
s, fligh
t
time decrea
s
es line
a
rly as
is sh
own by Equation (4).
n
i
C
C
dt
t
j
t
a
t
x
i
i
1
)
(
exp
)
(
Re
)
(
(4)
Routin
g st
rat
egy TEEN p
r
otocol i
s
rea
c
tive wi
rele
ss sen
s
o
r
n
e
twork an
d de
si
gn, it is
real
-time, can
make a
rapid
re
spo
n
se to
emergen
cie
s
.
TEEN a
nd i
m
pleme
n
tatio
n
me
cha
n
ism
of
LEACH is very simila
r, but
the form
er is
the re
spon
se
type, whi
c
h
b
e
long
s to
the
netwo
rk a
c
tive
sen
s
o
r
. The
advantag
es
of TEEN pro
t
ocol: prot
o
c
ol suitabl
e for the appli
c
at
ion enviro
n
m
ent
need
s re
al-ti
m
e perceptio
n; by setting the hard
thre
shol
d and so
ft threshold o
f
2 paramete
r
s,
TEEN can g
r
eatly red
u
ce
the numbe
r of data
transmissio
n, en
ergy savin
g
and othe
r than
LEACH.
Disa
dvantage
s: n
o
t suitabl
e for application i
n
the sy
stem
of perio
dic
sa
mpling, it is a
s
if
the node
s in
the netwo
rk have
not re
ceived the
relevant th
re
shold, then th
e node
can
n
o
t
comm
uni
cate
with the base station,
use
r
also
without
any data network.
GEAR also can b
e
co
nsidere
d
as a
n
im
proved m
e
thod of Directed
Diffusi
on. The
proto
c
ol d
o
e
s
not u
s
e th
e intere
st to
the entire
net
work to b
r
o
a
d
ca
st, but th
e use of loca
tio
n
informatio
n, to a
spe
c
ific
a
r
ea
of inte
re
st is th
en
broa
dca
s
t, u
s
ing
l
o
catio
n
info
rmation
(a
s fa
r a
s
possibl
e to choo
se the sh
ortest path
)
and nod
e re
sidu
al ene
rgy
situation, the packet is sent
back to the
sin
k
no
de. T
he metho
d
can greatly
save ene
rgy, prolo
ng the
sen
s
o
r
net
work
lifetime.
One
hop
nei
ghbo
r n
ode
s
and Sin
k
i
s
t
he root n
ode
of the
multicast tree to
re
alize
th
e
sen
s
o
r
n
ode
s to the Sin
k
multi hop
pat
h. It is
cha
r
a
c
teri
zed
by t
he routing
de
cisi
on m
u
st t
a
ke
into con
s
ide
r
ation n
o
t only
to ea
ch
path
of the
ene
rg
y, but also
re
lates to
the
e
nd to
delay
a
nd
data packet to be sent pri
o
rity end [6]. The simulati
on re
sults sh
ow that, with the minimum
energy path
energy cons
umption m
e
a
s
ureme
n
t protocol
s, SAR ene
rgy con
s
ume
le
ss.
The
disa
dvantag
e
of the algori
t
hm is not suitable
for la
rge an
d freq
uent ch
ange
s in the network
topology.
Particle swa
r
m algorithm
behavio
r is a
ffect
ed by its optimal pbe
st and optim
al gbest,
this versio
n is calle
d the
gl
obal ve
rsi
on
of the PSO
al
gorithm. An
other i
s
th
e lo
cal version
of the
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 748
6
– 7494
7490
PSO algo
rith
m, in this
alg
o
rithm, the
p
a
rticle
be
havi
o
r i
s
not
affected by the
glo
bal optim
al g
best
effect, but
by the l
o
cal o
p
timal lb
est
adj
ace
n
t pa
rticl
e
s in
p
b
e
s
t an
d its optimal
topology
effect,
the experim
ental re
sults show that,
w in
between [0.8,1.2], PSO algorithm has fa
ster
conve
r
ge
nce spe
ed, and th
e whe
n
w>1.
2, the algorit
h
m
is ea
sy to fall into the local extremum.
A good routin
g proto
c
ol of wirel
e
ss sen
s
or
network sh
ould have the
following feat
ure
s
: a
dynamic
choi
ce of the sin
k
node capa
cit
y
. Obviously
, the sin
k
nod
e
directly affects the life cycl
e
of the
whol
e
sen
s
o
r
n
e
two
r
k life
cy
cle. I
n
the
pr
o
c
e
ss of informatio
n tran
smi
s
sio
n
, the
sin
k
n
o
d
e
with the hig
hest freq
uen
cy, the maximum ene
rgy
con
s
umptio
n. When a
sink no
de en
ergy
con
s
um
ption
is too larg
e, the
se
nsor ne
twork a
c
cordi
ng to t
he sin
k
nod
e ene
rg
y consumptio
n,
energy con
s
umption
of n
ode
s dyna
mi
cally, co
nv
ey informatio
n, the si
nk no
de bal
an
ce t
he
netwo
rk e
n
e
r
gy con
s
umpti
on, can p
r
olo
ng the se
nsor network lifetime.
3. Using
Improv
ed Ant Colon
y
to Optimization of
Rou
t
ing Protocol in WSN
The PEGASIS protocol is used to con
nect t
he sen
s
or
node
s in
the chain
structu
r
e.
Run
n
ing the
PEGASIS protocol
whe
n
each nod
e using
sig
nal to
measure the
intensity of al
l its
neigh
bor n
o
d
e
s, dista
n
ce, whe
n
determi
ning the
ne
arest neig
hbo
r and adj
ust th
e sen
d
ing
sig
nal
stren
g
th to e
n
su
re that on
ly the neighb
ors
hea
r. Secondly, cho
o
se only one n
ode chain a
s
first
of chain tra
n
smi
ssi
on to
the sink n
ode dat
a [7
]. The data colle
cted b
y
point-to-p
o
i
nt
transmissio
n, fusio
n
, an
d
finally se
nt to
the
re
n
d
e
z
vous poi
nt. Ad
vantage
s of
this
protocol i
s
:
redu
ce
the L
EACH g
ene
rated in th
e p
r
ocess
of
clu
s
ter
re
config
uration
overh
ead, an
d thro
ugh
the data fusion pro
c
e
ss to reduce the numbe
r of transcei
v
ers, whi
c
h
redu
ce
s en
ergy
con
s
um
ption.
The
wirele
ss routin
g p
r
ot
ocol
an
d MFl
ood
r
outin
g
algorith
m
i
s
a broad
ca
st t
r
adition
al
routing
proto
c
ol
s. It doe
s
not nee
d to
maintain
th
e
topology a
n
d
routin
g net
works, e
a
ch n
ode
use
s
a
data
packet b
r
oa
d
c
a
s
t relay
re
ceived, if t
he recei
p
t of the
dupli
c
ate p
a
cket is
disca
r
d
ed.
By setting th
e ap
propri
a
te
TTL
value
to
avoid
due
to
the diffusion
of
large are
a
and occu
py
too
much
cyb
e
r
sou
r
ce, the d
i
ffusion
conv
erge
nce, to
e
n
su
re th
e dat
a pa
cket only
after finite h
o
p
routing; i
n
a
ddition to
re
peat p
a
cket
detectio
n
, ea
ch
nod
e n
e
e
d
s to
maintai
n
a
data
pa
cket
seq
uen
ce nu
mber SEQ a
nd a routing
table, the original nod
e transmitting o
n
e packet SEQ is
adde
d by 1,
and the
SEQ
is a
dde
d to
the data
pa
cket IP he
ad,
the rem
a
inin
g nod
es by d
a
ta
packet will b
e
the SEQ re
corded i
n
the
routi
ng tabl
e
and re
peate
d
packet d
e
tection b
a
sed
on
the SEQ, as is sh
own by equation
5
.
T
2
.
0
|
)
(
cov
:
|
)
(
s
s
a
a
C
k
k
R
Rk
k
(5)
For a route, ant it more choice, stre
ngt
h
of ants in the path left by the phero
m
one i
s
large
r
, an
d the inten
s
ity of pheromo
n
e
attra
c
ts mo
re ant
s, so a
s
to form
a p
o
sitive feed
b
a
ck.
Thro
ugh thi
s
positive fee
d
back, ant
s
ca
n find the
sh
ortest
routin
g
path eve
n
tu
ally. In the ACO
algorith
m
, the
phe
rom
one
trail to
be
kno
w
n
as a
pa
ra
metric p
r
oba
bility model
o
f
the ph
erom
one
model
simul
a
tion. The p
hero
m
on
e m
odel
con
s
ist
s
of a se
rie
s
of model p
a
ram
e
ters, the
para
m
eters o
f
the model where the valu
e is call
ed the
phero
m
on
e value.
On the roa
d
of wirele
ss se
nso
r
networks by
proto
c
ol
and ant colon
y
algorithm. Although
many schola
r
s have
studie
d
the rout
in
g proto
c
ol in
wi
rele
ss
se
nsor
net
wo
rk
s of
sev
e
r
a
l cla
s
si
c,
but in the
rou
t
ing protocol
of
the effectiv
e ene
rgy utili
zation
re
se
a
r
ch re
sults are
not
ma
ny,
the
ant colony
algorith
m
i
s
introd
uced
to the
wi
rel
e
ss
sen
s
o
r
netwo
rk rout
ing al
gorith
m
to
chall
engin
g
. To find the optimal path in
wirele
ss s
e
n
s
or n
e
two
r
ks,
but also con
s
ide
r
the effective
use
of the se
nso
r
no
de e
n
e
rgy [8]. A good
wirel
e
ss
sen
s
o
r
net
wo
rk routin
g
pro
t
ocol can sav
e
energy effectively, but also can e
nha
nc
e the netwo
rk
scala
b
ility and reliability.
Acco
rdi
ng to
the ab
ove
discu
ssi
on, a
c
cordi
ng to t
he cha
r
a
c
teri
stics of the
wirel
e
ss
sen
s
o
r
n
e
two
r
ks
need
to
d
e
sig
n
n
e
w ro
uting a
pproa
ch, ant
colo
ny
algorith
m
i
s
o
ne of th
em. A
n
t
colo
ny algo
rithm is
derive
d
from the
ob
servatio
n
s
of t
he natu
r
al
world a
n
t swa
r
m beh
avior a
n
d
abstract. Ant
colo
ny fora
ging
can find
the sh
orte
st path bet
we
en the n
e
st
and foo
d
, which
depe
nd
s on
the ch
emical
sub
s
tan
c
e
s
in a ph
er
o
m
one, ant a
n
d
in between,
they relea
s
e a
pheromo
ne, provide the p
a
th for the later ants
wi
zard. Such beh
a
v
ior is very good for the re
ality
rea
c
tion, this
idea is al
so
suitable for wi
reless
sen
s
or
netwo
rks, as i
s
sh
own by Equation (6).
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Optim
i
zation of Routing Protocol in
Wire
less
Sensor
Networks b
y
I
m
pr
ove
d
… (T
ianshun
Hua
ng)
7491
n
i
n
j
j
ji
i
n
j
j
ji
i
i
i
k
k
f
k
k
f
k
Z
k
m
P
k
11
1
)
1
(
)
(
)
1
(
)
(
)}
(
|
)
(
{
)
(
(6)
The agreem
e
n
t calls for ne
twork ope
rati
on by the end user Sin
k
of the sen
s
o
r
no
des a
r
e
partitione
d in
to cluste
rs, the location inform
atio
n of the node
s are assig
ned
and notify ea
ch
clu
s
ter he
ad
node ID lo
go and cl
ust
e
rs [9]. The
sen
s
or n
o
d
e
can run
with low en
ergy
con
s
um
ption
and the sta
n
dby in two ways, and can
be in existence in the on
e of four states,
namely pe
rce
p
tion, forwa
r
d
i
ng, sen
s
in
g and forwa
r
din
g
dorm
a
n
c
y.
The clu
s
ter
head a
s
the
network ma
nagem
ent
ce
nter, energy cha
nge can
monitor
node
s, four state de
ci
sion an
d mai
n
tain
the
se
nso
r
. Based
on the two nod
e en
e
r
gy
con
s
um
ption,
delay
optimi
z
ation
pe
rformance in
dex
to calculate
the path
c
o
s
t
func
tion.
The
clu
s
ter he
ad
node
s u
s
ing t
he co
st functi
on as the lin
k cost.
Routin
g st
rat
egy TEEN p
r
otocol i
s
rea
c
tive wi
rele
ss sen
s
o
r
n
e
twork an
d de
si
gn, it is
real
-time, ca
n re
spo
nd ra
pidly to eme
r
gen
cie
s
. TE
EN and
LEACH
usi
ng th
e stru
ctu
r
e a
nd
operation of
multi clu
s
ter in the
same
way, is diffe
re
n
t
in the course of e
s
tablish
i
ng cl
uste
r, with
the sele
cted
clu
s
ter n
ode
s, cluste
r hea
d in addi
tion
to reali
z
e dat
a sched
uling
by TDMA wa
y,
also h
a
rd
clo
s
e to clu
s
ter
membe
r
broa
dca
s
ting d
a
ta
about the value and soft clo
s
e value
s
of
two paramet
ers. Hard clo
s
e value is m
onitore
d
data
can not be crosse
d with value, soft clo
s
e
value is spe
c
i
f
ied by range
monitori
ng d
a
t
a. In
the sta
b
l
e sta
ge
clu
s
ter, no
de th
ro
ugh the
sen
s
or
con
s
tantly a
w
are
of its
surroun
di
ng
s.
Whe
n
a
no
de
first
monito
ri
ng d
a
ta to
ha
rd
clo
s
ing
val
ue,
then op
en th
e tran
sceiver for data tra
n
smi
ssi
on,
a
nd the dete
c
t
i
on value
sto
r
ed in th
e no
de
internal va
ria
b
les in the S
V
, as is sh
own by Equation
(7).
1
2
2
(
)
s
i
n
(
)
(
()
)
c
o
s
(
)
()
22
2
2
l
j
Ll
l
jj
j
x
xx
x
qx
l
h
(7)
This pa
per in
trodu
ce
s th
e
con
c
e
p
t an
d
cha
r
a
c
te
ri
st
ic
s of
wi
rele
s
s
sen
s
o
r
net
w
o
rk
s,
t
h
e
analysi
s
an
d
desi
gn of net
work
routing
proto
c
ol
chall
enge, given
by the algorit
hm of wirele
ss
sen
s
o
r
net
wo
rk b
a
sed on
ant colo
ny o
p
timizati
on
al
gorithm. Th
e routing
algo
ri
thm con
s
id
eri
n
g
the ch
ara
c
te
ri
stics of the n
e
tw
ork, with
self-organi
zati
on, dy
nami
c
.
Pherom
one method con
s
i
der
the ene
rgy, round tri
p
time
, hops
and
other fa
ctors.
T
herefo
r
e m
o
re suita
b
le for
wirel
e
ss
sen
s
or
netwo
rk routi
ng.
The first ACO algo
rithm
ant system
pra
c
tical
pri
n
cipl
e and
a
l
gorithm m
o
del and
descri
be, ant system is b
e
st ex
plained a
s
the basi
c
id
ea that ACO meta heuri
s
ti
c algo
rithm. The
basi
c
prin
cipl
e of a
n
t
syst
em to
pave t
he
way fo
r
th
e
cha
r
a
c
teri
stics, meta he
uristi
c algo
rithm,
can th
e meta
heu
risti
c
fra
m
ewo
r
k p
r
op
ose
d
by AC
O, the alg
o
rit
h
m of a
s
soci
ation definitio
n of
combi
nato
r
ial
optimization
pro
b
lem
s
,
ACO m
e
ta h
euri
s
tic i
s
i
n
deed
a
com
pone
nt and
the
operation mo
de of the ba
sic [10]. In this fram
ework, the ant system is a
s
an example
of
algorith
m
, an
d ca
n be
de
ri
ved from th
e
other
algo
rith
ms, altho
ugh
other
de
rivative algorith
m
is
pre
s
ente
d
al
one, but the ACO meta h
euri
s
tic fr
am
ewo
r
k i
s
the best su
mma
ry of
their. Then
descri
b
e
s
se
veral vari
ants of typical A
C
O al
go
ri
thm
,
these
are
b
e
tter than th
e
perfo
rma
n
ce
o
f
the origin
al a
n
t colony sy
stem has g
r
e
a
tly increa
se
d.
In the static n
e
twork, al
so
can
not gua
rantee the int
egrity
of the
netwo
rk to
pol
ogy. After
the formation
of the network to
polo
g
y, may enc
o
u
n
t
er many un
expecte
d situ
ations, such
as
netwo
rk
nod
e
positio
n, net
work n
ode
s h
a
ve no e
nerg
y
, or is
subj
ect to certai
n ob
stacl
e
s
blo
c
ked
and
so
on,
will have a
n
im
pact
on the
n
e
twork to
pol
o
g
y, and n
e
twork ap
plicatio
n effect. In th
e
desi
gn of ro
u
t
ing proto
c
ol
sho
u
ld be full
y taken
into
accou
n
t the chara
c
te
risti
c
s of the netwo
rk
topology dyn
a
mics, in ord
e
r to red
u
ce the in
fluen
ce
brou
ght by the netwo
rk n
o
de failure.
4. Optimizati
on of Ro
utin
g Protocol in
Wireless Se
nsor Net
w
o
r
ks Ba
sed on
Impro
v
ed Ant
Colon
y
and
Particle S
w
a
r
m Algorithm
Perform
a
n
c
e
evaluation
of WSN
rout
ing pr
otocol can be
me
a
s
ured
from multiple
asp
e
ct
s. The
analysi
s
of
the routin
g
proto
c
ol,
he
re from
wh
e
t
her the top
o
logy of ro
uting
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 748
6
– 7494
7492
proto
c
ol
s, protocol p
e
rfo
r
mance,
the n
e
twork life time, need to
maintain mult
iple path
s
, ro
bust,
scalabl
e routi
ng p
r
oto
c
ol i
n
the n
e
two
r
k Fe
stiv
al, whether to p
r
o
v
ide QoS
su
pport fo
r m
o
bility
sup
port, or
whether to p
r
o
v
ide se
curity mech
ani
sm e
t
c. the summ
ary.
Location info
rmation
of m
any ro
uting
proto
c
ol
s for wirele
ss
se
nso
r
n
e
two
r
ks requi
re
sen
s
o
r
nod
es. Becau
s
e th
e WSN d
o
e
s
not have access to a mechani
sm simil
a
r to IP addre
ss,
and is di
stri
b
u
ted in a pa
rticula
r
area, they can m
a
ke the tran
sm
i
ssi
on of data
in some e
n
e
r
gy
saving metho
d
s usin
g
the
address
i
n
formation.
Th
erefore, if
kno
w
n in
du
ction
are
a
, u
s
ing
the
sensor locati
on, query informati
on
will be issued only
to the perc
ei
ved area, reduce the number
of data tran
smissi
on.
Becau
s
e th
e
traditional PS
O is often th
e se
arch in t
he middl
e of
the global
an
d local
best po
sition
, searchin
g cap
ability and conve
r
ge
n
c
e pe
rform
a
nce i
s
heavil
y depende
nt on
con
s
tant acceleratio
n
and
inertia weig
ht settings
, in orde
r to overcome the shortcomin
gs,
the
Gau
ss fun
c
ti
on wa
s intro
duced to the
PSO algorit
hm, is used to guide the
movement of
th
e
particl
es; GP
SO does n
o
t requi
re the i
nertia weight,
and co
nsta
n
t
accel
e
ratio
n
generated by
random
the number of
Gauss
distri
but
ion. Tension will
be
call
ed the PSO:S
PSO stretchi
ng
techni
que
an
d defle
ction
and
rej
e
ction
tech
nolo
g
y
applie
d to th
e PSO, to
tra
n
sform the
ta
rget
function, limit
the p
a
rticl
e
t
o
have
foun
d
a lo
cal mi
nim
u
m
solution
o
f
motion,
whi
c
h
help
s
to
h
a
ve
more of a cha
n
ce to find th
e global o
p
timal solutio
n
, as is
sho
w
n b
y
Equation (8
).
1
11
()
tt
t
t
ij
t
i
j
i
j
i
j
v
v
c
r
pbe
s
t
x
(8)
Cha
nge the
p
a
rticle i
nertia
weig
ht swarm algo
rithm
a kin
d
of dyn
a
mic. In the
algorith
m
,
the algo
rithm
cha
nge
ine
r
tia wei
ght op
e
r
ation
e
ffect,
by the evoluti
on
spee
d
of particl
e swarm
optimizatio
n and
p
a
rticl
e
swarm agg
re
gation
d
egr
e
e
com
p
rehe
n
s
ive de
ci
sion
. Compa
r
e
d
with
the LDW alg
o
rithm, the average n
u
mb
er of iteratio
ns at least
a
n
averag
e re
ductio
n
of 25%,
conve
r
ge
nce spe
ed an
d accuracy a
r
e im
proved o
b
vio
u
sly. Che
n
G
u
imin, on the
basi
s
of LDW,
and give
s three
kind
s of
nonline
a
r d
e
crea
sing in
er
tia weight
strategy, fou
nd that for
most
contin
uou
s o
p
timization p
r
oblem
s, the strategy
of d
e
crea
sing
co
ncave fun
c
tio
n
is better th
an
that of linear strategy, while t
he linear
strategy is bett
e
r than
the
co
nvex function
strategy.
Elitist ant system, ran
k
b
a
sed ant
syste
m
,
ant col
ony
system
and
the maximu
m
min ant
algorith
m
, ant colony optimization al
go
rithm acco
rd
i
ng to the basic ant colo
ny algorithm for its
sho
r
tco
m
ing
s
, improve th
e
perfo
rma
n
ce
of the al
gorit
hm. Re
gardle
s
s of is the
b
a
si
c ant
col
o
n
y
algorith
m
an
d improve
d
a
n
t colony al
g
o
rithm, is
the
single
sp
eci
e
s, sin
g
le al
g
o
rithm ba
se
d
on
pheromo
ne,
not made
fu
ll use
of pa
rallel a
nd
di
stribute
d
co
mputing a
n
d
other
excell
ent
cha
r
a
c
teri
stics of ant colo
n
y
algorithm, can not fu
lly reflect the com
p
lexity of real ant soci
ety.
Overweight
may lead to seriou
s net
work co
nge
stion
and even con
gestio
n
colla
pse loa
d
node
s
clo
s
e
r
to the ba
se
st
ation in
wirel
e
ss
sen
s
o
r
n
e
twork,
sen
s
or n
ode
s n
e
a
r
the
ba
se
sta
t
ion
is sel
e
cte
d
from one of the mole
cula
r aggr
e
gation
node, other node
s only throug
h the su
b
assembly co
mmuni
cate
s with
the sin
k
node
c
an, a
s
the
se
nsor node
s i
n
co
mmon u
s
e
m
any
different ant
in ant colony
algorith
m
b
e
twee
n ph
eromone i
n
tera
ction a
nd dy
namic upd
ate to
quickly find the optimal pat
h from the so
urce
nod
e to the sin
k
no
de
co
st small
so
n.
By using the
object o
r
ient
ed OM
NeT
+
+ as th
e sim
u
lation platfo
rm of multi ant colony
algorith
m
an
d efficient
acce
ss load
a
w
are
cro
s
s l
a
yer routing
proto
c
ol
ba
sed on
Desi
g
n
of
experim
ent. Perceived a
s
(0,0) to
(1
84
51,665
21)
sq
uare plan
ar monitori
ng
a
r
ea,
se
nsor n
ode
s
are
ran
domly
deploye
d
2
6
0
, simul
a
tion
time is
120
0s. Con
s
ide
r
in
g
the reality of
se
nsor
network
node, the no
de is the ma
ximum transmissi
on di
sta
n
ce i
s
set to 90m, the fixed sin
k
nod
e
s
locatio
n
is
(4
5125
4). Tran
smissio
n
of the data p
a
cket size is 1
0
2
4
B, buffer qu
eue len
g
th is set
of sensor no
d
e
s into 100 d
a
ta packets, the sou
r
ce no
de gene
rate
s data at a rate of 10packet
/
s.
The IEEE802
.11 protocol
use
s
MA
C la
yer. Ope
r
at
io
n of improve
d
pa
rticle
swarm o
p
timizat
i
on
algorith
m
and
ant colony al
gorithm to op
timize t
he pro
g
ram of wi
rel
e
ss se
nsor n
e
twork routin
g,
averagi
ng the
result
s of 20
experim
ents t
o
comp
are, in Figure 2.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Optim
i
zation of Routing Protocol in
Wire
less
Sensor
Networks b
y
I
m
pr
ove
d
… (T
ianshun
Hua
ng)
7493
Figure 2. Co
mpari
s
o
n
of Routin
g Proto
c
ol in Wi
rele
ss Sensor
Net
w
orks Based
on Improve
d
Ant
Colo
ny with Particle S
w
arm
Algorithm
From the
sim
u
lation results it is not diffi
cult
to see th
at, the numb
e
r of ant
s the
size of M
ant col
ony a
l
gorithm
cycl
es
(converge
nce
)
influ
e
n
c
e incre
a
ses
linearly
cha
n
ge, wh
en th
e
number of
ants is too la
rge (such
as cl
ose to the
scale of
the problem),
while the
stability and
global
search
is improved,
but th
e al
go
rithm
c
onve
r
gen
ce
rate
sl
ow. A
numb
e
r of
ants in
a
n
t
colony algorit
h
m m
choice, shou
ld
also consi
der th
e global
search ab
ility and convergence
spe
ed of two
indicators.
Thro
ugh theo
retical a
nalysis and sim
u
la
tion sho
w
tha
t, the particle swa
r
m optim
ization
algorith
m
for wirel
e
ss se
nsor netwo
rk
b
u
sin
e
ss
dyna
mic
sched
ulin
g, a MT
CH a
t
the en
d of t
h
e
subframe is
starting a M
T
CH frame
of a great
probability. According to
thi
s
conclusion,
onl
y
indicates the
MTCH o
c
cu
pied by the sub fra
m
e in
DSI in the
starting e
nd
sub fra
m
e o
r
subframe, gai
n scheduli
ng
informatio
n correspon
di
ng
to the user can be accu
ra
tely. Compared
with the
meth
od of i
ndi
cati
ng e
nd
su
b frame, i
ndi
cati
ng meth
od
st
arting
su
b fra
m
e's presen
ce in
MTCH the
e
nd of
sub f
r
a
m
e is not g
o
od po
siti
oni
n
g
problem. B
a
se
d on
the
combi
nation
o
f
advantag
es of
indication method
e
nd sub
f
r
ame
at the sa
me ti
me, throu
gh
further
analy
s
is
found, an
ant
colo
ny algo
ri
thm occu
pied
sub
fram
e n
u
mbe
r
not m
u
ch, the
ab
solute po
sition
of
the sub fram
es en
d sub frame indi
cate
s the i
ndex m
e
thod still ha
s som
e
red
u
n
dan
cy, this paper
prop
oses a e
nd sub frame
differen
c
e i
n
dex sche
me
points, to fu
rt
her
red
u
ce th
e overhea
d o
f
DSI informati
on.
5. Conclusio
n
The pa
per
prese
n
ts optimi
z
ation of rout
ing
protocol in wirele
ss
se
nso
r
net
works ba
sed
on imp
r
oved
ant col
ony an
d parti
cle
swarm al
gorith
m
. This p
ape
r u
s
es the a
n
t
colony al
gorithm
and imp
r
ove
d
particl
e swarm alg
o
rith
m to optimiz
e the routing
in wirele
ss sen
s
o
r
network,
according to t
he speci
a
l re
quire
ment
s of netwo
rk
, the
node
s in a
si
ngle ho
p del
a
y
and statisti
cs
obtaine
d MA
C laye
r
nod
e
acce
ss effi
ciency, lo
ad
p
a
ram
e
ters
of que
ue l
engt
h info
rmation
as
routing p
a
th sele
ction. Experim
ents
sh
ow that, t
he algorith
m
to solve the opti
m
al tran
smission
path that
ca
n meet th
e
need
s
of wi
reless
se
ns
or networks,
real-tim
e,
reli
ability and
l
oad
balan
cing
an
d othe
r re
qui
rements, to
en
sure the
even
t driven
wirel
e
ss sen
s
o
r
n
e
twork
quality of
serv
i
c
e.
Ackn
o
w
l
e
dg
ements
This p
ape
r i
s
supp
orted
by the natio
nal scie
nce
and te
chn
o
lo
gy sup
port p
r
og
ram
(SQ20
13BAY4553
).
Referen
ces
[1]
J Ibriq, I Mah
gou
b. Cluster-
B
ased r
outin
g
in
w
i
r
e
less s
ensor
net
w
o
rk
s: Issues and
chall
e
n
g
e
s
.
Performanc
e Evalu
a
tion
of Co
mp
uter T
e
lec
o
mmu
n
icati
on S
ystem
. 20
04; Jan: 759-
76
6.
[2]
Lon
g Che
ngz
h
i
, Luo Jia
npi
ng
, Xia
ng Manti
a
n, Yu
Guicai. Rese
arch on t
he Ant Col
o
n
y
Optimizatio
n
Algorit
hm for Load Ba
la
nce in
W
S
N.
JCIT
. 2
012; 7(1
8
): 425
- 433.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 748
6
– 7494
7494
[3]
Yunl
ian
g
Pa
n,
Jun
Du, W
e
it
ao H
u
. An
ant
colo
n
y
o
p
timi
zation-s
upp
ort
vector mach
in
e for gr
ou
nd
target optimal tracking.
IJACT
. 2013; 5(9): 69
6- 703.
[4]
D Gan
e
san,
R
Govind
an, S S
henk
er, D Estri
n
, Hig
hl
y
Resi
li
ent. Ener
g
y
Efficient M
u
lti
path
Routi
n
g
i
n
Wireless Sens
or Net
w
orks.
Mobil
e
Co
mputi
n
g and C
o
mmu
n
icati
on Rev
i
e
w
(MC2R)
. 200
2; 1(2).
[5]
Andi Mu
hamm
ad Il
yas, M N
a
tsir Rahm
an.
Econom
ic
Dis
patch T
hermal
Generator Us
ing Mo
difi
e
d
Improved Particle S
w
arm Opt
i
mization.
T
E
L
K
OMNIKA Ind
ones
ian
Jo
urn
a
l of E
l
ectric
al
Engi
ne
erin
g
.
201
2; 10(3): 45
9-47
0.
[6]
D Braginsky
, D Estrin.
Rumor
Routin
g Algor
i
t
hm for Sensor
Netw
orks
. in the Proce
edi
ng
s of the F
i
rst
W
o
rkshop o
n
Sensor N
e
t
w
or
ks and App
lic
ations (W
SNA), Atlanta, GA. 2002.
[7]
N Hi
gash
i
, H
Iba. Particle
s
w
a
rm optimization
w
i
t
h
Gaussian mutation.
Institute of El
ectrical
an
d
Electron
ics En
gin
eers
. 20
03; (4): 72-79.
[8]
Xi
on
g Ji
an-
qia
ng, H
u
a
ng J
u
-
hua,
Li
ao Qu
n.
Simul
a
tio
n
A
n
al
ysis
of C
ontr
o
lli
ng
Buffetin
g
Nois
e B
a
se
d
on Hy
br
id Particle S
w
arm Algorithm.
JCIT
. 2
012; 7(2
3
): 753
- 761.
[9]
Haij
ia
ng W
a
ng
, Shan
lin
Ya
n
g
. Electricit
y
C
onsum
ption
Pr
edicti
on B
a
se
d
on S
V
R
w
i
t
h
Ant Co
lo
n
y
Optimization.
T
E
LKOMNIKA Indo
nesi
an Jo
u
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2013; 1
1
(11):
692
8-69
34.
[10]
Banks A, Vi
nc
ent J, An
yak
o
ha C. A
rev
i
e
w
of
partic
l
e s
w
a
rm o
p
timiza
tion.
Part I: ba
ckgrou
nd
an
d
deve
l
op
ment N
a
tural C
o
mputi
n
g
. 200
7; 45(6)
: 467-48
4.
Evaluation Warning : The document was created with Spire.PDF for Python.