TELKOM
NIKA
, Vol.13, No
.1, March 2
0
1
5
, pp. 202~2
1
0
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i1.790
202
Re
cei
v
ed O
c
t
ober 1
5
, 201
4; Revi
se
d Ja
nuary 13, 20
1
5
; Acce
pted Janua
ry 2
8
, 20
15
Modified Greedy Physical Link Scheduling Algorithm
for Improving Wireless Mesh Network Performance
Nach
w
a
n Mu
fti Adria
n
s
y
ah*, Muhama
d Asv
i
al, Bagio Budiardj
o
Electrical E
ngi
neer
ing D
e
p
a
rtment, Univers
i
tas Indon
esi
a
Kampus Bar
u
UI Depok 1
6
4
2
4
, Indon
esia
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: nach
w
a
n
.muf
ti@ui.ac.i
d
A
b
st
r
a
ct
T
he alg
o
rith
m
to alloc
a
te
me
sh active l
i
nk t
o
radi
o reso
ur
ce timeslot i
n
w
i
reless
mesh
netw
o
rk
(W
MN) is inve
stigated. T
h
is
pap
er pr
o
pose
s
the nove
l
me
thod to al
locat
e
multip
le li
nks
in on
e timeslo
t
for
improvi
ng the
w
i
reless mes
h
netw
o
rk throug
hp
ut vi
a spatia
l time div
i
sion
mu
ltip
le
access (ST
D
MA)
protoco
l
. T
h
e
throu
ghp
ut i
m
p
r
ove
m
e
n
t is
o
b
tain
ed
by
mo
difyin
g gr
ee
dy
bas
ed
al
gorit
hm that
is w
i
d
e
ly
know
n as
a
l
o
w
complex
i
ty a
l
gorit
hm.
W
e
p
r
opos
e
and
i
n
v
e
stigate
n
e
w
p
a
ra
meters
in
t
he
gre
edy
bas
ed
alg
o
rith
m th
at can b
e
us
ed
a
s
sched
uli
ng c
ontrol
par
am
et
ers, i.e. interfe
r
ence
w
e
i
ght, sched
uli
ng
w
e
i
ght,
and th
e su
m
of link
’
s
degr
e
e
. Simulati
on
results in
dic
a
te that this ap
proxi
m
ati
on i
n
creases n
e
tw
ork
perfor
m
a
n
ce i
n
throug
hput a
n
d
len
g
th of sch
edu
lin
g per
f
o
rma
n
ce cl
ose
d
to the up
per b
oun
d perfor
m
a
n
ce
that is achiev
e
d
by the al
gorit
hm th
at
uses the phys
i
cal i
n
terferenc
e mod
e
l.
Ke
y
w
ords
: W
M
N, ST
DMA, g
r
eedy a
l
gor
ith
m
, link sch
ed
ul
ing
1. Introduc
tion
The me
sh o
r
multi-hop to
pology cure
n
t
ly is
seen a
s
a strong
candid
a
te top
o
logy to
improve
wire
less net
work cap
ability and ha
s
a n
u
m
ber
of advantage
s an
d
use
d
for m
any
appli
c
ation
s
[
1
]–[3].
R
e
ce
ntly, w
i
r
e
les
s
me
s
h
n
e
t
w
o
rk
ha
s
e
m
er
ge
d
as
an
in
ter
e
s
t
ing
r
e
se
ar
c
h
area
an
d
gai
ning
attention
from
re
se
archer to imp
r
ov
e this tipical
n
e
twork in
different
a
s
pe
cts [1]
[4][5].
Specifically to improve r
adio resource efficiency utilization
in the multi-hop wireless mesh
netwo
rk
(WM
N
), Nelson a
nd Kleinrock
[6] introdu
ce
the spatial ti
me division
multiple acce
ss
(STDMA
). Th
e STDMA p
r
o
t
ocol a
c
cess
enabl
es to all
o
cate m
u
ltipl
e
links o
r
nod
e tran
smi
ssio
n
s
in one
time
sl
ot as long
a
s
those
con
c
u
rre
nt
tran
smi
ssi
on
s do
no
t degrade
si
g
nal qu
ality. The
mesh lin
k scheduli
ng prob
lem is prove
d
as NP har
d probl
em [7] and there a
r
e
many algo
rithms
develop
ed in STDMA proto
c
ol acce
ss [7]–[11]. T
he crucial obj
ectiv
e
of the mesh
link sched
uli
ng
optimizatio
n probl
em is to
maximize the
numbe
r of
links t
hat
is all
o
cat
ed in t
i
me
slot
re
sou
r
ce
s.
Gupta a
nd
Kumar [1
2] introdu
ce t
w
o
interferen
ce
model
s for the link
scheduli
n
g
cap
a
city eval
uation, i.e. p
r
otoc
ol an
d
physi
cal inte
rferen
ce m
o
d
e
l. Due to it
s sim
p
licity, the
proto
c
ol
inte
rfere
n
ce m
o
del i
s
widel
y use
d
by
rese
arche
r
s t
o
devel
op t
he ST
DMA
link
sched
uling
al
gorithm
s [1
3]–[15]. On
the
otherh
and,
th
e alg
o
rithm
s
t
hat a
r
e
devel
oped
in
physi
cal
interferen
ce
model i
s
pro
v
en to have better
pe
rf
orma
nce tha
n
the other
algorith
m
s
wi
th
comparable
developed i
n
protocol
i
n
terference
model [16],[
17], but always in
higher
comp
utationa
l co
mplexity. For thi
s
re
ason, we n
eed
to de
sig
n
m
e
sh lin
k
sched
ul
ing al
gorith
m
in
physi
cal interf
eren
ce m
odel
with low com
put
ational
co
mplexity and a good p
e
rfo
r
mance.
Gree
dy ba
se
d algo
rithm i
s
kno
w
n to
have
a lo
w
compl
e
xity algorithm fo
r reso
urce
allocation. Brar et al int
r
od
uce
gre
edy p
h
ysical
algo
ri
thm [18] as
STDMA alg
o
rithm in physi
cal
interference
model [19],[20]. T
he best performance
of physi
cal S
T
DMA lin
k scheduli
ng given by
Gore
et
al [1
0], namely
si
gnal to
inte
rfe
r
en
ce
ratio
g
r
aph li
nk sch
e
dule
(SGLS
)
algorith
m
. S
G
LS
algorith
m
p
r
o
v
ide the
be
st
perfo
rman
ce
of me
sh li
nk
sched
uling
by
iterativ
ely
ch
eck SINR fo
r
all
links that
sch
edule
d
in
the
same
time
slo
t, so th
at
this
algorith
m
h
a
ve hig
h
e
s
t net
work thro
ugh
put
kno
w
n
so far.
The time co
mputational
complexity of
SGLS algorit
hm is re
porte
d
.
In this pa
per,
we p
r
op
ose
a ne
w ap
pro
a
ch i
n
the g
r
eedy ba
se
d
STDMA lin
k
sched
uling
algorith
m
tha
t
can be u
s
e
d
to improve
mesh n
e
two
r
k perfo
rma
n
ce. The impro
v
ement of mesh
netwo
rk th
ro
ughp
ut is a
c
hieve
d
by i
n
trodu
ci
n
g
n
e
w
sched
uli
ng pa
ram
e
te
rs th
at ca
n
be
controlled. Th
e target of propo
sed alg
o
ri
thm is
highe
r network thro
ughp
ut than gree
dy physi
cal
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Modified Greedy
Phy
s
i
c
a
l
Link
Scheduling Algorithm
for Improv
ing
....
(Nac
hwan Mufti A.)
203
algorith
m
[18
]
and
clo
s
ed
to to the
be
st re
sult
th
at is give
n by
physi
cal inte
rferen
ce
ba
se
d
sched
uling p
e
rform
a
n
c
e [21].
The p
r
o
b
lem
formulatio
n of
STDMA li
nk
sched
uling
is
descri
bed
a
s
follows. The
i
nput of
link sched
ulin
g algorithm is a communication grap
h
,
, i.e. the
mesh topolo
g
y that
con
s
i
s
ts
of all the n
o
d
e
s a
nd th
e a
c
tive co
mmu
nicatio
n
lin
ks with
signal
to noi
se
ratio
(SNR)
above
the
minimum threshold,
. The set of nodes is rep
r
e
s
en
ted by
,
,…,
and the set of
active comm
unication links is represen
ted by
,
,…,
. The
link sched
ulin
g algorithm
s
intend to
allocate th
e dif
f
erent me
sh
active
com
m
unication li
nks
, that is geog
ra
phica
lly
sep
a
rate
d, to
the
certai
n ti
meslot. S
o
t
hat the
si
gn
al
to
interfe
r
e
n
ce and
noi
se ratio (SINR) o
f
all
links in thi
s
ti
meslot
sh
oul
d sati
sfy the
minimum th
re
shol
d
. If link
is all
o
cated i
n
the
same
times
l
ot with
link
, this
con
d
ition mu
st always
satisfy of ph
ysical i
n
terfe
r
en
ce m
odel
requi
rem
ent [12],
ij
ij
c
o
kV
kj
ki
P
d
SI
NR
P
N
d
(1)
The o
u
tput o
f
the mesh li
nk
sched
ulin
g algo
rithm i
s
the m
e
sh l
i
nk
sched
ule
.
,
,
,…,
, where
is the numbe
r of time slots to complete the schedul
ing task. In
wirel
e
ss me
sh netwo
rk lin
k sche
duling
optimizatio
n probl
em, we
have an obje
c
tive function
to
maximize
net
work th
rou
g
h
put (2
). The
constraint
s are
:
all links a
r
e
sched
uled
at least o
ne al
o
ng
the length
of sched
uling
(3
), ea
ch n
ode
can
not tr
an
smit and recei
v
e in one tim
e
slot
(4), a
n
ode
can
tran
smit/
r
eceive o
n
ly to/from on
e
other
nod
e
in
one tim
e
sl
o
t
(5-6), a
nd a
ll tran
smissio
n
s
must satisfy physical interfe
r
en
ce mo
del requireme
nt (7).
The inte
ge
r li
near p
r
og
ram
m
ing
(ILP)
re
pre
s
entatio
n of
STDMA
li
n
k
sched
uling
probl
em
as
follows
:
Objective:
max
(2)
Con
s
trai
nts:
∑
1
∀
∈
(3)
∑
∈
∑
∈
1
∈
,
∈
(4)
∑
∈
1
∀
∈
,∀
(5)
∑
∈
1
∀
∈
,∀
(6)
∑
∈
∖
∀
∈
,
∀
(7)
This pa
pe
r is org
ani
zed
as follows. In se
ction 1, we provid
e
related works in the
research area and probl
em form
ulation. In se
ction 2 and
3, we illust
rate the proposed l
i
nk
sched
uling al
gorithm d
e
scription an
d then followe
d
by resea
r
ch
method. Fin
a
lly, simulation
results an
d di
scussio
n
, and
con
c
lu
sion a
r
e
presented i
n
se
ction 4 a
nd 5 re
spe
c
ti
vely.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 1, March 2
015 : 202 – 2
1
0
204
2. The Propo
sed Algori
t
h
m
The b
a
si
c of t
he g
r
ee
dy ph
ysical
algo
rith
m for ST
DM
A proto
c
ol
acce
ss
(GP
-
ST
DMA) i
s
introdu
ce
d b
y
Brar et al [18]. The princi
ple
of this algo
rithm is to perfo
rm sch
eduli
n
g by
interferen
ce
evaluation
fo
r ea
ch
lin
k. To
analyze
interfe
r
en
ce
, the GP-ST
D
MA al
gorit
hm
transfo
rm
s the commu
nica
tion grap
h into a conflict g
r
aph. The lin
k with the largest num
ber
of
interference will
be scheduled
firs
t
after link’s interfer
ence sorting. The lin
k’s interference i
n
comm
uni
cati
on gra
ph is
repre
s
e
n
ted b
y
degree
of
node in the co
nflict gra
ph. Gree
dy algo
ri
thm
is
kno
w
n to
p
r
ovide l
o
w co
mputational
complexity
, bu
t not provide
optimal p
e
rfo
r
man
c
e.
On t
he
otherh
and, G
o
re et al [10]
introdu
ce S
G
LS-ST
D
MA
algorithm th
at is provid
e the near
opti
m
al
perfo
rman
ce
but high
er in
compl
e
xity. In this p
ape
r, we
sho
u
ld a
c
hieve
netwo
rk p
e
rfo
r
ma
n
c
e
near to the p
e
rform
a
n
c
e a
c
hieve
d
by [10]
but with lowe
r algo
rith
m compl
e
xity.
In this sectio
n, the improved
greedy p
h
y
sical
sp
atial time division
multiple a
c
cess i
s
prop
osed. In contrast with
the
basi
c
gre
edy physi
cal algorith
m
in
[18], the prop
ose
d
algo
rith
m
do not pe
rform the grap
h tran
sform
a
tion
. The c
onfli
ct free sch
eduli
ng is gu
ara
n
t
eed an
d de
ci
ded
based on three interfe
r
en
ce pa
ramet
e
rs: the block
p
a
rtition evalu
a
tion, the su
m of link’s de
gree
evaluation, a
nd the sched
uling wei
ght evaluation. After that, the
physi
cal interference model is
use
d
to guara
n
tee SINR re
quire
ment.
From
th
e co
mmuni
cation grap
h
,
, the set of active
comm
uni
cati
on lin
ks
are
evaluated
so
that the can
d
idate of
lin
ks that ca
n be
allocate
d in
one time
slot can b
e
defin
ed.
The area of the me
sh net
work i
s
divide
d to some
bl
o
c
ks pa
rtition, whe
r
e thi
s
pa
rtition method
is
use
d
p
r
eviou
s
ly in [22][23
]. The num
b
e
r of bl
ock t
hat is
use
d
i
n
this
pape
r
is
10
10
100
partition blo
c
ks. Norm
atively, the link
can b
e
allo
cated in the same timesl
ot with link
if
satisfy re
quirements a
s
fol
l
ows (Fig.1
):
-
Nod
e
in different blo
c
k p
a
rtition with its interfe
r
en
ce
node
s (no
de
, and node
)
-
Nod
e
in different bl
ck p
a
rtition with its interferen
ce n
ode
s (no
de
and nod
e
)
i
v
j
v
ij
e
kl
e
k
v
l
v
Figure 1. Mesh active link i
n
four blo
ck p
a
rtition
For inte
rfere
n
ce q
uantificati
on, the su
m of link’s d
egre
e
and
the interfere
n
ce
weig
ht
are
defined. T
h
e
sum
of link’
s de
gre
e
of
link
is defin
ed a
s
whe
r
e
is d
egree of n
ode
. The lin
ks with
highe
r
mea
n
s that th
ose
links on
den
se network
topology in th
e evaluated
a
r
ea b
e
cau
s
e t
h
is lin
k is
c
o
n
necte
d with
many of the li
nks. So that, the
links with hig
her
is given prio
rity sch
ed
uling firstly in a gree
dy way
.
The oth
e
r qu
antification
of
interfe
r
en
ce
is rep
r
e
s
ente
d
by the i
n
terferen
ce
wei
g
ht that is
derived from the distance of two links. For link
dan link
, the inte
rfere
n
ce weig
ht
is
defined a
s
:
m
a
x
,
(8)
So that, the sche
duling
wei
ght can b
e
de
fined as follo
ws,
′
1
(9)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Modified Greedy
Phy
s
i
c
a
l
Link
Scheduling Algorithm
for Improv
ing
....
(Nac
hwan Mufti A.)
205
The sch
eduli
ng wei
ght
′
, is used a
s
th
e interferen
ce paramete
r
whe
r
e hig
h
e
r
′
means lon
g
e
r dista
n
ce
of two links,
so that the link with high
′
have a
high
prob
ability to
sched
ule. To
give high
lin
k sch
edul
i
ng
efficien
cy, scheduli
ng
start
s
from
lin
k wi
th
minimum scheduling weight to maximum value.
The final
eval
uation i
s
to
g
uara
n
tee SI
NR of lin
ks that
are
allo
cate
d
in pa
rticular timeslot
sho
u
ld b
e
satisfying mini
mum thresh
old of SINR. Con
s
id
er to
node
re
ceiv
er
for SINR
guarantee
with physi
cal int
e
rferen
ce mo
del, E
quation
10 ca
n be de
rived from Equation 7.
(10
)
whe
r
e th
e total interfe
r
e
n
ce
expe
rien
ced i
n
n
ode
so
urced
from a n
u
mb
er
of
node
r
e
pr
es
e
n
t
ed
a
s
as
follows,
∑
(11
)
So that, with sub
s
titute (1
0) to (11
)
ca
n be found t
he req
u
ireme
n
t of link allo
cation to satisfy
physi
cal interf
eren
ce
con
s
train in Equatio
n 7, as follows:
∑
(12
)
The pseud
ocode of the propo
sed
al
gori
t
hm is de
scri
bed bel
ow,
Proposed Algorit
hm:
Input:
,
→
,
,…,
,
Output:
Transmission schedule
,
,…
,
Steps of algorith
m
:
1. Sort
,
,…,
in decreasing orde
r based o
n
interference d
e
g
ree
2.
Candidate links generation
w
i
th bl
ock partition criteria evaluation
3. Initializat
ion:
;
4.
∅
5.
Select one link from
, i.e.
; allocate
in
;
←
6. Sort
in increasin
g order
′
7.
Select one candi
date e.g.
to be
e
v
aluated
w
i
th the
Ph
y
s
ical Interfe
r
ence Model.
If
, allocate
to
, then
⋃
8.
Return to
follow
i
ng candidates, re
peat step 7 to 8
until all candidat
es for
are evalu
a
ted
9.
, repeat step 5 t
o
9
w
here
until lin
k scheduling co
mpleted
10.
Compute link scheduling perform
a
n
ce
The a
s
ympt
otic time
co
mplexity ana
lysis i
s
u
s
e
d
to an
alyze the
comp
utational
compl
e
xity param
eter in the ce
rtain time con
s
tr
aint
. In the propose
d
algo
rith
m, computati
onal
compl
e
xity can be analy
z
e
d
from ea
ch p
r
ocess in the
algorith
m
.
The set of active links
is sorted ba
se
d on the su
m of the link’s deg
ree re
q
u
ire
s
Oe
log
e
operation
s
. The so
rting result is defin
ed as
as the input of th
e sch
edulin
g
algorith
m
. For each link el
ement
in
, block partition
evaluation is
perfo
rmed in
orde
r to
gene
rate ca
n
d
idate links that can be allocate
d con
c
urrently with
link
. The block partition
evaluation requi
re
s
e
operatio
ns. T
he cal
c
ulatio
n of sche
dul
ing weig
ht take
s
e
operation
s
. So that, the in
put of link
scheduli
ng ge
n
e
ration
re
quires
el
o
g
e
e
e
e
operation
s
.
In the alg
o
rit
h
m, the first step i
s
to
sel
e
ct a li
nk i
n
with a l
a
rg
e sum of the li
n
k
’s
degree to all
o
cate to the
first time sl
ot and
re
ad
its can
d
idate
links. Thi
s
pro
c
e
ss
req
u
i
res
|
|
. Furthermore, sortin
g the
can
d
idate li
nks
an
d sel
e
ct one lin
k
with highe
st sche
duling
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 1, March 2
015 : 202 – 2
1
0
206
weig
ht to be
SINR evaluat
ed requi
re
s
|
e
|
log
|
e
|
time, where
|
e
|
|
e
|
a
nd in the worst case
|
e
|
e
.
The phy
sical
interfere
n
ce
evaluation for
m
candidat
e links req
u
ire com
p
lexity
m
whe
r
e
m
is the numbe
r of links that is a
llocated to
the same timeslot and
m
|
e
|
. This
pro
c
e
ss i
s
re
peated
until a
ll links have b
een
compl
e
te
ly assi
gned.
So that, the complexity of this
p
r
oc
es
s
is
.
.
The time
co
mplexity in o
ne cy
cle al
go
rithm is
|
|
|
|
log
|
|
.
, w
h
er
e
|
|
~
. This process repeated for all links in
.
So that,
the
total time c
o
mplexity is
log
. F
u
rthe
rmo
r
e, i
n
the
wo
rst
case
of the to
t
a
l computatio
nal
compl
e
xity is ap
proximated
to
e
ee
l
o
g
em
≅
e
log
e
≅
Oe
.
3. Rese
arch
Metho
d
The m
odel
of link
sche
duli
ng p
r
o
c
e
s
s d
epicte
d
in
Fig
u
re
2. The
pu
rpo
s
e li
nk scheduli
n
g
in the wirel
e
ss me
sh net
works i
s
to maximize the
nu
mber of lin
ks
that are allo
cated in a time
slot
as a ra
dio resource.
Figure 2. Mesh link sch
edul
ing simul
a
tion
modeling
In this p
ape
r, the wi
rele
ss
mesh
net
work
pe
rforman
c
e mea
s
u
r
ed i
n
length
of scheduli
n
g
and network throug
hput.T
he
le
ngth of sched
uling
is
define
d
a
s
multiplicatio
n
of the
num
b
e
r of
timeslot that is req
u
ired to compl
e
te sch
edulin
g (
), with the timeslot
duration
(
), as
follows
:
(13
)
If
is the number of links that sch
e
dule
concurrently in timeslot
, the
n
link sche
dul
e in
times
l
ot
represe
n
ted by
,
,…,
.
Link sch
edul
e repeat
s pe
riodi
cally with
the fixe
d
pattern
alo
n
g
the
network ope
ratio
n
, so that
a p
a
ir of transmitter-rec
e
iver that trans
m
its
in
times
l
ot
will communi
cate
again in time
slot
,
2
…
and so o
n
.
Usi
ng the Sh
anno
n’s form
ula [24], the achi
evable d
a
ta rate for link
∈
, i.e.
ha
s
the uppe
r bo
und value a
s
:
l
o
g
1
(bps)
(14
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Modified Greedy
Phy
s
i
c
a
l
Link
Scheduling Algorithm
for Improv
ing
....
(Nac
hwan Mufti A.)
207
So that, the amount of bit
s
tran
sfe
rre
d
in one time
sl
ot is
, where
is timesl
ot
duratio
n and
is the bandwidth allo
ca
tion for wirel
e
ss mesh ne
twork. Such that, average
throug
hput fo
r all tran
smission lin
ks as f
o
llows:
∑
∑
∑∑
,
(bps)
(15
)
whe
r
e
is the numbe
r of active comm
u
n
icatio
n links
and
,
is the upper b
oun
d o
f
data rate
times
l
ot-
and
is the numbe
r of links tran
smi
tted concurrently in timeslot-
.
The me
sh
lin
k
scheduli
ng
perfo
rman
ce
is obt
ai
ned
b
y
Monte-Ca
rl
o sim
u
lation
method.
The statio
na
ry node po
siti
ons
are
ran
d
o
mly gene
rat
ed in a
squ
a
re are
a
for e
a
ch iteration a
n
d
then the process to evalu
a
te t
he cu
rre
nt mesh top
o
l
ogy is sta
r
te
d. For ea
ch t
opolo
g
y in one
iteration, all
of active lin
ks a
r
e
sched
uled
b
a
sed
on the al
go
rithm whi
c
h i
s
evaluated.
The
simulatio
n
was
repe
ated i
n
100
0 iterations
with
the
different p
o
si
tions of n
ode
s an
d differe
nt
netwo
rk top
o
l
ogie
s
.
4. Results a
nd Discu
ssi
on
For comp
ari
s
on, the basi
c
TDMA an
d some al
go
rith
ms that are
p
r
eviou
s
ly pro
posed for
STDMA prot
ocol a
c
ce
ss
are u
s
ed a
s
the perfo
rma
n
ce referen
c
e. The arb
o
ri
cal lin
k sche
dule
(ALS-ST
D
MA
) algo
rithm is base
d
on p
r
otocol
inte
rfe
r
en
ce mo
del
and propo
sed in [13]. The
gree
dy phy
si
cal
(GP-ST
DMA) alg
o
rith
m is
ba
sed
o
n
physi
cal
int
e
rferen
ce m
o
del an
d p
r
op
ose
d
in [18]. The SINR graph l
i
nk sch
edule
(SGLS-
ST
DMA) algo
rithm is based on pure phy
sical
interferen
ce
model an
d propo
sed in [10
].
From p
r
evio
us work, the evaluation
of link sch
edulin
g algo
rithm with si
mulation
para
m
eter
se
tting in Table
1 gen
erally
starts
from 30
node
s, be
ca
u
s
e the
com
m
unication g
r
a
p
h
,
not formed
completely in the num
ber of
node
s belo
w
than 30 no
d
e
s. Gen
e
ral simulation
para
m
eter
se
tting is sho
w
n
in Table 1.
Table 1. Simulation pa
ram
e
ters
setting
Parameters
S
y
mbol
V
alue
Band
w
i
dth
W
10 MHz
Frame d
u
ration
T
f
10 ms
OFDM
s
y
mbol d
u
ration
T
s
25 µs
Transmission pow
e
r
P
10 mW
Path loss expone
nt
β
4
Noise po
w
e
r
N
0
-90 dBm
Communication threshold
c
20 dB
Interfer
ence thre
shold
i
10 dB
Ar
ea cover
e
d
R
x
R
886 x 8
86 m
2
The sim
u
lati
on re
sults
sh
ow that the SG
LS-ST
D
M
A
algorithm
has b
e
tter th
roug
hput
than the oth
e
r
algo
rithm
s
that are
sim
u
lated (Fi
gur
e
3). Thi
s
is
du
e to this alg
o
r
ithm iterative
l
y
che
c
ks SINR and efficient
ly utilize the interferen
ce
margi
n
for in
cre
a
si
ng the
mesh net
wo
rk
cap
a
city. Th
e rang
e of
the SG
LS-ST
D
MA th
ro
u
g
h
put
value
s
a
r
e between
9.504 Mbp
s
(30
node
s) to
0.961 Mb
ps
(1
10 no
de
s). T
hat is a
n
up
per b
oun
d capa
city of throug
hput in t
h
e
physi
cal int
e
rf
eren
ce
mo
del
. Fro
m
Fig.
3
,
also
can
be
see
n
the
wi
re
less me
sh
ne
twork
cap
a
cit
y
decrea
s
e
in e
x
ponentially negative cha
r
acteri
stic. Th
i
s
i
s
b
e
cause
the net
work
capa
city network
is divided o
n
a numbe
r of active links in
the network.
The GP-ST
D
MA algorithm
that also u
s
i
ng t
he phy
sical interferen
ce model i
s
proved ha
s
better p
e
rfo
r
mance in
through
put a
nd
length
of
sch
edulin
g than
the ALS-ST
DMA that is u
s
ing
the protocol i
n
terferen
ce
model
. F
r
om
Figure 3
an
d
Figure 4
al
so
ca
n b
e
see
n
that the ST
DMA
algorith
m
s p
r
ovides bette
r perform
an
ce
than TDMA
as the ba
sic
acce
ss p
r
oto
c
ol in wirele
ss
me
s
h
ne
tw
ork
.
In the throug
hput and le
n
g
th of sched
u
ling pa
ramete
r, the pro
p
o
s
ed algo
rithm
provide
s
the perfo
rma
n
ce b
e
tter th
an co
nventio
nal GP-S
T
D
MA algorithm
and clo
s
e t
o
SGLS-ST
D
MA
algorith
m
pe
rforman
c
e a
s
the best
re
su
lt in this
re
se
arch area
so
far. The p
r
op
ose
d
algo
rith
m
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 1, March 2
015 : 202 – 2
1
0
208
provide
s
th
ro
ughp
ut in a
n
averag
e 7.5
1
% better tha
n
GP-ST
D
M
A
algo
rithm a
nd 7.4
8
% wo
rse
than SGLS
-STDMA al
gorit
hm pe
rform
a
nce. In th
e a
v
erage
length
of sched
ulin
g pa
ramete
r,
the
prop
osed alg
o
rithm provid
es 9.98% lo
wer than
the G
P
-STDMA an
d 8.99% high
er than SGL
S
-
STDMA. F
r
o
m
this re
sult,
the p
r
o
p
o
s
e
d
alg
o
ri
thm
i
s
p
r
ove
n
to
have b
e
tter
p
e
rform
a
n
c
e
than
GP-STDMA algorithm a
nd
clo
s
ed to SG
LS
-STDMA algorithm p
e
rfo
r
man
c
e.
Figure 3. Net
w
ork throug
h
put perfo
rma
n
ce
Figure 4. Len
gth of sch
edu
ling perfo
rma
n
ce
30
40
50
60
70
80
90
10
0
11
0
0
1
2
3
4
5
6
7
8
9
10
N
u
m
ber
of
nod
es
N
e
t
w
or
k
T
h
r
oughput
(
M
bps
)
TD
M
A
AL
S-ST
D
M
A
GP
-
S
T
D
M
A
SG
L
S
-ST
D
M
A
P
r
opo
s
e
d
a
l
gor
i
t
h
m
30
40
50
60
70
80
90
100
110
0
1
2
3
4
5
6
N
u
m
ber
of
nod
es
Le
ngt
h
os sch
edu
l
i
n
g
(
m
s)
TD
M
A
AL
S-ST
D
M
A
GP
-
S
T
D
M
A
SG
L
S
-ST
D
M
A
P
r
opos
ed algor
it
hm
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Modified Greedy
Phy
s
i
c
a
l
Link
Scheduling Algorithm
for Improv
ing
....
(Nac
hwan Mufti A.)
209
The adva
n
ta
ge of our
alg
o
rithm is i
n
the aspe
ct of comp
utationa
l compl
e
xity.
From the
compl
e
xity analysi
s
in Section 2, we
obtain
that the propo
sed algorith
m
have compl
e
xity
log
≅
. We can state that the ne
w sched
uling
para
m
eters that are introd
uce
d
in the
algorith
m
ca
n improve th
e ba
sic gree
dy algo
ri
thm
without
add
com
p
lexity signifi
cantly. For
comp
aration referen
c
e, ALS-STDMA al
gorithm com
p
lexity is
log
as reporte
d in
[13] , where
is the numb
e
r
of link,
is the numbe
r of node,
is the thickne
ss of
grap
h, and
is the maxim
u
m nod
e de
g
r
ee. Th
e GP-STDMA alg
o
rithm com
p
lex
i
ty is
and th
e SGLS-
STDMA complexity is
[10]. So, it can be co
ncl
ude
d that the propo
sed al
go
rithm giving
perfo
rman
ce
better than t
he conventio
nal greedy p
h
ysical algo
ri
thm in simila
r co
mputatio
nal
compl
e
xity. In othe
rsi
de, t
he p
e
rfo
r
man
c
e
of t
he
pro
posed
algo
rithm cl
osed to
the b
e
st
re
sult
that is achi
eved in the SGL
S
-STDMA alg
o
rithm in lo
wer co
mplexity.
5. Conclusio
n
This pa
per shows th
e o
p
portunity to
i
n
crea
se
wi
rel
e
ss m
e
sh n
e
t
work
perfo
rmance via
improvin
g greedy physi
ca
l algorithm fo
r STDMA p
r
o
t
ocol a
c
cess.
The improve
m
ent of network
throug
hput a
nd length of
sched
uling in
low com
put
ational compl
e
xity can be
achi
eved by an
extensio
n of interferen
ce p
a
ram
e
ters th
at is
de
rived from ge
ometri
cal no
de p
o
si
tions. We sh
o
w
that three pa
ramete
rs that
are intro
d
u
c
ed in th
is pa
per can be e
x
ploited in a gree
dy way to
improve
wirel
e
ss network perfo
rman
ce
clo
s
e to t
he b
e
st re
sult of mesh lin
k sch
edulin
g algo
ri
thm
performanc
e
in lower c
o
mplexity.
Ackn
o
w
l
e
dg
ement
The first aut
hor is a le
ct
ure
r
in Unive
r
sita
s Tel
k
o
m
that pursu
ing do
ctoral
degree in
Universitas Indonesia. Thi
s
work
is partly financed by Dikti
through
BPPDN schol
arship schem
e
.
Referen
ces
[1]
IF
Akyild
iz, X
W
ang, W
W
ang. W
i
reless m
e
sh n
e
t
w
orks:
a surve
y
.
Els
e
vier J. Com
p
ut. Networks
.
200
5; 47: 445
–
487.
[2]
R Brun
o, M C
onti, E Gre
gori
.
Mesh N
e
t
w
or
ks: Commodit
y
Multihop
Ad
Hoc Net
w
orks.
C
o
mm
un.
Mag. IEEE
. 2005: 123
–1
31.
[3]
MH Satria, J Yunus, E Supr
i
y
anto. Emerg
enc
y Pr
en
atal
T
e
lemonitori
ng
S
y
stem in W
i
reless Mes
h
Net
w
ork.
TEL
K
OMNIKA
. 2014
; 12(1): 123
–13
4.
[4]
L Hui. A Nov
e
l
QoS Routi
ng A
l
gorit
hm in Wir
e
less Mes
h
Ne
t
w
orks.
T
E
LKO
M
NIKA Indon
e
s
. J. Electr.
Eng.
201
3; 11(
3): 1652
–1
664.
[5] E
Borcoci.
W
i
reless Mes
h
Ne
tw
orks T
e
chnol
ogi
es: Architec
tures , Protocol
s , Resource M
ana
ge
me
nt
and
Appl
icatio
ns
. 200
9. [Online].
Avail
abl
e:
http://
w
w
w
.
i
a
ri
a.org/confer
en
ces20
09/fil
e
sICW
MC
09/Eu
g
enBorc
o
ciT
u
torial.p
df. [Accessed: 10-Oct-
201
4].
[6]
R Nelso
n
, L Kleinr
ock. Spati
a
l T
D
MA: A
Collis
io
n-Free multih
op Ch
an
n
e
l Access Protocol.
IEEE
T
r
ans. Commu
n.
1985; COM-
33(9).
[7]
P Bjorklund, V
Peter, D Y
u
an.
Res
ource
Opti
mi
z
a
t
i
o
n
of S
p
atial
T
D
MA i
n
Ad H
o
c R
adi
o
Netw
orks: A
Colu
mn Ge
ner
ation
Ap
proac
h
. INF
O
COM 200
3. T
w
e
n
t
y
-
S
econ
d A
nnu
a
l
Joi
n
t C
onfer
ence
of th
e
IEEE Compute
r
and Comm
un
icatio
ns
. IEEE
Societi
e
s. 200
3; 00(C).
[8] J
Gronkvist.
Assign
ment Met
hods for S
pati
a
l R
euse T
D
M
A
. Mo
b
i
le
an
d Ad
H
o
c
N
e
t
w
o
r
ki
ng
a
n
d
Comp
uting (M
obiHOC). 2
000
: 119–1
24.
[9]
O Somarriba,
J Z
ander.
Rob
u
st ST
DMA sched
uli
ng i
n
Mu
lti-ho
p W
i
reles
s
Netw
orks for singl
e-no
d
e
positi
on p
e
rturbatio
ns.
200
9 IEEE 20th Int. S
y
mp. Pers. In
door Mo
b. Rad
i
o Commu
n
.
20
09: 566
–5
71.
[10]
AD Gore, A Karan
d
ikar. Li
nk
Schedu
lin
g Al
gorithms for W
i
reless Mes
h
Net
w
orks.
Comm
un. Surv.
Tutorials, IEEE
. 2011; 13(
2): 258–
27
3.
[11]
W
Chen, C
L
ea. A No
de-B
a
sed T
i
me Sl
ot Assignm
ent
Algorit
hm for
ST
DMA W
i
re
less Mes
h
Net
w
orks.
IEEE Trans. Veh. Technol.
201
3; 62(1): 272
–2
8
3
.
[12]
P Gupta, PR Kumar. T
he
Capacit
y
of W
i
rel
e
ss Net
w
o
r
ks.
IEEE Trans. Inf. Theory.
2000;
46(2): 388
–
404.
[13]
S Rama
natha
n, EL Ll
o
y
d.
Sched
uli
ng
alg
o
rithms for m
u
ltih
op ra
di
o n
e
t
w
o
r
ks.
IEEE/ACM Trans.
Netw
.
1993; 1(2): 166–
17
7.
[14]
T
Moscibroda,
R W
a
ttenhof
er. Col
o
rin
g
Unstru
cture
d
Radi
o N
e
t
w
or
ks Categ
o
ries
and S
ubj
ect
Descript
o
rs.
Distrib. Comput.
200
5; 21(4): 27
1–2
84.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 1, March 2
015 : 202 – 2
1
0
210
[15]
A Behzad, I Rubi
n. On t
he Performanc
e o
f
Graph-base
d
Schedu
lin
g Al
gorithms for Packet Rad
i
o
Net
w
orks.
IEEE Glob. Telecomm
un. Conf.
2
003; 6: 34
32–
3
436.
[16]
J Grönkvist, A Hanss
on.
Co
mp
ariso
n
b
e
tw
een gra
p
h
-
base
d
an
d i
n
terferenc
e-b
a
s
ed ST
DMA
sched
uli
n
g
. Proc. 2nd ACM Int. Sy
m
p
. Mob.
ad hoc Net
w
.
Comp
ut. - MobiHoc ’0
1. 20
01:
255.
[17]
K Jain, J P
a
d
h
y
e, V
N
Pa
dma
nab
ha
n, L Qiu.
Im
pact Of Interference On
M
u
lti-h
op Wire
le
ss Netw
ork
Performanc
e.
Procee
din
g
AC
M Mobicom. 2
003: 66
–8
0.
[18]
G Brar, DM Blou
gh, P Santi
.
Comp
utatio
n
a
lly Efficie
n
t Sched
uli
ng w
i
th the Physical
Interference
Mode
l for T
h
roug
hp
ut Impro
v
ement in W
i
r
e
less Mes
h
N
e
tw
orks.
Proceedi
ngs of the
12th a
nnu
a
l
intern
ation
a
l co
nferenc
e on M
obil
e
comp
utin
g and n
e
t
w
ork
i
ng (ACM). 200
6: 2–13.
[19]
L Bad
i
a, A Ert
a
, L L
enzi
n
i, F
Rossetto, M Z
o
rzi.
A Physic
a
l Mod
e
l Sc
hed
uler for M
u
lti-H
op W
i
re
less
Netw
orks Base
d on Loc
al Info
rmati
o
n
. 5th IEEE Int. Conf. M
ob. Ad Hoc Sens. S
y
st
.
2008:
213–
22
2.
[20]
D Yang,
X F
ang, N Li, G Xue.
A Simple
Greedy Alg
o
rit
h
m for Li
nk Sched
uli
ng w
i
th
the Physica
l
Interference M
ode
l
. Globa
l T
e
lecommu
nic
a
ti
ons Co
nfere
n
c
e
(GLOBECOM). IEEE. 2009
.
[21] AD
Gore.
On W
i
reless Li
nk
Sched
uli
ng a
n
d
F
l
ow
Control
. Indian Institut
e of T
e
chnolo
g
y
B
o
mba
y
.
200
8.
[22]
O Goussevska
ia, YA Os
w
a
ld,
R Wattenhofe
r
.
Comp
lexity i
n
geo
metric SINR.
Proc. ACM MobiHoc.
200
7: 100
–1
09
.
[23]
W
Liu,
D Mi
ao,
X C
hen,
W
W
ang. A
N
o
vel
L
i
nk
Sc
hed
ul
ing
Alg
o
rithm for
Spatia
l R
eus
e
in W
i
re
les
s
Net
w
orks. IEEE Veh. T
e
chnol. Conf. (V
T
C
Fall). 20
12: 1–
5.
[24]
CE Sh
an
non.
A Mathem
atic
al T
heor
y
of
Commun
i
cati
o
n
.
Bell Syst.
Tech. J.
1
948
: 27(Ju
l
y
and
October): 379
–
423,6
2
3
–65
6.
Evaluation Warning : The document was created with Spire.PDF for Python.