Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
9
, No
.
3
,
J
un
e
201
9
, pp.
1997
~
20
05
IS
S
N:
20
88
-
8708
,
DOI:
10
.11
591/
ijece
.
v
9
i
3
.
pp1997
-
20
05
1997
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
An
a
pp
roxim
atio
n
d
elay
between
c
ons
ecutive
r
eque
sts
f
or
c
on
gesti
on
c
ont
rol
in
u
nicast
CoAP
-
bas
e
d
g
roup
c
ommun
icat
i
on
Chanwi
t
S
uwann
apon
g
,
C
h
at
c
ha
i
Khu
nb
oa
Depa
rte
m
ent
of C
om
pute
r
Engi
n
ee
ring
,
Khon K
a
en
Univer
si
t
y
,
T
hai
l
and
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
ug
2
, 2
01
8
Re
vised
Dec
18
, 2
01
8
Accepte
d
Ja
n
1
7
, 2
01
9
Thi
s
rese
ar
ch
p
rese
nts
a
wa
y
t
o
avoi
d
net
wor
k
conge
stion
du
ring
unic
ast
CoAP
-
base
d
g
roup
comm
uni
ca
t
ion
using
inc
rea
sed
d
el
a
ys
bet
wee
n
conse
cutive
req
u
ests
(DCR)
in
L
oW
PAN
borde
r
route
rs
to
li
m
it
req
uest
send
rat
es.
It
al
so
provide
s
a
wa
y
to
det
ermine
DCR
val
ues
that
are
s
uit
able
for
var
ious
net
work
group
size
s
with
diffe
ring
node
c
ounts.
The
optim
al
DCR
is
obta
in
ed
using
the
least
square
s
appr
oximati
on
m
et
hod
and
the
rel
a
ti
v
e
m
ini
m
um
.
Result
s
from
expe
rimentation
show
s
a
positi
ve
r
elati
on
,
that
is,
an
inc
re
ase
in
grou
p
size
necess
it
ates
an
inc
re
ase
in
DCR
val
ue
.
Ex
per
iments
in
var
ious
group
s
iz
es
show
fav
ora
ble
net
work
pe
rform
anc
e
and
support
the
proposed
conge
s
ti
on
cont
rol
m
ethod using
DCR.
Ke
yw
or
d
s
:
Appro
xim
ation
CoA
P
Group c
omm
un
ic
at
ion
Un
ic
ast
W
i
reless se
nso
r netw
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
:
Chatc
hai
Khu
nboa
,
Dep
a
rtem
ent
of
Com
pu
te
r
Enginee
rin
g,
Kho
n Kaen
Un
iversity
,
123
M
oo
16 Mi
tt
apap
Rd
.
, N
a
i
-
Muan
g, Mua
ng D
ist
rict
, K
hon Kae
n 4
0002
, T
haila
nd
.
Em
a
il
: chatc
hai@kku.ac.t
h
1.
INTROD
U
CTION
To
day,
the
use
of
co
ns
trai
ne
d
de
vices
is
beco
m
ing
com
m
on
place.
The
ir
abili
ty
to
con
nect
to
the
internet
br
in
gs
about
the
c
oncept
of
the
I
nter
net
of
T
hi
ng
s
(IoT
).
T
he
se
dev
ic
es
a
r
e
ty
pical
ly
us
ed
f
or
colle
ct
ing
i
nfo
rm
ation
f
r
om
sens
ors
int
o
sta
ndar
d
net
works
[
1],
[
2].
I
nter
net
of
Things
c
om
e
in
tw
o
cat
egories:
Non
-
IP
an
d
I
P
-
ba
sed.
Non
-
IP
sy
stem
s
do
not
r
el
y
on
IP
a
ddr
esses
for
com
m
un
ic
at
ion
.
Such
no
n
-
IP
prot
oc
ols
inclu
de
the
Bl
ueto
oth
lo
w
e
nergy
an
d
Zig
bee
prot
oco
ls
.
IP
-
base
d
syst
e
m
s,
wh
ic
h
re
qu
i
re
IP
addresses
, c
omprise
prot
oco
ls
su
c
h
as
6Lo
W
PAN a
nd T
hr
e
ad.
To
s
upport
IP
-
base
d
com
m
un
ic
at
ion
[
3],
the
In
te
r
net
En
gine
erin
g
Tas
k
Force
(IETF
)
has
dev
el
ope
d
the
Const
rain
ed
A
ppli
cat
ion
Protoc
ol
(C
oAP),
a
we
b
com
m
un
ic
at
i
on
protoc
ol
s
pecial
ly
desig
ned
for
const
raine
d
de
vices.
It
is
si
m
i
la
r
to
HTTP
[
4]
,
bu
t
w
orks
on
top
of
the
Use
r
Diag
ram
Pr
oto
c
ol
(
U
DP
)
in
s
te
ad
of
the
Tra
ns
m
issi
on
C
ontrol
Protoc
ol
(TC
P)
.
This
el
i
m
i
nates
the
ne
ed
f
or
a
th
ree
-
w
ay
hands
ha
ke
and
th
e
need
to
m
ai
nta
in
a
connecte
d
sta
te
,
m
aking
CoA
P
ideal
f
or
const
raine
d
de
vices.
The
protoc
ol
al
so
suppo
rts
m
achine
-
to
-
m
a
chine
(M
2M)
c
omm
u
nicat
ion
.
M2M
i
nvolv
e
s
a
se
nsor
de
vi
ce
sen
ding
dat
a
thr
ough
a
wireless
netw
ork
to
an
ap
plica
ti
on.
T
he
ap
plica
ti
on
c
onve
rts
the d
at
a
into
m
edia
th
at
is
show
n
to
the u
ser
,
al
lo
wing
f
or
i
m
m
ediat
e interact
ion an
d dec
isi
on
m
aking.
Ty
pical
CoA
P
com
m
un
ic
at
io
n
is
one
-
to
-
on
e.
A
cl
ie
nt
exc
hanges
m
essages
with
a
se
r
ve
r
us
in
g
th
e
fo
ll
owin
g
ste
ps
as
sho
wn
i
n
Figure
1
:
1)
th
e
us
er
init
ia
te
s
a
GET
c
omm
a
nd
t
hro
ugh
a
w
eb
brow
se
r
to
r
equ
e
st
data
f
ro
m
the
CoA
P
se
rv
e
r.
2)
The
Co
AP
cl
ie
nt
node
rec
ei
ves
this
GE
T
com
m
and
f
r
om
the
bro
wse
r,
a
nd
sen
ds
a
request
m
essage
to
the
CoA
P
ser
ver
through
U
DP.
3)
Wh
en
the
s
erv
e
r
receives
the
request
m
e
ssage,
it
send
s
a
re
ply
m
essage
cont
ai
nin
g
the
requested
data
ba
ck
to
the
cl
ie
nt
.
4)
T
he
cl
ie
nt
sen
ds
the
data
to
the
browser.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
3
,
June
201
9
:
1997
-
2005
1998
Figure
1.
F
undam
ental
CoA
P
com
m
un
ic
at
ion
In
m
any
cases,
on
e
-
to
-
m
any
com
m
un
ic
at
ion
is
re
qu
i
red.
Su
c
h
insta
nces
include
sim
ult
aneous
data
colle
ct
ion
fro
m
m
u
lt
iple
sensor
nodes
f
or
ve
rificat
io
n
and
pr
ocessin
g,
a
nd
node
gro
upin
g.
C
oAP
gro
up
com
m
un
ic
at
ion
is
thu
s
im
portant
to
I
oT
,
a
nd
is
on
e
i
ns
ta
nce
of
the
gro
up
c
oncept
being
i
ncorpo
rated
int
o
data
com
m
un
ic
at
ion
.
U
nfo
rt
un
at
el
y,
Co
AP
do
es
no
t
pr
ovide
a
way
to
ha
nd
le
ne
tw
ork
congesti
ons
th
at
m
a
y
arise,
su
c
h
as
wh
e
n
one
cl
ie
nt
node
exc
ha
ng
e
s
m
essages
with
m
ulti
ple
serv
e
r
no
des
usi
ng
un
ic
ast
f
or
basic
com
m
un
ic
at
ion
or
f
or
re
sou
rce
obser
vatio
n.
Su
c
h
a
pro
blem
is
wo
rt
h
in
vestigat
in
g
[
5
],
[
6
]
.
As
a
res
ult,
a
stud
y
by
I
.
I
sh
a
q
et
al
[
7
]
pro
po
ses
the
use
of
a
C
oAP
-
base
d
gr
oup
c
omm
un
ic
at
ion
m
et
ho
d
cal
le
d
un
ic
as
t
wh
e
re
a
le
vel
of
intel
li
ge
n
ce
is
intro
duce
d
to
facil
it
at
e
reso
urce
m
anag
e
m
ent,
m
aking
it
po
ssible
to
us
e
a
sing
le
request
m
essage to
m
anag
e
r
es
ources
.
Group
com
m
un
ic
at
ion
th
rou
gh
Co
AP
is
hi
gh
ly
reli
able
and
ef
fecti
ve,
but
if
on
e
cl
ie
nt
node
sen
ds
un
ic
ast
s
to
m
ulti
ple
serv
e
r
node
s
in
rap
i
d
su
ccessi
on,
th
e
res
pons
e
m
essages
from
tho
se
ser
ver
s
would
be
equ
al
ly
rap
i
d,
and
would
res
ult
in
inevita
bl
e
congesti
on.
I
f
the
rate
of
r
equ
e
st
m
essages
bein
g
sent
is
no
t
m
od
erated,
w
he
n
t
he
netw
ork
reac
hes
it
s
ca
pacit
y
thres
hold,
it
w
ou
l
d
sta
r
t
to
d
isc
ar
d
m
essages,
ca
us
in
g
data
loss,
as
sh
own
in
F
igu
re
2.
To
av
oid
unic
ast
con
ge
sti
on,
a
delay
between
co
ns
ec
utiv
e
req
ue
sts
(DC
R)
is
introd
uced
to
ke
ep
the
rate
of
request
m
essages
to
within
th
e
lim
it
s
of
the
netw
ork
a
nd
th
e
recipie
nt
no
de
s.
I
f
le
ft
unco
nt
r
olled,
the
bur
de
n
on
t
he
netw
ork
ca
us
e
d
by
ra
pid
m
essages
cou
l
d
res
ult
in
the
a
f
or
em
entione
d
data loss
and e
ven n
et
wor
k
c
ollapse.
Figure
2
.
Co
A
P
-
ba
sed
gr
oup com
m
un
ic
at
ion
un
ic
ast
c
onge
sti
on
This
resea
rch
aim
s
to
pr
ovid
e
a
so
luti
on
to
the
unic
ast
congesti
on
pro
ble
m
by
add
in
g
D
CR
into
the
gateway
(
Lo
W
PAN
bo
rd
e
r
r
ou
te
r)
t
o
lim
it
the
m
essage
send
rate
to
within
a
n
acce
pta
ble
tolera
nce.
It
al
s
o
pr
ese
nts
a
way
to
determ
ine
DCR
values
f
or
dif
fer
e
nt
gro
up
siz
es
with d
if
fer
e
nt
node
de
ns
it
ie
s
us
in
g
th
e
le
as
t
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
An a
pp
r
oxim
at
ion
delay
b
et
w
een c
onsecuti
v
e req
uests
f
or
c
ongestio
n
c
ontrol in… (
Chanwit S
u
wann
apon
g
)
1999
sq
ua
res
a
ppr
oxi
m
ation
m
et
ho
d.
Op
ti
m
u
m
DCR
values
a
re
ob
ta
i
ned
from
the
relat
ive
m
ini
m
u
m
,
and
t
est
e
d
us
in
g
t
he
Co
oj
a sim
ulator.
The
rem
ai
nd
er
of
t
his
pap
e
r
i
s
or
ga
nize
d
as
fo
ll
ows:
Sect
io
n
2
pro
vid
es
ba
ckgr
ound
inf
orm
ation
on
CoA
P
,
unic
ast
gro
up
c
omm
u
nic
at
ion
in
IoT
,
an
d
le
ast
squ
ares
a
ppr
ox
im
at
ion
.
Sect
io
n
3
pr
ese
nts
t
he
resu
lt
s
and
a
naly
sis,
wh
ic
h
co
ver
s
t
he
m
e
tho
dol
ogy
fo
r
fi
nd
i
ng
t
he
DCR,
the
si
m
ula
ti
on
of
th
e
DCR
values,
and
a
descr
i
ption o
f
t
he results.
Th
e
la
st sect
ion
,
Se
ct
ion
4,
su
m
m
a
rizes the
re
sea
r
ch.
2.
RESEA
R
CH MET
HO
D
2.1.
A
b
ackgr
ou
n
d
of
C
oAP
The
C
onstrai
ne
d
Applic
at
ion
Protoc
ol
(CoAP)
is
a
sp
eci
al
iz
ed
we
b
tr
a
ns
fe
r
protoc
ol
desig
ne
d
f
or
const
raine
d
node
s
an
d
co
nst
rained
netw
orks.
It
s
uppo
rts
Io
T
us
in
g
li
gh
twei
ght
m
essages,
an
d
util
iz
es
relat
ively
litt
le
serv
er
resou
rc
es
and
po
wer
c
om
par
ed
to
ot
her
pr
oto
c
ols.
It
is
able
to
su
pport
requests
from
a
la
rg
e
gro
up
of
cl
ie
nts.
De
sig
ne
d
by
the
I
nter
net
E
ng
i
neer
i
ng
Ta
sk
F
or
ce
(
IETF
),
Co
AP
c
om
pr
ise
s
tw
o
l
ay
ers:
1)
Me
ssa
ging
Lay
er
to
ha
nd
l
e
UD
P
for
co
m
m
u
nicat
ion
,
and
2)
Re
qu
es
t/
Re
sp
onse
La
ye
r
to
handle
m
et
ho
d
and res
ponse
c
od
e
s.
CoA
P
is
sim
ilar
to
HT
TP
i
n
that
t
hey
both
use
the
RE
STful
W
e
b
Se
rv
ic
e,
a
f
unda
m
ental
web
te
chnolo
gy
a
va
il
able
on
e
ve
ry
platf
or
m
w
her
e
a
se
rv
e
r
node
c
an
creat
e
res
ources
f
or
us
e
i
n
the
U
RI,
a
nd
cl
ie
nt
nodes
ca
n
acce
ss
th
os
e
serv
e
r
node
re
so
urces
t
hro
ugh
f
our
ty
pes
of
m
e
tho
d
co
des
:
GET,
P
UT,
P
OS
T
,
and
DELET
E
[
8
]
.
A
nother
si
m
il
arity
is
that
CoA
P
can
se
nd
var
i
ous
ty
pes
of
payl
oa
ds
,
and
ca
n
ind
ic
a
te
the
ty
pe
of
payl
oa
d
us
in
g
XML,
J
SON,
a
nd
CB
OR
et
c.
Am
i
ds
t
the
sim
i
la
riti
es,
howe
ver,
a
disti
nct
dif
f
eren
ce
betwee
n
the t
w
o
pr
oto
c
ols is that Co
AP
c
omm
un
ic
at
es thr
ough
UDP
w
hile HTTP uses
T
CP. An a
dv
a
nt
age of
UDP,
a
c
onne
ct
ion
le
ss
pr
otoc
ol,
is
that
da
ta
can
be
se
nt
ver
y
quic
kly
without
the
ne
ed
f
or
a
t
hr
ee
-
way
hands
ha
ke
that
is re
qu
i
red by
TCP.
Ther
e
are
f
ou
r
ty
pes
of
m
e
ssages
de
fine
d
in
C
oAP:
1)
Co
nf
irm
able,
2)
Non
-
c
onfi
rm
a
ble,
3)
Ackn
ow
le
dg
e
m
ent,
and
4)
Re
set
.
In
t
he
Re
qu
est
la
ye
r,
Confirm
able
and
N
on
-
c
onfirm
able
m
essages
can
be
sent.
Wh
il
e
in
the
Re
spo
ns
e
la
ye
r,
an
Ac
knowle
dgem
ent
m
essage
can
be
pi
ggybac
ke
d.
T
he
reli
abili
ty
of
CoA
P
m
essages
are
def
i
ned
as
Co
nf
irm
ab
le
(CO
N)
a
nd
N
on
-
co
nfi
rm
a
ble
(
NON
).
C
ON
is
us
e
d
t
o
sen
d
reli
able
m
essa
ges,
with
def
a
ult
tim
eou
ts
and
e
xpone
ntial
back
-
offs
between
retra
ns
m
issi
on
s.
The
re
cei
vi
ng
serv
e
r
se
nds
a
n
Ackn
ow
le
dge
m
ent
m
essage
(
ACK
)
us
in
g
the
sam
e
m
essage
ID
as
t
he
cl
ie
nt'
s
req
ue
s
t.
The
m
essage
se
nd
/
recei
ve
dia
gr
a
m
is
sh
own
i
n
Fig
ur
e
3
(a
).
Me
ssages
w
hose
reli
abili
ty
is
of
l
ow
im
po
rtanc
e
(unr
el
ia
ble
)
c
an
be
sent
a
s
N
on
-
c
onfir
m
able,
in
w
hi
ch
case
the
ser
ver
will
no
t
res
pond
with
a
n
ackno
wled
gem
ent
m
essage
bu
t
will
reco
r
d
the
m
essage
ID
to
pr
eve
nt
duplica
ti
on.
In
the
e
xam
ple
in
Figure
3
(
b),
t
he
m
essage
ID
is
[0x
7d34]
.
I
f
the
se
rv
e
r
rec
ei
ves
a
non
-
c
onfirm
able
m
es
sage
but
is
un
able
to
process
it
, it w
ou
l
d respo
nd
with a Reset
m
essage
(RST
) [
4].
Figure
3
.
(a
)
R
el
ia
ble m
essage tran
sm
issi
on
,
(
b)
U
nr
el
ia
ble
m
essage tra
nsm
issi
on
2.2.
U
nicas
t g
roup co
mmun
ic
at
io
n in I
oT
Group
c
omm
u
nicat
ion
is
ess
entia
l
to
m
any
Io
T
a
ppli
cat
i
on
s
,
pa
rtic
ularl
y
in
con
st
rain
ed
net
works
su
c
h
as
l
ow
-
power
an
d
l
os
sy
netw
orks
(LL
N)
.
Ty
pical
ly
,
in v
ery
la
rg
e
ne
tworks
,
c
onstr
ai
ned
d
e
vices n
eed
t
o
be
gro
up
e
d
a
nd
c
on
t
ro
ll
ed
usi
ng
gro
up
c
omm
and
s
t
o
sim
plify
m
anag
em
e
nt
an
d
a
void
c
ongestio
n
[
6
]
.
Durin
g
un
ic
ast
com
m
un
ic
at
io
n,
on
e
sen
der
no
de
se
nd
s
m
essages
to
m
ulti
ple
receivers,
a
nd
m
us
t
wait
fo
r
an
ACK
m
essage
from
each
of
the
re
cei
vin
g
nodes
.
This
can
le
ad
to
congesti
on,
especial
ly
wh
en
the
receive
r
an
d
sen
der
dif
fer i
n t
heir
m
essage
sen
ding a
nd
re
cei
vin
g ca
pab
il
it
y.
Ish
a
q
et
al
[
5
]
su
gg
e
ste
d
a
un
ic
ast
gro
up
com
m
un
ic
at
i
on
m
et
ho
d
with
the
aim
of
segm
enting
m
anag
em
ent
task
s
to
al
lo
w
ea
sie
r
acce
ss
to
r
eso
ur
ces
.
A
gro
up
of
no
des
bein
g
m
anag
ed
is
cal
le
d
an
E
ntit
y.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
3
,
June
201
9
:
1997
-
2005
2000
The
gro
up
of
resou
rces
belo
ng
i
ng
to
an
E
ntit
y
is
cal
le
d
Entit
y
Mem
ber
s.
Re
s
ource
usa
ge
or
m
anage
m
ent
com
m
and
s
ca
n
be
se
nt
to
an
Entit
y
us
ing
a
sing
le
Co
AP
r
equ
e
st.
A
n
E
nt
it
y
Ma
nag
er
(
EM)
is
the
coll
ect
ion
of
m
anag
em
ent
too
ls
for
eac
h
Entit
y.
An
E
M
is
able
to
m
anag
e
a
n
E
ntit
y
and
it
s
res
ourc
es
on
a
Co
A
P
ser
ver
in LL
Ns. Cl
ie
nts in
the
Inter
ne
t can se
nd c
om
m
and
s to
t
he EM
to
c
on
t
ro
l
Entit
ie
s.
Hou
et
al
[
8
]
su
ggest
e
d
in t
he
ir p
a
per
t
he use
o
f
a
r
es
ource
-
or
ie
nted protoc
ol call
ed
Seal
H
tt
p
to so
l
ve
the
issue
with
un
ic
ast
-
base
d
gro
up
c
omm
un
ic
at
ion
.
Its
pr
oc
ess
util
iz
es
C
OMBIN
E
an
d
BR
AN
CH
i
ns
te
ad
of
the
EM,
a
nd
al
lows
nodes
the
abili
ty
to
sel
f
-
j
oi
n
a
nd
l
eave
groups
.
Com
par
ison
s
wer
e
m
ade
be
tween
Seal
Http
an
d
RESTf
ul
proto
cols,
an
d
URI
perform
ance
was
i
m
pr
ov
e
d
by
add
in
g
sp
at
iotem
po
ral
at
tribu
te
s
i
n
the
sta
nd
a
rd
U
RI
for
dynam
i
c
gr
ou
p
re
qu
e
sts
of
physi
cal
resour
ces
.
Th
e
researc
h
fin
ds
that
Seal
Http
has
bette
r
a
ver
a
ge e
nergy c
on
s
umpti
on du
rin
g group c
omm
un
ic
at
ion
in
the
w
e
b of t
hings
(
W
oT) tha
n
C
oAP
.
A
pap
e
r
by
Q
ua
kasse
et
al
[
9
]
proposes
a
w
ay
to
so
l
ve
ne
t
work
co
ngest
ion
in
C
oAP
by
i
m
pr
ovin
g
the
delay
and
adap
t
the
be
ha
vior
of
the
s
olu
ti
on
to
netw
ork
co
ndit
ion
s
.
The
pa
per
al
so
su
ggest
s
the
us
e
of
delay
s
betwee
n
unic
ast
requ
est
s
dep
e
nd
i
ng
on
the
li
nk
de
la
y
and
the
est
i
m
at
ed
gr
oup
s
iz
e.
The
res
ults
sho
w
i
m
pr
oved
p
e
rfor
m
ance in
te
r
m
s o
f
res
pons
e
tim
e and
pac
ke
t l
os
s.
2.3.
Le
ast s
qu
ares
ap
p
ro
xi
mat
i
on
An
ass
um
ption
m
ade
in
this
r
esearch
is
that
there
m
ay
be
disco
ntin
uiti
es
in
the
DCR
values
at
the
sp
eci
fic
ra
nge
that
we
f
ocu
s
on,
an
d
that
th
ere
co
uld
be
da
ta
m
easur
em
e
nt
errors
duri
ng
ex
per
im
ents
[1
0
].
This
resea
rc
h,
therefo
re,
us
es
an
ap
pro
xim
ation
functi
on
to
best
co
ns
tr
uct
a
gr
a
phic
al
li
ne
of
t
he
DCR
values
in r
el
at
io
n
to
th
e size
of the
c
om
m
un
ic
at
ion
gro
up, as
m
entio
ne
d
i
n
Sect
i
on
1.
The
le
ast
squar
es
m
et
ho
d
pro
du
ce
s
the
best
functi
on
f
or
da
ta
app
r
oxim
ati
on
beca
us
e
it
aver
a
ges
ou
t
data
er
rors
t
o
a
m
ini
m
u
m
.
When
the
f
unct
ion
is
draw
n
out
a
s
a
gra
ph,
the
li
ne
will
not
pa
ss
thr
ough
ea
c
h
data
po
i
nt,
but
will
traver
se
th
rough
the
vicinit
y
cl
os
e
to
the
m
.
The
funct
i
on
pr
oduces
a
s
m
oo
th
li
ne
that
appr
ox
im
at
es
r
eal
values.
T
he
resu
lt
in
g
f
unct
ion
is
co
ncise
and
is
in
depend
e
nt
of
data
siz
e.
Diff
e
re
ntial
an
d
integrati
on
cal
culat
ion
s
ca
n
e
asi
ly
be
app
li
e
d.
T
he
re
su
lt
in
g
f
unct
ion
s
a
re
po
ly
nom
ia
l,
a
nd
it
s
c
orrectn
ess
is
dep
e
ndent
on t
he deg
ree
of poly
no
m
i
al
s [
1
1
].
The
le
ast
s
qu
a
r
es
m
e
tho
d
of
f
un
ct
io
n
a
ppr
oxi
m
ation
ca
n
be
app
li
e
d
in
m
any
ways.
For
our
pur
poses
we
use
one
ty
pe
of
le
ast
s
quares
cal
le
d
the
discrete
le
ast
sq
ua
res
m
et
hod.
T
he
in
put
da
ta
is
non
-
c
onti
nuous,
and
t
he
a
ppr
oxi
m
ation
is
in
t
he
f
or
m
of
t
he
po
ly
no
m
ia
l
n
p
with
the
highe
st
degree
of
n
that
pro
du
ce
s
m
ini
m
u
m
m
ea
n
s
quare
d
e
rror, sho
wn in
E
qu
at
ion
(
1
)
.
2
2
0
1
1
00
NN
n
i
i
i
i
i
n
i
ii
w
f
x
p
x
w
f
x
a
a
x
a
x
(1)
w
he
n
01
n
x
,
x
,
.
.
.
,
x
are
no
t
con
ti
nu
ou
s
,
and
01
n
w
,
w
,
.
.
.
,
w
re
pr
ese
nt
weig
hted
dat
a,
if
Nn
,
the
er
r
or
ca
n
be
el
i
m
inate
d
us
i
ng
a
poly
no
m
i
al
fo
rm
ula.
I
n
this
case,
the
num
ber
of
disc
onti
nuous
data
is
m
or
e
im
po
rtant
tha
n
the d
e
greee
of
appr
ox
im
at
ion
. F
or c
onti
nuou
s d
at
a,
E
quat
io
n
(
1
)
ca
n be e
xpan
de
d
as
foll
ows
[1
2
]:
1
01
0
0
0
0
N
N
N
N
j
j
j
n
j
i
i
i
i
n
i
i
i
i
i
i
i
i
i
a
w
x
a
w
x
a
w
x
w
x
f
x
(2)
wh
e
n j
=
0,1,
...
,
n o
r
wh
e
n j
=
0
0
1
0
01
0
0
0
0
N
N
N
N
n
i
i
i
i
n
i
i
i
i
i
i
i
i
i
a
w
x
a
w
x
a
w
x
w
x
f
x
(3)
wh
e
n j
=
1
1
2
1
1
01
0
0
0
0
N
N
N
N
n
i
i
i
i
n
i
i
i
i
i
i
i
i
i
a
w
x
a
w
x
a
w
x
w
x
f
x
(4)
wh
e
n j
=
n
12
01
0
0
0
0
N
N
N
N
n
n
n
n
i
i
i
i
n
i
i
i
i
i
i
i
i
i
a
w
x
a
w
x
a
w
x
w
x
f
x
(5)
This
ca
n be e
xpress
ed
as
po
ly
no
m
ia
l E
qu
at
io
n
(
6
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
An a
pp
r
oxim
at
ion
delay
b
et
w
een c
onsecuti
v
e req
uests
f
or
c
ongestio
n
c
ontrol in… (
Chanwit S
u
wann
apon
g
)
2001
01
n
n
f
x
a
a
x
a
x
(6)
Eq
uation
(
5
)
c
an be e
xpresse
d
as
a m
at
rix.
X
a
f
(7)
Ma
tric
es X
a
nd
f,
when
all
wei
gh
te
d data
ar
e eq
ual
1
i
w
,
can
be ex
pr
es
sed
as
0
1
2
0
0
1
1
1
1
1
1
1
2
3
1
1
1
1
1
1
1
2
1
2
2
1
1
1
1
1
m
m
m
m
m
n
i
i
i
i
i
i
i
i
i
i
i
m
m
m
m
m
n
i
i
i
i
i
i
i
i
i
i
i
m
m
m
m
m
n
n
n
n
n
n
i
i
i
i
i
i
i
i
i
i
i
x
x
x
x
x
f
x
a
a
x
x
x
x
x
f
x
a
a
x
x
x
x
x
f
x
2
4
b
ac
(8)
w
he
n
t
his
is
expresse
d
as
an
e
qu
at
i
on,
we
get
the
li
ne
ar
f
unct
io
n
01
i
y
a
a
x
,
the
quad
rati
c
f
un
ct
io
n
2
0
1
2
ii
y
a
a
x
a
x
,
the
cu
bi
c
functi
on
23
0
1
2
3
i
i
i
y
a
a
x
a
x
a
x
,
t
he
qu
a
rtic
f
unc
ti
on
2
3
4
0
1
2
3
4
i
i
i
i
y
a
a
x
a
x
a
x
a
x
,
an
d
so
on.
Be
cause
the
data
in
this
researc
h
is
ob
ta
ine
d
f
ro
m
exp
e
rim
entat
io
n,
the
le
ast
s
quares
m
et
ho
d
i
s
bette
r
s
uited
to
the
ta
sk
t
ha
n
inter
pola
ti
on
because
ex
pe
r
i
m
ent
data
m
a
y
con
t
ai
n
erro
rs
cau
s
ed
by
the
m
easur
em
ent
proc
ess.
Th
e
res
ul
ti
ng
gra
ph
li
ne
need
not
pass
ever
y
data point,
bu
t
sh
oul
d be a
res
ult o
f
the
least
total
error.
3.
RESU
LT
S
A
ND AN
ALYSIS
This
resea
rc
h
pro
po
ses
a
n
al
te
rn
at
ive
f
orm
of
un
ic
ast
w
he
re
the
L
oWP
AN
bor
der
rou
te
r
m
anag
es
cl
ie
nt
req
uest
m
essages,
as
sh
ow
n
in
F
ig
ure
4,
an
d
us
es
a
n
a
si
m
ple
DCR
-
base
d
so
l
ution
based
on
le
isure
as
def
i
ned
i
n
RF
C725
2
[
13
]
to
le
ssen
c
onges
ti
on
.
T
he
m
eth
od
set
s
th
e
li
m
it
rate
at
the
bor
der
router
whe
n
un
ic
ast
m
essa
ges
are
co
ntin
uous
ly
sent
to
serv
er
s
without
wait
ing
f
or
respon
se
m
es
sages,
th
us
av
oidi
ng
congesti
on.
Figure
4
.
Re
qu
est
m
essage
m
anag
em
ent throug
h
a
Lo
WPAN
borde
r
r
ou
te
r
The
best
DCR
valu
e(s
)
for
e
ach
se
r
ver
gro
up
siz
e
a
re
obta
ined
from
app
r
oxim
a
ti
on
ba
sed
on
t
he
fo
ll
owin
g
proc
ess:
1)
Fin
ding
DCR
based
on
le
isure.
T
he
resu
lt
s
are
the
DCR
values
a
nd
a
ver
a
ge
res
pons
e
tim
e
in
diff
e
re
nt
gro
up
siz
es.
2)
T
he
values
are
the
n
put
into
a
le
ast
squ
ares
ap
pro
xim
at
ion
f
unct
ion.
3)
T
he
relat
ive
m
ini
m
um
s
that
pr
oduc
e
the
le
ast
aver
age
r
esp
onse
tim
es
in
each
gro
up
siz
e
are
ob
ta
ine
d.
T
he
detai
ls
of each
step a
r
e ex
plaine
d
in
Sect
ion
s
3.1. t
o 3.3.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
3
,
June
201
9
:
1997
-
2005
2002
3.1.
D
C
R
c
alculat
i
on
ba
se
d
on
Lei
sure
We
can
fin
d
th
e
values
of
DC
R
based
on
le
isure
[8
]
.
Give
n
the
ser
ver
gro
up
siz
e
(G),
t
he
ta
rg
et
data
trans
fer
rate (R
),
a
nd the
esti
m
at
ed
respo
nse
size (S
), t
he DCR
can
b
e
ca
lc
ulate
d
us
in
g
t
he
E
quat
ion (
9)
:
11
lo
w
e
r
b
o
u
n
d
L
e
is
u
r
e
SG
D
C
R
G
R
(
G
)
(
9
)
In
our
ex
per
im
ents,
t
he
val
ue
of
G
was
betw
een
3
a
nd
23,
S
was
ap
pro
xi
m
at
ely
80
byte
s,
an
d
R
was
set
to
a
co
ns
e
r
vative
8kbits/
s
(1
kB/s)
.
T
he
lowe
r
bound
for
the
le
is
ur
e
was
betwee
n
120
m
s
and
83
m
s.
Using
E
quat
io
n
(
9
)
,
we
see
a
decr
ease
in
D
CR
values
sta
r
ti
ng
from
12
0
m
s.
W
e
the
refor
e
us
ed
a
DC
R
ran
ge
betwee
n
0
-
12
0
and
m
easur
e
d
the
aver
a
ge
r
esp
on
se
ti
m
es
for
dif
fer
e
nt
gro
up
siz
es
(G
=
3,
5,
10,
15
,
20,
and 25
node
s) by r
unni
ng sim
ulati
on
s
u
si
ng the C
ooj
a sim
ulator. Fig
ur
e
5 s
hows
t
he res
ults in C
ooj
a.
Figure
5
.
Re
spon
s
e ti
m
es f
or
diff
e
re
nt grou
p si
zes as a
fun
c
ti
on
of the
dela
y betwee
n
c
onsecuti
ve req
ues
ts,
evaluate
d usin
g
C
ooj
a sim
ulator
3.2.
D
C
R app
rox
im
at
i
on
s
To
est
im
at
e
th
e
value
of
the
op
ti
m
u
m
DCR,
we
us
e
t
he
data
f
ro
m
Sect
ion
3.1
to
pl
ot
a
gr
a
ph
us
in
g
the
le
ast
sq
uares
m
et
ho
d.
T
he
gr
ap
h
is
the
best
m
at
he
m
at
i
cal
rep
rese
ntati
on
of
the
relat
ion
s
hi
p
betwee
n
DCR
and ave
ra
ge
r
e
sp
onse
tim
es wi
th m
ini
m
iz
ed
eff
ect
s
from
d
at
a error, as
sho
wn in Fi
gure
6
.
3.3.
Rel
at
i
ve m
ini
mum for
each gr
ou
p
siz
e
The
relat
ive
m
ini
m
u
m
that
p
rod
uces
the
le
ast
aver
age
res
pons
e
ti
m
e
in
each
gro
up
is
determ
ined
from
the
gr
ap
h
of
DCR
va
lue
s
from
Sect
ion
3.2.
G
r
oups
c
om
pr
isi
ng
3
a
nd
5
ser
ve
rs
yi
el
d
a
DCR
of
0.02
s
.
The
value
inc
r
eases
f
or
gr
oup
siz
es
10,
15,
20,
a
nd
25
to
0.0
3
s
,
0.04
s,
0.0
9
s
an
d
0.10
s,
res
pecti
ve
ly
,
as
sh
ow
n
in
Fi
gur
e 7
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
An a
pp
r
oxim
at
ion
delay
b
et
w
een c
onsecuti
v
e req
uests
f
or
c
ongestio
n
c
ontrol in… (
Chanwit S
u
wann
apon
g
)
2003
Figure
6
.
DCR
appr
oxim
a
ti
on
for
eac
h g
rou
p si
ze
Figure
7
.
Re
la
ti
ve
m
ini
m
u
m
f
or each
grou
p
s
iz
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
3
,
June
201
9
:
1997
-
2005
2004
Figure
6
an
d
Figure
7
s
how
the
res
ults
of
tho
se
e
xperim
ents.
I
n
e
ver
y
exp
e
rim
ent,
the
Lo
WPA
N
bor
der
r
oute
r
s
ends
co
ns
ec
utive
re
qu
est
m
es
sages
to
se
rv
e
r
nodes
at
the
sa
m
e
rate.
Each
exp
e
rim
ent
yi
el
ded
a
si
m
il
ar
patte
rn
for
each
gro
up
siz
e.
The
init
ia
l
par
t
of
t
he
grap
hs
s
how
a
hi
gh
a
ver
a
ge
res
pons
e
ti
m
e
bec
aus
e
the
DCR
valu
e
was
low,
re
su
lt
ing
in
requ
est
m
essages
bein
g
sent
to
o
rap
idly
an
d
c
ausin
g
co
ngest
ion
as
respo
ns
e
m
ess
age
pac
kets
f
r
om
m
ul
ti
ple
serv
e
r
no
des
overwhelm
the
bo
r
der
r
ou
te
r
.
B
ut
as
the
D
CR
value
increases
,
the
m
essage
sen
d
r
at
e
beco
m
es
m
or
e
s
uited
to
th
e
networ
k'
s
capaci
ty
,
resu
lt
in
g
in
lo
wer
a
nd
lowe
r
aver
a
ge
res
ponse
tim
es
un
ti
l
a
m
ini
m
u
m
aver
age
res
pons
e
tim
e
is
reache
d.
Af
te
r
this
poi
nt
the
res
pons
e
tim
e
sta
rts to
i
ncr
ea
se agai
n
as t
h
e
DCR i
nc
reases
due to
the
hi
gher
d
el
ay
inse
r
te
d
into
the
border
rou
te
r.
3.4.
D
C
R
es
timat
i
on
Af
te
r
t
he
relat
ive
m
ini
m
u
m
is
found
in
S
ect
ion
3.3
,
the
y
are
v
erifie
d
throu
gh
e
xpe
rim
entation
.
In
orde
r
t
o
e
valuate
t
he
pe
rfor
m
ance
of
our
pro
posed
so
luti
on,
we
us
e
sim
ulati
ons.
T
o
determ
i
ne
the
perform
ance
t
he
of
the
DC
R
inserted
bet
ween
c
on
s
ecu
ti
ve
un
ic
ast
r
equ
e
sts
for
group
com
m
un
ic
at
ion
betwee
n
a
sin
gl
e g
at
eway a
nd
m
ult
iple ser
ve
rs,
we use t
he
Cooja
netw
ork
sim
ulator.
The
fo
ll
owin
g
ind
ic
at
ors
a
re
obser
ve
d
f
rom
the
si
m
ulatio
ns
:
t
he
a
ver
a
ge
re
spo
ns
e
ti
m
e,
the
tim
e
ta
ken
by
ser
vers
to
res
pond
to
un
ic
ast
c
omm
un
ic
at
io
n,
t
he
hop
c
ount
(t
he
nu
m
ber
of
rout
ers
each
pac
ket
go
e
s
thr
ough
to
rea
ch
it
s
de
sti
n
at
ion),
a
nd
the
pa
cket
loss
rati
o.
The
set
ti
ng
s
f
or
t
hese
e
xperi
m
ents
are
detai
le
d
in
Table
1.
Table
1.
E
val
ua
ti
on
E
xperim
e
nts S
et
ti
ngs
No
d
e T
y
p
e
Z1
m
o
te
Co
n
tik
i versio
n
Media Ac
cess
Co
n
trol (M
AC)
Co
n
tik
i 3.0
Carrie
r
Sen
se
Mult
ip
le Acc
ess
(CSM
A)
Rad
io
Frequ
en
cy
(
RF)
IE
E
E
8
0
2
.15
.4 ch
an
n
el 15
Ro
u
tin
g
Proto
co
l:
RPL
Co
AP
Grou
p
size:
3
no
d
es
5
no
d
es
1
0
no
d
es
1
5
no
d
es
2
0
no
d
es
2
5
no
d
es
DCR
:
0
.02
s
0
.02
s
0
.03
s
0
.04
s
0
.09
s
0
.10
s
The
e
xp
e
rim
e
nts
in
vo
l
ve
se
nd
i
ng
re
qu
est
m
essages
f
ro
m
a
Lo
WPAN
bor
der
r
ou
te
r
t
o
a
group
of
serv
e
r
nodes
wh
il
e
util
iz
ing
the
DCR
va
lu
es
obta
ined
i
n
Sect
ion
3.2
to
con
t
ro
l
the
m
essage
rate.
F
or
each
gro
up
siz
e
,
a
total
of
50
sim
ulati
on
s
a
re
ca
rr
ie
d
out.
T
he
resu
lt
s
c
onfirm
the
ef
fecti
veness
of
ou
r
pro
pos
e
d
so
luti
on,
as
s
how
n
i
n
Ta
ble
2.
Table
2.
Su
m
m
ary o
f
the
Resu
lt
s o
f
t
he
E
xper
i
m
ents
Grou
p
Size
DCR
Ho
p
cou
n
t (
h
)
Res
p
o
n
se ti
m
e
(
s)
Pack
et los
s (%)
Min
Av
erage
Max
Min
Av
erage
Max
Min
Av
erage
Max
3
0
.02
1
1
.00
1
0
.17
0
.17
0
.17
0
0
0
5
0
.02
1
1
.00
1
0
.26
0
.26
0
.26
0
0
0
10
0
.03
1
1
.20
2
0
.78
0
.79
0
.79
0
0
0
15
0
.04
1
1
.73
3
1
.57
1
.58
1
.59
0
0
0
20
0
.09
1
1
.65
3
2
.64
2
.67
2
.69
0
.2
0
.3
0
.4
25
0
.10
1
1
.80
3
3
.90
3
.91
3
.93
0
.7
0
.8
1
.0
In
te
rm
s
of
res
pons
e
ti
m
e,
th
e
resu
lt
s
sho
w
that
for
sm
alle
r
gro
up
siz
es
(
3
an
d
5),
a
DC
R
of
0.0
2
s
pro
du
ces
an
a
ve
rag
e
res
pons
e
tim
e
of
0.1
7
s
and
0.2
6
s,
res
pecti
vely
.
T
he
aver
a
ge
respo
nse
tim
e
increases
as
the num
ber
of
neig
hbors inc
r
eases, a
res
ult of
a
higher n
um
ber
o
f
c
olli
sion
s
in
a s
har
e
d m
ediu
m
.
Re
gardin
g
pac
ket
loss
,
a
n
im
portant
i
nd
ic
at
or,
gr
oups
c
on
ta
ining
le
ss
th
an
20
node
s
e
xp
e
rience
an
aver
a
ge
pac
ket
loss
of
0%
,
or
no
l
os
s.
I
n
gro
ups
la
r
ger
th
an
20
no
des,
t
he
per
ce
ntage
sta
rts
to
inc
rea
se
to
0.3%,
an
d
kee
ps
i
ncr
easi
ng
as
the
gro
up
s
iz
e
gro
ws.
Thi
s
is
du
e
to
the
highe
r
node
de
ns
it
y
causi
ng
m
or
e
colli
sion
s
betw
een
gro
up
m
e
m
ber
s.
Anothe
r
possible
f
act
or
is
that
in
la
r
ger
gro
up
s
,
the
re
is
a
higher
a
ver
a
ge
hop
co
unt.
Co
m
m
un
ic
at
ion
be
tween
the
L
oWP
A
N
bord
e
r
router
an
d
the
serv
e
r
re
qu
i
res
m
or
e
hops
,
re
su
lt
ing
in m
or
e dro
ppe
d
m
essages
befor
e
they e
ven
t
ually
r
each
the
ir d
est
inati
on.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
An a
pp
r
oxim
at
ion
delay
b
et
w
een c
onsecuti
v
e req
uests
f
or
c
ongestio
n
c
ontrol in… (
Chanwit S
u
wann
apon
g
)
2005
4.
C
O
NC
L
US
I
O
N
This
rese
arc
h
pro
po
ses
a
m
e
ans
to
a
void
netw
ork
c
onge
sti
on
in
l
ow
powe
r
a
nd
l
os
s
y
networ
ks
durin
g
unic
ast
gro
up
c
ommun
ic
at
io
n
by
i
ntr
oducin
g
de
la
ys
between
consecuti
ve
re
qu
e
sts
to
lim
i
t
the
m
essage
sen
d
rate.
It
al
so
pr
esents
a
way
to
obta
in
the
pr
op
e
r
DCR
val
ue
f
or
eac
h
ne
twork
group
si
ze
by
us
in
g
t
he
def
a
ult
le
isure
pe
ri
od
in
the
re
quest
/resp
on
se
la
ye
r
of
t
he
C
oAP
prot
oco
l,
a
nd
ap
plyi
ng
th
e
le
ast
sq
ua
res
appr
ox
i
m
ation
m
et
hod
to
f
i
nd the
re
la
ti
ve
m
ini
m
um
D
CR
.
To
e
valuate
t
o
the
vali
dity
an
d
perform
ance
of
this
m
et
hod,
t
he
C
ooj
a
sim
ula
tor
is
us
e
d
to
sim
ulate
diff
e
re
nt
DCR
values
in
net
works
unde
r
va
rio
us
co
nd
it
io
ns
as
def
i
ned
by
each
ex
per
i
m
ent.
The
res
ults
are
analy
zed
a
nd
evaluate
d.
E
xperim
ents
confirm
that
the
m
et
ho
d
pro
posed
in
t
his
re
s
earch
is
e
ff
ect
ive
f
or
avo
i
ding
net
w
ork
co
ngest
io
n
caused
by
uni
cast
group
c
om
m
un
ic
at
ion
in
Co
AP
,
with
al
l
m
anag
e
m
e
nt
bein
g
virtu
al
ly
on
th
e
send
e
r
side
.
I
t
is
able
to
eff
e
ct
ively
red
uce
the
com
m
un
ic
at
ion
w
orkl
oad
on
t
he
ser
ver
s
ide
of
the Co
AP p
ro
t
oco
l.
I
t i
s
su
it
a
ble for
gr
oup
c
omm
un
ic
at
ion
an
d p
r
ov
i
des
e
xcell
ent scal
ab
il
ity.
REFERE
NCE
S
[1]
N.
L
.
Moham
ed
,
et
al
.,
“
Perform
anc
e
o
f
Non
-
Uniform
Dut
y
-
C
y
c
le
dCont
iki
MA
C
in
W
ireless
S
ensor
Networks,
”
Inte
rnational
Jo
urnal
of El
e
ct
ri
c
al
and
Comput
er
Engi
n
ee
ring
(
IJE
CE
)
,
v
ol
/i
ss
ue:
7
(
2
)
,
pp
.
942
-
949,
2016
.
[2]
S.
Navulur
,
et
al
.,
“
Agric
ult
ura
l
Mana
gement
t
hrough
W
ire
le
ss
Sensors
and
Inte
rne
t
of
Thi
ngs
,
”
Inte
rnationa
l
Journal
of
Elec
t
rical
and
Computer
Eng
ine
ering
(
IJE
CE
)
,
v
ol
/i
ss
ue:
7
(6)
,
pp
.
349
2
-
3499
,
201
7.
[3]
M.
Kovatsc
h,
et
al
.,
“
A
Low
-
Po
wer
CoAP
for
C
onti
ki
,
”
IEEE
Eighth
Inte
rnation
al
Confe
renc
e
on
Mobil
e
Ad
-
Ho
c
and
Sensor Sy
st
e
ms
,
Vale
n
ci
a
,
sp
ai
n
,
pp.
855
-
860,
2011.
[4]
Inte
rne
t
Eng
ineeri
ng
T
ask
Force
(I
ET
F),
“
Group
Com
m
unic
at
ion
for
the
C
onstrai
ned
Appl
ic
a
ti
on
Protoco
l
(CoAP
),
”
2014
.
Ret
ri
eve
d
f
rom
: https:/
/
to
ols.iet
f
.
org/ht
m
l/
rfc
739
0
.
[5]
I.
Ishaq
,
e
t
al
.
,
“
Flexi
ble
Unic
a
st
-
Based
Group
Com
m
unic
at
ion
for
CoAP
-
Ena
ble
d
Devices,”
S
ensors
,
v
ol
/i
ss
ue:
14(6),
pp
.
9833
–
9877,
2014
.
[6]
M.
Ti
loca,
et
a
l
.
,
“
Axiom
:
DTLS
-
base
d
sec
ure
IoT
group
comm
unic
at
ion,
”
ACM
Tr
ansacti
ons
on
Embe
dd
ed
Computing
Syst
e
ms
,
vol
/i
ss
ue:
16
(3),
pp
.
1
-
29
,
20
17.
[7]
I.
Ishaq
,
e
t
al
.
,
“
Expe
riment
al
ev
al
ua
ti
on
of
unicast
and
m
ultic
ast
co
ap
group
co
m
m
unic
at
ion,
”
S
ensors
,
vol
/i
ss
ue
:
16(7),
2016
.
[8]
C.
Hou,
et
al
.,
“
SeaHtt
p:
A
res
ourc
e
-
ori
ented
protoc
ol
to
exten
d
REST
sty
le
f
or
web
of
thi
ng
s,”
Journal
of
Computer
Scien
ce
and
Techno
lo
gy
,
vo
l
/i
ss
ue:
29
(2),
pp
.
205
-
215
,
2014.
[9]
F.
Ouaka
ss
e
a
nd
S.
Rakr
ak,
“
An
Adapti
v
e
Soluti
on
for
Congesti
on
Control
in
Co
AP
-
base
d
Group
Com
m
unic
at
ions,”
Inte
rnat
ional
Journal
of
Adv
a
nce
d
Computer
Sci
en
ce
and
App
li
cations
,
vol
/
issue:
8(6),
pp
.
234
-
239,
2017
.
[10]
C.
Fan
and
C.
Liu,
“
A
novel
al
gorit
hm
for
ci
rcle
c
urve
fit
t
ing
base
d
on
the
le
ast
sq
uar
e
m
et
hod
b
y
t
he
point
s
of
the
Newton’s
rings,
”
Inte
rnationa
l
Confe
renc
e
on
Computers,
Comm
unic
a
ti
ons
and
Syste
ms
,
G
y
ongsan,
Korea
,
pp.
256
–
260,
2015
.
[11]
I.
K.
Ba
y
er
and
S.
Soni
“
Le
ast
Square
Approxim
at
ion
Techni
q
ue
for
Ene
rg
y
Conserva
ti
on
in
W
ire
le
ss
Senso
r
Networks
,”
Int
ernati
onal
Con
fe
renc
e
on
Control,
Instrum
ent
ati
on
,
Comm
unic
ati
on
an
d
Computati
on
al
Technol
ogi
es
(
ICCICCT
)
,
Ta
m
il
nadu,
Ind
ia,
pp
.
886
-
889
,
2014.
[12]
J.
L
.
Buch
ana
n
a
nd
P.
R
.
Turn
er,
“
Nu
m
eri
ca
l
Met
hod
and
An
aly
si
s
,
”
New York
,
McGraw
-
Hill
,
1
992.
[13]
Inte
rne
t
Engi
n
e
e
ring
Ta
sk
Force
(IE
TF),
“
The
Constrai
ned
Appli
ca
t
ion
Protocol
(CoAP
),
”
2014
.
Ret
ri
eve
d
from
:
htt
ps://
tool
s.i
et
f
.
org/ht
m
l/
rfc
725
2
.
BIOGR
AP
HI
ES OF
A
UTH
ORS
Ch
anw
it
Su
w
a
nn
apo
ng
rec
ei
v
ed
a
B.
Eng.
and
M.E
ng.
degr
e
e
in
computer
engi
nee
r
ing
from
Khon
Kae
n
Un
ive
rsit
y
,
Tha
i
land
in
2011
an
d
2013
respe
ct
i
vely
.
He
has
publi
shed
m
an
y
publi
c
at
ions
in
t
he
ar
ea of
W
ir
eless
Sensor Net
work,
Ad Hoc
Net
works
and
Sm
art Agric
ul
ture.
Ch
atc
hai
Khu
n
boa
is
an
associa
t
e
profe
ss
or
in
the
Dep
art
m
ent
of
Com
pute
r
E
ngine
er
ing
a
t
Khon
Kae
n
Univer
sit
y
,
Tha
i
la
nd
.
Dr.
Khunboa
r
ec
e
ive
d
a
B
.
Eng
from
Khon
Kae
n
Univer
sit
y
in
1992,
his
MS
c
degr
ee
in
T
elec
om
m
unic
at
ions
from
Univer
sit
y
of
Pitt
sburgh
in
2000,
and
Ph
D
degr
ee
in
Inform
at
ion
T
ec
hno
log
y
from
Ge
orge
Mason Unive
rsity
in
2005
.
Hi
s res
ea
rch
in
te
r
ests
inc
lud
e
W
ireless Sensor Ne
twork
,
Int
ern
e
t
of
Things
and
Softwar
e
def
ine
d
Netwo
rks (SD
N).
Evaluation Warning : The document was created with Spire.PDF for Python.