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.
10
,
No.
4
,
A
ugus
t
2020
,
pp.
432
2
~
43
30
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
10
i
4
.
pp
4322
-
43
30
4322
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Novel ho
listic arc
hitectu
re for anal
ytical op
eration
on senso
ry data
rela
yed as cl
oud s
ervi
ce
s
Manujak
shi B
.
C
.
1
,
K
.
B
.
R
am
esh
2
1
Depa
rtment of
Com
pute
r
Scie
n
ce
and
Engi
ne
ering,
Presiden
c
y
Univer
sit
y
,
Indi
a
2
Depa
rtment of
El
e
ct
roni
cs
and
I
nstrum
ent
at
ion
Engi
ne
eri
ng,
RV
Coll
ege
o
f
Eng
i
nee
ring
,
Ind
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Oct
17
, 201
9
Re
vised
Feb
4
,
2020
Accepte
d
Fe
b
26
, 202
0
W
it
h
inc
rea
sing
adopt
ion
of
t
he
sensor
-
base
d
appl
icati
on
,
th
ere
is
an
expone
nt
ia
l
rise
of
the
sensor
y
d
at
a
that
ev
ent
u
all
y
ta
k
e
the
shap
e
of
t
he
b
ig
dat
a
.
How
eve
r
,
the
pr
ac
t
icalit
y
of
execut
ing
hi
gh
end
ana
l
y
t
i
c
al
op
era
t
ion
over
th
e
r
esourc
e
-
constr
ai
n
ed
bi
g
da
ta
h
as
n
eve
r
b
ei
ng
stud
ie
d
cl
ose
l
y
.
After
rev
ie
win
g
exi
sting
app
roa
che
s,
it
is
expl
ore
d
that
t
her
e
is
no
cost
-
eff
e
ct
iv
e
sc
hemes
of
big
d
at
a
an
aly
tics
ov
er
l
arg
e
sca
l
e
s
ensor
y
data
proc
essiing
th
at
ca
n
b
e
di
rectl
y
used
as
a
servi
ce
.
Therefore,
t
he
pro
psoed
s
y
stem
int
rodu
ce
s
a
holi
st
ic
arc
h
it
e
ct
ur
e
where
stre
ame
d
data
af
te
r
per
form
ing
ext
r
ac
t
ion
of
knowe
dge
ca
n
b
e
offe
red
in
the
fo
rm
of
servic
es.
Im
ple
m
ent
ed
in
MA
TL
AB,
th
e
proposed
stu
d
y
uses
a
v
er
y
sim
pli
st
i
c
appr
oac
h
consid
eri
ng
en
erg
y
co
nstrai
ned
of
the
sensor
nodes
to
find
that
proposed
s
y
stem
offe
rs
bet
te
r
accura
c
y
,
red
u
ce
d
m
ini
ng
dura
ti
on
(i.
e
.
faster
response
ti
m
e),
a
nd
red
uce
d
m
emor
y
d
epe
nden
cie
s
to
prove
th
at
i
t
offe
rs
cost
eff
ective a
n
aly
t
i
ca
l
soluti
on
in
c
ontra
st
to exi
stin
g
s
y
st
em.
Ke
yw
or
d
s
:
An
al
yt
ic
s
Bi
g
data
Energy
Re
so
urces
Sensors
Copyright
©
202
0
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
:
Ma
nuj
a
ks
hi B
. C
.
,
Dep
a
rtm
ent o
f C
om
pu
te
r
Scie
nce a
nd
E
ng
i
ne
erin
g,
Pr
esi
de
ncy
Un
i
ver
sit
y,
Be
ng
al
uru,
Ka
rn
at
a
ka,
India
.
Em
a
il
:
m
anu
j
a
ks
hi
bc
2014@
gm
ai
l.co
m
1.
INTROD
U
CTION
The
a
doptio
n
of
t
he
se
ns
or
in
co
ns
ist
ently
increasi
ng
i
n
m
any
com
m
e
rcial
as
well
as
dom
estic
app
li
cat
io
ns
in
pr
ese
nt
ti
m
es
[1
]
.
Ba
sic
al
ly
,
the
se
nsors
are
pro
gr
am
m
ed
t
o
c
ollec
t
sign
if
ic
antly
la
rg
e
quantit
y
of
a
n
en
vironm
ental
data
wh
ic
h
is
ac
cum
ulate
d
over
certai
n
sin
k
by
the
pro
cess
cal
le
d
as
data
aggre
gation
[
2]
.
From
the
te
chn
ic
al
sta
nd
a
rds
an
d
c
on
c
ept
of
sens
or
netw
ork,
it
is
belie
ve
d
that
e
ve
ry
s
ens
or
has
a
sp
eci
fic
al
locat
ion
of
pa
rtic
ular
am
ount
of
resou
rces
(es
pecial
ly
ener
gy)
w
hile
pe
rfor
m
ing
the
proce
s
s
of
data
transm
i
ssion
as
w
el
l
as
idle
sensing
[
3].
It
will
ob
vi
ou
sly
m
ean
that
a
senso
r
s
pontane
ously
degrad
es
it
s
energy
le
vel
in
eac
h
unit
of
durati
on of
th
ei
r
ope
rati
on. Th
ere
f
or
e,
w
he
n
se
nsors
are
de
plo
ye
d
in
m
as
sively
connecte
d
en
vi
ronm
ent,
the
y
accum
ulate
a
la
rg
e
r
stream
of
data
wh
ic
h
is
re
quire
d
t
o
be
pr
ocesse
d
to
o.
Usu
al
ly
,
a
n
al
gorithm
respo
ns
ible
t
o
do
th
at
is
em
bed
de
d
within
a
se
nsor
it
sel
f.
T
his
will
m
ean
that
m
or
e
the
data
to
be
processe
d,
the
sens
or
will
de
pl
et
e
m
or
e
am
ou
nt
of
e
nergy
b
y
them
sel
ves
and
eve
ntua
ll
y
it
will
sta
rt
ne
gativel
y
aff
ect
in
g
the
com
plete
netw
ork
t
oo.
T
his
i
s
the
case
ho
w
ene
rg
y
e
ff
ic
ie
ncy
is
c
onnect
ed
wit
h
the
analy
ti
cal
op
e
rati
on
over
the
big
data,
wh
ic
h
re
quires
i
m
m
ediat
e
at
t
ention
from
the
app
li
cat
io
n
vi
abili
ty
per
s
pecti
ve.
A
par
t
f
ro
m
this,
there
are
m
a
ny
oth
e
r
pro
bl
e
m
s
too
.
For
an
exam
ple,
w
hen
a
n
in
vestigat
ion
is
carrie
d
ou
t
t
ow
a
r
ds
big
dat
a
analy
ti
cal
aspect,
it
is
nece
ssary
that
t
he
da
ta
unde
r
c
on
si
der
at
io
n
s
houl
d
hav
e
al
l
the
inh
er
ent
pro
per
ti
es
of
it
that
can
be
us
e
d
as
evide
nce
that
inp
ut
data
is
of
bi
g
data.
Unfortu
natel
y,
extracti
ng
s
uc
h
f
or
m
of
li
ve
stream
of
dat
a
with
al
l
the
char
ect
e
risti
cs
of
t
he
bi
g
da
ta
is
no
t
feasible
f
ro
m
research
view
point,
w
hic
h
is
al
so
one
of
the
cor
e
im
ped
i
m
ents
towa
r
ds
a
su
ccessf
ul
res
earc
h
work
ove
r
se
ns
ory
big
dat
a
[
4].
A
nothe
r
essenti
al
pr
oble
m
associat
ed
with
the
res
earch
-
ba
sed
a
naly
sis
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
Novel
holi
sti
c
ar
c
hitec
ture fo
r
analyti
cal
op
er
atio
n on sen
s
or
y
data rel
aye
d
…
(
Ma
nujak
sh
i B.
C.
)
4323
of
the
big
data
is
that
al
l
the
big
data
is
require
d
to
be
sto
red
at
s
pecific
sect
or
ov
e
r
the
distrib
uted
st
orage
un
it
s
of
the
cl
oud
da
ta
centers
.
Unfortu
natel
y,
al
l
the
existi
ng
com
m
ercial
pr
act
ic
es
of
bi
g
data
m
anag
em
ents
cal
ls
fo
r
first
st
or
i
ng
t
he
ra
w
data
over
the
s
tora
ge
unit
s
an
d
the
n
a
naly
ti
cal
schem
es
are
ap
plied
ove
r
the
bi
g
data
in
or
der
t
o
obta
in
m
ined
inf
orm
ation
.
The
fi
nally
ob
ta
ined
m
ined
inform
at
ion
is
then
st
or
e
d
in
data
war
e
hous
e
[
5]
.
Su
c
h
c
hai
n
of
proce
ss
es
not
on
ly
captu
res
m
axim
u
m
infr
ast
r
uctu
res
bu
t
a
lso
is
com
pu
ta
ti
on
al
ly
expensi
ve
process
f
or
w
hich
reas
on
t
he
knowle
dge
extracte
d
a
re
just
st
or
e
d
a
nd
it
s
util
iz
at
ion
i
s
a
gain
ba
sed
on
the
sk
il
ls
of
the
us
ers.
At
pr
es
ent,
there
are
r
esearch
wor
k
be
ing
car
ried
ou
t
ov
e
r
big
data
ap
pro
aches
an
d
it
s
i
m
pr
ov
em
ent
[6
-
8],
but
in
re
al
it
y
they
are
m
uch
in
infancy
sta
ge
and
r
equ
i
res
m
or
e
t
i
m
e
in
order
to
ex
pec
t
the
antic
ipated
outc
om
e.
Ther
e
is
al
so
no
den
yi
ng
the
fa
ct
that
there
e
xisti
ng
good
num
ber
of
m
ining
/
anal
yt
ic
al
too
ls
to
carry
out
the
t
ask
of
m
ining
bu
t
unf
or
tu
na
te
ly
,
there
is
al
ways
certai
n
le
vel
of
ba
rr
ie
r
s
in
order
t
o
ex
pect
m
axi
m
iz
ed
ou
t
com
e
and
f
ull
-
fled
ge
accu
rac
y.
Althou
gh
va
rio
us
researc
h
paper
s
ad
vo
cat
es
t
he
us
a
ge
of
distri
bu
te
d
s
of
t
war
e
fr
am
ewo
r
k
Ha
doop
a
nd
Ma
pR
edu
ce
but
the
re
are
al
so
researc
he
r
who
has
ex
plici
tl
y
hig
hlig
hte
d
lim
i
ta
ti
on
of
it
wh
ic
h
is
ye
t
an
unso
l
ved
prob
le
m
t
il
l
date
[9
]
.
Ther
e
f
or
e, this
p
aper
pr
ese
nts
a n
ov
el
and un
iqu
e m
od
el
ing
of
an
al
yt
ic
al
ap
pr
oach
that c
on
si
ders sens
ory
d
at
a
as
an
in
pu
t
w
hi
ch
after
pr
oce
ssing
yi
el
ds
a
bette
r
f
orm
of
knowle
dge
tha
t
has
higher
ac
cur
acy
.
The
m
od
el
in
g
is
carrie
d
out
in
s
uch
a
w
ay
that
a
sen
s
or
with
ou
t
dis
sipati
ng
m
or
e
am
ou
nt
of
re
so
urces
sho
uld
offer
su
pp
or
ta
bili
ty
of
be
tt
er
for
m
of
big
dat
a
analy
ti
cal
o
per
at
io
n.
Op
ti
m
iz
at
ion
of
the
stora
ge
is
ano
the
r
sign
ific
a
nt
ta
r
get
of
t
he
pro
po
s
ed
syst
em
.
The
orga
nizat
ion
of
t
he
propose
d
pa
pe
r
is
as
f
ollows:
Se
ct
ion
2
discu
ss
es
ab
ou
t
syst
e
m
desig
n,
al
on
g
with
assum
ption
an
d
strat
egies
use
d
f
ollow
e
d
by
discuss
io
n
of
res
ult
analy
sis i
n
sect
ion
3. Finall
y, t
he
c
on
cl
us
ive
rem
ark
s a
re
pro
vid
e
d
in
sect
io
n
4.
This
sect
ion
discusse
s
ab
out
the
recent
work
bein
g
carried
ou
t
to
wards
de
velo
pi
ng
analy
ti
cal
app
li
cat
ii
on
s
usi
ng
big
data
con
ce
pt
as
an
extensi
on
t
o
our
pr
i
or
work
[10].
Most
rec
ently
,
the
disc
us
si
on
carried
out
by
Zo
u
et
al
.
[11]
hav
e
sta
te
d
i
m
po
rtance
of
big
data
appr
oa
ch
over
forest
ry
data.
Sim
i
lar
f
or
m
of
disc
us
sio
n
of
t
he
big
dat
a
analy
ti
cs
co
ns
ide
rin
g
a
ca
se
stu
dy
of
di
saste
r
m
anag
e
m
ent
was
ca
rri
ed
out
by
S
hah
et
al
.
[
12]
.
Co
ns
i
de
rati
on
of
t
he
case
st
ud
y
of
healt
hcar
e
s
ect
ion
was
se
en
in
the
w
ork
of
Had
i
et
al
.
[13]
wh
e
re
a
ne
twork
opti
m
izati
on
-
base
d
ap
proac
h
has
be
en
d
isc
us
sed
with
res
pect
to
li
near
pro
gr
am
m
ing
and
fair
ness
s
chem
e.
Issu
es
and
c
halle
nges
relat
ed
to
t
he
big
data
ag
gregati
on
co
nne
ct
ed
to
sens
or
y
a
pp
li
c
at
ion
hav
e
be
en
discusse
d
by
Bo
ub
ic
he
et
al
.
[
14
]
w
her
e
va
rio
us
strat
egies
have
be
e
n
pr
ese
nted
.
Th
e
work
o
f
Ca
o
e
t
al
[1
5]
ha
ve
pr
ese
nted
c
onne
ct
ion
bet
ween
energy
eff
ic
ie
ncy
an
d
big
da
ta
and
con
cl
ud
e
d
that
there
is
po
te
nt
ia
l
need
t
o
be
co
ns
ide
re
d
as
they
ar
e
ye
t
open
issu
es.
Jab
bar
et
al
.
[
16
]
have
pr
ese
nted
a
discuss
io
n
of
f
ra
m
ewo
r
k
that
de
al
s
with
the
proce
s
sin
g
big
data
in
existi
ng
syst
e
m
al
on
g
wit
h
highli
gh
ts
of
the
ef
fecti
ve
ne
ss.
Jin
dal
et
al
.
[
17
]
hav
e
use
d
f
uzzy
-
l
og
ic
towa
rd
s
de
ve
lop
in
g
a
n
a
naly
ti
cal
so
luti
on
c
onsid
erin
g
healt
hc
ar
e
-
base
d
a
ppli
cat
ion
em
ph
asi
zi
ng
over
th
e
cl
assifi
er
desig
n.
The
wor
k
of
Me
ng
et
al
.
[1
8]
have
pr
ese
nted
a
f
ram
ewo
rk
for
i
m
pr
ovin
g
qua
li
ty
of
ex
per
ie
nce
by
ad
op
ti
ng
co
nvolu
ti
on
neural
netw
ork
excl
usi
vely
m
eant
f
or
hi
gh
-
dim
en
sion
al
pro
blem
s.
Pu
thal
[
19]
hav
e
us
ed
a
la
tt
ic
-
based
m
od
el
for
form
ulati
ng
access
-
based m
eth
od
ology t
owa
rd
s
stream
s o
f big
data f
oc
us
i
ng on v
olu
m
e and v
el
ocity
issues
i
n
big
data
consi
de
rin
g
healt
hca
r
e
app
li
cat
ions.
Ex
plici
t
stud
y
towa
rd
s
se
nsor
y
data
and
it
s
scal
abili
ty
app
r
oach
was
disc
us
se
d
by
Ra
ff
e
rty
et
al
.
[20].
A
noth
er
re
view
car
ri
ed
out
by
Ri
zw
an
et
al
.
[21]
ha
ve
em
ph
asi
zed
over
the
nano
c
omm
un
ic
at
ion
as
pect
ov
e
r
big
data
c
on
si
der
i
ng
case
stu
dy
of
healt
hc
are
s
ect
or
.
Al
-
Ali
e
t
al
.
[
22]
hav
e
pr
ese
nted
a
wo
r
k
w
her
e
energy
eff
ic
ie
ncy
factor
is
e
m
ph
asi
zed
ov
e
r
us
in
g
busi
ne
ss
intel
li
gen
ce
ov
e
r
the
analy
ti
cal
app
li
cat
io
n
of
big
data.
Ma
ga
rin
o
et
al
.
[2
3]
hav
e
use
d
a
ge
nt
-
based
a
ppr
oa
ch
f
or
in
vestigat
ing
the
sle
ep
-
base
d
data
ob
ta
ine
d
f
ro
m
sensor
s
and
I
oT
ap
pl
ic
at
ion
s
with
higher
acc
ur
ac
y.
Ma
rj
a
ni
et
al
.
[24]
has
al
so
high
li
gh
te
d
var
i
ous
op
e
n
-
en
d
is
su
es
c
onnecte
d
with
t
he
use
of
big
data
m
ining
ove
r
I
oT.
Parwez
et
al
.
[2
5]
ha
ve
use
d
un
s
uper
vise
d
appr
oach
o
f
cl
as
sific
at
ion
over th
e
m
ob
il
e
net
work
f
or
analy
ze
cal
l
record
s.
W
a
ng
et
al
.
[2
6]
hav
e
presente
d
a
pr
e
dicti
ve
fr
a
m
ewo
r
k
f
or
f
oreca
sti
ng
th
e
pri
ce
of
el
ect
ric
it
y
on
the
basis
of
t
he
cl
assifi
cat
ion
as
well
as
sel
ect
ion
of
po
te
ntial
featu
res.
S
un
et
al
.
[27]
ha
ve
pr
esented
a
discussi
on
ab
ou
t
the
analy
ti
cal
appro
ac
h
over
t
he
net
wor
k
com
m
un
it
y
i
n
I
oT.
The
w
ork
of
Li
et
al
.
[
28]
ha
s
carried
ou
t
stud
y
to
wards
geog
r
ap
hic
da
ta
wh
ere
the
op
ti
m
iz
ation
appr
oach
is
use
d
us
in
g
ope
n
so
urce
distrib
uted
s
of
t
war
e f
ram
ewo
r
ks
. S
im
il
ar l
ine o
f
resea
rc
h
w
ork
is al
so
carri
ed
ou
t
by Y
ue
et al
. [
29
]
w
ho
h
av
e
dev
el
op
e
d
bi
g
data
analy
ti
cs
for
ide
ntifyi
ng
we
b
e
ven
ts
as
so
ci
at
ed
with
s
ens
or
y
data.
Ji
ang
et
al
.
[30]
hav
e
us
e
d
hi
dd
e
n
m
arko
v
m
od
el
ing
f
or
m
on
it
ori
ng
t
he
be
ha
viour
of
am
bient
assist
ed
li
ving.
Ther
e
f
or
e,
t
he
re
ha
s
been
va
rio
us
a
ppr
oach
es
to
w
ard
s
bi
g
data
analy
ti
cs
wh
ic
h
is
pro
ve
n
to
offe
r
si
gn
ific
ant
be
ne
fici
al
m
ining
op
e
rati
on
w
hil
e the
nex
t
sect
ion o
utli
nes
t
he researc
h
iss
ues
.
The
i
den
ti
fie
d researc
h
iss
ues
are
as
foll
ows:
Existi
ng
a
ppr
oa
ches
of
the
da
ta
analy
ti
cs
ass
ociat
ed
with
bi
g
data
does
n’
t
include
th
e
co
m
pr
ehen
sive
a
nd
inh
e
ren
t
pro
ble
m
s in
it
p
rior to
processi
ng.
Stor
a
g
e
facto
r
is
not
f
ound
t
o
be
ad
dr
e
ssed
i
n
a
ny
big
data
relat
ed
ap
proa
ch
without
w
hi
ch
a
pp
li
cabil
it
y
of analy
ti
cs ca
nnot
be
scal
a
ble and
pr
act
ic
al
too
.
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.
10
, No
.
4
,
A
ugus
t
2020
:
4322
-
4330
4324
Exclusi
ve
c
onnecti
vity
an
d
i
m
pact
stud
y
of
e
nergy
co
ns
t
raint
ov
e
r
resour
ce
co
ns
trai
ne
d
nodes
are
not
stud
ie
d ov
e
r
a
discrete scal
e
of a
ppro
ac
hes
.
Stud
ie
s
pro
ving
the
co
st
ef
fec
ti
ven
ess of
the
p
rese
nted
s
olu
t
ion
t
ow
a
r
ds
im
pro
ving
the
pe
r
form
ance
of
bi
g
data analy
ti
cs
are less em
ph
a
siz
ed ov
e
r
e
xis
ti
ng
syst
em
.
Ther
e
f
or
e,
the
sta
tem
ent
of
pro
blem
of
th
e
pro
pose
d
st
udy
ca
n
be
sta
t
ed
as
“
Develo
ping
a
c
os
t
ef
fe
ct
iv
eness
in
modeli
ng
anal
yt
ic
al
too
l
for
complex
sen
sory
stream
of
bi
g
data
is
quit
e
cha
ll
en
ging
in
or
de
r
to b
et
te
r
perf
orma
nce
of the
cl
oud
s
ervi
ces
.”
This
pa
rt
of
th
e
researc
h
wor
k
is
an
e
xtensi
on
of
our
pr
i
or
m
od
el
[31]
and
[
32
]
w
hile
the
f
ocus
of
th
e
pro
pos
ed
ap
proac
h
i
s
towa
rd
s
e
vo
lving
up
with
a
ho
li
sti
c
arc
hitec
ture
that
can
use
sens
ory
data
as
a
serv
ic
e
wi
th
cost
eff
ect
iv
e
desig
n
i
m
plem
entat
ion
.
Co
ns
ide
rin
g
a
case
stud
y
of
inte
rn
et
-
of
-
thi
ngs
(IoT),
the
im
ple
m
entat
ion
is
car
rie
d
ou
t
us
in
g
a
naly
ti
cal
research
m
et
ho
dol
ogy.
T
he
picto
rial
represe
ntati
on
of the
pro
pose
d
syst
em
i
m
ple
m
entat
ion
is as
Fig
ur
e
1.
S
y
n
t
h
e
t
i
c
Y
i
e
l
d
o
f
S
e
n
s
o
r
y
D
a
t
a
S
t
o
r
a
g
e
o
p
t
i
m
i
z
a
t
i
o
n
A
n
a
l
y
t
i
c
a
l
O
p
e
r
a
t
i
o
n
G
e
n
e
r
a
t
i
o
n
o
f
S
e
n
s
o
r
y
D
a
t
a
D
e
v
e
l
o
p
i
n
g
a
s
c
a
l
a
b
l
e
s
t
o
r
a
g
e
s
y
s
t
e
m
A
s
i
m
p
l
i
f
i
e
d
k
n
o
w
l
e
d
g
e
m
i
n
i
n
g
F
r
a
m
e
w
o
r
k
f
o
r
d
a
t
a
a
n
a
l
y
t
i
c
s
D
a
t
a
b
a
s
e
m
a
n
a
g
e
m
e
n
t
G
a
t
e
w
a
y
s
e
r
v
e
r
n
o
d
e
s
C
o
n
s
t
r
u
c
t
T
r
e
e
P
e
r
f
o
r
m
s
o
r
t
i
n
g
o
f
b
r
a
n
c
h
E
x
t
r
a
c
t
i
o
n
o
f
c
a
n
d
i
d
a
t
e
f
r
e
q
u
e
n
t
p
a
t
t
e
r
n
Figure
1
.
P
r
op
os
e
d
m
et
ho
dolog
y
of a
naly
ti
c
s
Accor
ding
to
Figure
1,
it
st
at
es
that
propose
d
syst
em
i
s
basical
ly
a
l
evel
-
based
a
nd
t
op
-
do
wn
appr
oach
w
he
n
it
com
es
to
the
arc
hitec
tu
re
de
sig
n.
T
he
top
le
vel
of
the
arc
hitec
tur
e
is
ab
ou
t
sy
nt
hetic
gen
e
rati
on
of
the
sens
o
ry
da
ta
wh
ic
h
is
f
ur
t
her
f
ollo
we
d
up
by
c
on
st
ru
ct
in
g
a
scal
able
stora
ge
s
yst
e
m
.
This
fir
st
le
vel
of
operati
on
l
eads
to
ge
ner
a
ti
on
of
knowle
dg
e
.
T
he
sec
ond
le
vel
of
operati
on
is
res
pons
i
ble
for
pe
rfo
rm
in
g
analy
ti
cal
operati
on
c
on
si
der
i
ng
a
real
-
t
i
m
e
scenar
io
of
im
ple
m
enta
ti
on
.
T
he
a
naly
sis
is
carried
out
co
ns
ide
rin
g
an
e
f
fecti
ve
data
bas
e
m
anag
em
ent
syst
e
m
with
i
nclusi
on
of
gat
eway
syst
e
m
of
I
oT,
diff
e
re
nt
va
rients
of
se
rv
e
rs,
an
d
sens
ors
no
des.
T
he
t
hird
le
vel
of
ope
rati
on
is basical
ly
m
eant
for
pe
rfor
m
in
g
analy
ti
cal
op
er
at
ion
w
her
e
tr
ee
-
base
d
m
echan
ism
is
util
ized
f
or
the
t
opologica
l
con
st
r
uction
of
t
he
IoT
env
i
ronm
ent
fo
ll
ow
e
d
by
the
so
rtin
g
of
th
e
br
a
nc
h.
Fi
nally
,
the
pro
pose
d
syst
em
m
akes
us
e
of
the
frequ
e
nt
patte
rn
as
th
e
m
ining
al
go
rithm
wh
ere
the
ou
tc
om
e
sh
ows
that
accurate
kn
ow
le
dg
e
e
xtr
act
ion
.
In
te
re
sti
ng
thin
g
is
th
at
the
propose
d
syst
em
pe
rfor
m
s
enh
a
ncem
ent
of
t
he
existi
ng
f
re
quent
patte
r
n
l
ogic
by
add
i
ng
up
al
l
the
le
vels
of
operati
on
pr
i
or
to
ap
plyi
ng
fr
e
qu
e
nt
patte
rn
-
base
d
m
ining
al
gorithm
s.
Th
e
core
idea
of
t
he
pro
po
s
ed
syst
em
i
s
that
it
sh
ou
l
d
us
e
the
e
xtra
ct
ed
knowle
dg
e
in
the
form
of
a
naly
zed
se
ns
ory
data
in
the
f
or
m
of
cl
ou
d
-
base
d
se
rv
ic
e
s.
He
nce,
t
he
inf
or
m
at
ion
ob
ta
ine
d
from
the
sens
ory
fiel
ds
as
re
pr
ese
ntati
ves
of
bi
g
da
ta
are
not
on
l
y
op
ti
m
ally
st
or
e
d
in
cl
oud
datace
nte
rs
but
are
al
s
o
re
la
ye
d
in the f
or
m
o
f services
.
2.
SY
STE
M DESIGN
The
co
re
purpoe
of
the
propose
d
syst
e
m
design
is
to
e
volve
up
with
an
in
novative
arch
it
ect
ural
fr
am
ewo
r
k
tha
t
is
capab
le
of
offer
i
ng
sens
or
y
data
in
the
f
o
rm
of
knowle
dg
e
that
ca
n
be
relay
ed
in
th
e
form
of
cl
oud
ser
vices.
H
oweve
r,
there
a
re
vari
ou
s
a
sp
ect
s
t
hat
are
c
onsid
erd
i
n
the
im
plem
entat
ion
ph
a
s
e
of the
pro
pose
d
st
ud
y.
Th
is
s
ect
ion
discusse
s abo
ut the es
s
entia
l i
nfor
m
ation
i
nclu
de
d
in
the syste
m
design
.
2.1
.
Assu
m
pt
i
on
and
de
pende
nc
ie
s
The
pr
im
ary
assum
ption
of
t
he
propose
d
sy
stem
is
to
cons
ider
th
at
the
ne
twork
c
onne
c
te
d
bet
ween
the
us
e
r
te
rm
i
nal
an
d
the
s
erv
ic
e
pr
ov
i
de
r
is
eff
ic
ie
ntly
config
ur
e
d
a
nd
is
hi
gh
ly
s
afe
so
that
no
ne
of
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
Novel
holi
sti
c
ar
c
hitec
ture fo
r
analyti
cal
op
er
atio
n on sen
s
or
y
data rel
aye
d
…
(
Ma
nujak
sh
i B.
C.
)
4
325
the
arti
fact
s
ar
e
the
res
ultants
of
sec
ur
it
y
breac
h.
The
us
e
r
te
rm
inal
is
b
asi
cal
ly
a
gate
way
syst
e
m
wh
ic
h
is
directl
y
co
nn
ect
e
d
t
o
ba
se
sta
ti
on
f
or
ag
gr
e
gatin
g
ov
e
rall
se
ns
or
y
data.
T
he
s
econda
ry
ass
um
pt
ion
of
the
pro
pose
d
syst
e
m
is
that
there
are
la
rge
nu
m
ber
data
in
the
form
of
stream
that
are
arr
an
ge
d
in
dynam
ic
qu
e
ue
syst
e
m
wh
e
re
ce
rtai
n
a
dap
ti
ve
m
anage
m
e
m
t
of
queu
ing
syst
e
m
is
assum
ed
to
be
e
xecu
te
d.
The
te
rtia
ry
assum
ption
of
the
pro
posed
syst
e
m
is
tha
t
al
l
the
se
ns
ors
hav
e
a
sta
ti
c
rate
of
ene
r
gy
dissi
patio
n
w
hile
at
tem
pting
to
perform
any
fo
rm
of
info
rm
ation
f
orwa
rd
i
ng
pr
oc
ess.
T
he
pr
im
e
dep
en
de
ncy
factor
ass
oc
ia
te
d
with
the
pro
posed
syst
e
m
is
that
it
con
side
rs
involvem
ent
of
def
i
niti
ve
nu
m
ber
of
sen
so
rs
to
be
arr
a
nge
d
in
the
form
of
cl
us
te
rs
a
nd
perform
ing
th
e
process
of
da
ta
aggreg
at
i
on.
Anot
her
si
gnific
ant
de
penden
cy
of
the
pr
opos
e
d
syst
e
m
is
that
as
the
stud
y
is
hig
hly
depend
e
nt
upon
dy
nam
ic
senso
r
y
data
wh
ic
h
is
not
feasil
ble
to
be
obta
ined
f
or
t
he
a
naly
sis
an
d
the
refor
e
it
dem
and
s
a
pro
gr
am
m
atic
m
e
chan
ism
fo
r
yi
el
din
g
sens
or
y
data in
d
ynam
ic
o
r
der. Co
ns
ide
rati
on
of this a
ssu
m
ption
a
nd d
e
pe
ndencies
.
2.2
.
Implem
ent
ati
on
s
tr
at
e
gy
Dev
el
op
i
ng
a
rob
us
t
im
ple
mentat
ion
pla
n
f
or
ens
uri
ng
se
ns
ory
data
as
a
ser
vice
is
de
finite
ly
not
an
easy
ta
s
k
a
nd
the
re
a
re
va
rio
us
esse
ntia
l
facto
rs
that
are
re
quire
d
t
o
be
c
onside
r
ed
wh
il
e
de
ve
lop
in
g
a
cost
-
e
ff
ect
iv
e
analy
ti
cal
m
od
el
.
Fo
ll
owin
g
are
t
he
strat
egies
that
are
i
nvolv
e
d
in
des
ign
a
nd
de
velo
pm
ent
of prop
os
e
d
sy
stem
.
Dev
el
op
i
ng
ho
li
sti
c a
rch
it
ect
ur
e
us
in
g
m
od
ul
ar a
ppro
ac
h
The
pro
pose
d
syst
e
m
de
m
an
ds
to
be
f
orm
ulate
d
in
the
f
or
m
of
a
ho
li
s
ti
c
arch
it
ect
ur
e
:
ho
we
ve
r,
const
ru
ct
in
g
holi
sti
c
arch
it
ect
ur
e
is
a
c
ha
ll
eng
in
g
ta
sk
as
there
a
re
m
any
issues
t
o
be
or
gan
iz
e
d
a
nd
addresse
d.
T
he
refor
e
,
the
pro
po
s
ed
r
esearc
h
work
will
con
sid
e
r
m
od
ula
r
ap
proac
h
w
he
re
bigge
r
pro
blem
s
will
be
s
plit
into
sm
al
le
r
ve
rsion
a
nd
the
n
gro
up
e
d
bac
k
tog
et
her.
T
he
pro
posed
syst
e
m
theref
ore
sp
li
t
the
com
plete
i
m
ple
m
entat
ion
into
three
pha
ses
i.e.
gen
e
rati
on
of
sens
or
y
big
data,
pe
rfor
m
ing
op
ti
m
i
zat
io
n
of
sto
ra
ge
syst
e
m
,
and
pe
rform
ing
analy
ti
cal
op
erati
on
on
the
top
of
it
.
A
ll
these
three
m
od
ules
are
groupe
d
tog
et
he
r
t
o
c
onstruct a
holi
sti
c arc
hitec
ture.
Con
si
der
at
io
n of
bi
g data p
roblem
s
It
is
ne
cessa
ry
that
the
i
m
ple
m
entat
io
n
of
the
pr
opos
e
d
syst
e
m
do
h
a
ve
consi
der
at
io
n
of
t
he
pr
ob
le
m
s
associat
ed
wi
th
the
big
data.
The
c
or
e
iss
ue
s
associat
ed
with
the
big
da
ta
is
that
the
da
ta
are
la
rg
e,
un
st
ru
ct
ur
e
d,
a
nd
is
chall
eng
i
ng
to
be
reposit
ed
in
the
SQ
l
base
d
stora
ge
syst
e
m
.
Ther
efore,
the
propose
d
s
yst
e
m
ado
p
ts
a
m
echan
ism
wh
ere
inter
net
-
of
-
thi
ngs
(IoT)
is
con
side
red
a
s
a
case
stud
y
with
the
prese
nce
of
gateway
syst
e
m
,
database
,
and
local
se
nsors
(
or
I
oT
de
vice).
Th
e
pro
po
s
ed
syst
em
dev
el
ops
an
ex
plici
t
m
e
chan
ism
j
us
t
to
ens
ur
e
a
n
ef
fecti
ve
transm
i
ssion
of
dat
a
a
s
well
as
co
m
pr
e
he
ns
ive
ana
ly
ti
cs
m
app
in
g wit
h
t
he real
-
ti
m
e p
ro
blem
.
Strate
gic
in
volvem
ent o
f
e
ne
r
gy constrai
nt
Ther
e
are
var
i
ou
s
reas
ons
f
or
e
ne
rg
y
diss
ipati
on
for
a
sens
or
or
I
oT
de
vice
w
her
e
the
ene
r
gy
consum
ption
is
directl
y
pr
oport
io
nal
to
the
da
ta
trans
m
issi
on
proce
ss.
He
nce,
the
essenti
al
fact
is
that
if
t
he
process
involve
d
in
da
ta
processi
ng
and
analy
sis
is
m
ade
li
gh
twe
igh
t
tha
n
am
ou
nt
of
e
nergy
that
is
al
locat
ed
f
or
s
uch
ta
sk
ca
n
be
co
ntr
olled
t
o
so
m
e
extent.
Ther
e
f
or
e,
the
pro
po
se
d
syst
em
c
on
siders
a
fixe
d
budget
of
ene
r
gy
al
locat
ion
in
te
rm
s
of
const
raint
and
us
e
s
a
tree
-
base
d
m
echan
ism
a
l
ong
with
sim
plifie
d
m
ining
a
ppr
oa
ch usin
g fr
e
qu
ent p
at
te
r
ns
i
n order
to si
m
pli
fy the
process
involve
d
i
n
se
nsory
data a
naly
ti
cs.
All
the
abov
e
m
entioned
three
points
are
co
ns
ide
re
d
as
the
co
r
e
strat
egic
im
ple
m
entat
ion
in
the
pr
opos
e
d
syst
em
wh
e
re
the
pr
im
e
l
og
ic
is
to
en
s
ur
e
that
a
n
e
f
fecti
ve
m
echan
ism
of
know
le
dg
e
extracti
on
ta
ke
s
place
with
great
er
reli
abili
ty
.
Ther
e
f
or
e,
the
im
ple
m
enta
ti
on
of
propos
ed
syst
em
is
carrie
d
ou
t
us
in
g
an
al
yt
ic
al
research
m
et
ho
d
wh
e
r
e
it
beco
m
es
feasible
f
or
pe
rfor
m
ing
an
extensi
ve
disc
ov
e
ry
of
knowle
dge
from
the
sens
or
y
data
with
an
ass
ur
a
nce
t
hat
that
they
are
hi
gh
ly
e
ne
rg
y
e
ff
ic
ie
nt
a
s
well
as the
outc
om
e
of
know
le
dge
del
ivery is
quit
e accu
rate i
n
it
s context.
2.3
.
Framew
or
k
for
sens
ory
-
inf
or
mat
i
on
as ser
vices
This p
a
rt of th
e fr
am
ewo
r
k
is
d
esi
gne
d
on th
e b
asi
s of the f
act
that existi
ng
for
m
s o
f
the
sens
or
y data
are
m
assivel
y
la
rg
e
i
n
it
s
si
ze
an
d
dim
ension
w
her
ea
s
t
he
e
xisin
g
res
earch
-
ba
sed
a
ppr
oach
es
offe
rs
le
ss
e
m
ph
asi
s
ov
e
r
the
data
com
plexity
.
A
pa
rt
from
this,
th
e
volum
e
of
t
he
se
nsory
da
ta
is
so
high
that
it
abno
rm
ally
sat
ur
at
es
the
sto
rag
e
unit
s
of
cl
ouds
w
hich
are
again
enc
ountere
d
with
un
str
uctu
red
data.
Hen
ce
,
an
a
nal
yt
ic
al
fr
a
m
ewo
rk
is
de
sig
ned
for
this
pur
pos
e
wh
ic
h
is
furt
her
cl
assifi
e
d
in
orde
r
to
ca
rry
ou
t
three e
xclusi
ve
operati
ons a
s foll
ows
as
show
n
in
Fig
ure
2
.
Yiel
d of sen
se
d
in
f
or
m
at
ion
As
t
he
pro
po
s
ed
syst
em
is
basical
ly
a
f
r
a
m
ewo
r
k
t
herefo
re
it
has
a
de
pe
nd
e
ncy
of
stream
ed
inf
or
m
at
ion
.
Fo
r
this
pur
pose
,
any
sta
nd
a
rd
dataset
can
be
util
iz
ed
as
the
idea
is
j
us
t
to
offer
m
axi
m
ized
data
in
the
form
of
stream
.
Ho
wever,
offli
ne
se
nsory
data
c
onsi
der
at
io
n
will
la
ck
the
pro
ble
m
s
that
is
associat
ed
with
the
real
env
ir
onm
ent
towards
this
data
will
be
m
issi
ng
.
Th
eref
or
e
,
the
pro
posed
syst
e
m
will
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.
10
, No
.
4
,
A
ugus
t
2020
:
4322
-
4330
4326
pro
gr
am
m
a
ti
ca
ll
y
gen
erate
se
ns
or
data
with
an
ai
d
of
C
onti
ki
f
or
car
ryi
ng
ou
t
t
his
e
xp
e
ri
m
ent.
Ba
sic
al
ly
,
it
is
on
e
f
or
m
of
t
he
op
e
n
s
ourc
e
syst
e
m
that
offer
s
exte
ns
iv
e
assessm
ent
env
i
ronm
ent.
The
in
f
or
m
at
i
on
that
is
gen
e
rated
by
the
Con
ti
ki
is
con
si
der
e
d
t
o
be
sen
sory
data
that
bea
rs
al
l
the
char
ec
te
risti
cs
of
it
c
arr
ie
d
ou
t
pro
gr
am
m
a
ti
cal
ly
.
Stor
a
ge
opt
i
m
i
zat
ion
The
propose
d
syst
e
m
of
fer
s
a
property
of
el
ast
ic
it
y
of
the
cl
oud
sto
rag
e
s
yst
e
m
in
the
form
of
cl
oud
bu
c
ket
w
hich
is
basical
ly
a
form
of
director
y
of
the
sto
rag
e
on
the
t
op
of
the
sto
ra
ge
syst
em
us
ed
over
the
data
cente
r
s.
I
nteresti
ngly
,
the
pro
pose
d
syst
e
m
of
fer
s
s
tora
ge
facil
it
y
for
the
us
er
us
i
ng
t
he
cl
oud
buckets
and
not
t
he
da
ta
center
sto
ra
ge.
The
stu
dy
im
ple
m
entat
i
on
use
s
distrib
uted
data
base
m
anag
em
ent
syst
e
m
in
order
to
offe
r
involu
ntary
m
anag
em
ent
of
fau
lt
tolera
nc
e.
As
the
pri
or
sens
o
ry
data
ar
e
hig
hly
unstr
uc
ture
d
and
t
her
e
fore,
it
is
qu
it
e
chall
eng
in
g
to
pe
rfor
m
any
form
of
data
proce
s
sing
on
it
.
The
refor
e
,
this
pr
oble
m
is
m
it
igate
d
by
us
i
ng
H
base
that offers
bette
r
in
de
xing
m
echan
ism
towa
rds
al
l
the rows
of
the
dist
rib
uted
data
pre
se
nt
in
t
he
cl
oud
e
nv
ir
on
m
ent.
Furthe
r,
the
m
echan
ism
towards
us
i
ng
t
he
cl
ou
d
bu
c
ket
syst
em
s
offers
the
ap
pro
pri
at
e
us
a
ge
of
the
use
r
on
the
basi
s
of
th
e
act
ual
dem
and
s
of
da
ta
processi
ng.
The
pro
posed
syst
e
m
m
ai
ntains
a
uniqu
e
in
dex
i
ng
keys
ov
e
r
dif
fe
ren
t
ty
pe
s
of
the
distrib
uted
stora
ge
se
r
ver
s
that
act
ually
a
ssisy
s
in
faster
data
extracti
on
and
m
anag
em
e
nt
process
.
T
he
pri
m
e
m
e
chan
ism
of
this
syst
e
m
e
nab
l
e
s
the
processi
ng
of
unstr
uct
ured
to
str
uctu
r
ed
data
that
there
by
m
akes
it
su
it
able
for
ap
plyi
ng
a
ny
fo
rm
of analy
ti
cs on
it
.
Cost
-
ef
fecti
ve a
naly
ti
cal
o
perat
ion
This
par
t
of
t
he
im
ple
m
ent
at
ion
deals
wi
th
hi
gh
li
ghti
ng
the
analy
ti
cal
op
e
rati
on
that
pe
rfor
m
s
extracti
on
of
m
ined
knowle
dg
e
i
n
orde
r
to
relay
senso
ry
analy
zed
data
as
a
serv
ic
e.
T
he
patte
r
ns
ass
ociat
ed
with
the
tra
nsm
issi
on
of
t
he
sensory
data
is
ob
ta
ine
d
that
is
us
ed
for
extracti
ng
a
n
ex
plici
t
data
fr
om
the
se
ns
or
node
s
as
well
as
var
i
ou
s
othe
r
a
sso
ci
at
ed
in
form
ation
e.
g.
fr
e
quencies
of
a
tt
e
m
pts
of
retra
ns
m
issio
n
an
d
delay
of
tra
ns
m
itti
ng
i
nfor
m
at
ion
.
As
al
l
the
in
form
at
ion
ass
ociat
e
d
with
t
he
networ
k
processi
ng
is
ver
y
m
uch
im
portant
f
or
t
he
netw
ork
anal
yst
,
there
fore
t
hese
s
ort
s
of
l
at
ent
inf
orm
ati
on
ar
e
extracte
d
by
t
he
syst
e
m
fo
r
be
tt
er
pr
eci
si
on
m
anag
em
ent.
The
c
or
e
m
eain
g
of
th
e
kn
owle
dg
e
i
n
the
pr
opos
e
d
syst
e
m
is
basical
ly
the
tren
d
of
t
he
net
w
ork
patte
rn
of
transm
issi
on
that
is
form
ulate
d
by
the
s
ens
or
nodes/
dev
ic
es
is
the
m
od
e
of
diff
e
ren
t
ap
pl
ic
at
ion
.
The
propose
d
syst
em
can
m
ake
us
e
of
this
knowl
edg
e
in
order
t
o
i
m
pro
ve
the
oper
at
ion
of
va
rio
us
app
li
cat
ions
that
dep
e
nds
upon
the
se
nsor
y
data.
It
is
alr
ead
y
known
that
s
uppo
rtabil
it
y
of
the
unstr
uctured
data
that
f
urt
her
off
ers
be
tt
er
acce
ssibil
it
y
towards
the
m
os
t
discrete set
of
data that ac
t
ually
co
ntr
ols th
e
ov
e
r
head o
f da
ta
.
F
r
a
m
e
w
o
r
k
f
o
r
S
e
n
s
o
r
y
I
n
f
o
r
m
a
t
i
o
n
a
s
S
e
r
v
i
c
e
s
Y
i
e
l
d
o
f
s
e
n
s
e
d
i
n
f
o
r
m
a
t
i
o
n
S
t
o
r
a
g
e
o
p
t
i
m
i
z
a
t
i
o
n
C
o
s
t
E
f
f
e
c
t
i
v
e
a
n
a
l
y
t
i
c
a
l
o
p
e
r
a
t
i
o
n
Figure
2
.
Fr
am
ewor
k
f
or
se
nsory
-
i
nfor
m
at
ion
as
servic
es
2.4
.
Framew
or
k
for
adv
an
ced
an
alytic
s
This
par
t
of
the
i
m
ple
m
ent
at
ion
is
fo
cu
s
ed
on
dev
el
oping
ad
va
nced
analy
ti
cal
op
erati
on
w
hile
the
stud
y
al
so
e
m
ph
asi
zes
over
a
ddressin
g
the
existi
ng
pro
blem
associat
ed
with
rest
rict
ed
co
ntr
ol
ov
e
r
cl
oud
-
base
d
re
so
urces
as
wel
l
as
la
ck
of
co
m
pat
ibil
it
y
of
existi
ng
a
naly
ti
cal
op
erati
on
from
co
m
pu
ta
ti
on
al
cost
ef
fecti
ve.
The
c
or
e
st
ud
y
obj
ect
ive
of
t
hs
par
t
of
the
im
ple
m
entat
ion
is
to
offe
r
a
c
om
pr
ehe
ns
ive
desi
gn
of
t
he
us
e
r
-
bas
ed
kn
ow
le
dg
e
m
ining
a
ppro
a
ch.
T
he
sec
on
dar
y
obj
ect
ive
of
t
his
pa
rt
of
the
im
ple
m
entat
ion
will
be
to
ob
t
ai
n
var
i
ou
s
pote
ntial
patte
rn
s
associat
ed
wi
th
the
la
te
nt
c
onnecti
vity
of
the
patte
rn
s
a
m
on
g
var
i
ou
s
set
of
discrete
data
ge
ner
at
e
d
by
th
e
uniq
ue
se
nso
r
no
des.
The
c
or
e
l
og
ic
of
th
e
pro
posed
stu
dy
will
be
to
obta
in
th
e
ou
tp
ut
from
the
pr
i
or
m
odule
wh
ic
h
is
f
ur
t
her
s
ubj
ect
e
d
to
the
tree
m
echan
ism
fo
r
bette
r
analy
sis
of
te
h
data
i
n
the
f
or
m
of
no
des
and
e
dges.
T
he
proc
ess
tha
n
car
ry
out
an
ef
fecti
ve
tre
e
-
base
d
m
anag
em
ent
as
well
as
sorti
ng
of
the
br
a
nc
hes
wh
ic
h
is
finall
y
fo
ll
owe
d
by
the
fr
e
qu
ent
patte
r
ns
c
oncept
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
Novel
holi
sti
c
ar
c
hitec
ture fo
r
analyti
cal
op
er
atio
n on sen
s
or
y
data rel
aye
d
…
(
Ma
nujak
sh
i B.
C.
)
4327
that
ulti
m
a
te
l
y
le
ads
to
ge
ner
at
io
n
of
var
i
ou
s
m
ined
outc
om
es.
The
p
rim
e
nove
lt
y
of
this
par
t
of
the
i
m
ple
m
ent
at
ion
is
that
it
is
capab
le
of
est
ablishin
g
s
ign
ific
a
ntco
nnect
ivit
y
witi
n
the
dif
fer
e
nt
da
ta
in
the
f
or
m
of
node
s
al
ong
wi
th
co
ns
ide
rin
g
it
s
resp
ect
ive
con
te
xt.
T
he
pro
po
se
d
syst
e
m
al
so
pr
ese
nt
s
an
analy
ti
cal
fr
a
m
ewor
k
that
is
c
apab
le
of
pe
rfo
rm
ing
exp
li
ci
t
com
pu
ta
ti
on
of
diff
ere
nt
va
riants
of
the
est
im
at
es
connecte
d
to
t
he
nodes
.
A
part
fr
om
this,
the
com
plete
con
c
ept
is
dev
el
ope
d
in
su
c
h
a
wa
y
that
i
t
can
ac
tuall
y
su
pp
or
t
distri
bute
d
ap
plica
ti
on
al
ong
with
c
onnec
ti
vity
opti
on
of
bo
th
sin
gle
as
well
as
double
hop
com
m
un
ic
at
ion
syst
em
.
The
pr
ese
nted
stu
dy
carry
the
e
ntire
operati
on w
i
th
res
pect
to v
a
rio
us
dem
and
s o
f
t
he
app
li
cat
io
n
associat
ed
with
the
cl
oud
ser
vi
ces.
The
pro
pose
d
syst
e
m
a
lso
i
m
ple
m
ent
s
a
threshold
-
b
ase
d
m
echan
ism
in
order
to
fin
d
out
if
th
e
sel
ect
ed
patte
rn
is
s
upporte
d
by
th
e
present
t
raffic
of
com
m
un
ic
at
ion
channel.
The
im
ple
m
entat
ion
strat
egy
is
c
onti
nu
e
d
us
i
ng
a
novel
tr
ee
-
ba
sed
m
echan
is
m
wh
ic
h
offe
r
s
bette
r
local
iz
at
ion
of
an
ex
plici
t
data
as
we
ll
as
it
f
ind
s
a
uniq
ue
r
ou
te
of
c
omm
un
ic
at
ion
betwe
en
tw
o
act
ive
node
s
that
al
ways
sta
rts
f
ro
m
the
r
oot
no
de.
Furth
er
the
im
ple
mentat
ion
is
car
ried
out
by
du
al
ste
ps
of
ope
rati
on
wh
e
re
t
he
fi
rst
ste
p
is
to
co
ns
t
ru
ct
t
he
tree
str
uctu
re
wh
il
e
th
e
seco
nd
ste
p
i
s
to
ca
rr
y
out
a
naly
ti
cal
op
era
ti
on
.
Fi
gure
3
hi
gh
li
gh
ts
t
he
pr
opose
d
f
ram
ewo
r
k
wh
ic
h
ge
ne
rates
var
i
ous
data
SD1,
S
D
2,
…
.
SDn
in
distri
bu
te
d
m
ann
er
that
f
urt
her
u
ses
tree m
echan
ism
to
gen
e
rate
a
high
li
gh
t
connecte
d
tree
topolo
gy
with
ind
ivi
dua
l
tree
s
t1,
t2
,
….
T
n.
The
m
echan
is
m
su
pp
ort
s
tre
e
m
anag
e
m
ent
as
well
as
perform
s
so
rting
of
branc
h
fo
ll
owed
by
extracti
ng ca
ndidate
f
re
qu
e
nt
patte
rn,
wh
ic
h i
s u
lt
i
m
at
ely k
now
n
as
final
knowle
dge e
xtra
ct
ion
.
S
D
a
a
S
F
r
a
m
e
w
o
r
k
S
d
1
S
d
2
S
d
n
S
d
3
-
-
-
-
-
T
1
T
2
T
3
T
n
A
p
p
l
y
G
r
a
p
h
T
e
c
h
n
i
q
u
e
t
o
g
e
n
e
r
a
t
e
T
r
e
e
T
r
e
e
m
a
n
a
g
e
m
e
n
t
B
r
a
n
c
h
S
o
r
t
i
n
g
E
x
t
r
a
c
t
c
a
n
d
i
d
a
t
e
F
P
F
i
n
a
l
K
n
o
w
l
e
d
g
e
e
x
t
r
a
c
t
i
o
n
Figure
3
.
Fr
am
ewor
k
f
or
a
dv
a
nced analy
ti
cs
3.
RESU
LT
A
N
ALYSIS
This
sect
io
n
di
scusses
a
bout
t
he
outc
om
e
being
ob
ta
ine
d
from
the
i
m
ple
m
entat
ion
of
the
pr
opos
e
d
syst
e
m
.
The
scriptin
g
of
the
propose
d
sys
tem
is
carried
ou
t
in
M
AT
LAB
w
here
500
nodes
ha
ve
bee
n
consi
der
e
d
to
be
dis
per
se
d
i
n
a
sim
ulatio
n
area
of
1000
x120
0m
2
.
All
t
he
co
nf
i
gurati
on
of
th
e
sens
or
node
bear
s
the
c
ha
re
ct
erist
ic
s
of
M
EMSIC
nodes
wh
il
e
the
pr
opos
e
d
syst
em
has
bee
n
c
om
par
ed
with
the
e
xi
sti
ng
fr
e
qu
e
nt
patte
r
n
-
base
d
a
ppr
oa
ch
i
n
order
to
asse
ss
perform
ance
com
par
at
ive
a
naly
sis.
The
assessm
ent
ha
s
been car
ried
out with
r
es
pect
to effect
ive m
i
ning
durati
on, e
nergy
dep
le
ti
on, a
nd m
e
m
or
y consum
ption
.
The
st
ud
y
outc
om
e
as
show
n
i
n
Fig
ures
4
-
6
cl
early
shows
that
pr
opos
e
d
s
yst
e
m
is
bette
r
in
c
ontras
t
to
existi
ng
f
re
qu
e
nt
patte
r
n
appr
oach.
The
basic
reas
on
f
or
m
ining
dura
ti
on
is
that
existi
ng
f
reque
nt
patte
r
n
involves
c
ount
ing
it
em
s
wh
ic
h
ex
pone
ntial
ly
increases
w
her
eas
pro
po
s
ed
syst
e
m
counts
on
ly
knowle
dg
e
ob
ta
ine
d
he
nce
tim
e
ta
ken
is
ver
y
m
uch
reduced
.
P
r
opos
e
d
syst
em
inv
ol
ved
zer
o
rec
ursive
op
e
rati
on
wh
il
e
tree
-
ba
sed
a
ppro
ac
h
is
util
iz
e
d
for
bette
r
to
polo
gical
con
str
uct
of
an
Io
T
s
yst
e
m
that
ensu
res
bette
r
co
nt
ro
l
of
energy
reducti
on
wh
ic
h
can
no
t
be
ca
rr
ie
d
ou
t
in
existi
ng
syst
em
.
The
pr
op
os
e
d
syst
e
m
has
al
so
lowe
r
m
e
m
or
y
con
s
um
pt
ion
as
a
doption
of
tree
-
ba
sed
a
ppro
ac
h
reduces
the
de
pende
ncies
of
m
axi
m
iz
ed
locat
ion
of
data
sto
rag
e
w
hich
is
not
se
en
in
e
xisti
ng
appr
oach.
T
he
refor
e
,
pr
opose
d
syst
e
m
can
be
cl
ai
m
ed
to
offer
bette
r
a
nd
cost
-
ef
fecti
ve
a
naly
ti
cal
p
erfor
m
a
nce in co
ntrast
to ex
ist
in
g
syst
e
m
.
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.
10
, No
.
4
,
A
ugus
t
2020
:
4322
-
4330
4328
Figure
4. Com
par
at
ive
an
al
ys
is
of
ef
fecti
ve
m
i
ning
durati
on
Figure
5. Com
par
at
ive
an
al
ys
is
of ene
rg
y
de
pleti
on
Figure
6
.
Com
par
at
ive
analy
s
is of m
e
m
or
y con
s
um
ption
4.
CONCL
US
I
O
N
At
present
,
the
re
are
var
i
ous
form
s
of
ap
proach
e
s
that
ha
s
bee
n
sta
te
d
and
cl
ai
m
ed
to
ef
fecti
vely
m
itigate
the
prob
le
m
s
of
loa
d
balanci
ng
in
cl
oud
en
vir
onm
ent.
H
ow
e
ve
r,
m
it
igati
ng
the
m
assive
loa
d
ove
r
dynam
ic
an
d
distribu
te
d
env
i
ronm
ent
of
pri
vate
cl
oud
is
so
m
eth
in
g
w
hich
is
qu
it
e
chall
eng
i
ng
.
Ther
e
f
or
e,
t
he
pro
po
se
d
syst
e
m
has
intr
oduced
a
m
echa
nism
of
cost
-
e
ff
ect
ive
l
oad
ba
la
ncing
m
echan
is
m
us
in
g
a
naly
ti
cal
m
et
ho
ds.
T
he
sig
nificant
c
ontrib
utio
n
of
th
e
pro
posed
a
ppro
ac
h
is
as
fo
ll
ow
s:
i)
it
ov
e
rc
om
es
the
lim
i
ta
ti
on
of
cl
ou
d
-
base
d
con
tr
ol
syst
em
,
incom
patibil
ity
of
existi
ng
m
ining
m
od
el
s,
and
is
a
ppli
cable
ov
e
r
com
pr
eh
ensive
e
ven
t
processi
ng,
ii
)
the
stud
y
presents
a
si
m
plifie
d
and
ye
t
so
phist
ic
at
ed
us
ag
e
of
f
reque
nt
pat
te
rn
s
that
co
nnect
s
data
with
nodes
,
ii
i)
the
m
et
ho
d
al
so
use
s
tree
-
base
d
s
chem
e
capale
e
nough
for
s
us
ta
inin
g discrete
tra
ns
m
issi
on
over
sensor
y a
ppli
c
at
ion
.
R
E
F
E
R
E
N
C
E
S
[1]
Johns
on,
Bjorn,
and
Yanz
h
en
Qu
,
"A
holi
stic
m
odel
for
m
aki
ng
cl
oud
m
ig
rat
ion
d
ec
ision
:
A
conside
ra
ti
o
n
of
sec
uri
t
y
,
arc
h
it
e
ct
ure
and
bus
ine
ss
ec
onom
ic
s
,
"
In
2012
IEEE
10th
Int
ernati
o
nal
Symposium
on
Parall
e
l
and
Distribute
d
Proc
essing
wit
h
Applicat
ions
,
pp
.
435
-
441
,
2012
.
[2]
Baka
l
ash,
Reuv
en,
Gu
y
Shak
e
d,
and
Jos
eph
Caspi
,
"M
et
hod
of
and
appa
ra
t
us
for
dat
a
agg
reg
ation
uti
l
iz
in
g
a
m
ult
idi
m
ensiona
l
databa
se
an
d
m
ult
i
-
stage
da
ta
aggr
eg
at
ion
o
per
ations
,
"
U.S.
Pat
ent
Application
Publ
ic
a
ti
on
,
2005.
[3]
Roesic
ke
,
Bernd
,
and
Manfre
d
S
ei
denstr
ic
ker
,
"
Anal
y
tical
te
st
e
le
m
ent
with
wir
el
ess
data
tra
ns
m
ission
,
"
Publ
.
of
Appl
ic
a
ti
on
wit
h
search
report
-
European
Pa
te
nt
Office
,
20
05
.
[4]
Le
e
,
Ja
y
,
Behr
ad
Baghe
ri
,
and
Hung
-
An
Kao
,
"Rec
en
t
adv
ances
a
nd
tre
nds
of
c
y
b
er
-
ph
y
si
cal
s
y
st
e
m
s
and
big
dat
a
ana
l
y
t
ic
s
in
ind
ustria
l
informatics
,
"
In
Int
ernat
ional
proce
edi
n
g
of
in
t
con
fe
r
enc
e
on
industr
ial
in
formatic
s
(
INDIN
)
,
pp.
1
-
6
.
2014
.
[5]
Pati
l,
Shant
aku
m
ar
B.
,
Prem
j
y
oti
Pat
il,
and
R
oopashre
e
H
.
R.
,
"In
te
gr
at
ed
fr
amework
for
se
cur
e
and
ene
r
g
y
eff
icient
comm
unic
a
ti
on
s
y
stem
in
het
ero
gen
eo
us
sensory
application
,
"
Inte
r
na
ti
onal
Journal
of
El
e
ct
rica
l
and
Computer
Engi
n
ee
ring
(
IJE
C
E)
,
vol.
9
,
no
.
4
,
269
5
-
2702,
2019
.
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
Novel
holi
sti
c
ar
c
hitec
ture fo
r
analyti
cal
op
er
atio
n on sen
s
or
y
data rel
aye
d
…
(
Ma
nujak
sh
i B.
C.
)
4329
[6]
Burhanuddi
n,
M
.
A.,
Ronizam
Ism
ai
l,
Nur
ul
Izz
ai
m
ah,
Ali
Ab
dul
-
Jabba
r
Moham
m
ed,
and
No
rza
imah
Zainol
,
"A
naly
sis
of
Mobile
Servi
ce
Providers
Perform
anc
e
Us
ing
Nai
ve
Ba
y
es
Data
Mining
Te
chni
q
ue
,
"
In
te
rnation
al
Journal
of
Elec
t
rical
and
Computer
Eng
ine
ering
(
IJE
CE)
,
vol.
8,
no.
6
,
pp
.
5153
-
5161,
2018
.
[7]
Qin,
S.
Joe
,
"P
roc
ess da
t
a ana
l
y
t
i
cs
in the
er
a
of
b
ig
da
ta
,
"
AICh
E Journal
,
vo
l.
60,
no.
9
,
pp.
3092
-
3100
,
2014
.
[8]
Schnase
,
John
L.
,
Dani
el
Q.
Duff
y
,
Glenn
S.
Ta
m
kin,
Denis
Nade
au,
John
H.
Thomps
on,
Cristi
na
M.
Grieg
,
Mark
A.
McIne
rne
y
,
and
W
il
lia
m
P
.
W
ebste
r
,
"M
ERRA
ana
l
yti
c
serv
ic
es:
M
e
et
ing
th
e
big
d
a
ta
ch
al
l
enge
s
of
cl
imat
e
sci
ence
through
cl
oud
-
ena
bl
ed
c
li
m
ate
an
aly
t
ic
s
-
as
-
a
-
servic
e
,
"
Com
pute
rs
,
En
vi
ron
ment
and
Ur
ba
n
Syste
ms
,
vol
.
61
,
pp.
198
-
211
,
20
17
.
[9]
Grolinge
r,
Kat
ar
ina
,
Mich
ae
l
Ha
y
es
,
W
il
son
A.
Higashino,
Alex
andr
a
L
'
Heure
ux
,
David
S.
Alli
son,
and
Miria
m
A
.
M.
Capretz
,
"Ch
al
l
enge
s for
m
apr
educ
e
in
big dat
a
,
" In
2014
I
EEE
world
congre
ss
on
services
,
pp
.
182
-
189
,
2014
.
[10]
Manuja
kshi
B
.
C
.
,
R
amesh
K.
B.
,
"F
ramework
for
cost
-
eff
e
c
ti
ve
anal
y
tical
m
ode
ll
ing
for
sensor
y
d
ata
ove
r
cl
oud
env
ironment
,
"
Inte
r
nat
ion
al
Journal
o
f
El
e
ct
rica
l
and
Computer
Enginee
ring
(
IJE
CE
)
,
vol.
9,
no.
5,
pp.
3822
-
3832
,
2
019.
[11]
Zou,
W
e
it
ao
,
W
ei
peng
Jing,
Guangsheng
Chen
,
Yang
Lu,
and
Houbing
Song
,
"
A
surve
y
of
b
ig
dat
a
anal
y
tics
fo
r
sm
art
fore
str
y
,
"
I
EE
E
Acce
ss
,
vol
.
7
,
pp
.
46621
-
4
6636
,
2019
.
[12]
Shah,
S
y
ed
At
tique,
Durs
un
Zafer
Seker
,
Sufi
a
n
Ham
ee
d,
and
Dirk
Drahe
im
,
"The
Rising
R
ole
of
Big
Dat
a
Anal
y
tics
and
I
oT
in
Disaster
Mana
gement:
R
ec
en
t
Advanc
es,
Ta
xonom
y
and
Pros
pec
ts
,
"
IE
E
E
Acce
ss
,
vo
l.
7
,
pp.
54595
-
5461
4
,
2019
.
[13]
Hadi,
Moham
me
d
S.,
Ahm
ed
Q
.
La
we
y
,
Ta
isir
E
.
H
.
El
-
Gorash
i,
and
Jaa
f
ar
M
.
H
.
El
m
irgha
ni
,
"P
at
ie
nt
-
Cen
tr
i
c
Cel
lular
Ne
twor
ks Opti
m
iz
ation
using Bi
g
Da
ta
Anal
y
tics
,
"
IE
E
E
A
ccess
,
vol
.
7
,
pp.
49279
-
4929
6
,
2019
.
[14]
Boubic
he
,
Sabr
i
na,
Dj
al
l
el
Eddine
Boubi
che,
Az
eddi
ne
Bil
ami,
a
nd
Hom
ero
Toral
-
Cruz
,
"Big
d
ata
ch
al
l
enge
s
and
dat
a
aggr
ega
t
ion
strategie
s
in
wir
el
ess sensor n
etw
orks
,
"
IEE
E
A
c
ce
ss
,
vol
.
6
,
pp
.
20558
-
20571
,
2
018
.
[15]
Xianghui
Cao,
L
u
Li
u,
Yu
Cheng
,
Xuem
in
Shen
,
"
Towa
rds
Ene
rg
y
-
Ef
ficien
t
W
ir
el
ess
Networki
n
g
in
the
Big
Data
Era
:
A Surve
y
,
"
IEE
E
Comm
unications
Surve
ys
&
T
utorial
s
,
vol.
20,
no.
1
,
pp.
30
3
-
332
,
2017
.
[16]
Jabba
r,
Sohai
l,
Kale
em
R.
Mal
i
k,
Mudass
ar
Ahm
ad,
Om
ar
Aldabba
s,
Muham
m
ad
As
if,
Shehzad
Khali
d,
Kijun
Han,
and
S
y
ed
Hass
an
Ahm
ed
,
"A
m
et
hodolog
y
o
f
re
al
-
ti
m
e
d
at
a
fusion
for
lo
ca
l
iz
ed
big
data
anal
y
tics
,
"
I
EEE
Ac
c
ess
,
vol
.
6
,
p
p.
24510
-
24520
,
2018
.
[17]
Jindal
,
Anish,
A
m
it
Dua,
Nee
r
aj
Kum
ar,
As
hok
Kum
ar
Das,
Ath
ana
sios
V.
Vasil
akos,
and
Joel
J
.
P
.
C
.
Rodrigu
es
,
"P
rovidi
ng
healt
hca
re
-
as
-
a
-
servi
ce
using
fuz
z
y
r
ule
b
ase
d
big
da
ta
ana
l
y
t
ic
s
in
cloud
computing
,
"
IEEE
journal
o
f
biomedi
ca
l
and
heal
th
inf
orm
atics
,
vol
.
22
,
no
.
5
,
pp
.
1605
-
1618
,
2018
.
[18]
Meng,
Qian
y
u
,
Kun
W
ang,
Xia
om
ing
He,
and
Min
y
i
Guo
,
"Q
o
E
-
drive
n
big
d
ata
m
ana
gement
i
n
per
vasiv
e
edg
e
computing
env
ir
onm
ent
,
"
B
ig
Da
ta
Min
ing
and
A
naly
tics
,
vo
l.
1,
no.
3
,
pp
.
222
-
2
33
,
2018
.
[19]
Puthal
and
De
e
pak
,
"La
tt
i
ce
-
M
odel
ed
Inform
ation
Flow
Control
of
Big
Sensing
Data
Stre
ams
for
Sm
art
Hea
lt
h
Applic
a
ti
on
,
"
IE
EE
In
te
rnet
of
Things
Journal
,
v
ol.
6
,
no
.
2
,
pp
.
1
312
-
1320
,
2018
.
[20]
Raff
erty
,
Jos
eph
,
Jonath
an
S
y
nn
ott
,
Chr
is
D.
Nu
gent
,
Andrew
E
nnis,
Phili
p
A.
Cat
her
wo
od,
Ian
McChesne
y
,
Ia
n
Cle
la
nd
,
and
Sa
lly
McC
le
an
,
"A
Scal
abl
e
,
Resea
rch
Orien
te
d
,
Gene
ric
,
Senso
r
Data
Plat
form
,
"
IEE
E
Acce
ss
,
vol.
6
,
pp
.
45473
-
45484
,
2018
.
[21]
Riz
wan,
Ali,
Ahm
ed
Zoha
,
Ru
i
Zha
ng
,
W
asi
m
Ahm
ad,
Kam
ran
Ars
had,
Naja
h
Abu
Ali
,
Akram
Alom
ai
n
y
,
M
uhamm
ad
Ali
Im
ran
,
and
Qam
m
er
H.
Abbasi
,
"A
rev
ie
w
on
th
e
role
of
n
ano
-
c
om
m
unic
at
ion
i
n
future
he
al
th
care
s
y
stems
:
A bi
g
d
at
a
an
aly
t
ic
s pe
r
spec
ti
v
e
,
"
I
EEE
Ac
c
ess
,
vol
.
6
,
p
p.
41903
-
41920
,
2018
.
[22]
Al
-
Ali,
Abdul
-
R
ahman,
Im
ran
A.
Zualker
nan
,
Moham
m
ed
Rashid,
Ragi
ni
Gupt
a
,
and
Maz
in
Ali
kar
ar
,
"A
sm
art
hom
e
ene
rg
y
m
ana
gement
s
y
s
tem
using
IoT
an
d
big
dat
a
an
alytic
s
appr
oa
ch
,
"
I
EE
E
Tr
ansacti
o
ns
on
Consu
mer
El
e
ct
ronics
,
vol
.
63,
no.
4
,
pp.
42
6
-
434
,
2017
.
[23]
Garc
ía
-
Mag
ari
ñ
o,
Ivá
n
,
Raquel
Lacue
sta
,
an
d
Jaime
Ll
ore
t
,
"A
gent
-
base
d
sim
ula
ti
on
of
sm
art
beds
with
Inte
rne
t
-
of
-
Thi
n
gs for
expl
o
ring big
da
ta a
na
l
y
tic
s
,
"
IEEE
A
ccess
,
vol.
6
,
pp.
366
-
379
,
2017
.
[24]
Marja
ni
,
Mohs
en,
Fariza
Nasar
uddin,
Abdull
ah
Gani,
Ahm
ad
Kari
m
,
Ibra
him
Abake
r
T
arg
io
Hashem
,
Aisha
Siddiqa
,
and
Ibr
ar
Yaqoob
,
"Bi
g
IoT
data
anal
y
t
ic
s:
a
rch
i
te
c
tu
re,
opportun
it
i
es,
and
open
r
ese
arc
h
ch
al
l
enge
s
,
"
IEE
E
Acce
ss
,
vo
l.
5
,
pp
.
5247
-
52
61
,
2017
.
[25]
Parwez
,
Md
Sa
li
k,
Dand
a
B.
Rawa
t,
and
Mo
ses
Garuba
,
"B
ig
data
an
aly
t
ics
for
user
-
ac
t
iv
ity
an
aly
sis
and
user
-
anomal
y
d
e
te
c
ti
on
in
m
obil
e
wire
le
ss
net
w
ork
,
"
IE
EE
Tr
ansacti
ons
on
Industrial
Informatics
,
vol.
13
,
no.
4
,
pp.
2058
-
2065
,
2017
.
[26]
W
ang,
Kun,
Chenha
n
Xu,
Yan
Zha
ng,
Song
Guo,
and
Alber
t
Y.
Zoma
y
a
,
"Robus
t
big
dat
a
an
aly
t
ics
for
el
ec
trici
t
y
pric
e
fore
ca
sting
in the
sm
art
gri
d
,
"
IE
EE
Tr
ansacti
ons on
Bi
g
Da
ta
,
vo
l.
5,
no.
1
,
pp.
34
-
45
,
2017
.
[27]
Sun,
Yunchua
n,
Houbing
Song,
Antonio
J.
Jara
,
and
Rongfang
Bie
,
"Int
ern
e
t
of
thi
ngs
and
big
dat
a
ana
l
y
t
ic
s
fo
r
sm
art
and connecte
d
comm
unit
ies
,
"
IEEE
a
cc
ess
,
vol.
4
,
pp
.
766
-
7
73
,
2016
.
[28]
Li
,
Peng
,
Song
Guo,
Toshia
ki
Mi
y
a
za
k
i,
Xiaof
ei
Liao,
Ha
i
Jin,
Alber
t
Y.
Zoma
y
a
,
and
Kun
W
a
ng
,
"Tra
ffi
c
-
awa
re
geo
-
distri
but
ed
big
data
an
aly
t
i
cs
with
pre
d
ic
t
a
ble
job
completion
ti
m
e
,
"
IEEE
Tr
ansacti
ons
on
Parall
e
l
an
d
Distribute
d
S
yst
ems
,
vol
.
28
,
no
.
6
,
pp.
1785
-
179
6
,
2016
.
[29]
Yue,
Peng
,
Che
nxia
o
Zha
ng
,
Mingda
Zh
ang,
Xi
Zh
ai,
and
Li
ang
cun
Jiang
,
"A
n
SD
I
appr
oac
h
for
big
da
ta
ana
l
y
t
ic
s:
Th
e
c
ase
on
sensor
w
eb
eve
nt
de
tection
and
geopr
oces
sing
workflow
,
"
IEE
E
Journal
of
Sel
e
ct
ed
Topi
cs
in
App
l
ie
d
Earth
Obs
erv
ati
ons a
nd
Re
mot
e
S
ensi
ng
,
vol
.
8
,
no
.
10
,
pp
.
4720
-
4728
,
2015
.
[30]
Jiang,
Ping,
Jonatha
n
W
inkl
e
y
,
Can
Zha
o,
Robe
rt
Munnoch,
Gey
ong
Min,
and
La
ure
n
ce
T
.
Yang
,
"A
n
int
el
li
g
e
nt
informati
on
for
warde
r
for
h
ea
l
t
hca
re
big
da
ta
s
y
stems
with
dist
ribut
e
d
we
ara
b
le
sensors
,
"
IEE
E
systems
journal
,
vol.
10
,
no
.
3
,
pp
.
1147
-
1159
,
20
14
.
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.
10
, No
.
4
,
A
ugus
t
2020
:
4322
-
4330
4330
[
3
1
]
M
a
n
u
j
a
k
s
h
i
B
.
C
.
,
R
a
m
e
s
h
K
.
B.
,
“
S
D
a
a
S
:
F
r
a
m
e
w
o
r
k
o
f
S
e
n
s
o
r
D
a
t
a
a
s
a
S
e
r
v
i
c
e
f
o
r
L
e
v
e
r
a
g
i
n
g
S
e
r
v
i
c
e
s
i
n
I
n
t
e
r
n
e
t
o
f
T
h
i
n
g
s
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
E
m
e
r
g
i
n
g
R
e
s
e
a
r
c
h
i
n
C
o
m
p
u
t
i
n
g
,
I
n
f
o
r
m
a
t
i
o
n
,
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
p
p
.
3
5
1
-
3
6
3
,
2
0
1
6
.
[32]
M
a
n
u
j
a
k
s
h
i
B
.
C
.
,
R
a
m
e
s
h
K
.
B.
,
“
A
N
o
v
e
l
E
x
p
e
r
i
m
e
n
t
a
l
P
r
o
t
o
t
y
p
e
f
o
r
A
s
s
e
s
s
i
n
g
I
o
T
P
e
r
f
o
r
m
a
n
c
e
o
n
R
e
a
l
-
T
i
m
e
A
n
a
l
y
t
i
c
s
,”
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
O
n
-
l
i
n
e
C
o
n
f
e
r
e
n
c
e
,
p
p
.
4
6
-
5
5
,
2
0
1
8
.
BIOGR
AP
HI
ES OF
A
UTH
ORS
Manujak
s
hi
B
.
C
.
is
working
as
assistant
profe
s
sor
in
the
dep
artm
ent
of
computer
scie
n
ce
and
engi
ne
eri
ng,
pre
sidency
univ
ersi
t
y
,
Beng
al
uru
,
Karna
ta
k
a,
Indi
a.
She
h
as
co
m
ple
te
d
B
.
E
in
computer
sci
ence
and
engi
ne
eri
n
g
from
SJ
MIT,
Chit
rad
urg
a,
Ka
rna
ta
k
a,
Ind
ia,
a
nd
M.
Tech
i
n
computer
scie
nc
e
and
engi
ne
eri
n
g
from
UBD
TC
E,
Dava
nger
e,
K
arn
ataka
.
Her
are
as
of
int
ere
st
are
B
ig
d
at
a
and
cl
oud
computing
technique.
K.
B.
Ramesh
,
As
socia
te
Profes
sor
and
Hea
d,
Depa
rtment
of
El
e
ct
roni
cs
and
Instrum
ent
at
ion
Engg.
R
V
col
l
e
ge
of
Eng
i
ne
ering,
Beng
al
uru,
Karna
ta
k
a,
Indi
a.
Heh
as
compl
et
ed
PhD
in
Com
pute
r
Scie
n
ce
and
Engi
ne
er
ing
from
Kuvem
pu
Univer
sit
y
.
H
e
hasa
round
twe
nt
y
-
three
y
ears
(23)
of
teac
hing
expe
ri
ence
in
E
&I
Engg.
His
m
aj
or
res
ea
rch
ar
e
a
is
in
Com
puter
Scie
nc
e
and
Engi
ne
eri
ng
an
d
m
inorre
sea
r
ch a
rea
is i
n
Biom
ed
ic
a
l
Eng
ineeri
ng
/Bi
oinformatics
.
Evaluation Warning : The document was created with Spire.PDF for Python.