Indonesi
an
Journa
l
of
El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
13
,
No.
3
,
Ma
rch
201
9
, p
p.
1
2
0
8
~
1
220
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
3
.i
3
.pp
1
208
-
1
220
1208
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Multi
ple
e
rror
c
orre
ctio
n
towards
o
ptim
isation o
f
e
nergy
in
s
ensor
n
etwork
Sa
mi
r
ah
R
az
ali
, K
am
aruddi
n Mam
at
, Nor
Shahniz
a
K
am
al Ba
sh
ah
Facul
t
y
of
Com
pute
r and
Ma
them
at
ic
a
l
Sci
ences,
Univer
si
ti T
e
ch
nolog
y
of
MA
R
A,
Shah
Al
am, M
al
a
y
s
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
S
ep
1
5
, 201
8
Re
vised
D
ec
10, 2
018
Accep
te
d Dec
25, 201
8
H
y
brid
ARQ
(HA
RQ)
is
among
the
opt
imum
err
or
cont
rols
impl
emente
d
i
n
W
ire
le
ss
Sensor
Network
as
it
red
uce
s
the
over
h
e
ad
from
ret
ran
sm
ission
and
err
or
cor
r
ec
t
ing
code
s.
Th
e
adv
a
nce
m
ent
in
W
SN
inc
lude
s
th
e
u
sage
of
high
num
ber
of
nodes
and
the
in
crease
in
tr
aff
i
c
with
l
a
rge
data
tra
nsm
itted
among
the
node
s
h
ad
conc
ern
ed
th
e
nee
d
for
a
n
ew
appr
oa
ch
in
er
ror
cont
ro
l
al
gorit
hm
.
Th
is
pape
r
proposed
t
he
m
ult
ipl
e
err
or
cor
rec
t
ion
base
d
on
HA
R
Q
proc
ess
to
ai
d
th
e
cha
nges
in
ch
a
nnel
with
prope
r
err
or
cor
recti
on
assig
nm
ent
towar
ds
opti
m
i
s
ing
the
per
form
anc
es
of
W
SN
in
term
s
of
bit
err
or
ra
te
s
,
remai
ning
ene
rg
y
,
and
la
t
ency
f
or
diffe
ren
t
t
y
p
e
s
of
conge
stion
and
cha
nne
l
condi
ti
ons
.
In
thi
s
stud
y
,
we
have
deve
lop
ed
the
cha
nn
el
ada
ptatio
n
al
gorit
hm
that
c
an
ada
pt
to
sudden
cha
ng
es
an
d
demons
tra
te
d
the
opti
m
al
err
or
cor
re
ct
ing
code
s
as
well
a
s
adj
ustm
ent
on
the
tra
nsm
it
po
wer
for
the
give
n
ch
anne
l
c
ondit
ion
and
co
ngesti
on
pre
sen
t
ed.
From
the
res
ult
anal
y
s
ed
,
the
opt
imis
at
ion
bet
we
en
th
e
r
e
m
ai
ning
en
erg
y
and
Bit
Err
or
rates
h
appe
n
ed
on
the
b
asis
of
ada
pt
ing
to
the
s
e
different
cha
n
nel
cond
it
ion
an
d
conge
stion
to
m
ini
m
iz
e
red
undanc
i
es
appe
n
ded.
From
the
result
obta
ined,
we
conc
lud
ed
tha
t
b
y
using
m
ult
iple
err
or
cor
r
ec
t
ion
al
gori
thm
with
the
ai
d
of
adj
ustm
en
t
on
the
tr
ansm
it
powe
r,
the
rema
ini
ng
en
erg
y
c
an
be
opti
m
ised
to
get
her
wi
th
Bit
E
rror
r
at
es
a
nd
the e
x
ce
ss
ive
red
undancie
s
can be
r
educed.
Ke
yw
or
d
s
:
HA
R
Q
Kalm
an
f
il
te
r
Rem
ai
nin
g
en
e
rg
y
SN
R
c
ha
nn
el
a
dap
ta
ti
on
W
i
reless
s
e
nso
r
n
et
w
ork
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Kam
aru
ddin M
a
m
at
,
Faculty
of Com
pu
te
r
an
d
Ma
them
a
ti
cal
Scie
nces,
Un
i
ver
sit
i Tec
hnology
of MARA,
Sh
a
h Alam
, M
al
ay
sia
.
Em
a
il
: ka
m
ar@tm
sk
.u
itm
.ed
u.m
y
1.
INTROD
U
CTION
The
Wireless
Sensor
Net
wor
k
(
WSN)
is
ve
ry
cr
ucial
in
m
on
it
ori
ng
fiel
d
su
c
h
as
ha
bitat
m
on
it
or
in
g,
env
i
ronm
ental
,
agr
ic
ultu
ral,
m
ilit
ary,
and
tracki
ng
fiel
d.
Re
centl
y,
ther
e
are
so
m
e
e
m
erg
i
ng
a
pp
li
cat
ion
s
of
WSN
in
bi
g
da
ta
[1
]
,
[
2]
an
d
internet
of
th
ing
s
(IoT)
[3
]
,
[
4].
T
he
existi
ng
te
c
hnology
of
WSN
crit
ic
al
ly
ben
e
fits
in
te
r
m
s
of
cost,
sc
al
abili
ty
,
and
al
so
pro
vid
e
s
upports
to
wards
hum
an
-
w
ork
co
ns
trai
ned
w
he
n
m
on
it
or
ing da
nger
ous
places s
uch as
natu
ral
disaste
rs
a
nd
unf
rien
dly en
vir
on
m
ents [5]
.
As
W
S
N
is
energy
-
c
onstrai
ned
a
nd
e
rror
-
pro
ne,
resea
rc
her
s
hav
e
est
a
blishe
d
m
any
m
et
ho
ds
a
nd
ways
to
overc
om
e
these
pro
bl
e
m
s.
Re
search
es
bac
k
the
n
pr
ov
i
ded
the
m
eth
od
t
o
re
duce
energy
c
on
s
umpti
on
thr
ough
cal
ibr
at
ing
or
a
dju
st
ing
the
t
ran
sm
issi
on
powe
r
s
uch
a
s
by
est
im
at
ing
the
Si
gn
al
to
N
oise
Ra
ti
o
(S
NR
)
[
6]
and
Re
cei
ved
Sig
nal
Stren
gth
I
ndic
at
or
RSS
I
[
7]
us
in
g
Kalm
an
Fil
te
r
(K
F
)
in
order
to
a
dju
st
the
transm
issi
on
powe
r.
T
he
im
portant
as
pect
of
high
er
ror
rates
in
a
net
work
ca
nnot
be
cast
asi
de
al
though
m
ini
m
iz
ing
t
he
ener
gy
usa
ge
us
in
g
transm
issi
on
po
wer
c
ontr
ol
(TPC)
is
an
eff
ect
ive
m
et
hod
to
m
a
intai
n
the
li
fetim
e
as
hig
h
error
rates
cause
retra
ns
m
is
sion
to
fl
ood
the
netw
ork
an
d
this
consum
es
m
or
e
energy.
Thu
s
,
the
pro
blem
o
f
high
e
nergy
us
a
ge
nee
ds
to
be
ta
ckle
d
a
longside
with
high
er
ror
rate
s.
In
a
ddit
ion
,
so
m
e
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Multi
ple err
or
correcti
on tow
ar
ds
op
ti
mis
ation of e
ner
gy
i
n sens
or
netwo
r
k (
Samirah
Ra
za
li
)
1209
te
chn
iq
ues
in
i
m
ple
m
enting
error
co
rr
ect
io
n
sc
hem
es
suc
h
as
H
ARQ
ind
i
rectl
y
reduce
e
rror
rate
s
a
nd
po
s
sibly
reduc
e
the
energy
con
s
um
ption
by
reducin
g
the
f
loodin
g
of
retr
ansm
issi
on
in
a
congested
ne
twork
.
T
her
e
are
relat
ed
st
ud
ie
s
on
the
e
ff
ect
of HARQ
with
BC
H
in
Co
de
d
Di
visio
n
M
ulti
ple
Acess
(C
DM
A) WS
N
towa
rd
s
the e
rror rates a
nd e
ne
rg
y c
onsu
m
ption
[
8
]
,
[
9].
The
auth
ors
re
ported
on
hi
gh
Bi
t
Err
or
Ra
te
s
(BER)
at
low
hop
co
unt.
Th
us,
the
arch
it
ect
ur
e
or
node
dep
l
oym
ent
con
trib
utes
to
the
increase
in
er
r
or
rates
or
pac
ket
corrupti
on
in
the
pr
e
sence
of
inter
fer
e
nc
es
and
no
ise
s
pa
rtic
ul
arly
w
hen
the
nodes
are
locat
ed
within
the
t
ran
sm
issi
on
ra
ng
e
of
oth
e
r
node
s.
Pr
act
ic
al
ly
,
the
nodes
a
re
us
ua
ll
y
dep
loye
d
i
n
a
n
on
-
unif
or
m
m
ann
er
f
ollowi
ng
the
ge
ogra
ph
ic
al
str
uctu
r
e.
Th
us
,
there
can
be
var
ia
ti
ons
of
node
de
ns
it
ie
s
in
on
e
m
on
it
ori
ng
area.
Mo
re
ov
e
r,
a
pa
rt
f
r
om
interfer
e
nc
es
an
d
sig
nal
f
adin
g,
the
cha
nn
el
c
onditi
on
m
igh
t
change
ov
e
r
ti
m
e
du
e
to
re
fl
ect
ion
s
a
nd
refra
ct
ion
s
of
sig
nals
an
d
al
so
m
igh
t
be
con
t
rib
uted fr
om
the h
ar
dwar
e it
sel
f.
Thus
, t
he
im
ple
m
enta
ti
on
of
e
xisti
ng
error co
rr
ect
io
n
sc
hem
es
m
ig
ht not
be
reli
able
as
diff
e
re
nt
le
vels
of
c
ongestio
ns
m
igh
t
pr
e
sen
t
in
on
e
m
on
it
or
i
ng
area
wit
h
s
udde
n
c
ha
nges
of
int
erf
e
ren
ce
s
and
no
ise
ov
e
r
tim
e.
The
wo
r
k
of
[10]
intr
oduce
d
the
m
ulti
-
cod
i
ng
sc
he
m
es
fo
r
WSN
wh
e
re
diff
e
re
nt
cod
i
ng
schem
es
wer
e
i
m
ple
m
ente
d
at
the
nodes
and
the
sin
k.
Ba
sed
from
t
his
li
te
ratur
e,
it
was
dem
on
strat
ed
t
hat
the
m
ulti
-
c
od
i
ng
sc
hem
es
increase
d
the
l
ifet
i
m
e
of
a
ne
twork
.
H
ow
e
ve
r,
f
ur
t
her
st
udie
s
on
the
opti
m
al
cod
ew
ord
le
ngth
an
d
e
rror
co
rrec
ti
ng
ca
pa
bili
ty
that
corres
ponds
to
t
he
presence
of
no
is
es
are
need
e
d
i
n orde
r
to
full
y op
ti
m
iz
e the e
nergy
us
a
ge
a
nd er
ror
rates
in
a
n
et
work.
Diff
e
re
nt
co
di
ng
schem
es
and
er
ror
c
orrecti
ng
capa
bi
li
ty
will
app
end
dif
fe
ren
t
nu
m
ber
of
redu
nd
a
ncies
to
the
tra
ns
m
itt
ed
bits.
Ba
se
d
on
our
pre
vious
stu
dies
[
11
]
-
[
13]
,
we
de
m
on
strat
ed
tha
t
hig
h
redu
nd
a
ncy
co
ns
um
ed
too
m
uch
e
nergy
in
decodin
g
the
t
ran
sm
it
te
d
data.
Me
an
w
hile,
a
network
with
ba
d
conditi
on
a
nd
high
e
rror
rate
s
m
igh
t
no
t
ab
le
to
handle
la
rg
e
num
ber
of
erro
neous
bits
due
t
o
the
rel
at
ively
low
er
r
or
c
orre
ct
ing
capa
bili
ty
.
Thus,
in
t
his
pap
e
r,
we
propose
d
the
a
pp
ro
ac
h
to
op
ti
m
iz
e
the
energy
us
a
ge
with
BER
by
m
od
ify
ing
the
Hyb
rid
ARQ
(HARQ
)
proc
ess
ai
de
d
wit
h
m
ult
i
-
cod
i
ng
schem
es
and
powe
r
con
t
ro
l
that
c
an
ada
pt
to
th
e
changes
in
channel
co
ndit
ion
that
wer
e
est
i
m
at
ed
as
SN
R
us
i
ng
K
F.
Our
pro
po
se
d
al
go
rithm
m
erit
s
on
it
s
ca
pa
bili
t
y
to
a
dap
t
t
o
the
c
hanges
in
SN
R
in
w
hic
h
e
ver
y
SN
R
range
descr
i
be
di
ff
e
r
ent
erro
r
co
rr
e
ct
ion
use
d
a
nd
transm
it
po
we
r
cal
ibrati
on
to
m
ini
m
ise
the
higher
redu
nd
a
ncies
wh
e
n
t
hese
e
xc
essive
r
ed
unda
ncies
are
not
need
e
d
i
n
bette
r
S
NR
c
onditi
on
a
nd
vice
ve
rsa.
T
ra
ns
m
it
powe
r
cal
ibrati
on
ai
de
d
t
o
re
du
ce
the
inter
fe
ren
ce
s
an
d
no
ise
s
i
n
no
isy
c
hann
el
wh
e
n
S
NR
is
low
as
the
r
at
io
of
no
ise
s
will
rise
the
BER
sh
ar
pl
y.
W
e
pr
ese
nt
ed
the
com
par
i
so
ns
betwe
en
our
propose
d
al
gorithm
towards
the
increase
in
pe
r
form
ances
with
the
existi
ng
error
c
orrecti
on
us
e
d
in
Sect
ion
3
in
w
hich
we
able
to
re
duce
high
BER
in
lo
w S
NR c
onditi
on a
nd m
ai
ntain r
em
ai
nin
g
e
nerg
y i
n
hi
gh S
NR
conditi
on.
2.
RESEA
R
CH MET
HO
D
The
dev
el
op
m
ent
of
Mult
ipl
e
Error
Co
rr
e
ct
ion
(MEC
)
consi
sts
of
se
ver
al
m
et
ho
ds.
First,
we
pro
po
se
d
the
MEC
al
go
ri
th
m
s
based
on
t
he
Hy
br
id
AR
Q
(
HA
R
Q)
process.
Seco
nd,
we
ai
de
d
the
al
gorithm
by
cal
ibrati
ng
the
transm
it
powe
r
accor
di
ng
to
the
ME
C
assigne
d
for
dif
fer
e
nt
error
c
orrecti
ng
cod
e
s
at
diff
e
re
nt
SN
R
range.
Last
ly
,
we
inte
gr
at
ed
Kalm
an
Fil
ter
(
KF
)
to
est
i
m
at
e
th
e
SN
R
values
for
in
com
ing
transm
issi
on
in
order
t
o
a
void
w
ron
g
assi
gn
m
ent
of
er
r
or
c
orrecti
ng
cod
e
s
to
wards
the
s
udde
n
c
hangin
g
env
i
ronm
ent o
f
CDMA
WSN.
Figure
1
sho
w
s
the
ch
ron
ologica
l
de
velo
pm
ent
of
our
propose
d
al
gorithm
.
W
e
al
s
o
c
arr
ie
d
ou
t
t
hr
e
e
diff
e
re
nt
pr
el
i
m
inary
exp
e
rim
ents
com
pr
isi
ng
the
sim
ulati
on
of
C
DMA
WSN
with
no
error
c
orrecti
on,
with
a
range
of
low
e
rror
co
rr
ect
in
g
capab
il
it
y,
and
with
a
range
of
high
erro
r
correct
ing
ca
pa
bili
ty
.
W
e
sel
ect
ed
BC
H
and
RS
cod
e
s
as
the
error
c
orrecti
ng
cod
es
in
our
al
gorithm
based
on
the
prel
im
inary
exp
eri
m
ents
wh
e
re
BC
H
an
d
RS
cod
es
op
t
i
m
iz
ed
the
per
f
or
m
ance
of
CDMA
WSN
co
m
par
ed
to
the
Conv
olu
ti
onal
Cod
es
[11
]
,
[
12]
.
S
ome
stu
dies
al
s
o
agr
ee
d
t
hat
the
RS
[14]
a
nd
BC
H
c
od
es
[
15]
are
am
on
g
oth
e
r
op
ti
m
u
m
er
ror
correct
ing
c
od
es
wh
ic
h
pr
ov
i
de
reli
abili
ty
du
ri
ng
tra
ns
m
is
sion.
From
our
prel
i
m
inary
te
sti
ng
[11
]
-
[
13]
,
we
ob
s
er
ved
t
hat
in
a
lo
w
SN
R
c
onditi
on
with
high
BER
,
the
us
e
of
lo
w
er
ror
co
rr
ect
in
g
ca
pab
il
it
y
m
igh
t
no
t
be
able
to
so
l
ve
the
pro
blem
of
high
BER
.
Me
anwhil
e,
the
us
e
of
high
er
ror
co
rr
ect
in
g
capab
il
it
y
caus
ed
th
e
pro
blem
s
of
in
crem
ent
in
ene
rg
y
c
onsu
m
ption
as w
el
l
as
la
te
ncy
. W
e
al
so
f
ound
that
m
edium
err
or
c
orr
ect
ing
capab
il
it
y
fo
r
t
he
high
S
NR
conditi
on
m
igh
t
be
able
to
optim
iz
e
BER
per
form
ance
as
well
as
the
re
m
a
ini
ng
energy.
Be
sid
es
that,
the
us
e
of
co
deword
le
ng
th
co
rr
es
pondin
g
to
th
e
nu
m
ber
of
gen
e
rated
bits
in
the
netw
ork
is al
s
o crit
ic
al
. H
ig
he
r nu
m
ber
s
of c
od
e
w
ord
le
ng
t
hs
(m
or
e tha
n 127
) prom
ote m
or
e energy u
sage.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
1
2
0
8
–
1
2
2
0
1210
Figure
1.
Ch
rono
l
og
y
of the
develo
pm
ent o
f
Mult
iple Er
ror C
orrecti
on (
M
EC)
Fr
om
this
obs
erv
at
io
n,
we
c
on
cl
ud
e
d
t
hat
there
is
t
he
ne
ed
to
use
m
ulti
ple
cod
i
ng
s
chem
es
fo
r
diff
e
re
nt
ra
nge
of
S
NR.
We
al
so
c
on
si
der
e
d
s
om
e
of
othe
r
relat
e
d
w
or
ks
[16]
w
hic
h
dem
on
strat
ed
t
ha
t
the
m
ul
ti
-
cod
in
g
i
s
m
or
e
ene
rgy
-
eff
ic
ie
nt.
Ba
sed
on
this
de
du
ct
io
n,
we
pro
po
se
d
the
SN
R
cl
assifi
c
at
ion
al
gorithm
in
adap
ti
ng
t
o
the
changes
of
c
hannel
c
onditi
on.
T
his
al
gori
thm
fo
ll
ow
s
th
e
existi
ng
pro
cess
of
HA
R
Q
in whic
h
we
m
od
ifie
d t
he
al
gorithm
t
o
be
in
j
ect
ed w
it
h
MEC
m
od
ul
e and
transm
i
t
p
ower m
od
ule
. W
e
al
so
integrate
d
Kal
m
an
Fil
te
r
(K
F
)
to
est
i
m
at
e
the
SN
R
va
lues
to
av
oid
a
ssign
i
ng
wrong
erro
r
correct
ion
du
e
to sudd
e
n
cha
nges in a
noisy
ch
an
nel. Som
e
f
ie
ld test
s
[17
]
-
[
19]
obse
rv
e
d t
hat the sign
al
i
m
pair
m
ents is
tim
e
-
var
yi
ng.
I
n
a
ti
m
e
-
var
yi
ng
c
onditi
on,
t
her
e
is
a
need
to
c
on
ti
nu
ously
trac
k
the
c
ha
ng
es
w
it
hin
the
cha
nn
el
as
sign
al
s
f
re
qu
e
nt
ly
attenu
at
ed,
ref
le
ct
ed
,
an
d
r
efr
act
ed
.
I
n
ad
diti
on
,
t
he
ha
rdwar
e
it
sel
f
m
ig
ht
be
a
co
ntri
buti
ng
factor
to
t
he
changes
in
S
NR
as
dif
fer
e
nt
no
des
giv
e
s
out
diff
e
ren
t
SN
R.
T
hus,
channel
est
im
at
ion
i
s
su
bst
antia
l t
o
e
ns
ure t
hat prop
er error c
orrect
ing
c
odes assi
gned
to
a
de
fine
d
S
NR r
a
nge
w
hich
c
orres
ponds to
the
sudd
e
n
c
ha
ng
e
s
in
S
NR
w
hen
t
her
e
a
re
tr
ansm
issi
on
s
be
tween
nodes
in
any
giv
e
n
ti
m
e.
KF
is
am
ong
the
com
m
on
sta
te
est
i
m
at
or
wit
h
a
dv
a
nta
ges
su
c
h
that
l
ow
com
plexity
com
par
ed
to
othe
r
filt
ers
[
20
]
and
it
s
capab
il
it
y
to
pro
vid
e
bette
r
est
i
m
ation
f
or
Ga
us
sia
n
a
nd
li
near
m
od
e
ls,
and
li
m
i
ted
no
n
-
li
nea
rity
.
W
e
integrate
d
KF
pr
i
or
t
o
ge
ne
ra
ti
ng
data
an
d
a
fter
the
dec
oding
proce
ss.
In
order
to
m
ini
m
iz
e
the
com
pu
t
at
ion
al
ov
e
r
head
of
KF
eq
uat
io
n,
we
m
od
ifie
d
the
ACK/
NA
C
K
m
essage
wh
ic
h
ap
pe
nd
t
he
SN
R
val
ue
of
the
est
i
m
ation
init
ia
te
d
to
the
tr
ansm
itted
bits
from
the
sen
der.
T
his
is
to
ack
nowled
ge
the
recei
ver
on
the
pr
e
vious
est
im
at
ion
val
ue
a
nd
to
pr
e
par
e
t
he
transm
issi
on
if
there
a
re
s
udde
n
c
hanges
i
n
S
NR
val
ue
s
o
that
the
possibil
it
y
in
assig
ning
wro
ng
e
rro
r
correct
in
g
c
od
e
s
is
re
du
c
ed
if
the
sen
der
i
niti
at
es
ano
t
her
transm
issi
on
.
2.1.
SNR
Clas
sific
at
i
on
to
w
ards
th
e
Adap
tatio
n to the
Chan
ges
in
C
h
anne
l Conditi
on
We
di
vid
e
d
th
e
m
et
ho
dolo
gy
into
t
w
o
f
ollo
wing
so
l
utio
ns
based
on
the
un
s
ol
ved
pro
ble
m
s
of
hi
gh
BER
and
the
pro
blem
s
fr
om
high
e
nergy
us
a
ge
an
d
high
la
te
ncy
du
e
to
the
im
ple
m
entat
ion
of
hi
gh
er
r
or
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Multi
ple err
or
correcti
on tow
ar
ds
op
ti
mis
ation of e
ner
gy
i
n sens
or
netwo
r
k (
Samirah
Ra
za
li
)
1211
correct
ing
ca
pa
bili
ty
(F
igure
1).
I
n
m
et
ho
d
1,
we
pro
pose
d
the
c
ha
nn
el
ad
aptat
ion
al
gorit
hm
b
ased
on
H
ARQ
process
a
nd
cl
assifi
ed
the
ra
ng
e
of
SN
R
.
We
cal
culat
ed
the
ra
ng
e
of
S
NR
f
or
our
predefi
ned
t
opol
og
ie
s
of
rand
om
un
ifo
r
m
no
de
distribu
ti
on
a
nd
ra
ndom
non
-
unifo
rm
no
de
dis
tribu
ti
on.
A
r
andom
un
if
orm
no
de
distrib
ution
c
onsist
s
of
la
ye
re
d
topolo
gy
wit
h
ra
ndom
no
de
scat
te
ring
ar
ound
the
sin
k
w
it
h
un
if
or
m
distance
of
10
m
betwe
en
eac
h
node.
Me
anwhil
e,
th
e
ra
ndom
non
-
un
i
form
distribu
ti
on
co
ns
ist
s
of
la
ye
red
t
opologie
s
wh
e
re
the
dis
ta
nce
betwee
n
nodes
va
ried
between
10
–
100
m
.
The
SN
R
was
cal
culat
ed
f
ol
lowi
ng
the
sta
nd
a
rd
ise
d
e
qu
at
io
n bel
ow
[21]
:
(
)
=
−
(1)
w
he
re
denotes
the
receive
d
sign
al
po
wer
wh
il
e
denotes
the
noise
po
wer
in
w
hich
that
is
m
easur
ed
in
unit
s
of
powe
r
(
Watt
s
or
m
il
l
i
watt
s).
Exte
nd
i
ng
the
f
or
m
ula,
the
sig
nal
po
wer
can
be
obt
ai
ned
us
in
g
F
riis T
ra
ns
m
issi
on
equa
ti
on
as
sho
wn in E
quat
ion (
2)
[22]
:
P
=
(
P
T
G
T
G
R
)
(
λ
2
)
(
4π
R
)
2
(2)
w
he
re
P
T
is
tra
nsm
it
po
we
r,
G
T
is
the
gai
n
of
tra
ns
m
it
antenn
a,
G
R
is
the
gai
n
of
receivi
ng
ante
nn
a
,
a
nd
R
is
the d
ist
a
nce
be
tween
sen
der a
nd r
ecei
ver
.
T
he
noise
power,
is cal
culat
ed f
ollow
i
ng E
qu
at
ion
(
3)
[
23
]
.
N
0
=
k
T
s
B
(3)
w
he
re
k
de
note
s the B
oltzm
ann co
ns
ta
nt,
T
s
deno
te
s the syste
m
t
e
m
per
at
ure, a
nd
B
de
no
te
s
the
band
width.
Fr
om
the
cal
c
ulati
on
,
we
ob
ta
ined
the
act
ual
SN
R
that
fo
ll
ows
our
predefi
ned
arc
hi
te
ct
ur
e
an
d
par
am
et
ers
as
sh
ow
n
in
Ta
ble
1
(S
ect
io
n
2.
3).
W
e
ext
racted
an
d
ap
plied
the
cal
culat
ed
values
int
o
th
e
pr
e
-
bu
il
t
syst
e
m
id
entifi
cat
ion
m
od
e
l
usi
ng
MATL
AB.
From
the
extracte
d
data,
we
ob
ta
i
ned
the
m
at
rices
from
the
best
fit
m
od
el
us
ing
the
input
-
outp
ut
m
od
el
fr
om
the
syst
e
m
identific
at
ion
too
l;
m
a
tric
es
A
(
-
0.236
468303
4960
98),
B
(
0.113
429553
384058
),
an
d
C
(
3.594
501242
4461
3).
Table
1
sho
ws
the
e
xtracted
inf
or
m
at
ion
of
act
ual
SN
R
a
longside
with
the
est
i
m
at
ed
SN
R
obta
ine
d
us
in
g
Kalm
a
n
Fil
te
r
eq
uation
as
sh
ow
n
in
E
qu
a
ti
on
(4)
a
nd E
quat
ion (
5).
X
k
=
A
x
k
−
1
+
B
u
k
−
1
+
w
k
−
1
(4)
=
+
(5)
w
he
re
A
is
the
s
ta
te
transiti
on
m
od
el
app
li
ed
to
the
pr
e
vi
ou
s
sta
te
x
k
−
1
,
B
is
the
con
t
ro
l
-
i
nput
m
od
el
app
li
e
d
to
the
co
ntr
ol
vecto
r
u
k
−
1
,
an
d
w
is
the
process
noise
.
de
note
s
the
obse
r
vatio
n
m
od
el
wh
ic
h
m
aps
the
true
sta
te
s
pace
into
t
he obse
r
ved s
pace a
nd
.
Table
1
s
hows
the
act
ual
SN
R
obta
ined
wit
h
the
est
im
a
te
d
SN
R
values
aft
er
we
hav
e
cal
culat
ed
t
he
m
easur
em
ent
and
i
nnovat
ive
gain
f
r
om
KF
eq
uation.
We
then
div
ide
d
the
SN
R
range
into
five
di
sti
nct
gro
up
s:
S
NR
lw
(S
NR
l
ow
est
)
with
the
range
of
S
NR
v
≤
5
,
SN
R
l
(SNR
lo
w)
with
6
≤
S
NRv
≤
20,
SNR
m
(S
NR
m
edium
)
with
21
≤
S
NR
v
≤
35,
S
NR
h
(SN
R
High)
with
36
≤
S
NR
v
≤
50,
an
d
SN
R
hg
(S
NR
highest)
wh
e
re
SN
Rv
≥
51.
Ba
sed
on
pr
e
vious
stu
dies
a
nd
real
-
tim
e
te
stin
g
[24]
,
m
os
t
stud
ie
s
a
gr
ee
d
that
the
li
nk
i
s
in
a
good
qu
al
it
y
w
hen
the
S
NR
va
lue
is
m
or
e
th
an
40
dB.
Th
us,
we
set
the
be
nch
m
ark
of
S
NR
in
high
c
onditi
on
(S
NR
h
)
w
hich
is
con
si
der
e
d
as
good
li
nk
qual
it
y.
Accordi
ng
to
t
he
stu
dy
[25]
,
SN
R
va
lue
belo
w
5
dB
is
consi
der
e
d
as
bad
li
nk
qual
it
y
as
the
Pac
ke
t
Deli
ver
y
Ra
ti
o
(
PD
R
)
is
ze
r
o.
Acc
ordin
g
t
o
s
om
e
stud
ie
s,
S
NR
al
on
e
is
no
t
s
uffici
ent
to
in
dicat
e
co
ng
est
i
on
w
hen
high
nu
m
ber
of
nodes
pr
ese
nt
in
on
e
m
on
it
ori
ng
a
rea
.
Th
us
,
we
im
ple
m
ented
the
congesti
on
det
ect
ion
base
d
on
inc
reasin
g
node
de
ns
it
y
as
a
net
wor
k
wi
ll
be
congeste
d
w
he
n
node
de
ns
it
y
increases
a
bove
certai
n
po
i
nt.
A
netw
ork
is
assum
ed
to
be
co
ng
e
ste
d
wh
e
n
there
are
m
or
e
than
32
node
s
in
on
e
gi
ve
n
area
base
d
on
a
pr
e
vious
stu
dy
[26]
.
The
P
DR
an
d
thr
oughputs
st
arted
to
decre
ase
ind
ic
at
in
g
co
ngest
io
n
as
the
no
de
de
ns
it
y
reaches
45.
T
her
e
fore
,
we
no
ti
ce
d
that
th
e
netw
ork
sta
rted
to
c
ongest
w
hen
it
reac
he
d
48
nodes
f
or
our
pr
e
-
def
i
ne
d
to
polo
gy
(
30
0
m
x
300
m
)
of
the
m
on
it
or
ing
are
a.
We
e
xten
de
d
t
he
cl
asses
f
r
om
pr
e
vi
ous
S
NR
r
an
ge
m
entioned
wh
e
re
we
i
nd
ic
at
ed
t
hat
th
e
netw
ork
is
e
xtr
e
m
el
y
con
gest
ed
(
EC)
wh
e
n
the
no
de
de
ns
it
y
is
m
or
e
tha
n
27.92
53,
m
edium
con
geste
d
(MC)
wh
e
n
t
he
ra
nge
of
node
de
nsi
ty
is
betwee
n
16.75
52
a
nd
22.
3402,
an
d
non
-
co
ngest
ed
(
NC)
w
hen
th
e
no
de
densi
ty
is less t
han 11.
1701.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
1
2
0
8
–
1
2
2
0
1212
Table
1.
T
he
val
ues of
act
ual
SN
R a
nd esti
m
at
ed
S
NR
with
increasin
g dist
ance
betwee
n nodes
Distan
ce (
m
)
Tr
an
s
m
it
Power
(
d
B
m
)
Actu
al SNR
(dB
)
Esti
m
at
ed
SNR
(dB
)
10
0
4
1
.97
6
4
4
1
.98
4
3
20
0
3
1
.44
0
3
3
1
.44
8
2
30
0
2
5
.27
7
1
2
5
.28
5
0
40
0
2
0
.90
4
3
2
0
.91
2
2
50
0
1
7
.51
2
4
1
7
.52
0
3
60
0
1
4
.74
1
1
1
4
.74
9
0
70
0
1
2
.39
7
9
1
2
.40
5
8
80
0
1
0
.36
8
2
1
0
.37
6
1
85
0
9
.44
6
7
9
.45
4
6
90
0
8
.57
7
9
8
.58
5
8
100
0
6
.97
6
4
6
.98
4
3
2.2.
Sele
ction
of
the
Op
timal Err
or Co
rrec
tin
g Codes
an
d
Tr
an
smit
P
ower
Me
thod
2
wa
s
div
ide
d
int
o
three
s
ub
-
m
e
thods
of
Me
thod
2.1,
Me
thod
2.2,
a
nd
Me
thod
2.3
.
Fo
ll
owin
g
the
cl
assifi
cat
ion
of
SN
R
ra
nge
in
Me
tho
d
1,
we
si
m
ulate
d
the
error
c
orrecti
on
co
des
of
BC
H
an
d
RS
with
a
var
i
et
y
of
error
co
rr
ect
in
g
capa
bi
li
ti
es
su
ch
that
1
≤
t
≤
10
.
W
e
i
m
ple
m
ented
two
ty
pes
of
error
correct
ing
c
odes
(BCH
an
d
RS
cod
es
)
f
ollo
wing
the
ra
ng
e
of
co
dewo
rd
l
eng
t
h,
n
denot
ed
as
15
≤
n
≤
127
f
or
rand
om
un
iform
distribu
ti
on
and
63
≤
n
≤
127
f
or
ra
ndom
no
n
-
unif
orm
no
de
distri
buti
on.
The
ide
a
is
to
si
m
u
la
te
each
error
c
orrecti
ng
co
de
in
obta
inin
g
the
m
os
t
op
ti
m
al
cod
e
with
co
rr
es
pondin
g
co
dewo
rd
le
ng
t
h
and
e
rror
c
orre
ct
ing
capa
bili
ty
in
or
de
r
to
a
dap
t
to
diff
e
re
nt
channel
co
ndit
ion
s
.
For
in
sta
nce,
our
pro
po
s
ed
al
gorithm
i
m
pl
e
m
ented
high
error
co
rr
ect
in
g
ca
pab
il
it
y
f
or
lo
wer
S
NR
t
o
si
gn
ific
a
ntly
re
du
ce
the
t
oo
high
BER
w
her
eas
for
a
go
od
net
work
c
onditi
on,
we
im
ple
m
e
nted
a
m
uch
lo
wer
er
ror
c
orre
ct
ing
ca
pa
bili
t
y.
Th
e
reason
is
that
wh
e
n
t
he
S
NR
co
nd
it
io
n
is
good
(
good
BE
R
perf
or
m
ance),
t
oo
hi
gh
er
ror
c
orrec
ti
ng
ca
pab
il
it
y
m
igh
t no
t be
ne
cessary as retr
ansm
issi
on
is enou
gh
to
s
olv
e
the er
r
or
s
. Th
i
s r
ed
uces th
e over
hea
d
of
e
nc
od
i
ng
and d
ec
odin
g p
ro
ces
ses
of the
ECC
as w
el
l as
the lat
ency ca
us
e
d by the c
om
pu
ta
ti
on
of t
hat ECC
.
Table
2
an
d
Ta
ble
3
s
how
t
he
li
st
s
of
te
ste
d
BC
H
an
d
RS
with
res
pecti
ve
cod
e
w
ord
le
ngth
a
nd
er
r
or
correct
ing
ca
pa
bili
ty
.
The
c
od
e
s
wer
e
sim
ula
te
d
with
CDMA
WSN
to
st
ud
y
t
he
eff
ect
s
of
B
ER
an
d
rem
ai
nin
g
ene
r
gy
as
well
as
l
at
ency
an
d
to
fin
d
the
m
os
t
op
ti
m
al
err
or
c
orrecti
ng
co
de
s
f
or
e
ve
ry
ra
nge
of
SN
R
that
ha
d
been
cl
assifi
ed
.
W
e
al
s
o
a
dded
the
t
ran
sm
i
t
power
cal
ibr
at
ion
to
our
pr
opos
e
d
al
go
rithm
to
m
axi
m
iz
e
re
m
ai
nin
g
ene
rg
y
and
re
du
ce
no
ise
s
in
co
nges
te
d
area
(Met
hod
2.2).
I
n
a
conde
ns
e
d
ne
twork
wh
e
re
the
no
des
are
s
o
cl
os
e
with
eac
h
oth
e
r
a
nd
ha
ving
a
good
range
of
tra
ns
m
issi
on
coverage
,
i
m
ple
m
enting hig
h
tra
ns
m
it
p
ow
e
r
m
igh
t n
ot
b
e
necessa
ry.
Table
2.
T
he
L
ist
o
f Si
m
ulated
BC
H
codes
Er
ror
Co
rr
ectin
g
Co
d
es
n
k
Er
ror
Co
rr
ectin
g
Cap
ab
ilities, t
BCH
7
4
1
15
5
3
31
26
1
31
16
3
31
11
5
63
57
1
63
45
3
63
24
7
63
18
10
127
120
1
127
92
5
127
64
10
255
247
1
255
179
10
511
502
1
511
466
5
511
421
10
Table
3.
T
he
L
ist
o
f Si
m
ulated
RS
co
des
Er
ror
Co
rr
ectin
g
Co
d
es
n
k
Er
ror
Co
rr
ectin
g
Cap
ab
ilities, t
RS
7
3
1
15
11
2
15
5
5
31
27
2
31
13
9
31
11
10
63
59
2
63
45
9
63
43
10
127
123
2
127
113
7
127
107
10
255
251
2
255
235
10
511
507
2
511
499
6
511
491
10
In
a
ddit
ion,
in this
co
ndense
d
netw
ork,
the nod
e
s
m
igh
t
be
interfe
rin
g
with
the
tran
sm
iss
ion
r
an
ge
of
ano
t
her
nod
e
s
[8
]
,
[
9]
.
T
his p
r
om
otes
the
inc
rease
in
i
nterf
e
ren
ces an
d
no
i
ses
an
d
ye
t
the r
ed
uctio
n
of
tr
ansm
i
t
powe
r
m
igh
t
be
a
ble
to
red
uc
e
the
Mult
iple
Access
In
te
r
fer
e
nce
(MA
I)
fr
om
the
CDMA
arch
it
ect
ure.
I
n
a
conde
ns
e
d
network
wh
e
re
ge
ner
al
ly
the
hi
gh
nu
m
ber
of
node
s
with
high
tran
sm
i
t
powe
r
m
igh
t
le
ad
t
he
MAI
to
pea
k
[27
]
,
[
28]
,
cau
sin
g
the
pac
ket
to
be
c
orrupted
an
d
i
ncr
ease
the
BE
R.
The
re
du
ct
i
on
i
n
tra
ns
m
it
powe
r
m
igh
t
be
able
to
sli
gh
tl
y
incre
ase
the
rem
ai
ni
ng
e
nergy
[29]
and
the
or
et
ic
a
ll
y
m
igh
t
be
ab
le
to
red
uce
M
AI
as
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Multi
ple err
or
correcti
on tow
ar
ds
op
ti
mis
ation of e
ner
gy
i
n sens
or
netwo
r
k (
Samirah
Ra
za
li
)
1213
well
.
H
ow
e
ver,
we
al
so
s
ugge
ste
d
that
the
transm
it
po
wer
reducti
on
m
us
t
be
opti
m
a
l
as
too
m
uch
re
duct
i
on
of
tra
ns
m
i
t
power
will
al
so
c
auses
pac
ket
dro
p
as
the
sig
na
l
powe
r
decr
e
ases
w
he
reas
t
he
distance
bet
wee
n
nodes
in
the
non
-
ra
ndom
un
i
form
distribu
ti
on
in
crease
to
100
m
.
The
transm
it
po
we
r
can
be
cal
culat
ed
by
us
in
g
the
Fr
ii
s
Transm
issi
on
equ
at
io
n
s
how
n
in
Eq
uatio
n
(
2).
E
qu
at
io
n
(6)
denotes
the
si
m
pl
if
ie
d
eq
uation
t
o
ob
ta
in
tra
ns
m
it pow
e
r fr
om
the r
ecei
ve
d
si
gn
al
p
owe
r:
=
Pr
(
4
2
)
(6)
w
he
re
Pr
is t
he re
cei
ved
sig
nal a
nd
is t
he dist
ance
betwee
n n
od
e
s.
2.3.
Measurem
en
t Mo
dels
Table
4
sho
w
s
the
par
am
eter
s
def
ine
d
for
te
sti
ng
an
d
si
m
ulati
on
of
the
pro
pose
d
work.
We
i
m
ple
m
ented
the
m
ini
m
u
m
distance
of
10
m
f
or
unif
or
m
di
stribu
ti
on
an
d
100
m
for
non
-
un
i
form
distrib
ution.
Non
-
unif
or
m
di
stribu
ti
on
is
re
ferred
to
as
th
e
distance
betw
e
en
nodes
(m
i
gh
t
not
be
t
he
sam
e
fr
om
on
e
node
to anothe
r) res
ulti
ng
i
n dif
fere
nt levels
of
node den
sit
y pr
es
ent w
it
hi
n one
m
on
it
or
ing are
a.
Table
4.
Param
et
ers
de
fine
d f
or sim
ulatio
n m
od
el
Para
m
eter
Valu
e
Min.
d
ist
.
b
etw
een two
no
d
es
No
ise
Tr
an
s
m
it
Po
wer
(
p
t
)
Mon
ito
ring
ar
ea (m
e
te
r
2
)
Path
los
s p
ara
m
e
te
r
(α)
Pay
lo
ad
,
H
eader
NACK/AC
K
(
)
Er
ror
Detec
tio
n
Er
ror
Co
rr
ectio
n
Nu
m
b
e
r
o
f
Nod
es
Nu
m
b
e
r
o
f
Bits (bi
ts)
1
0
m
AW
G
N
Pow
er a
llo
ca
tio
n
d
ep
en
d
in
g
o
n
SN
R
estima
tio
n
(
-
10
d
Bm
to
0 d
Bm
)
3
0
0
m
x 3
0
0
m
3
.5
1
2
8
,256 b
it/
pkt
8
bits
(
a
d
d
itio
n
a
l 6 b
its a
s NACK
/AC
K wer
e
a
p
p
en
d
ed
with
SN
R
a
n
d
up
d
a
te valu
e fro
m the
receiver
)
CRC
-
3
0
(
CDMA
c
o
m
p
lian
ce
)
Prop
o
sed
M
EC (
v
a
riation
on
the BC
H and
RS E
CC
)
4
,16
,32
,48
,6
4
,80
1
0
0
0
0
,
2
0
0
0
0
,
3
0
0
0
0
,
4
0
0
0
0
,
5
0
0
0
0
The
a
ppli
cat
ion
of
100
m
(as
the
m
axi
m
um
distance
between
no
des)
s
upports
t
he
c
om
m
on
sens
or
nodes
s
uc
h
as
Mi
caZ
and
T
el
os
B
with
t
he
transm
issi
on
range
up
t
o
100
m
.
W
e
a
dd
e
d
Additi
ve
W
hite
Gau
s
sia
n
N
ois
e
(AWGN)
a
nd
Ra
yl
ei
gh
Fa
ding
to
stu
dy
the
im
pacts
bet
we
en
t
hese
tw
o
inter
ven
ti
ons
.
The
powe
r
al
locat
ion was
cali
br
at
ed
acc
ordin
g
t
o
the
S
NR r
a
nge as
w
el
l as e
r
ror
c
orrecti
on s
chem
es.
We
dev
el
op
e
d
our
m
easur
e
m
ent
m
od
el
to
m
easur
e
the
pe
rfor
m
ances
of
CDMA
WSN
in
te
rm
s
of
rem
ai
nin
g
energy,
BER
,
an
d
la
te
ncy.
The
f
ol
lowing
eq
uations
wer
e
us
e
d
to
ob
ta
in
th
e
m
easur
em
ent
resu
lt
s.
The
e
xpressi
on of r
em
ai
nin
g e
nergy incl
ud
i
ng
decodin
g
e
ne
rg
y ca
n be
de
note
d
a
s:
E
Ec
c
=
Hop
x
No
pa
c
k
e
t
x
(
No
bits
+
(
No
bit
s
x
0
.
75
)
+
E
De
c
(7)
w
he
re
Hop
co
rr
es
ponds
to
nu
m
be
r
of
hops
of
t
he
pr
e
def
i
ned
la
ye
red
arc
hitec
ture
,
No
pa
c
k
e
t
is
the
num
ber
of
pack
et
s
in
vo
l
ve
d
in
the
tran
s
m
issi
on
,
No
bit
s
is
nu
m
ber
of
ge
ner
a
te
d
bits
w
hich
for
researc
h
(1000
0
bits),
a
nd
Ene
rgy
De
c
is t
he dec
od
i
ng e
nergy cal
cul
at
ed
in a
s
s
how
n
in
equati
on
(
8).
E
De
c
=
(
2
nt
+
2t
2
)
(
E
a
ddit
ion
+
E
multiplic
a
t
ion
)
(8)
w
he
re
n
de
note
s
the
c
odew
ord
le
ngth
of
t
he
sel
ect
ed
e
rro
r
c
orrecti
ng
c
od
e
s
a
nd
t
is
the
e
rror
co
rr
e
ct
ing
capab
il
it
y
of
t
ha
t
error
c
orrect
ing
c
odes.
We
ta
bu
la
te
d
t
he
t
est
ed
co
des
in
wh
ic
h
wer
e
use
d
in
the
cal
c
ul
at
ion
of
decodin
g
e
ne
rg
y
in
Ta
ble
2
an
d
Ta
ble
3.
W
e
al
so
m
od
el
le
d
the
BER
fo
rm
ulati
on
in
BPSK
m
od
ulat
ion
as
fo
ll
ows:
BER
=
1
2
(
1
−
√
E
b
N
0
E
b
N
0
+
1
)
(9)
Lat
ency
is
cal
c
ulate
d
fo
ll
ow
i
ng
E
qu
at
io
n
(
10)
,
wh
e
re
is
t
he
pro
pag
at
io
n
delay
.
Pr
opa
gatio
n
de
la
y
ref
er
s
to
th
e
tim
e
ta
ken
b
e
tween
de
par
t
ure
of d
at
a from
the
se
nd
e
r
a
nd arr
ival
of d
at
a at
the
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
1
2
0
8
–
1
2
2
0
1214
receiver
.
Me
an
wh
il
e,
is
the
tr
ansm
issi
on
delay
or
al
s
o
kn
own
as
the
packet
iz
at
ion
delay
in
wh
ic
h
ca
n
be
de
fine
d
as
t
he
a
m
ou
nt
of
tim
e
require
d
to
tra
ns
m
it
all
of
the
pac
ket'
s
bits
into
t
he
li
nk.
U
su
al
ly
,
the
transm
issi
on
delay
is
affe
ct
ed
by
data
-
ra
te
of
the
li
nk.
Pr
opa
gatio
n
de
la
y
was
cal
culat
ed
by
di
vid
in
g
th
e
distance
bet
we
en
sen
de
r
an
d
receiver
with
prop
a
gatio
n
s
peed
of
t
he
m
edia.
The
tra
nsm
issi
on
delay
was
cal
culat
ed by d
ividin
g
the
len
gth
of
pack
et
i
n bit
s w
it
h t
he t
ran
sm
issi
on
r
a
te
o
f
the
pr
e
de
f
ined netw
ork
.
La
te
ncy
=
D
el
a
y
Pr
opagat
ion
+
D
el
a
y
Tran
sm
ission
(10)
2.4.
Opt
im
al Erro
r Correc
ti
on
Schemes
f
or
t
he pr
oposed
Multiple E
rr
or
Correcti
on
(
MEC
)
Table
5
sho
ws
the
set
ti
ng
s
of
optim
al
err
or
correct
io
n
sc
hem
es
fo
r
the
propose
d
Mu
lt
iple
Err
or
Correct
ion
(M
EC)
al
gorit
hm
.
This
set
ti
ng
is
optim
al
if
the
arc
hitec
ture
f
ollows
our
pr
e
def
i
ned
to
po
l
ogie
s
i
n
wh
ic
h
the
def
a
ult
num
ber
of
bits
is
10
000
bi
ts
with
inc
reas
ing
num
ber
of
nodes
(
betwee
n
4
a
nd
80
no
de
s).
I
n
this
stu
dy,
the
CDMA
a
r
c
hitec
ture
im
ple
m
ented
Mult
i
-
Ca
rr
ie
r
(MC
-
CDMA)
cha
nnel
acce
ss
m
et
hod
a
nd
BPSK
m
od
ula
ti
on
as
ta
bu
la
te
d
in
Table
4.
Fr
om
the
si
m
ula
ti
on
resu
l
ts
acqu
ire
d,
th
e
BC
H
cod
es
with
cod
e
w
ord
le
ng
th
of
12
7
can
no
t
outpe
rfo
r
m
RS
cod
es
with
the
sam
e
cod
e
word
le
ngth
w
he
n
the
netw
ork
conditi
on
is
e
xt
rem
e
ly
con
ge
s
te
d
due
to
re
dund
a
ncy
a
dded
by
the
BC
H
c
odes
is
hi
gh
e
r
t
han
the
RS
c
odes.
I
n
so
m
e
stud
ie
s,
m
or
e
red
unda
nc
ie
s
m
igh
t
lower
the
BER
.
H
ow
e
ve
r,
in
ou
r
case,
too
h
ig
h
r
edun
dan
cy
a
ppende
d
to
the
transm
itted
bits
in
a
con
ge
ste
d
net
work
fl
oode
d
the
netw
ork
causi
ng
m
or
e
bits
to
be
co
rru
pted
.
Th
us,
we
obser
ve
d
that
RS
code
w
it
h
cod
e
w
ord
le
ng
t
h
of
127
i
s
op
ti
m
u
m
fo
r
an
extrem
el
y
congeste
d
co
ndit
ion
.
Fo
r
a
m
ediu
m
congested
co
nd
it
io
n,
we
a
ppli
ed
BC
H
co
des
f
or
both
,
hi
gh
S
NR
and
t
h
e
highest
SNR
with
good
li
nk
c
ondi
ti
on
wh
e
re
BE
R
is
relat
ively
low.
Hi
gh
e
r
a
ppen
de
d
bits
m
i
gh
t
not
c
orrupt
the
bits
as
m
uch
as
wh
e
n
the
SN
R
is
low.
Fo
r
a
non
-
co
ngest
ed
conditi
on
,
we
app
li
ed
retra
nsm
issi
on
for
hi
gh
e
r
SN
R
with
lo
w
BER
present
in
t
he
netw
ork
as
the
retra
nsm
issi
on
is
enough
to
co
rr
ect
the
erron
e
ous
bits.
This
is
also
to
a
dd
up
t
hat
there
wer
e
on
ly
a
fe
w
node
s
that
m
igh
t
interfe
re
with
the
tran
s
m
issi
on
range
in
w
hich
if
the
re
we
re
colli
sion
s
or
interfe
re
nce
tha
t
can
cor
r
upt
the
bits,
the
er
r
or
s
m
igh
t
no
t
be
too
si
gn
ific
ant
com
par
ed
to
the
conditi
on w
it
h hig
her n
ode
de
ns
it
y.
Table
5.
O
pti
m
al
error co
rr
ect
ing
c
odes
for M
EC
SNR class
es ex
ten
sio
n
Er
ror
Co
rr
ectin
g
Co
d
es
Co
d
ewo
rd len
g
th
,
n
Inf
o
r
m
atio
n
Bits,
k
Er
ror
Co
rr
ectin
g
Cap
ab
ility,
t
Co
n
d
itio
n
: E
C
Cas
e 1: ECSNR
lw
RS
127
113
7
Cas
e 2: ECSNR
l
RS
127
115
6
Cas
e 3: ECSNR
m
RS
127
117
6
Cas
e 4: ECSNR
h
RS
127
119
4
Cas
e 5: ECSNR
hg
RS
127
121
3
Co
n
d
itio
n
: M
C
Cas
e 1: MCSNR
lw
RS
63
53
5
Cas
e 2: MCSNR
l
RS
63
51
6
Cas
e 3: MCSNR
m
RS
63
49
7
Cas
e 4: MCSNR
h
BCH
63
36
5
Cas
e 5: MCSNR
hg
BCH
63
24
7
Co
n
d
itio
n
: NC
Cas
e 1: MCSNR
lw
BCH
127
64
10
Cas
e 2: NCSNR
l
RS
127
121
3
Cas
e 3: NCSNR
m
BCH
31
11
6
Cas
e 4: NCSNR
h
RS
31
17
7
Cas
e 5: NCSNR
hg
Retran
s
m
iss
io
n
wi
th
I
R
Table
6
s
how
s
the
op
ti
m
al
tra
ns
m
it
po
we
r
c
orres
pondin
g
t
o
dif
fer
e
nt
cl
as
ses
a
nd
co
nges
ti
on
pr
e
sent
base
d
in
our
a
bovem
entione
d
arc
hitec
ture.
The
tra
ns
m
i
t
powe
r
with
hig
h
SN
R
w
ere
sign
ific
a
ntly
red
uce
d
t
o
20%
f
or
e
xtre
m
el
y
con
geste
d
c
onditi
on,
40%
for
m
ediu
m
con
gested
,
and
50%
f
or
non
-
co
ngest
ed.
I
n
the
extrem
el
y
con
gested
co
ndit
ion,
the
tra
nsm
it
po
we
r
wa
s
not
reduce
d
too
m
uch
as
com
par
ed
to
the
non
-
congeste
d
co
ndit
ion
in
or
der
to
avo
id
inc
re
m
ent
in
the
noise
pr
ese
nt.
Th
is
is
becau
se
the
signa
l
pow
er
will
red
uce
the
t
ransm
it
po
wer
cal
ibrated
.
T
he
lo
wer
powe
r
wil
l
be
pea
k
the
noise
,
wh
ic
h
le
ad
to
t
he
incr
e
m
ent
in
BER
.
For
no
n
-
congeste
d
co
ndit
ion
,
t
he
tra
nsm
it
po
wer
ca
n
be
re
du
ce
d
unti
l
50
%
a
s
ha
ving
hi
gh
SN
R
ind
ic
at
e
high si
gn
al
po
wer wit
h l
ow
noise
power.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Multi
ple err
or
correcti
on tow
ar
ds
op
ti
mis
ation of e
ner
gy
i
n sens
or
netwo
r
k (
Samirah
Ra
za
li
)
1215
Table
6.
O
pti
m
al
tran
sm
it
p
ower
for M
EC
SNR class
es ex
ten
sio
n
Tr
an
s
m
it
po
wer
c
a
lib
ration
Co
n
d
itio
n
: E
C
Cas
e 1: ECSNR
lw
Maintain
Cas
e 2: ECSNR
l
Maintain
Cas
e 3: ECSNR
m
Red
u
ctio
n
by
1
0
%
Cas
e 4: ECSNR
h
Red
u
ctio
n
by
1
0
%
Cas
e 5: ECSNR
hg
Red
u
ctio
n
by
20%
Co
n
d
itio
n
: M
C
Cas
e 1: MCSNR
lw
Maintain
Cas
e 2: MCSNR
l
Red
u
ctio
n
by
1
0
%
Cas
e 3: MCSNR
m
Red
u
ctio
n
by
2
0
%
Cas
e 4: MCSNR
h
Red
u
ctio
n
by
3
0
%
Cas
e 5: MCSNR
hg
Red
u
ctio
n
by
4
0
%
Co
n
d
itio
n
: NC
Cas
e 1: NCSNR
lw
Red
u
ctio
n
by
1
0
%
Cas
e 2: NCSNR
l
Red
u
ctio
n
b
y
20
%
Cas
e 3: NCSNR
m
Red
u
ctio
n
by
3
0
%
Cas
e 4: NCSNR
h
Red
u
ctio
n
by
4
0
%
Cas
e 5: NCSNR
hg
Red
u
ctio
n
by
5
0
%
3.
RESU
LT
S
A
ND AN
ALYSIS
We
colle
ct
ed
the
res
ults
base
d
on
the
ex
per
i
m
ent
of
MEC
i
n
te
rm
s
of
incre
m
ent
in
no
de
densi
ty
and
com
par
ed
the
non
-
cha
nnel
ada
ptati
on
us
in
g t
he
existi
ng error
c
orrecti
on s
chem
es. F
igu
re
2
sho
ws
the a
ver
a
ge
rem
ai
nin
g
energy
against
SNR
between
ME
C
with
BC
H
(127,
k)
t
=
10
, RS
(1
27
,
121)
t
=
3,
BC
H
(3
1,
k)
t
=
7,
a
nd
RS
(
31,
k)
t
=
7.
The
existi
ng
R
S
(
31,
k)
t
=
7
ha
ve
the
hi
gh
e
st
con
sta
nt
rem
ai
nin
g
e
nergy
th
r
oughout
the
SN
R.
The
highest
rem
ai
nin
g
e
nergy
f
or
MEC
is
ob
ta
in
ed
w
he
n
the
S
NR
is
at
the
good
c
onditi
on.
Eve
n
though
RS
(
31,
k)
t
=
7
ha
d
the
hi
gh
e
st
rem
a
ining
e
ne
rg
y
in
a
no
n
-
co
ngest
ed
c
onditi
on,
the
hi
gh
e
r
redu
nd
a
ncy a
dded
b
y t
his c
ode w
as
not
pr
act
ic
al
f
or h
i
gher
node de
ns
it
y.
Figure
2. A
verage Rem
ai
nin
g Ene
rg
y a
gain
s
t SNR
In
a
dd
it
io
n,
M
EC
cann
ot
re
duce
BER
as
f
urt
her
reducti
on
of
the
rem
ai
n
ing
e
nergy
wa
s
too
m
uch
du
e
t
o
draw
ba
cks
bet
wee
n
the
capa
bili
t
ie
s
of
er
ror
correct
ions
th
at
add
e
d
re
dund
a
ncies
wit
h
m
or
e
com
plexiti
es
a
nd
require
d
hig
he
r
e
nergy
w
hen
bette
r
er
r
or
co
rr
ect
io
n
is
us
e
d.
H
oweve
r
,
MEC
op
ti
m
ized
the
rem
ai
nin
g
ene
rg
y
in
high
S
NR
as
to
o
powerfu
l
e
rror
c
orrecti
on
sche
m
es
wer
e
no
t
necessa
ry.
I
ns
t
ead,
t
he
retransm
issi
on
was
en
ough
to
handle
the
er
rors
w
hich
do
no
t
us
ed
to
o
m
uc
h
ene
rg
y
as
co
m
par
ed
to
ECC
du
e
to
decodin
g
a
nd
c
om
pu
ta
ti
on
ove
r
hea
d.
H
ow
e
ve
r,
f
or
lo
w
SN
R
(
wh
e
r
e
BER
is
relatively
ver
y
high)
,
it
is
su
bst
antia
l
to
correct
the
e
rror
s
t
o
reduce
too
m
any
rese
nd
pac
kets
f
r
om
the
sen
der.
A
hi
gher
num
ber
of
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
1
2
0
8
–
1
2
2
0
1216
flo
od
e
d
retra
nsm
issi
on
s
in
th
e
netw
ork
not
on
ly
us
es
ene
r
gy
but
al
s
o
res
ources
a
s
well
w
hich
m
igh
t
r
end
e
r
the
ne
tw
ork
c
ausin
g
m
al
fu
nc
ti
on
.
The
rem
ai
nin
g
e
nergy
for
lo
w
SN
R
was
le
sse
r
c
om
par
ed
to
ot
he
r
S
NR
range
as
hi
gh
e
r
er
ror
c
orrecti
on
capa
bili
ty
was
us
e
d.
T
hu
s,
we
can
co
nc
lud
e
t
hat
the
optim
isa
ti
on
bet
ween
rem
ai
nin
g
ene
r
gy and BER b
y
m
eans o
f
a
da
ptati
on
to
the c
hannel co
ndit
ion
it
sel
f.
I
n
lo
w
S
NR,
the
re
m
ai
nin
g
energy m
igh
t no
t be a
ble to be
b
ooste
d
as
m
uch
as
wh
e
n
the
SN
R i
s h
i
gh.
This lo
w
er
r
or
correct
ing capa
bili
ty
cod
e
s
m
igh
t
not
able
to
ha
ndle
the
hi
gh
er
ror
rates
if
lo
wer
er
ror
c
orr
ect
ing
ca
pa
bili
ty
is
us
e
d
in
l
ow
S
NR
wh
e
re BER is
high.
Figure
3
s
how
s
the
c
om
par
ison
of
a
ver
a
ge
la
te
ncy
against
SN
R
betwee
n
MEC
an
d
e
xisti
ng
c
odes
.
The
e
xisti
ng
c
od
i
ng
sc
hem
es
ha
ve
c
onsta
nt
la
te
ncy
thr
ou
ghout
the
SNR
due
to
the
abse
nce
of
ch
ann
e
l
ad
aptat
io
n.
T
he
la
te
ncy
changed
with
SN
R
range
due
to
the
cha
ng
e
s
in
error
c
orrecti
on
us
e
d
thr
ough
ou
t
the
SN
R
as
s
how
n
in
the
fig
ur
e
.
The
la
te
ncy
corres
ponded
t
o
the
a
pp
e
nde
d
bits
in
the
ne
twork
wh
e
re
higher
app
e
nded
bits
or
re
dunda
ncy
causes
hi
gh
e
r
la
te
ncy
as
la
rg
e
num
ber
of
bits
prolo
nged
the
decodin
g
tim
e
of
the
receiv
ed
i
nfor
m
at
ion
.
I
n
this
stu
dy,
B
CH
rec
orde
d
higher
la
te
ncy
due
to
it
s
c
om
plexit
y
and
higher
nu
m
ber
of
re
dundancies
.
H
ow
e
ve
r,
this
c
orres
ponds
to
the
fact
t
hat
BC
H
ha
ve
be
tt
er
error
co
r
recti
on
capab
il
it
y
that
the
RS
cod
e.
F
or
insta
nce,
BC
H
(31,
k)
out
perform
ed
RS
(31,k)
with
the
sa
m
e
t
=
7
in
te
rm
s
of
BER
even
t
hough
the
la
te
ncy
was
too
hi
gh.
H
ow
e
ver,
the
us
e
of
m
ult
i
-
co
ding
m
igh
t
able
to
op
tim
i
ze
the
la
te
ncy
as
the
l
at
ency
will
no
t
al
ways
be
h
ig
h
thr
ough
ou
t
the
SN
R
.
Wh
en
hig
h
redu
nd
a
nc
y
is
no
t
need
e
d,
th
e
la
te
ncy w
il
l be
reduce
d by us
i
ng m
uch
lo
wer capa
bili
ty
co
de
s that a
pp
e
nd
ed
m
uch
lo
we
r
r
e
dundancy.
Figure
3. A
verage Late
ncy ag
ai
ns
t S
NR
Figure
4
il
lustr
at
ed
the
ave
ra
ge
BER
again
s
t
increasin
g
no
de
de
ns
it
y.
Th
e
MEC
sh
owe
d
the
lo
west
BER
com
par
ed
to
existi
ng
s
chem
es
fo
r
the
non
-
c
onge
ste
d
conditi
on
of
l
ess
than
11.17
01
node
de
ns
it
y
wh
e
n
SN
R
reac
hing
30
dB.
The
re
aso
n
is
that
B
CH
(31,
k)
wa
s
i
m
ple
m
ented
fo
r
m
ediu
m
S
NR
in
non
-
c
on
gested
env
i
ronm
ent
a
s
su
ch
re
dund
ancy
sign
ific
a
ntly
so
lve
the
BER
and
does
no
t
flo
od
the
network
in
th
e
non
-
congeste
d
en
vironm
ent.
How
ever,
f
or
hi
ghe
r
node
densi
ty
,
BC
H
(31,
k)
was
im
pr
act
ic
a
l
becau
se
of
to
o
high
redu
nd
a
ncy. How
e
ve
r,
the exi
sti
ng
RS (
12
7, k)
d
i
d
not ou
t
perform
BER
of
MEC
for
no
n
-
c
ongeste
d
co
nd
it
i
on
(≤
11.
1701)
.
Figure
5
s
ho
ws
the
ave
ra
ge
rem
a
ining
e
nergy
again
st
node
de
ns
it
y.
The
net
wor
k
achieve
d
the
highest
rem
ai
nin
g
e
nergy
in
congeste
d
co
ndit
ion
f
or
m
edium
SN
R
as
we
us
e
d
m
edium
err
or
c
orr
ect
ing
capab
il
it
y
wh
i
ch
opti
m
ise
d
the
nee
d
betwe
en
rem
ai
nin
g
e
nergy
an
d
re
du
nd
a
ncy
ad
de
d
to
the
netw
ork.
Ther
e
was
m
uch
higher
er
ror
c
orre
ct
ing
ca
pa
bili
t
y
us
e
d
f
or
lo
w
co
ng
e
ste
d
c
on
diti
on
a
s
the
r
e
m
ai
nin
g
e
nergy
wa
s
sli
gh
tl
y
lower
than
the
existi
ng
RS
(
31,
k).
Howe
ver,
in
com
par
ison
to
Figure
4
w
here
BER
of
MEC
was
at
it
s
lowest
w
he
n
no
de
den
sit
y
at
11.17
01,
th
e
higher
rem
ain
in
g
en
er
gy
of
RS
pro
ve
d
tha
t
the
existi
ng
RS
did
no
t
op
ti
m
iz
e
th
e
BER
and
re
m
ai
nin
g
en
er
gy
.
Me
anwhil
e,
there
was
sli
gh
t
increm
ent
in
the
rem
ai
nin
g
energy
from
6
28946.3
3
J
f
or
node
de
ns
it
y betwee
n 16.75
52 and
22.
3402.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Multi
ple err
or
correcti
on tow
ar
ds
op
ti
mis
ation of e
ner
gy
i
n sens
or
netwo
r
k (
Samirah
Ra
za
li
)
1217
Figure
4. A
verage BER
a
gain
st Node
d
e
ns
it
y (for
SN
R=
30
)
Figure
5. A
verage Rem
ai
nin
g Ene
rg
y a
gain
s
t Node
de
ns
it
y
Figure
6
sho
w
s
the
Percen
ta
ge
increm
ent
in
rem
ai
nin
g
energy
of
MEC
ag
ai
ns
t
SN
R
com
par
e
d
with
BC
H
(
127,k
)
for
t=
7,
RS
(
127,k
)
f
or
t
he
sam
e
t
and
t
he
c
om
par
ison
of
inc
rem
ent
in
rem
ai
nin
g
e
nergy
betwee
n
existi
ng
RS
a
nd
BC
H.
F
ro
m
the
graph,
it
can
be
ob
s
er
ved
t
hat
as
the
SN
R
inc
r
eased,
th
e
pe
rc
entage
increm
ent
in
rem
ai
nin
g
ene
rgy
al
so
increased.
Be
sides
that
,
the
per
ce
ntag
e
of
increm
ent
of
rem
ai
nin
g
e
nergy
of
MEC
is
higher
t
han
t
he
e
xi
sti
ng
RS
(
127,
k)
a
nd
BC
H
(
127,
k)
w
he
n
SN
R
ac
hieving
betwee
n
20
dB
and
50 d
B
.
It
is
al
so
point
ed
ou
t
t
hat
the
RS
(
127,
k)
ha
s
a
higher
pe
rc
entage
of
inc
re
m
ent
durin
g
t
he
low
S
NR
of
20
dB
beca
us
e
MEC
has
m
or
e
powerfu
l
error
c
orrecti
on
sc
hem
es
wh
ic
h
use
d
m
or
e
energy.
Howe
ver,
th
e
diff
e
re
nce
bet
ween
t
he
per
c
entage
i
ncr
em
ent
of
MEC
a
nd
R
S
is
quit
e
sm
a
ll
.
Thi
s
al
so
c
orrespo
nds
to
the
gr
a
ph
in
Fig
ur
e
4
w
hich
de
m
on
strat
ed
tha
t
du
ri
ng
t
he
lowest
S
NR,
MEC
achieve
d
bette
r
BER
than
the
existi
ng
RS
w
hich
res
ults
in
lowe
r
rem
ai
nin
g
e
ne
rg
y.
T
hu
s
,
a
co
nclu
sion
was
m
ade
w
her
e
MEC
ha
d
op
ti
m
ise
d
the
perform
ances
betwee
n
th
e
BER
and
rem
ai
nin
g
ene
r
gy
and
fo
r
t
he
lowest
SN
R
an
d
f
or
highes
t
SN
R, MEC
p
e
r
form
ed
m
or
e t
ow
a
r
ds
op
ti
m
is
at
ion
of r
em
ai
nin
g
e
ne
rg
y a
nd
lat
ency.
Evaluation Warning : The document was created with Spire.PDF for Python.