TELKOM
NIKA
, Vol.13, No
.1, March 2
0
1
5
, pp. 230~2
3
7
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i1.1194
230
Re
cei
v
ed
No
vem
ber 3
0
, 2014; Re
vi
sed
Jan
uar
y 6, 20
15; Accepted
Jan
uary 20, 2
015
Heterogenous Traffic Performances on Local Multipoint
Distribution Service System in Rainy Environment
Naema
h
Mu
barak
a
h*, Suherman, Yulianta Sireg
a
r, Arman Sani
Dep
a
rtment of Elecrica
l Eng
i
n
eeri
ng, Univ
ers
i
t
y
of Sumater
a
Utara
Meda
n-Ind
one
sia 20
15
5
*Corres
p
o
ndi
n
g
author, e-ma
i
l
:naem
ah.mub
a
raka
h@gm
ail.
com
A
b
st
r
a
ct
Local
m
u
ltipoint distribution s
e
rvice (
L
MDS)
system
provides high
bitrates services
up t
o
40 Mbps
oper
ating
i
n
mi
lli
meter
b
ands.
T
he r
adi
o sys
tem c
a
n
be
e
m
p
l
oy
ed to
ser
v
e h
e
tero
gen
o
u
s traffics i
n
a
n
y
app
licati
ons
s
u
ch
as v
i
de
o
surve
ill
ance,
pu
blic
i
n
tern
et hotsp
ots a
nd
other
servi
c
es. How
e
ver,
th
e
imple
m
entati
o
n
in tropica
l ar
ea faces trans
miss
ion pr
ob
le
ms w
h
ich d
e
g
r
ade si
gna
l re
ceptio
n, such as
heavy r
a
in. T
h
i
s
pap
er exa
m
i
nes the
perfor
m
a
n
ce
of het
er
oge
no
us traffics in LMDS
in a
rainy e
n
vir
o
n
m
ent.
Some tec
h
n
i
qu
es to t
a
ckle
rai
n
p
l
ob
le
ms
w
e
re i
m
ple
m
ente
d
a
n
d
the
effe
cts w
e
re ass
e
s
ed. It w
a
s pr
ov
e
n
that PHY a
nd
MAC lay
e
rs co
ntribute
t
o
p
e
rformanc
es e
n
h
ance
m
ent o
n
r
a
iny
envir
on
ment as
lon
g
as
th
e
system us
es the suit
abl
e sc
hed
uler. A p
a
cket-base
d
sc
hed
uler w
a
s p
r
opos
ed by c
o
nsid
erin
g pack
e
t
impact on q
ual
ity. T
he results show
ed
that the pro
pose
d
techn
i
qu
e is
ab
le to improve v
i
de
o perfor
m
a
n
c
e
abo
ut 0.02
% o
v
er the
maxi
mum s
i
gn
al to
no
ise (
m
SNR)
sc
hed
uler,
ma
int
a
in v
o
ice
deco
d
in
g rate as w
e
ll a
s
the prior
i
ty fair (PF
)
schedul
er
and incr
eas
e data dec
od
ibi
lit
y 0.2%.
Ke
y
w
ords
:
PH
Y impr
ove
m
ent
, heteroge
no
us
traffics, schedulin
g, OF
DM, LMDS
1. Introduc
tion
Local multipo
i
nt distrib
u
tio
n
se
rvice
(L
MDS)
system
offers hi
gh d
a
ta rate
con
n
e
ctivity in
freque
ncy b
a
nd 20
-40
GHz [1]. The
ra
dio sy
stem
i
s
cap
able
of p
r
oviding
broa
dban
d services
su
ch
as internet hot
spot
s
compli
mentin
g the
ex
isting
WiFi networks, publi
c
se
rvices, audi
o and
digital video
bro
a
d
c
a
s
tin
g
, video
su
rveillance
an
d
other pote
n
t
ial se
rvice
s
.
Ho
weve
r, radio
prop
agatio
n i
n
ultra
hig
h
freque
ncy
ban
ds fa
ce
s
se
ri
ous chan
nel
attenuation.
The atten
uati
on i
s
worse
n
whe
n
the
radi
o
system o
perating in
tropi
ca
l area
whi
c
h ha
s
high
rain inte
nsity
[2].
Therefore, im
provem
ent works a
r
e re
qui
red to enh
an
ce the perfo
rm
ance.
The exi
s
ting
wo
rks on
cap
a
sity imp
r
oveme
n
ts v
a
ry fro
m
p
h
y
sical
(P
HY) layer t
o
appli
c
ation l
a
yer. OF
DM
system generates hi
gh
peak-to-averag
e power ratio (PAPR) and is
sen
s
itive to f
r
eque
ncy
offset [3]. Hen
c
e,
PHY im
prov
ements mai
n
l
y
deal
with P
APR redu
ctio
n a
s
well
as frequenc
y
offs
etting. PAPR redu
c
t
ion falls
into three
c
a
tegories [
3
]: distortion metho
d
s
by
usin
g clippi
n
g
and filterin
g, distortionl
e
ss meth
od
s whi
c
h avoid
sign
al modifi
cation throug
h, fo
r
instan
ce, pro
per si
gnal m
appin
g
, and codi
ng meth
ods
by in
cre
a
sin
g
the sig
nal de
codibili
ty.
Some te
chni
que
s d
eal
with freq
uen
cy
offset proble
m
have
bee
n
highlig
hted i
n
[4] . On
e of
the
prop
osed techniqu
es to o
v
erco
me su
b
-
ca
rri
er fr
e
q
u
ency offset is by adaptively distributing
the
power level a
m
ong
subcarrier.
The
ad
a
p
tive po
we
r a
llocatio
n
(AP
A
) adj
ust
s
th
e po
we
r l
e
vel o
n
sub
c
a
rri
er fre
quen
cy ba
se
d on the ex
pe
cted bit error
rate (BER) [5].
The allo
catio
n
schem
e is the main O
F
DM imp
r
ov
ement meth
o
d
locating in
mediu
m
acce
ss (MAC)
layer.
Th
e allocation scheme
in
cl
u
d
e
s
freq
uen
cy
sele
ction m
e
thod a
s
well
as
use
r
sch
edul
ing alg
o
rithm
.
Dynamic
subcarrier
all
o
catio
n
(DS
A
) tech
niqu
e
is on
e of the
freque
ncy
all
o
catio
n
meth
ods which dy
namically a
s
signs the
ch
an
nel fre
que
ncy
acco
rdin
g to
the
cha
nnel
stat
e inform
ation
(CSI) [6] . User
sche
dul
ers have
be
en inten
s
ivel
y prop
osed
by
resea
r
chers
sin
c
e it i
s
ap
plicatio
n-d
e
p
enda
nt. Ro
u
nd robin
(RR), first in fi
rst
out (FIF
O) [
7
],
maximum si
gnal to noise (mSNR) [
8
], and prop
or
tional fair
(PF) sch
edul
ers [9] are use
r
sched
uler ex
ample
s
.
In uppe
r laye
rs, OF
DM im
provem
ents
depe
nd on
e
nd user m
e
th
ods. End
use
r
or h
o
st
method con
s
ists of the tran
spo
r
t layer prot
ocol improvem
ents and the a
pplication la
yer
techni
que
s. User datag
ra
m
protoc
ol (UDP
) imp
r
ov
ement by
usi
ng inte
r-fram
e
ret
r
an
smi
s
sion
[10] is an e
x
ample of transport laye
r method,
wh
ile the use of ratele
ss
code [11] is t
he
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Heteroge
nou
s Traffic Pe
rform
ances o
n
Local Mu
ltipoi
nt Distrib
u
tion
.... (Naem
ah Muba
ra
kah
)
231
appli
c
ation l
a
yer method
. Since these layers sol
u
tions a
r
e in
depe
ndent o
f
the underl
y
ing
netwo
rks, the
y
are out
of the discu
ssi
on
scope.
This
pap
er
examine
s
th
e pe
rform
a
n
c
e
s
of
v
a
rio
u
s t
r
af
f
i
cs
o
n
an
LM
DS
sy
st
em
b
y
integratin
g PHY and MA
C layers sol
u
tions. Th
e pape
r al
so p
r
opo
se
s the
packet sch
e
d
u
ler
whi
c
h i
s
abl
e
to differentia
te prio
ritize
d
packet
s
to e
n
han
ce L
M
DS
system
ope
rating in tropi
cal
area. In o
r
d
e
r
to sim
u
late
the tropi
cal
environ
ment,
the rain
parameter
s of t
he sele
cted
city:
Medan – Ind
o
nesi
a
, are in
cluded in the e
v
aluation met
hod.
2. Packet-ba
sed Sched
u
ler for Hete
ro
genous Tr
affics
The pa
cket-b
ase
d
sched
ul
er is p
r
opo
se
d bas
ed on the maximum
sign
al to noise (mSNR)
sched
uler
bu
t has th
e ca
pability of pri
o
ritizin
g
vide
o, voice a
n
d
data pa
cket
s am
ong
oth
e
r
packet
s
. The
mSNR
sched
uler allo
cate
s band
width b
a
se
d on PHY
prope
rty: sig
nal to noise ratio
(SNR).
The sched
uler so
rts
user
s ba
sed on the qu
ality of their
sign
als. The
highe
st prio
ri
ty
belon
gs to u
s
er with the hi
ghe
st SNR. If
n
i
,
denotes S
N
R of i mobile
use
r
s
with n reso
urce
s, if
there a
r
e j mobile u
s
e
r
s at given scheduli
ng ti
me, the the sche
duling
pri
o
rity is given
by
Equation (1) [8]:
)
,
(
max
arg
_
_
n
i
j
mSNR
mobile
Selected
(1)
The mSNR sched
uler d
o
e
s not consi
der t
he type
s of the packets, which potentially
redu
ce
s th
e
quality of the
real
-time vid
eo an
d vo
i
c
e
.
In ord
e
r to
overcome th
e drawba
ck,
the
prop
osed
pa
cket-ba
s
e
d
sche
dule
r
ran
ks th
e u
s
e
r
s based
on p
a
cket types
prio
rity ord
e
r.
The
prio
rity order
follows the following
seq
u
e
n
ce: (i
). video packet type I; (ii) video packet type P/B;
(iii). higher v
o
ice packet
size; (i
v). higher data packet si
ze; (v
).
lower voice
packet si
ze;
(vi).
lowe
r data p
a
cket si
ze. The prio
rity is applie
d by
so
rting the mSNR o
u
tput
in
Equation (1
), by
usin
g Equatio
n (2).
)
_
_
(
_
mSNR
mobile
Selected
Sort
mobile
Selected
(2)
The packet priority has
been applied
in band assignm
ent based on SNR (BABS)
sched
uler [1
2]; howeve
r
, the sched
ul
er sort
s
mSNR
re
sult on
ly based
on
traffic si
ze.
The
prop
osed sch
edule
r
also improve
s
fra
m
e-b
a
sed
[13] and non-sorting sch
edu
ler [14] which
are
aimed only fo
r video traffic.
3. Rese
arch
Metho
d
3.1. Rain intensit
y
and lo
ss calcula
tio
n
Rain
cau
s
e
s
fading on
OF
DM sy
stem, for in
stan
ce, rain attenuati
on re
ache
s 8
0
dB on
29 GHz OF
DM system wit
h
5,7 km lin
k [Salehudin].
Poor rain atte
nuation ma
ke
s it is importa
nt
to measure t
he inten
s
ity and take i
n
to cal
c
ulatio
n. Rain inten
s
ity measurement
is perf
o
rme
d
in
three
differen
t
location
s in
Medan
City - Indone
sia: P
adan
g Bulan,
Polonia a
nd
Sampali u
s
in
g
Hellman measurement unit
[8]. The measured
rain
int
ensity is expressed
by millimeter per hour
(mm/h). T
he
spe
c
ific
path l
o
ss A (dB/
km
) is th
en
cal
c
ulated by u
s
i
ng Equatio
n
3 wh
ere
R i
s
rain
intensity (mm
/
h):
)
(
)
(
x
b
aR
x
A
(3)
L
o
dx
b
x
aR
A
)
(
(4)
The
rain
loss
in a p
r
o
pag
ation p
a
th with
l
engt
h
of L
(km) i
s
exp
r
e
s
sed by Eq
uati
on 4. IT
U-
R Rec.P.530
-10 [15] is
used for
rain i
n
tensity va
lida
t
ion. The foll
owin
g graph
in Figu
re 1 i
s
a
result of me
a
s
ureme
n
t an
d cal
c
ul
ation
with att
enu
ation pa
ram
e
ters taken
from I
T
U-R.83
8-1
[16]
for R
0,01
in three area
s: Polonia, Sampali
,
and P
adan
g
Bulan at freq
uen
cy 30 G
H
z. The ave
r
a
ge
values of thre
e area
s a
r
e referred to as
Medan.
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 : 230 – 2
3
7
232
Figure 1. Rai
n
loss at 30 GHz
3.2. PHY la
y
e
r model
The phy
sical
para
m
eter
o
f
the OFDM
netwo
rks
foll
ows spe
c
ifica
t
ions d
e
scri
b
ed in [17]
whe
r
e related
values in Ta
ble 1 we
re u
s
ed in cal
c
ul
ation wh
eneve
r
requi
re
d.
Table 1. Physical layer p
a
rameters
Parameter
Units
Formula
Value
Transmit Po
w
e
r
into Antenna
dBW
Ptx: transmit po
wer per ca
rrier
0
Transmit antenn
a gain
dBi
Gt:Gant
15
Freque
nc
y
GHz
f: Transmit freq
u
enc
y
30
Path Length
Km
d: Hub to Subscri
ber Station Ran
g
e
2
Field Margin
dB
Lfm : Antenna Mi
s-Alignment
-1
Free
-Space Loss
dB
FSL = -92.4
5
-20l
og(f)
-20log(d
)
-128,013
Total Path Loss
dB
Ltot = FSL + LF
M
-129,013
Receiver Antenn
a Gain
dBi
Gr =
Gant
30
Effective Bandw
i
d
th
MHz
BRF : Receiver
Noise Band
w
i
dth
80
Receiver Noise Figure
dB
NF : Effective No
ise Figure
5
Thermal Noise
dBW/MHz
10log(kTo)
-143,85
S
y
stem Loss
dB
Lsy
s
=
G
t+Ltot+
G
r
-84,013
Received Signal Level
dBw
RSL=Ptx+Ls
y
s
-84,013
Thermal Noise P
o
wer Spectr
al
density
dBW/MHz N
0
=10log(kTo
)
+
N
F
-138,859
Carrier t
o
Noise r
a
tio
dB
C/N = RSL-
No-1
0log(BRF
)
35,8151
3.3. MAC, Ne
t
w
o
r
k and tr
anspor
t la
y
e
r models
This p
ape
r
model
s OF
DM MAC with
uplink
and
downlin
k slot
s a
s
shown i
n
Figu
re 2.
Uplin
k fra
m
e
s
a
r
e d
e
vide
d by equ
al slots an
d sha
r
ed amo
ng
u
s
ers. Th
e ba
ndwi
d
th re
qu
est
mech
ani
sm i
s
assu
med
pe
rformed
in
a
specifi
c
cha
n
n
e
l so that
the
main frame
s
are
de
stined
f
o
r
uplin
k allo
cati
ons
only. In orde
r to
simp
lify the timing, piggyba
cki
n
g
ban
dwi
d
th
requ
est i
s
u
s
ed.
Network layer is assume
d to have a sin
g
l
e
base statio
n usin
g dire
ct
routing meth
od.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Heteroge
nou
s Traffic Pe
rform
ances o
n
Local Mu
ltipoi
nt Distrib
u
tion
.... (Naem
ah Muba
ra
kah
)
233
Figure 2. OF
DM MAC mo
del
In real
world,
pa
cket tran
smissi
on
s in
cl
ude
origi
nal and retran
sm
itted
pa
ckets.
Modelli
ng
the
situ
ation in
sim
u
lation results co
mpl
i
cated
ban
dwidth calcul
ation. The
r
efo
r
e, this
re
sea
r
ch
simplifie
s the
transpo
rt laye
r mod
e
l by ap
plying Use
r
Datagra
m
Prot
ocol
(UDP) to
all users. Thi
s
is acce
ptable
as video a
nd
voice traffics thro
u
gh the int
e
rnet a
r
e no
w mostly using
UDP.
3.4. Application la
y
e
r mo
del
The a
pplications
or t
r
affics are
assu
me
d to be i
n
upl
ink di
re
ction
s
. Traffics in
cl
ude vide
o,
voice an
d da
ta. Video traffics are gen
e
r
ated by us
i
n
g their tra
c
e
s
which co
ntai
n numbe
rs
of
bytes that a
r
e
tran
smitted i
n
a
spe
c
ific ti
me. Vi
deo tra
c
e
s
u
s
e
akiyo
_
cif.yuv video
traces a
s
u
s
ed
in [
14
]. The
voice t
r
a
c
e
s
a
r
e ta
ke
n
from a
n
Ada
p
tive Differe
ntial Pulse
Cod
e
Mo
dul
ation
(ADPCM)
co
des
whi
c
h
h
a
s 3
2
kbp
s
rates. T
he d
a
ta traffics a
r
e ge
ne
rated
rand
omly with
rand
om tran
smissi
on time.
Traffics are g
enerated an
d
rando
mized
with di
fferent
starting p
o
int in the each si
mulation
iteration. Ev
ery u
s
e
r
g
e
nerate
s
different traffic i
n
differe
nt a
llocatio
n
. By doin
g
thi
s
, the
rand
omn
e
ss
of the simulat
i
on ba
ckgro
u
nd is a
c
hieve
d
and sim
u
lat
i
on re
sults a
r
e fair enou
gh.
3.5. Simulation Design
Simulation
s
are i
m
plem
e
n
ted by
usi
ng a
n
o
b
je
ct oriente
d
p
r
og
rammi
ng
langu
age,
followin
g
stag
es a
s
sh
own in Figure 3.
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 : 230 – 2
3
7
234
Figure 3. Simulation de
sig
n
Radi
o timing
is divided
i
n
to small
slo
t
s an
d
den
oted a
s
tran
smissi
on
fram
es. T
r
affi
c
trace
s
,
rain
lo
ss calculation
s
a
n
d
po
ssi
bl
e ava
ila
ble
b
and
width
with
hun
dreds ite
r
ation
s
fo
r A
P
A
method [5], DSA [6] and JSPA [6] opti
m
izations are determi
ned
at firs
t place.
In each iterati
on,
every node i
s
assi
gne
d a traffic tra
c
e ra
ndomly with
random
startin
g
time. Base
station colle
cts
use
r
requ
est
s
withi
n
the
same
uplin
k
frame.
Th
ose
req
uest
s
a
r
e then p
r
o
c
e
s
sed b
a
sed
on
rand
omly ge
nerate
d
avail
able b
and
wi
dth. T
he allo
cation i
s
o
b
served u
s
in
g
RR, FIFO, P
F
,
mSNR, BABS, maximum proportional rate c
o
ns
trai
nt
s
algorithm (MPCA) [18] and the propos
ed
sched
uler. T
hose allo
cati
on is th
en i
n
forme
d
to
use
r
throug
h
downlink f
r
ames.
User
then
transmist
s bytes acco
rdin
g
to the allocat
ed ban
dwidt
h
.
Cha
nnel int
r
o
duces li
nk lo
ss,
while
ba
se st
ation i
n
trodu
ce
s buffe
r loss, both l
o
sse
s
are
recorded. Th
e total of succe
ssfully rece
ived byte
s wit
h
in the same
allocation fra
m
e is calculat
ed
to determine network
utility. Mobile users
ar
e
set
up to 10 nodes and bandwidth 1 M
H
z is
approximated
to generate
maximum 10
Mbps in
clea
r sky.
3.6. The ev
aluate
d param
e
ter
s
In orde
r to measure net
work imp
a
ct
to the
hetero
geno
us traffics, thi
s
pap
e
r
co
nsi
ders
Peak Signal
to Noi
s
e
Ratio (PS
N
R) fo
r video,
an
d
de
codi
ng
ra
tes voi
c
e
an
d data
traffics.
Problem
relie
s on d
e
termi
nation of PSNR valu
es. T
he only mod
e
l available t
o
cal
c
ulate P
S
NR
from si
mulati
on is
given b
y
[18], where
the re
ceive
d
video is re
co
ntructe
d
b
a
se
d on n
u
mb
er
of
su
ccessfully received byte
s, by
usi
n
g
the Evalvid
tools.
Th
e
tools then
co
mpares it to
the
origin
al vide
o
,
the PSNR v
a
lue i
s
th
en
o
b
tained.
Re
search
faces terri
ble
pro
b
le
m a
s
the
ma
n
ual
pro
c
e
ss i
n
[1
9] sho
u
ld be
repe
ated for thous
and ite
r
ation
s
. Since the gen
era
t
ed video tra
c
e
s
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Heteroge
nou
s Traffic Pe
rform
ances o
n
Local Mu
ltipoi
nt Distrib
u
tion
.... (Naem
ah Muba
ra
kah
)
235
contai
ns of I and P frame
s
, where I frames h
a
ve
mo
re si
gnificant impact
s
to video qu
ality ra
ther
than P frame
s
, this pap
er
modelle
d the
PSNR cal
c
ul
ation by usin
g the followin
g
mathemati
c
al
approximatio
n:
n
m
n
P
Max
n
p
T
n
p
m
I
Max
m
i
T
m
i
PSNR
)
_
,
,
(
)
_
,
,
(
(5)
m
i
,
is the re
ceive
d
bytes of the m
th
I-frame,
m
i
T
,
is the transmitted bytes and
I
Max
_
is
maximum PSNR. Th
e sam
e
para
m
eters apply for P-frames.
De
codi
ng rates of voice a
nd data traffics
ar
e
cal
c
ul
ated ba
sed
o
n
the pe
rcent
age of the
su
cc
es
sf
ully
r
e
ceiv
e
d
by
t
e
s
.
4. Results a
nd Analy
s
is
After runni
ng
a-thou
sa
nd
iteration
s
, re
cord
ing th
e receive
d
bytes and
cal
c
ul
ating the
netwo
rk para
m
eters,
it
is found that rain ca
uses t
he PS
NR pe
rforma
nces o
f
video
traffics
decrea
s
e
up
to 1.7 dB, the voice
de
co
ding rate
s
do
wn to a
bout
0.4%, and th
e data d
e
cod
i
ng
rates are red
u
ce
d up to
1.
3% from
clea
r sky pe
rform
ances. It is
al
so p
r
oven
tha
t
PHY and M
A
C
layers methods
s
u
c
h
as
A
PA, DSA and J
SPA are ab
le to decrease pack
et loss rates
c
a
used by
rain lo
ss in d
i
fferent deg
re
es a
s
sho
w
n
in Figure 4a
. Although those m
e
thod
s are not a
b
le
to
achi
eve cl
ear sky
perfo
rma
n
ce
s; D
SA, A
PA, and JSP
A
redu
ce
pa
cket lo
ss caused by rain u
p
to
0.08%, 0.21% and 0.24%
sub
s
eq
uentl
y
.
Even though
the PHY method
s have positive
impa
cts on the o
v
erall perfo
rmance; the
approp
riate
sche
dule
r
sele
ction i
s
im
portant as
P
H
Y
and MA
C lay
e
rs imp
r
ovem
ents
on
rain l
o
ss
will be insigni
f
icant if the chos
en
scheduler i
s
poor.
Figure 4b
shows that the
DSA, APA, a
nd
J
SPA methods
wors
en pac
k
et loss
when the
s
y
s
t
em employ
s
RR and FIFO
s
c
hedulers
,
whic
h
are n
o
t suita
b
le for hete
r
o
geno
us traffics. The
sched
ulers cau
s
e the pa
cket loss dropp
ed up
to
21%.
The be
havio
r of the sche
d
u
lers vari
es f
r
om o
ne to a
nother to
wa
rds the
hete
r
o
gene
ou
s
traffics. As
seen o
n
Figu
re 4b, it co
ul
d be ex
p
e
ct
ed that RR
and FIFO
re
sult po
or traffic
perfo
rman
ce
s. However,
RR pa
cket lo
ss imp
a
ct
i
s
di
stribute
d
am
ong vide
o, voice a
nd d
a
ta
as
sho
w
n
in
Fig
u
re
5. T
he m
o
st affe
cted
traffic
by the
packet l
o
ss in RR sche
du
ler i
s
voi
c
e
which
has the lo
we
st decoding
rat
e
amo
ng oth
e
r
sch
edul
er
s.
The affe
cted t
r
affics by the
packet lo
ss in
FIFO sche
d
u
ler a
r
e vid
eo and d
a
ta
which have
the lowe
st performan
ces am
ong o
t
her
sched
ulers.
(a)
(b)
Figure 4. Net
w
or
k pe
rform
ances
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 : 230 – 2
3
7
236
Mean
while,
MPCA sche
duler b
ehav
es contra
dict
ively toward
s heterog
eno
us traffics.
MPCA is frie
ndly to voice traffic but not to
video and data. Most MPCA packe
t loss o
c
curs o
n
video an
d d
a
ta pa
ckets.
As the results, PS
NR
an
d data d
e
co
ding rate of
MPCA sche
d
u
ler
drop
ped to a
bout 28.8 dB
and 94.3%.
On the other hand, mSNR, PF, BABS and the
propo
sed
sche
dulers sho
w
con
s
i
s
tent
respon
se to h
e
terog
eno
us t
r
affics whe
r
e
video, voice a
nd data have
balan
ce pe
rfo
r
man
c
e
s
. The
mSN
R
sc
he
du
le
r fo
r
in
s
t
anc
e
,
ge
n
e
r
a
tes 2
9
.
5 d
B
vide
o PSNR in
av
erag
e, alm
o
st
simil
a
r to
PF
,
BABS and the proposed
schedulers,
wi
thout sacri
s
fying voice
and data pe
rformances whi
c
h
rea
c
he
d 98.3
%
and 96,8%
deco
d
ing rates.
a. video trafiic
b. Voice traffic
c. data traffic
Figure 5. Traf
fic perfo
rman
ce
s
Among the
examined
schedul
ers, the
pro
p
o
s
ed
p
a
cket-b
ased sched
uler su
ccessfully
increa
se
s the
PSNR
pe
rformance u
p
to
29.8 dB, in
a
v
erage
0,06
dB high
er th
a
n
mSNR vide
o
traffics. Th
e
prop
osed
schedul
er al
so
maintain
s
voice traffic d
e
co
ding
rate
s a
s
high
as PF
sched
uler a
n
d
improve
s
d
a
ta decodin
g
rates u
p
to 0. 2% highe
r than mSNR.
6. Conclusio
n
This research has p
r
oven
that PHY and MAC
laye
rs meth
od
s a
r
e able to re
duce rain
impact.
Ho
wever, p
r
op
er
sched
uling
schem
e
sh
oul
d be
carefull
y sele
cted
to
wo
rk p
r
op
erly.
So
me
sc
he
du
le
rs
, su
ch
as
RR
, ar
e
n
o
t s
u
ita
b
l
e
for
rainy e
n
viron
m
ent, whil
e o
t
her
sched
ule
r
s
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Heteroge
nou
s Traffic Pe
rform
ances o
n
Local Mu
ltipoi
nt Distrib
u
tion
.... (Naem
ah Muba
ra
kah
)
237
are not frien
d
ly to hetero
geno
us traffi
cs. Fo
r ex
am
ple, MPCA is suitable for
voice but not
fo
r
video and d
a
ta.
Sc
heduler
mSNR, PF,
BABS, and the
pr
opos
e
d one generate balanc
e
d traffic
perfo
rman
ce
s that are applicable for use in
rain
y heteroge
n
ous traffic e
n
vironm
ent. The
prop
osed
schedul
er i
s
a
b
le to in
crea
se vide
o pe
rforman
c
e
ab
out 0.06 dB
over the
m
S
NR
sched
uler, g
e
nerate
s
voi
c
e
decoding
rat
e
as g
ood
as
PF sched
uler, and improve
data de
codi
n
g
rate 0.2% hig
her than mS
NR.
Since
re
se
ach u
s
e
s
UDP
i
n
tra
n
sport l
a
yer,
vario
u
s p
r
otocol
asse
ssment
could
be of th
e
future work
s
.
Referen
ces
[1]
Jun Y, Z
ang, Ben Leta
i
ef, K. Cross-la
yer
A
daptive R
e
s
ource Ma
nag
e
m
ent for W
i
reless Packet
Net
w
ork
w
i
t
h
OFDM Signa
lin
g
.
IEEE
Transac
tions on Wireless Communic
a
tions.
200
4; 5(
11).
[2]
Mubar
akah, N
aema
h
, Rika
W
A
, Mul
y
a
d
i
M. Performanc
e of Su
bcarri
er
and P
o
w
e
r All
o
catio
n
OF
DM
on Mi
llim
eter
Wave.
T
E
LKOMNIKA (T
elec
ommunic
a
tio
n
Co
mp
uting
Ele
c
tronics
and
C
ontrol)
. 2
0
13:
167-
174.
[3]
Ju
w
o
no, F
ilber
t H, Rand
y
SP
, Dadan
g G. A Stud
y
o
n
Pea
k
-to-Averag
e Po
w
e
r Rati
o in
DW
T
-
OF
DM
S
y
stems.
T
E
L
K
OMNIKA (T
eleco
m
mu
nic
a
ti
on C
o
mp
uting
Electron
ics a
nd C
ontro
l)
. 201
4;
12(
5):
395
5-39
61.
[4]
Sari EF
. A Mo
del to
inv
e
stig
ate Performa
n
c
e of
Orthog
o
nal Fre
que
nc
y Cod
e
Div
isio
n
Multipl
e
xi
ng.
T
E
LKOMNIKA (T
eleco
m
mu
ni
cation C
o
mputi
ng Electro
n
ics
and C
ontrol)
. 2
012; 10(
3).
[5]
Naem
ah M, Soeh
ar
w
i
nto, F
a
khru
ddi
n RB. Opti
mizing OF
DM Do
w
n
l
i
n
k
Performanc
e on LMDS
Sy
s
t
e
m
.
Intern
ation
a
l Jo
urna
l of Engin
eer
ing
Rese
arch & T
e
chno
logy (IJER
T
)
. 2013; 2(11)
.
[6]
Moha
nram C,
Hash
ya
m S. Joint Su
bcarri
er
and P
o
w
e
r Al
locati
on i
n
C
h
ann
el-A
w
a
re Q
ueu
e-A
w
ar
e
Sched
uli
ng for
Multius
e
r OF
DM.
IEEE
Transaction on Wireless Comm
unic
ations.
20
07; 6
(
9).
[7]
Dhro
na P, A
li
NA, Hassa
ne
in
H.
A p
e
rfor
ma
nce stu
d
y of s
c
hed
uli
ng
al
go
rithms
in
Poi
n
t-to-Multip
oin
t
WiMAX networks
. Proc. Local Comput
er Ne
t
w
orks C
onfer
ence, (Montre
al, Que., Oct.). 2008: 84
3-
850.
[8]
Choi
J. Distri
buted
be
amfor
m
ing
usin
g a
cons
e
n
sus
al
gorithm for c
o
oper
ative r
e
la
y
net
w
o
rks
.
Co
mmun
icati
o
ns Letters, IEEE
15.4. 201
1: 368-3
70.
[9]
Mouli
n
H. Pro
portio
nal
sch
e
duli
ng,
s
p
lit-pr
oofness, an
d merge-
proofn
e
ss.
Ga
m
e
s and
Econ
om
i
c
Behav
ior
. 20
08
; 63(2): 567-5
8
7
.
[10]
Suherm
an S,
Al–Aka
idi M.
An efficie
n
t n
egativ
e ackn
o
w
l
e
d
gem
ent–
b
a
sed tra
n
sport
protoco
l
in
802.1
1
me
dia
streamin
g.
Inte
rnatio
nal
Jour
n
a
l of A
d
H
o
c
and
Ubi
q
u
i
tou
s
Co
mp
utin
g
. 201
4; 16(
3):
161-
171.
[11]
Suherm
an S,
Mar
w
an
AA. Increas
ing
up
li
nk bro
a
d
ban
d
vide
o stream
i
ng pr
otoco
l
p
e
r
formance
in
W
i
MAX net
w
o
r
k
.
Internation
a
l
Journa
l of Internet Protocol T
e
chn
o
lo
gy
7.3. 201
3: 176-
185.
[12]
W
oo C, Eunsu
ng O, Daesik H.
Simp
le dy
n
a
mic subc
arrie
r
allocati
on w
i
th CQI feedbac
k reductio
n
for
OFDMA syste
m
s
. Vehic
u
l
a
r T
e
chnolog
y, IE
EE
T
r
ansactio
n
s on 57.
5. 20
08: 329
9-3
303.
[13]
Kang
SH, Z
a
k
hor A.
Pack
et
sched
uli
ng
al
g
o
rith
m for w
i
re
less vi
de
o stre
amin
g.
Proc. International
Packet Vide
o
W
o
rkshop, Pittsburg
h
, PY. 2002.
[14]
Suherm
an, A
l
-Akaid
i M.
Ad
justin
g W
i
MA
X
for a
d
edic
a
t
ed s
u
rvei
lla
nc
e n
e
t
w
ork.
In
te
rn
a
t
io
na
l
Journ
a
l of Elec
trical an
d Co
mputer Eng
i
n
eeri
n
g
. 201
3; 3(4).
[15]
IT
U-R. Propagati
on d
a
ta
and pre
d
i
ction me
th
o
d
s requ
ired
for the desig
n of
terrestrial li
ne-
of-sight s
y
stem
s.
Recommendation ITU-R
. 20
01; P.530-
10.
[16]
IT
U-R. Specific attenuati
on m
ode
l for
rain for
use in pre
d
icti
on metho
d
s.
Rec. ITU-R
. 1999; P.838-1.
[17]
CY Ch
u, KS
C
hen. Effects of
Rain
F
a
d
i
ng
o
n
the
Effici
e
n
c
y
of th
e
Ka-B
and
LM
DS S
ystem in
th
e
T
a
i
w
a
n
Area.
IEEE Transactions on Vehic
u
lar Technology
. 2005; 5
4
(1)
:
9-19.
[18]
Yibi
ng L,
Xu Z
,
F
ang Y. Impro
v
ed D
y
n
a
mic S
ubcarri
er Al
loc
a
tion Sc
he
dul
i
ng for SC-F
DM
A S
y
stems
.
Journ
a
l of Infor
m
ati
on
& Co
mputatio
na
l Scie
nce.
201
2; 9(1
2
): 3529
–3
537.
[19]
Ke CH, Shi
eh
CK, H
w
a
n
g
W
S
, Z
i
viani
A. An Evalu
a
tio
n
F
r
ame
w
ork for M
o
re Re
alistic S
i
mulati
ons of
MPEG Video T
r
ansmissi
on.
J
ourn
a
l of Infor
m
ati
on Sci
enc
e and En
gi
neer
ing.
20
08.
Evaluation Warning : The document was created with Spire.PDF for Python.