Int
ern
at
i
onal
Journ
al of Ele
ctric
al
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
9
, No
.
5
,
Octo
ber
201
9
, pp.
3759
~3
771
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v9
i
5
.
pp3759
-
37
71
3759
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Trust m
odel g
enetic node
reco
very based
on cloud
theory f
or
un
derwater acou
stic sens
or netwo
rk
Buddes
ab
,
Th
ri
veni J
,
Venugopal
K
R
Depa
rt
m
ent
o
f
C
om
pute
r
Scie
n
ce a
nd
Engi
n
ee
rin
g
,
Univer
si
t
y
Vi
sv
esva
ra
y
a
Coll
e
ge
of Engin
ee
rin
g,
Banga
lor
e
Univ
e
rsit
y
,
Ind
i
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ja
n 1
1
, 2
01
9
Re
vised
A
pr 8
,
201
9
Accepte
d
Apr
1
6
, 201
9
Underwat
er
Ac
oustic
Sensor
Networks
[UA
SN
s
]
are
be
co
m
ing
a
ve
r
y
growing
rese
arch
topi
c
in
the
f
i
el
d
of
W
SN
s.
U
AS
Ns
are
har
m
ful
b
y
m
an
y
at
t
ac
ks
su
ch
as
Jam
m
ing
at
ta
cks
at
the
ph
y
si
ca
l
lay
er
,
Coll
ision
att
ac
ks
at
th
e
dat
a
l
ink
lay
e
r
a
nd
Dos
at
ta
cks
at
the
n
et
work
l
a
y
er.
UA
SN
s
has
a
unique
cha
ra
cteri
sti
c
such
as
unre
li
able
comm
unic
at
ion,
m
obil
ity
,
and
computat
ion
of
under
wat
er
sensor
net
work.
Bec
ause
of
th
is
the
tr
adition
al
sec
ur
i
t
y
m
ec
hani
sm
,
e.
g.
cr
y
ptogr
aphic,
enc
r
y
pt
ion,
aut
h
oriz
a
ti
on
and
authentica
t
ions
are
not
suit
abl
e
for
UA
SNs.
Man
y
trust
m
ec
h
an
ism
s
of
T
W
SNs
[Te
rre
str
ia
l
W
ire
le
ss
Sensor
Networks
]
had
proposed
to
U
AS
Ns
and
fai
led
to
provide
sec
u
rity
for
UA
SN
s
envi
ronm
e
nt,
due
to
d
y
na
m
ic
net
work
struct
ure
an
d
wea
k
l
ink
conn
ec
t
ion
be
twee
n
sensors
.
In
thi
s
pape
r,
a
novel
Trust
Mode
l
Gene
tic
Algorithm
base
d
on
Cloud
The
or
y
[T
MC]
for
UA
S
N
s
has
bee
n
proposed.
Th
e
TMC
-
GA
suggested
a
gen
et
i
c
node
re
cove
r
y
al
gorit
hm
to
improve
the
T
MC
net
work
in
te
rm
s
of
bette
r
net
work
l
ifeti
m
e,
residual
ene
rg
y
and
to
tal
energ
y
consu
m
pti
on.
Also
en
sures
tha
t
senso
r
nodes
ar
e
par
ticipating
in
the
rer
ou
ti
ng
in
the
routi
ng
d
isc
over
y
and
per
fo
rm
s
well
in
te
rm
s
of
succ
essful
pa
ck
et
deli
ver
y
.
Sim
ulatio
n
result
provid
es
that
th
e
proposed
TMC
-
Gene
tic
node
re
cove
r
y
al
gori
th
m
outpe
rform
s
compare
d
to
othe
r
relate
d
wo
rks
in
te
rm
s
of
t
he
num
ber
of
h
ops,
end
-
to
-
end
del
a
y
,
tot
a
l
ene
rg
y
consum
pti
on,
residu
al
en
erg
y
,
routi
ng
ov
erh
ea
d
and
ne
twork
li
f
et
ime
.
Ke
yw
or
d
s
:
Cl
oud
T
heory
Mod
el
Data s
ec
ur
it
y
T
W
S
N
UASN
W
i
reless
Senso
r
N
et
w
or
k
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
:
Buddesa
b,
Dep
a
rtm
ent o
f C
om
pu
te
r
Scie
nce a
nd E
ng
i
ne
erin
g,
Un
i
ver
sit
y Vis
ves
var
ay
a C
ollege
of Engine
e
rin
g,
Ba
ng
al
or
e
,
Ba
ng
al
or
e
Unive
r
sit
y, 560001
-
I
nd
ia
.
Em
a
il
:
ton
nur
21@
gm
ai
l.co
m
1.
INTROD
U
CTION
UASNs
are
s
pe
ci
al
ly
design
e
d
to
operate
a
nd
m
on
it
or
unde
r
wate
r
suc
h
as
the
oc
ean
,
la
ke,
et
c
a
nd
com
m
un
ic
at
ion
fr
e
qu
e
ncy
of
operati
on
is
be
tween
10
Hz
an
d
1
M
Hz
UASNs
ha
ve
un
i
qu
e
c
har
act
erist
ic
s
su
c
h
as
ac
ousti
c
cha
nn
el
c
om
m
un
ic
at
ion
,
w
eak
li
nk
c
onne
ct
ion
,
dynam
ic
netw
ork
,
a
nd
m
ob
il
i
ty
.
Du
e
to
this
pro
vid
in
g
sec
uri
ty
to
UASNs
is
no
t
as
easy
as
T
W
S
Ns
.
I
n
U
ASN,
net
work
struc
ture
op
e
rates
fir
st
unde
r
water
a
nd
se
nds
t
he
in
form
a
ti
on
to
the
loc
al
sta
ti
on
a
nd
then
f
orward
t
o
the
ba
se
sta
t
ion
[
1].
U
nder
water
acoust
ic
sens
or
netw
ork
(UAS
N)
deals
with
the
propagati
on
of
un
derwate
r
s
ound
an
d
interact
ion
of
m
echan
ic
al
wa
ves
that
ge
nerat
e
no
ise
with
water.
A
UAS
N
has
m
any
s
m
al
l
sensor
no
des
w
hich
are
placed
unde
rw
at
er
th
at
relat
ively
coope
rate
in
ga
therin
g
i
nform
at
ion
ab
ou
t
the
en
vir
onm
e
nt
an
d
sat
isfyi
ng
th
e
chall
eng
i
ng
re
qu
i
rem
ents.
Aco
us
ti
c
W
ir
el
ess
Sensor
N
et
w
ork
co
ns
is
ts
of
sensors
an
d
ve
hicle
s
wh
ic
h
f
or
m
an
auto
no
m
ou
s
ne
twork
f
or
c
ol
la
borati
ve
m
on
it
or
i
ng
ta
sk
s
.
The
a
ppli
cat
ion
s
of
U
AS
N
are
Mo
nito
rin
g
heav
y
instru
m
ents,
con
t
ro
ll
in
g
wea
ther
poll
utio
n,
Cl
i
m
at
e
Re
p
or
ti
ng,
natural
disturbance
s
causes,
pre
dicti
on,
Ma
rine
Lif
e
St
ud
y,
O
cea
n su
r
veill
ance a
nd di
saste
r
preve
nti
on [2].
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.
5
,
Oct
ober
20
19
:
3
7
5
9
-
3
7
7
1
3760
W
i
reless
sen
s
or
netw
orks
a
re
sens
or
node
s
wh
ic
h
are
sp
at
ia
ll
y
distri
bu
te
d
over
a
netw
ork
that
colla
borate
with
eac
h
oth
e
r
t
o
c
ollec
t
inf
orm
at
ion
f
ro
m
one
node
to
a
nothe
r
node.
W
SN
s
ha
ve
a
qual
it
y
to
sense
ph
ysi
cal
conditi
on,
e.
g.
weathe
r
te
m
per
at
ur
e,
pr
e
ssu
re
,
a
nd
sound.
WSNs
c
o
-
operati
vely
colle
ct
detect
ed
inf
or
m
at
ion
to
the
local
sta
ti
on
an
d
pass
to
t
he
central
sta
ti
on
.
WSNs
arc
hitec
ture
in
vo
l
ves
the
OSI
m
od
el
wh
ic
h
i
nclu
des
t
he
a
ppli
cat
ion
la
ye
r,
t
ran
s
port
la
ye
r
,
net
wor
k
la
ye
r
,
data
li
nk
la
ye
r,
a
nd
ph
ysi
cal
la
ye
r.
WSNs
ar
e
wi
dely
us
e
d
in
t
he
m
any
areas
su
c
h
as
ci
vili
an,
i
ndus
try
,
a
nd
arm
y,
and
so
m
e
of
the
WSN
s
app
li
cat
io
ns
are
Fire
Detec
ti
on
,
I
ndus
t
rial
Pr
ocess
M
onit
or
i
ng,
Mo
ni
toring
Tra
ff
ic
and
Hea
lt
h
care
m
on
it
or
ing, etc
[3].
Cl
oud
the
or
y
m
od
el
is
a
cl
oud
te
ch
no
l
og
y
firm
sp
eci
al
iz
ing
base
d
on
cl
oud
c
om
pu
ti
ng
co
nce
pts.
Cl
oud
theo
ry
con
ce
pt
r
un
on
the
internet
as
a
base
platfo
rm
to
pr
ovi
de
ser
vice
to
a
us
er
from
s
erv
e
rs.
Cl
oud
m
od
el
m
ai
ntains
central
rem
ote
sta
tio
ns
to
sa
ve
data
and
ap
plica
ti
on
a
nd
m
anage
the
data
to
th
e
us
er
.
Cl
oud
m
od
el
pr
ovide
se
rv
ic
es
to
a
us
er
on
dem
and
on
the
inter
net
with
th
e
secur
e
net
work
acce
ss
f
or
a
sh
are
of
c
onfig
urabl
e
and
c
om
puti
ng
res
ources
su
ch
a
s
ser
ver
s
,
net
wor
ks,
da
ta
sto
ra
ge
,
ap
plica
ti
ons,
an
d
app
li
cat
io
n
se
r
vices. Clo
ud M
od
el
pro
vid
e
lo
wer o
per
at
in
g cost a
nd m
ai
nt
ai
n
ace c
os
t t
o
pro
vid
er
s [4].
Terr
est
rial
W
i
r
el
ess
S
ens
or
N
et
work
s
[T
WSNs]
are
sp
eci
al
ly
design
e
d
to
pro
vid
e
ser
vic
es
based
on
la
nd
.
T
WSNs
inclu
de
'
ai
r'
h
as
the
pr
im
ary
co
m
m
un
ic
at
ion
channel
to
suppl
y
con
nected
i
nfor
m
at
ion
fro
m
on
e
node
to
a
nothe
r.
T
W
S
Ns
m
a
inly
con
sist
of
send
in
g
an
d
r
ecei
vin
g
dev
ic
es
to
send
an
d
receive
inf
orm
at
ion
and
ca
rr
ie
s
da
ta
or
voic
e
in
form
ation
via
el
ect
ro
m
agn
et
ic
rad
io
wa
ve
s.
The
ty
pica
l
tradit
ion
al
s
ecur
it
y
m
echan
ism
s
hav
e
bee
n
s
uggested
to
pro
vid
e
sec
ur
it
y
to
Terre
stria
l
W
i
reless
Se
nsor
Netw
ork
s
uch
a
s
Crypto
gr
a
phy,
Au
t
hen
ti
cat
io
n, Co
nf
i
den
ti
al
it
y, an
d Av
ai
la
bi
li
ty
.
A
Tr
us
t
Ma
na
gem
ent
Syst
e
m
(TMS)
has
been
rece
ntly
su
ggest
e
d
to
pro
vid
e
e
ne
rg
y
eff
ic
ie
nt
a
nd
handle
at
ta
cks
in
m
any
la
ye
rs
and
im
pr
ov
e
d
node
co
op
e
rat
ion
in
s
uccessful
pack
et
deli
ve
ry,
local
iz
at
ion
of
node.
I
n
the
T
r
us
t
Ma
na
gem
e
nt
Schem
e
for
sens
or
netw
ork
s
secu
rity
(TMS),
t
he
f
orward
ing
node
is
c
hoose
s
base
d
on
t
he
be
h
avi
or
of
ne
ig
hbor
no
des,
a
nd
the
direct
tr
ust
com
pu
ta
ti
on
is
perf
or
m
ed.
The
TM
S
m
ain
ta
ins
a
databa
se
at
each
node
le
ve
l
to
preser
ve
the
tr
us
t
.
T
he
routin
g
m
echan
ism
can
perform
back
a
nd
fort
h
pro
pag
at
io
n
be
cause
of
wh
ic
h
there
is
m
assive
a
m
ou
nt
of
energy
loss
and
al
so
si
nce
the
node
delive
rs
the
pack
et
t
he dire
ct
tru
st
will
also
get inc
rem
ented
w
hich
is in
correct
.
The
TM
S a
ppr
oach ha
s the
fo
ll
ow
in
g disad
va
ntages
,
a.
The
tr
us
t
c
ompu
ta
ti
on
dep
e
nds
on
t
he
E
xpect
ed
Tra
ns
m
i
ssion
C
ount
(E
TC)
w
hich
is
t
he
di
rect
tru
st
f
or
the no
des.
b.
Ther
e
is a l
ot of e
nergy wa
sta
ge du
e
to bac
k and f
or
t
h pro
pa
gation i
nv
olve
d
in
the
r
ou
ti
ng
path.
c.
The
t
ru
st
m
echan
ism
do
es
not
ta
ke
i
nto
co
ns
i
der
at
io
n
t
he
bit
er
ror
rate
on
the
physi
cal
m
e
diu
m
wh
ic
h
is
a
sign
ific
a
nt im
pact in the
un
de
rw
at
e
r
envir
on
m
ent.
d.
The
TMS
al
go
rithm
trust
m
e
asur
e
is
not
go
od
beca
us
e
ab
norm
al
pack
et
forw
a
r
ding
is
cause
d
by
not
only
m
al
ic
iou
s
at
ta
cks
bu
t
al
s
o
un
r
el
ia
ble
acou
sti
c
cha
nn
el
a
nd
un
ce
rtai
n
unde
rw
at
er
en
vir
onm
ent,
e.g.,
oce
an
currents
.
Nodes
i
n
the
w
irel
ess
sen
s
or
netw
ork
a
re
pron
e
to
fail
ur
e
du
e
to
e
nergy
dep
le
ti
on,
ha
rdwar
e
fail
ure,
and
c
omm
un
ic
at
ion
li
nk
e
rrors,
et
c.
"Fault
node
recove
ry
al
gorithm
"
dete
ct
s
a
fau
lt
node
wh
e
n
s
om
e
s
ens
or
nodes
s
hutd
own
due
to
lo
w
ba
tt
ery
ener
gy.
The
ge
netic
al
gorithm
is
us
ed
to
rec
over
th
e
dep
le
te
d
node
in
the
netw
ork.
The
node
de
pleti
on
m
ay
be
due
to
no
avail
able
po
wer,
or
it
m
ay
reach
the
t
hr
es
ho
l
d
le
vel.
T
he
m
a
in
aim
to
identif
y
the
dea
d
nodes
a
nd
the
n
f
ind
few
no
des
to
be
re
place
d
am
on
g
t
hem
so
t
hat
the
c
os
t
of
m
ai
ntenan
ce
i
s
reduce
d
an
d
ov
e
rall
netw
ork
li
feti
m
e
is
per
i
od
ic
al
ly
increased
.
D
ur
i
ng
TMC
trust
m
od
el
evo
l
ution,
t
he
fo
ll
owin
g
t
wo
con
ce
r
ns
hav
e
been
ta
ke
n
i
nto
c
onsiderati
on
to
pr
ov
i
de
a
bette
r
QoS
to
netw
ork
in term
s o
f
t
he netw
ork
li
feti
m
e and
resid
ua
l ener
gy
a.
The
n
et
w
ork
li
fetim
e
has
ta
ken
into
acc
ount
to
i
m
pr
ove
the
ove
rall
li
fetim
e
of
the
syst
e
m
by
prov
i
di
ng
eff
ic
ie
nt
ene
r
gy
power
to
t
he
netw
ork
no
de
s
so
that
they
can
pa
rtic
ipate
in
the
r
ou
ti
ng
and
it
dec
reas
es
the d
ea
d n
od
e
s
in
the
n
et
wor
k.
b.
The
re
sid
ual
energy
of
the
ne
twork
will
be
m
ai
ntained
w
el
l
in
the
TMC
m
od
el
by
pro
ving
s
uffici
ent
energy
to
node
s
wh
ic
h
are
suffe
rin
g
f
ro
m
insu
f
fici
ent
ene
r
gy
powe
r
to
pa
rtic
ipate
in
the
routin
g
an
d
da
ta
com
m
un
ic
at
ion
.
We
propose
a
TMC
ge
netic
node
rec
ov
e
r
y
m
od
el
to
im
pro
ve
TMC
m
od
el
i
n
te
rm
s
of
netw
ork
li
fetim
e,
the
residu
al
e
nergy
of
netw
ork,
QoS
an
d
al
so
reduces
dead
nodes
as
wel
l.
The
co
ntri
bu
t
ion
s
a
s
fo
ll
ows.
T
he
pro
po
se
d
m
od
el
first
identif
ie
s
the
fau
lt
y
wh
ic
h
a
re
be
l
ow
the
de
fi
ned
th
res
ho
l
d
le
vel.
In
the n
et
w
ork
after
so
m
e
it
er
at
ion
an
d
data com
m
un
ic
at
ion
betwee
n
s
our
ce
and
de
sti
nation
ap
ply
the
ge
netic
process
-
sel
ect
ion,
cr
os
s
over
,
m
utatio
n
to
t
he
nodes.
The
n
it
rec
ov
e
rs
t
he
fa
ulty
node
s
with
t
he
re
quire
d
energy
w
hose
values
a
re
on
e
from
the
gen
et
ic
process.
He
nce
the
rec
ove
red
nodes
ca
n
be
use
d
for
r
er
ou
ti
ng
in
the
netw
ork
an
d
al
s
o
ca
n
pa
rtic
ipate
in
the
data
pac
ket
tran
sm
issio
n
from
so
urc
e
to
desti
natio
n
th
us
i
m
pr
ove
netw
ork
r
esi
du
al
e
ne
rg
y a
nd
netw
ork
li
feti
m
e.
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
Tru
st
m
od
el
ge
netic
nod
e
rec
overy
based
on
cloud t
heory f
or
un
derw
ate
…
(
Buddes
ab
)
3761
The
rem
ai
nd
er
of
this
pap
e
r
is
org
a
nized
as
f
ollow
s:
In
Se
ct
ion
2
,
w
e
rev
ie
w
the
r
el
at
ed
w
ork
.
In
Sect
ion
3
,
the
syst
em
mo
del
a
nd
desi
gn
a
re
pr
ese
nt
ed.
Sect
i
on
4
pro
vid
es
a
det
ai
le
d
desc
ription
of
the
propos
e
d
TMC
-
G
A
tr
us
t
m
od
el
.
Sect
io
n
5
t
he
pro
pos
ed
al
gorithm
s.
Sect
io
n
6
pr
esents
t
he
sim
ulati
on
resu
lt
s a
nd r
e
la
te
d
pe
rfo
rm
ance an
al
ysi
s.
Fi
na
ll
y, con
cl
us
io
ns
a
re
dr
a
w
n
i
n Sect
ion 7
.
2.
RELATE
D
W
ORK
Gan
e
riwal
et
.al
.
[
5]
ha
ve
anal
yz
ed
Re
puta
ti
on
-
base
d
F
ram
e
work
f
or
im
plem
enting
s
om
e
trustwo
rthy
sens
or
nodes
a
t
ru
ntim
e
based
on
the
nei
ghbor
no
de
be
ha
vior.
Ya
o
et
al
.
[6
]
ha
ve
in
ve
sti
gated
the
s
ens
or
netw
ork
secu
ri
ty
with
trust
m
anag
em
ent
m
et
ho
ds.
"PL
US
:
Param
et
erized
and
L
ocal
iz
ed
trUst
m
anag
em
en
t
Schem
e"
fo
r
protect
ion
of
sens
or
netw
or
k
to
pro
vid
e
us
er
tru
ste
d
e
nv
i
ronm
ent
ha
s
bee
n
propos
ed
to
gu
a
ra
ntee
the
netw
ork
sec
ur
i
ty
issues
eff
ic
ie
ntly
.
Feng
e
t
al
.
[7
]
ha
ve
integrate
d
the
appr
oach
of
node
s
beh
a
vior
strat
e
gies
an
d
m
od
if
ie
d
evide
nce
theo
ry.
Va
rio
us
trust
factor
s
a
nd
c
oeffici
ent
are
co
ns
ide
red
to
get
trust
val
ues
of
direct
an
d
i
ndirect
trust.
NB
BTE
us
e
d
F
uz
zy
log
ic
m
et
ho
ds
are
us
e
d
to
fetch
the
node
trust
le
vels and
D
-
S
evide
nce t
heor
y base
d on the
le
vels of tr
us
t
worthi
ness
Re
n
et
al
.
[8
]
ha
ve
handled
the
in
form
at
i
on
relat
ed
to
t
ru
st
with
a
se
cur
e
an
d
e
ff
ic
i
ent
way
for
Un
at
te
nde
d
W
i
reless
Senso
r
Netw
ork
(
U
W
SN
s
)
an
d
p
r
ov
i
de
a
tru
st
data
stora
ge
with
sc
al
able,
m
or
e
eff
ic
ie
nt
and
tr
u
st
ge
nerat
ion
.
Bo
ukerc
he
et
al
.
[9
]
hav
e
intr
oduce
d
a
trust
and
re
puta
ti
on
m
anag
e
m
ent
schem
e
s
with
the
m
ob
il
e
agen
ts
run
ning
on
each
node.
A
gen
t
-
base
d
Tr
ust
and
Re
puta
ti
on
Ma
na
gem
e
nt
schem
e
(
ATRM)
syst
e
m
cal
cula
te
s
trust
an
d
re
pu
ta
ti
on
us
i
ng
the
m
ob
il
e
agen
t.
Shai
kh
et
al
.
[10]
hav
e
e
va
luate
d
the
gro
up
of
sens
or
nodes
trust
that
pro
vi
des
m
et
ho
ds
to
identify
the
at
ta
cker
'
s
m
ali
ci
ou
s
se
nsor
node
s
al
ong
wit
h
sm
all
pr
e
ve
ntion m
e
chan
ism
d
eg
re
e.
Ha
n et
al.
[
11
]
hav
e
prop
os
e
d Att
ack
-
Re
sist
a
nt Trust m
od
el
b
ase
d
on M
ulti
di
m
ension
al
tr
us
t M
et
rics
(A
RTM
M)
th
a
t
ensures
a
m
o
re
eff
ic
ie
nt
an
d
reli
able
U
ASNs
net
work.
D
om
ing
o
et
al
.
[
12
]
ha
ve
intr
oduce
d
Ra
y
-
theo
ry
-
ba
sed
m
ulti
path
Ra
yl
ei
gh
unde
rw
at
er
m
od
el
for
ef
fici
ent
da
ta
com
m
un
icati
on
in
U
nde
rw
at
e
r
W
i
reless
C
omm
un
ic
at
ion
Ne
tworks
(
U
W
C
N)
.
S
ound
pr
opagati
on
cal
c
ul
at
ion
,
tra
ns
m
i
ssion
l
os
ses
be
tween
transceive
rs
a
nd
in
vestigat
e
d
bit
erro
r
rate
ef
fects
has
bee
n
c
onside
red
to
i
m
ple
m
ent
effe
ct
ive
com
m
un
ic
at
ion
.
Lim
et
a
l.
[1
3]
hav
e
intr
od
uced
a
syst
em
at
ic
m
et
ho
d
f
or
sens
or
net
work
cal
le
d
"Pr
ovenan
ce
base
d
Tr
us
tw
ort
hin
es
s
Asse
ss
m
ent"
adopted
wh
ic
h
ge
ner
at
es
a)
data
it
em
trust
sco
re
val
ues
b)
eval
uation
of
netw
ork n
od
e
tru
st s
co
re
valu
e.
Jia
ng
et
al
.
[14
]
hav
e
desc
ri
be
d
pro
vid
i
ng
tr
us
t
in
the
W
S
Ns
an
d
Ef
fici
ent
Distrib
uted
Trust
Mod
el
(EDTM
)
eval
ua
te
s
the
trustwort
hin
ess
of
se
ns
or
nodes
m
or
e
preci
sel
y.
Su
bject
ive
lo
gic
prob
a
bili
ty
den
sit
y
functi
on
c
on
si
der
e
d
ass
essin
g
direct
c
omm
un
ic
at
ion
tr
us
t,
data
pa
ck
et
trust,
li
n
k
com
m
un
ic
at
ion
tr
us
t,
netw
ork
or
no
de
e
nergy
tr
ust
.
Jia
ng
et
al
.
[
15
]
hav
e
prov
i
ded
the
saf
et
y
of
U
nd
e
r
water
M
ob
il
e
Senso
r
Netw
ork.
This
researc
h
i
nclud
e
s
the
neig
hbor
nodes
m
on
it
or
l
os
s
a
nd
error
of
data
pack
et
s
,
se
ns
or
no
de
energy
c
onsu
m
ption,
T
r
us
t
le
v
el
are
c
on
si
de
red
to
e
val
uate
the
tr
us
t
e
vid
e
nce
a
nd
direct
trust,
i
ndirect
Trust
,
histor
ic
al
tr
us
t,
reco
m
m
end
at
ion
tru
st,
data
t
ru
st
al
s
o
been
stud
ie
d
f
or
tr
ust
cal
culat
ion
.
Ju
el
on
g
et
al
.
[16]
hav
e
intr
oduce
d
L
ARP
us
e
d
to
m
ini
m
iz
e
the
netw
ork
dela
y
in
te
rm
s
of
t
ran
sm
issi
on
,
e
ns
uri
ng
le
ss
e
nerg
y
consum
ption
in
the
netw
ork
and
pro
vid
e
bette
r
netw
ork
com
m
un
ic
ati
on
qu
al
it
y.
Chen
et
al
.
[17
]
hav
e
pro
po
se
d
to
i
de
ntify
at
ta
cked
m
al
iciou
s
se
nsor
nodes
i
n
va
rio
us
eve
nts.
It
detect
s
m
a
li
cio
us
sens
or
no
de
s
and
no
ti
f
ie
s t
he net
work al
on
g wit
h diff
e
re
nt ev
e
nt tr
us
t rati
ng.
Qu
et
al
.
[
18]
hav
e
intr
oduce
d
the
im
pr
ov
e
m
ent
of
t
ru
st
m
echan
ism
s
with
the
def
e
nding
m
et
ho
ds.
"
W
at
ch
dog
a
nd
Tr
us
t
m
echan
ism
"
has
bee
n
pr
opos
e
d
to
analy
ze
the
sec
ur
it
y
of
net
wor
k
f
ro
m
the
insi
de
a
nd
"Def
e
ndin
g
m
et
hods
"
us
ed
to
analy
ze
tr
ust
m
echan
ism
and
watc
hdog
in
te
rm
s
of
netw
ork
li
fet
i
m
e.
Han
et
al
.
[
19]
hav
e
i
ntrod
uc
ed
to
e
nsure
locat
ion
sec
uri
ty
of
the
netw
ork.
This
rese
arch
pro
vid
es
node
locat
ion
i
n
te
r
m
s
of
good
ac
cur
acy
a
nd
bet
te
r
reli
abili
ty
fr
om
the
def
i
ne
d
tr
us
t
m
od
el
wh
ic
h
m
ini
m
i
ze
the
unknow
n
node
s
local
iz
at
ion
error
a
nd
e
nh
ances
lo
cal
iz
at
ion
acc
ur
acy
.
Kim
e
t
al
.
[2
0]
hav
e
pro
pos
ed
to
analy
ze
prop
e
r
trust
m
od
el
s
with
the
ri
gh
t
com
m
un
ic
at
ing
an
a
ppro
a
ch
to
get
the
reli
able
path
f
r
om
so
ur
ce
node
to
destin
at
ion
no
de.
T
he
fo
ll
owin
g
m
et
hods
are
c
on
sidere
d
to
im
ple
m
ent
trust;
t
hey
are
the
de
gr
ee
of
each
node
is
cal
culat
ed
to
decide
w
hethe
r
to
c
omm
un
ic
at
e
or
not
a
nd
tr
us
t
le
vel
of
the
sens
or
node
is
est
i
m
at
ed
to com
m
un
ic
at
e
within t
he
tra
ns
m
i
ss
ion ra
nge.
Ba
ti
sh
et
al
.
[21]
hav
e
pro
pos
ed
"Ef
fici
ent
Weig
hted
T
ru
s
t
Evaluati
on
S
yst
e
m
fo
r
W
ir
el
ess
Sensor
Netw
orks
(
WT
ES
-
WSN)”
t
o
sign
ific
a
ntly
increase
the
net
work
ef
fici
enc
y
and
m
ini
m
ize
the
tr
us
t
e
val
uation
cost.
Sa
rko
dee
et
al
.
[2
2]
ha
ve
desc
ribe
d
th
at
reputat
ion
an
d
tr
us
t
ar
e
div
ide
d
int
o
var
io
us
cl
ass
es
or
cat
egories.
Re
fer
e
ncin
g
m
et
ho
d
is
us
e
d
i
n
pro
vid
i
ng
resea
r
cher
s
with
a
to
ol
that
m
akes
it
easi
er
to
ref
e
r
ence
the
featu
res
on
reputat
ion
a
nd
tr
us
t
in
a
m
uch
ea
sie
r
wa
y
than
if
re
fere
ncin
g
ha
s
to
be
di
rected
to
severa
l
un
c
oor
di
nate
d resou
rces.
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.
5
,
Oct
ober
20
19
:
3
7
5
9
-
3
7
7
1
3762
So
la
nki
et
al
.
[
23
]
ha
ve
rev
ie
wed
the
existi
ng
data
sec
ur
it
y
arch
it
ect
ure
a
nd
key
m
anag
e
m
ent
syst
e
m
in W
SN.
T
he
pro
posed
sec
ur
it
y
te
chn
i
qu
e
off
ers
a
f
ru
it
f
ul
s
ol
ution
to
sec
ur
e
data
tra
ns
m
issi
on
am
on
g
t
he
low
batte
ry li
fe a
nd ti
ny d
e
vices,
use
d
i
n WS
N due t
o
it
s lo
w
ti
m
e an
d
s
pace c
om
plexiti
es.
3.
SY
STE
M MO
DEL A
N
D DESIGN
3.1
.
S
ys
te
m m
od
el
UASNs
C
onsist
s
of
a
m
ulti
-
hop
ho
m
og
e
ne
ou
s
netw
ork
t
hat
is
init
ia
ll
y
al
l
the
underwate
r
se
ns
or
nodes
ha
ve
an
equ
al
am
ount
of
i
niti
al
ener
gy
le
vel,
the
sa
m
e
capab
il
it
y
of
c
omm
un
ic
ation
,
c
om
pu
ta
ti
on,
a
nd
stora
ge.
I
n
the
U
ASN,
t
her
e
are
n
ti
ny
se
nsor
no
des,
de
note
d
by
∈
S
,
where
=
{
}
=
1
.
Each
se
ns
or
node
si
is
ra
ndom
l
y
dep
loye
d
at
the
po
sit
io
n
with
the
com
m
un
ic
at
ion
range
r.
T
he
tw
o
neig
hbor
node
s
can
directl
y
com
m
un
ic
at
e w
it
h
ea
ch
oth
e
r
on
ly
w
he
n
t
he
distance d
(
,
)
betwe
en
tw
o
neig
hb
or
no
des
si
and
s
j
sat
isfie
s d(
,
)
≤ r
.
In
t
he
U
ASNs
netw
ork
c
on
s
iderin
g
3
ki
nds
of
node
,
the
y
are
trust
or
,
trustee
a
nd
rec
omm
end
e
d.
Wh
e
n
a
sens
or
node
A
wa
nts
to
get
the
trust
of
an
oth
e
r
sensor
no
de
B,
the
eval
uated
node
B
is
nam
e
d
as
a
trustee
no
de.
Ba
sed
on
com
m
un
ic
at
ion
be
hav
i
or
s
betwe
en
nei
ghbor
node
s,
a
trust
or
can
judge
w
he
ther
a
trustee
is
tr
us
t
worthy
or
not.
If
t
her
e
a
re
direct
com
m
un
ic
at
ion
s
betwee
n
trust
or
a
nd
tr
us
te
e,
direct
tr
us
t
is
cal
culat
ed.
I
ndirect
trust
is
com
pu
te
d
based
on
oth
e
r
sens
or
nodes
rec
om
m
end
at
ions.
T
he
sens
or
node
wh
ic
h
is pro
vid
in
g re
com
m
end
at
io
ns i
s n
am
ed
as a
r
ec
omm
end
er
.
3.2
.
S
ys
te
m d
esi
gn
Durin
g
Softwa
re
dev
el
op
m
ent,
Syst
em
Design
is
c
onside
r
ed
a
crit
ic
al
a
nd
esse
ntial
phase
of
the
dev
el
op
m
ent
phase.
The
desi
gn
pa
rt
of
the
grow
t
h
will
ha
ve
m
or
e
lo
gical
data
an
d
the
oret
ic
al
ideas
to
fu
l
fill
the
requir
em
ents
of
f
unct
ion
a
l
and
no
n
-
func
ti
on
al
re
qu
ire
m
ents
of
de
vel
op
m
ent.
Sm
all
syst
e
m
s
are
e
asy
t
o
desig
n
wit
h
m
i
nim
u
m
req
uire
m
ents
wh
e
reas
la
rg
e
syst
em
r
equ
i
red
t
o
inte
gr
at
e
with
sub
syst
e
m
s
serv
ic
es
to
sat
isfy req
uire
m
ents. S
yst
em
arch
it
ect
ure
will
b
e the
out
put
of syste
m
d
esi
gn in so
ftwa
re
dev
el
op
m
ent.
4.
TMC
-
G
A
T
R
US
T
MO
DEL
4
.1
.
O
bj
ec
tiv
es
In
t
he
TMC
m
et
hod,
t
he
f
or
wardin
g
node
befor
e
bei
ng
s
el
ect
ed
unde
r
goes
a
l
ot
of
te
s
t/
par
am
et
ers
m
easur
e
w
hich
is
resp
onsible
for
picki
ng
the
fo
r
wa
rd
i
ng
node.
It
ta
kes
int
o
co
ns
ide
rati
on
both
di
rect
trusts
as
well
as
in
direc
t
trust
in
t
he
tr
us
t
com
pu
ta
ti
on.
D
ur
in
g
t
he
t
ru
st
c
om
pu
ta
ti
on
pac
ket
los
s,
energy
c
on
s
um
pt
ion
and
t
he
pack
et
error
rate
is
ta
ken
w
hich
im
pr
oves
the
accu
racy.
Th
e
r
ou
ti
ng
proces
s
in
t
he
pr
opos
e
d
m
et
hod
m
ai
ntains
a
fie
ld
known
as
‘
Node
in
Ro
ute'
wh
ic
h
s
upport
s
the
inform
at
i
on
of
al
l
the
traver
se
d
no
des
duri
ng
the ro
uting p
rocess.
4.2.
T
MC
-
ge
netic n
od
e
rec
ov
er
y me
thod
ology
The
str
uctu
re
of
TMC
-
Ge
net
ic
No
de
recov
ery
is
show
n
i
n
Fig
ure
1
T
MC
-
GA
tru
st
com
pu
ta
ti
on
m
et
ho
dolo
gy c
on
sist
s
of th
ree
p
a
rts.
1
Trust
ev
i
den
ce
g
e
ner
at
io
n.
2
Direct t
r
us
t cal
culat
ion
.
3
Re
com
m
end
at
ion an
d
i
nd
i
rect t
ru
st cal
c
ulati
on.
4
Gen
et
ic
Node r
ecov
e
ry
Algori
thm
.
4.3.
Tr
us
t
e
vi
dence
gener
ati
on
In
c
urren
t
tru
s
t
m
od
el
s,
packet
loss
is
widel
y
us
ed
as
an
im
po
rtant
trust
m
et
ric.
Ho
we
ve
r,
getti
ng
sta
ti
sti
cs
li
ke
the
nu
m
ber
of
rep
li
ed
AC
K
pac
kets
as
trust
evide
nc
e
is
no
t
reli
able
in
an
unde
rw
at
e
r
env
i
ronm
ent.
A
no
vel
trust
m
od
el
nam
ed
TMC
,
in
w
hic
h
pac
ket
loss
i
s
analy
zed
la
ye
r
by
la
ye
r
to
avo
i
d
th
e
influ
e
nce
from
an
unreli
able a
coust
ic
ch
a
nn
e
l and dy
nam
ic
netw
ork
st
ru
ct
ur
e
.
4.3.1
.
Pac
ket
l
os
s
ba
se
d
trus
t
e
vidence
The
pac
ket
loss
is
ob
ser
ve
d
and
eval
uated
betwee
n
OSI
la
ye
rs
fr
om
the
ph
ysi
cal
la
ye
r
to
the
trans
port
la
ye
r.
The
aco
us
ti
c
channel
is
ch
aracte
rized
by
high
bit
er
ror
and
te
m
po
ra
ry
loss
of
co
nne
ct
ivit
y.
High
bit error
easi
ly
causes
s
erio
us
data p
ac
ket loss
, whic
h ca
n be calc
ulate
d
as
fo
ll
ows.
First, the
p
at
h
l
os
s
of
the aco
us
ti
c si
gn
al
is
obtai
ne
d
as:
(
,
)
=
(
)
(1)
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
Tru
st
m
od
el
ge
netic
nod
e
rec
overy
based
on
cloud t
heory f
or
un
derw
ate
…
(
Buddes
ab
)
3763
wh
e
re
is
the
transm
issi
on
di
sta
nce,
is
t
he
freq
ue
ncy
of
the
ac
ousti
c
sign
al
,
(
)
is
the
abs
orptio
n
coeffic
ie
nt
a
nd
0
is
a
norm
al
izati
on
c
onsta
nt,
is
t
he
e
nergy
pro
pag
at
io
n
c
oe
ff
ic
ie
nt
a
nd
in
ge
ner
al
,
f
or
t
he
sp
he
rical
s
pr
ea
ding,
=
2
.
Ba
sed
on
the
pa
th
loss
(
,
)
and
t
he
powe
r
s
pectr
al
den
sit
y
(
)
,
the
sign
al
-
to
-
noise
rati
o
(SNR)
of the ac
ousti
c
channel ca
n be
calc
ulate
d by
(
,
)
=
/
(
,
)
(
)
△
(2)
The
total
pow
er
sp
ect
ral
de
ns
it
y
dep
e
nds
on
the
fou
r
kinds
of
nois
es.
i.e
ocea
n
backg
rou
nd
no
ise
(tu
rbulen
ce),
s
hippin
g,
wav
e
s a
nd therm
al
n
oise [1
2].
1.
Ocea
n back
gro
und n
oise (
t
urb
ulence
noise
)
10 l
og N
t
(f) =
17
-
30 l
og f
2.
Sh
i
pp
i
ng noi
se
10
l
og N
s
(
f)
=
40
+
20(s
-
0.5
)
+
26 lo
g
f
-
60 log(f
+
0.0
3)
3.
Wa
ves
10lo
g N
w(
f)
=5
0+
7.5w
1/
2
+20 lo
g
f
40lo
g(f+0.4
)
4.
The
rm
al
n
oise
10 lo
g N
th
(f
)=
-
15
+
20 l
og f
Wh
ere
N
t
h(f)
=
ther
m
al
n
oise
Bi
nar
y
Ph
a
se
Sh
ift
Keyi
ng
(
BPSK
)
m
od
ul
at
ion
is
use
d
t
o
tra
ns
m
it
the
data
as
the
physi
cal
la
ye
r
m
edium
. H
enc
e the
packet
lo
ss can be
co
m
pu
te
d usi
ng,
P
loss
−
ph
y
=
(
1
−
1
2
er
fc
√
SN
R
)
b
(3)
w
he
re
P
loss
-
phy
=
path
l
os
s
for ph
ysi
cal
lay
er
,
b
=
num
ber
of
bits patt
er
n
Figure
1.
TMC
-
G
A
t
ru
st
com
pu
ta
ti
on m
et
hodo
l
og
y
4.3.2
.
M
A
C
l
ay
er
p
acke
t
lo
ss
The
M
AC Lay
er
pack
et
l
os
s c
an
be ob
ta
in
ed
as b
el
ow
(4
)
−
=
(4)
w
he
re
N
ack
=
Nu
m
ber
of A
C
K packets
,
N
s
end
=
N
um
ber
of
pack
et
s
sen
d
P loss
-
m
ac
=L
oss o
f
M
AC p
ac
kets
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.
5
,
Oct
ober
20
19
:
3
7
5
9
-
3
7
7
1
3764
4.3.3.
Tr
ansp
ort la
yer pack
et lo
s
s
The
Tra
nsport
la
ye
r
pack
et
lo
ss
is
treat
ed
as
zero
be
cause before
the
node
sen
ds
a
pack
et
to
the
node
it
d
ynam
ic
ally
determ
ines n
ei
ghbors
each
ti
m
e.
P loss
-
tra
n=0
(5)
4.3.4
.
Ne
twor
k la
yer
In
the
netw
ork
histor
ic
al
trac
e
is
ob
ta
ined
wh
ic
h
is
respo
ns
ible
f
or
fin
din
g
or
pr
e
dicti
ng
the
pac
ket
loss
with
t
he
com
pr
om
ise
d
no
de
s
us
i
ng
a
li
near
m
od
el
ov
er
a
pe
rio
d
f
or
a
tim
e
series
data
us
in
g
F
A
RIM
A
m
et
ho
d
[
24
].
T
he
e
qu
at
io
n gi
ves
t
he
total
pa
cket loss
P
l
o
s
s
= P
l
o
s
s
-
m
a
c
-
-
P
l
o
s
s
-
p
r
e
-
P
l
o
s
s
-
p
h
y
(6)
4.4. Tr
us
t
e
vidence
–
pa
c
ket
error
The
Tr
us
t
Evi
den
ce
dep
e
nds
upon
the
er
ror
rat
e
of
the
pa
cket.
The
pa
c
ket
error
rate
is
com
pu
te
d
by
check
i
ng the
dat
a it
e
m
an
d num
ber
o
f bit
s
w
hich we
re lo
st
ov
e
r
a
p
e
rio
d [
13
]
.
=
1
−
2
(
0
.
5
−
∫
(
)
)
(7)
Wh
e
re
v
d
is
the
com
pu
te
d
da
ta
it
e
m
nu
m
e
rical
.
x
is
th
e
sense
d
da
ta
it
e
m
s
m
ean
value
.
f
(
x)
is
t
he
dat
a
it
e
m
s p
robab
il
i
ty
d
ensity
fun
c
ti
on
.
(
)
=
1
√
2
−
(
−
)
2
2
2
(8)
4.5. Tr
us
t
e
vidence
–
e
nerg
y
cons
u
mpti
on
The
e
nergy c
onsu
m
ption
s
is
giv
e
n by the
f
ol
lowing
(9),
=
1
−
°
(9)
w
he
re
E
res=
res
idu
al
e
nergy
of nod
e
,
E
o
=i
niti
al
en
er
gy
The param
et
er to
ta
l ener
gy c
onsu
m
ption i
s gi
ven
a
s foll
ows
TE
c
=
∑
=
1
(10)
w
he
re
l
=
nu
m
ber
of
li
nks
,
Ei
=Energy c
onsum
ed
by the
i
th
li
nk
The
e
ne
rg
y
c
onsu
m
ed
by
the
th
i
li
nk
gi
ven
by,
w
her
e
E
nergy
re
qu
i
red
f
or
da
ta
transm
issi
on
is
m
or
e
than
t
he
e
nerg
y neede
d f
or
da
ta
g
ene
rati
on.
E
c
=2E
tx
+ E
amp
d
γ
(11)
w
he
re
E
tx
=
ene
rg
y
require
d for
data tran
sm
is
sion
,
E
amp
=ene
rg
y
require
d for
data g
e
nerat
ion
d
=dist
a
n
ce
between t
w
o
inte
r
m
ediat
e n
ode
,
γ
=en
vir
on
m
ent f
act
or
0.1≤γ≤
1
4.6. Tr
us
t
rec
ommen
datio
n
comput
at
i
on
The
tr
us
t
rec
om
m
end
at
ion
s
are
com
pu
te
d
base
d
on
the
f
ollow
i
ng
m
easur
e
.
I
f
m
they
rep
rese
nt
the
nu
m
ber
of n
ei
ghbo
rs,
t
hen the
r
es
ul
ti
ng
t
ru
st
s ar
e cal
c
ulate
d
[14].
E
x
=
1
∏
=
1
En=min
{
1
∑
,
1
}
=
1
Ee=m
i
n(
1
∑
,
1
)
=
1
(12)
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
Tru
st
m
od
el
ge
netic
nod
e
rec
overy
based
on
cloud t
heory f
or
un
derw
ate
…
(
Buddes
ab
)
3765
4.7. Direc
t
tr
ust
c
omput
at
i
on
The val
ue
of
di
rect com
m
un
ic
at
ion
tr
us
t i
s c
o
m
pu
te
d
us
i
ng
belo
w
e
qu
at
i
ons [2
5
]
.
=
¯
xi
⁻ =
1
∑
=
1
E
ni
=
√
2
1
|
−
|
E
ei
=root (S
2
-
E
n
i
2
}
(13)
Wh
e
re
n
=
nu
m
ber
o
f
at
tri
bute
,
x
i
= at
tri
bu
t
e v
al
ue
The
a
bove
v
al
ues
a
re c
om
pu
te
d base
d on pa
cket loss
, pack
et
error, a
nd ene
rg
y c
onsu
m
ed.
5.
ALGO
RITH
MS
5.1
.
TM
S
a
lg
orithm
Algorithm
1
:
T
MS Ro
utin
g A
lgorit
hm
Step
1
:
TMS t
akes
in
pu
ts
as So
ur
ce
Node, Desti
nation N
od
e
, a
nd
Transm
issi
on
Rang
e
.
Step
2
: First, the
S
ource
Node wil
l be
respo
ns
ible
for
c
ollec
ti
ng
ne
ighbor
node
li
st wh
ic
h
is
within
com
m
un
ic
at
ion
range
pro
vid
ed.
Step
3
: Stop t
he p
ro
c
ess if t
he
S
ourc
e nod
e
-
neig
hb
or no
de
li
st co
ntains t
he desti
nation n
ode.
Step
4
:
If
t
he neig
hbor
node
li
st d
oes
not inclu
de
th
e d
est
inati
on
node
the
n
cal
c
ulate
the
Direct t
r
us
t
of
each
of the
nei
ghbor
no
des ba
sed o
n
E
tx
valu
e.
TE
tx=
∑
/
−
0
E
tx
.=N
ACK/
N
gen
Step
5
:
Pick
a
node
wh
ic
h
has
t
he
highest
trust
value
a
nd
re
peats
the
pr
oc
ess
unti
l
the
destinat
ion
is
reache
d.
5.2. TM
C
-
r
outing
a
l
gori
thm
Algorithm
2
: T
MC
Rou
ti
ng
A
lgorit
hm
Step
1
:
TMC
takes i
nput a
s S
ource
Node, Desti
na
ti
on
N
od
e
, a
nd
C
omm
un
ic
at
io
n.
Tra
ns
m
issi
on
Ra
ng
e
.
Step
2
:
First, the
S
ource
Node wil
l be
respo
ns
ible
for
c
ollec
ti
ng
ne
ighbor
node
li
st wh
ic
h
is
wit
hin
i
n
com
m
un
ic
at
ion
range
pro
vide
d.
Step
3
:
Stop the
proc
ess if t
he
S
ourc
e nod
e
-
neig
hbor no
de
li
st co
ntains t
he desti
na
ti
on no
de.
Step
4
:
If
t
he neig
hbor
node
li
st d
oes
not inclu
de
th
e d
est
inati
on
node
the
n fin
d n
ei
ghbour.
N
odes u
si
ng the
NI
R l
ist
.
Step
5
:
For t
he new
neig
hbors
gen
e
rate t
rust evi
de
nce.
Step
6
:
For t
he new
neig
hbors c
ompu
te
direct t
ru
s
t.
Step
7
:
For the
n
e
w n
ei
ghbors
c
om
pu
te
tr
us
t rec
omm
end
at
ions
us
ing i
ndirect
tr
ust
m
easur
e.
Step
8
:
Pick
up a
node
which
has
th
e h
ig
hest tr
us
t
m
easur
e.
Step
9
:
Re
peat pro
ce
ss unti
l t
he des
ti
nation
is
reac
hed.
5.3
.
TM
C
-
ge
netic n
od
e
rec
ov
er
y alg
orit
h
m
Nodes
i
n
the
w
irel
ess
sens
or
netw
ork
a
re
pron
e
to
fail
ur
e
du
e
to
e
nergy
dep
le
ti
on,
ha
rdwar
e
fail
ure,
and
c
omm
un
ic
at
ion
li
nk
e
rrors,
et
c.
"Fault
node
recove
ry
al
gorithm
"
dete
ct
s
a
fau
lt
node
wh
e
n
s
om
e
s
ens
or
nodes
s
hutd
own
due
to
lo
w
ba
tt
ery
ener
gy.
The
ge
netic
al
gorithm
is
us
ed
to
rec
over
th
e
dep
le
te
d
node
in
the
netw
ork. The
nod
e
d
e
pleti
on
m
ay
b
e due t
o no avail
able
po
wer o
r
it
m
ay
r
each the
th
res
hold le
vel.
Algorithm
3
:
T
MC
G
eneti
c
N
od
e
Reco
ver
y
Algorithm
Step
1
:
Sour
ce
No
de, Desti
natio
n N
od
e
, a
nd Tr
a
nsm
issi
on
Ran
ge
acts as a
n
in
pu
t.
Step
2
:
The So
ur
ce
Nod
e
w
il
l fi
nd
th
e set o
f nodes
within t
ran
sm
issi
on r
a
nge
known
as
ne
ig
hbor no
des.
Step
3
:
If
t
he neig
hbor
nodes ha
ve
th
e d
est
inati
on
node
the
n
st
op the
process
.
Step
4
: If
t
he neig
hbor
nodes d
o no
t
hav
e
the
destin
at
ion
node
t
he
n
c
om
pu
te
the
neig
hbor
Node
s u
si
ng the
NI
R l
ist
.
Step
5
: Fo
r
the
n
e
w n
ei
ghbors ge
nerat
e tru
st e
vid
e
nc
e
Step
6
: Gen
et
ic
proce
ss is trig
ge
red f
or eve
ry T ite
r
at
ion
s t
o
rec
ov
er th
e
f
a
ulty
no
des
Step
7
: Find t
he
c
ount
o
f
num
ber
of
fau
lt
y n
odes in
the
netw
ork. F
aulty
n
odes
are
set o
f
Nodes
wh
ic
h do not
hav
e
s
uffici
ent am
ou
nt of e
ne
r
gy
require
d for tra
ns
m
issi
on
.
Step
8
:
If
no
fau
lt
y
nodes
in
t
he ne
twork
, th
e
g
e
ne
ti
c algo
rithm
stop
s
.
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.
5
,
Oct
ober
20
19
:
3
7
5
9
-
3
7
7
1
3766
Step
9
:
If
a
f
a
ulty
nod
e
ex
ist
s t
hen g
e
ner
at
e a
set
known
as c
hro
m
os
o
m
es
are generate
d wh
ic
h de
pends
on
the num
ber
of
fau
lt
y n
odes.
T
he
c
hrom
os
ome
size i
s equa
l t
o
the
num
ber
of f
a
ulty
nod
e
s.
Step
10
: C
on
si
der a se
t of f
our
c
hrom
os
om
es {CZ1,
CZ2,CZ
3,
CZ
4}
.
Step
11
: Su
ppos
e
the s
et
o
f
no
des
{
5,9,15,2
0,2
3,35,
40,45,5
5,60
}
. T
he
siz
e
of
fa
ul
ty
n
od
e
s is Te
n
the
n
t
her
e
are
f
our
ch
r
omoso
m
es
wh
ic
h
are
ra
ndom
ly
g
ener
at
e
d
with
t
he
val
ues
of
ze
ro
a
nd
ones
as
giv
e
n
by
the foll
owin
g v
al
ues.
CZ1={0
,0,1,0
,
1,1,0,
1,1,0}
CZ2={1
,1,0,0
,
0,1,0,
1,0,1}
CZ3={0
,0,1,0
,
1,1,0,
1,0,1}
CZ4={1
,1,0,0
,
0,1,0,
1,1,0}
Step
12
:
Sele
ct
b
est
tw
o
c
hrom
os
om
e
s.
Step
13
:
The
n
a
pp
ly
c
r
os
s
ov
e
r
as
sho
wn in Fi
gure
2
cro
ss
over
fu
nction
a
nd m
u
ta
tio
n p
r
ocess
as
s
how
n
in
Figure
3
M
utati
on
f
un
ct
io
n.
Step
14
: R
eplace
no
de
s whose
value
i
s 1
.
Step
15
:
Gen
et
ic
al
gor
it
h
m
stop
s a
nd
TMC
r
outi
ng a
lgorit
hm
co
ntinu
e
s to
ex
ec
ut
e f
or
rest
of it
erati
on
Figure
2
.
Cr
os
s
ov
e
r funct
io
n
Figure
3
.
Muta
ti
on
functi
on
6.
PE
RFO
R
MANC
E E
V
ALU
ATIO
N
6.1
.
Q
oS
pa
r
am
eters
used
t
o
e
va
lu
a
te
of
t
he TM
C, TM
S, and
TM
C
-
GA a
lg
orit
hm
s
Si
m
ulati
on
of
TMC
is
perf
or
m
ed
agai
ns
t
TMS
a
nd
e
va
luate
d
perf
orm
ance
with
the
fo
ll
owin
g
par
am
et
ers,
1.
En
d
to
End t
o Dela
y (Ro
und
Trip Ti
m
e)
2.
Nu
m
ber
of Ho
ps
3.
Total
En
er
gy
Con
s
um
ption
4.
Re
sidu
al
E
ne
rgy
5.
Rou
ti
ng
O
verh
ead
6.
Netw
ork
Li
feti
m
e
Si
m
ulati
on
is pe
rfor
m
ing
w
it
h
m
ulti
ple so
ur
c
es and
destinat
ion
nodes
in th
e n
et
w
ork.
T
he
n
um
ber
of
nodes
in
the
s
yst
e
m
set
to
30
,
init
ia
l
batte
ry
ener
gy
set
to
1000
MJ,
a
nd
Transm
issi
o
n
ran
ge
of
sen
sor
nodes
set
to
40
m
.
Mult
iple
so
ur
ces
a
nd
the
destinat
ion
is
def
i
ned
i
n
the
sim
ulatio
n
there
fore
dat
a
transm
issi
on
ta
kes
place bet
wee
n ea
ch
s
ource a
nd
destinat
io
n.
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
Tru
st
m
od
el
ge
netic
nod
e
rec
overy
based
on
cloud t
heory f
or
un
derw
ate
…
(
Buddes
ab
)
3767
6.2
.
Sim
ulat
i
on
s
e
tup
We
c
onduct
a
si
m
ulati
on
in
Ma
tl
ab
to
e
valuate
the
p
e
rform
ance
of
TM
C
-
G
A
T
he
se
nsor
net
work
is
set
up
wh
e
re
the
en
vir
on
m
ental
con
diti
on
ne
eds
to
be
m
o
nitor
e
d.
T
he
de
plo
ye
d
se
nsor
gathers
the
da
ta
,
a
nd
then
this
data
is
sent
to
the
Ser
vice
Pr
ovid
ers.
Th
e
CSPs
then
orga
nize
the
inform
ati
on
acc
ordin
g
to
the
sp
eci
fic
fiel
ds.
The
us
e
rs
can
appro
ac
h
the
store
d
secu
re
data
in
the
cl
oud
ov
e
r
the
int
ern
et
.
T
he
ha
r
d
dis
k
40GB
an
d
RA
M
is
set
to
35
GB.
The
O
pe
rati
ng
Syst
em
is
W
i
ndows
8.
1/XP
is
us
ed
with
the
J
AVA/J2E
E
cod
i
ng
la
ngua
ge.
T
he
i
nteg
r
at
ed
de
velo
pme
nt
en
vir
onm
e
nt
is
Net
Be
a
ns
6.9.1
with
M
YSQL
databas
e.
Th
e
visu
al
i
nterf
ace
is set t
o
c
omm
and li
ne pr
om
pt.
6.3
.
Per
fo
r
m
an
ce
Analy
sis
6.3.1.
E
nd to
e
nd
t
o del
ay
(r
ou
n
d t
ri
p
t
im
e)
En
d
to
E
nd
de
la
y
def
ine
d
as
tim
e
ta
ken
by
the
route
re
quest
(RREQ
)
pack
et
s
to
trav
el
from
the
so
urce
to d
est
i
nation node
a
nd f
or
a
r
oute
re
ply
(RREPL
Y
)
to
c
om
e
back from
the
destin
at
ion
t
o
s
ource
no
de.
En
d
to
en
d
del
ay
m
easur
es
one
r
ound
trip
ti
m
e
ta
ken
by
source
a
nd
d
est
in
at
ion
to
trave
rs
e
the
netw
ork
within
the
transm
issi
on
range
to
deli
ver
data
pac
ke
ts
in
the
wireless
sens
or
net
work.
He
re
En
d
to
En
d
to
de
la
y
i
s
cal
culat
ed by a
tim
er.
E2E
delay = tst
ar
t
-
tst
op
En
d
to
en
d
de
la
y
si
m
ula
ti
on
is
perform
e
d
again
st
thre
e
def
ine
d
al
gorithm
s.
In
th
e
si
m
ulati
on
com
par
ison
res
ults of
TM
S,
T
MC
, an
d
TMC
-
ge
netic
nod
e
r
ecov
e
ry alg
or
it
hm
, TMC, and TMC
-
Ge
netic
n
ode
recovery
al
gor
it
h
m
has
ta
ken
le
ss
tim
e
co
m
par
e
d
to
TMS
becau
se
in
T
MS
routing
al
gorithm
ro
utin
g
m
a
y
unde
rgo
bac
k
a
nd
forth pr
op
a
gation
t
o
fetch
nex
t n
ode in the
n
et
wor
k.
TM
S r
outi
ng algor
it
h
m
m
a
y visi
t
m
or
e
than
one
node
in
the
r
ou
ti
ng
process
w
her
e
as
TMC
an
d
T
MC
-
Genet
ic
node
rec
overy
a
lgorit
hm
do
es
no
t
visit
nodes
twic
e
i
n
the
net
wor
k.
TMC
al
gorithm
m
ini
m
izes
end
to
e
nd
delay
to
guara
ntee
the
qu
al
it
y
com
m
un
ic
at
ion
cha
nnel
fro
m
so
ur
ce
t
o
de
sti
na
ti
on
a
nd
m
or
e
ener
gy
e
ff
ic
ie
nt.
Fi
gur
e
4
E
nd
-
to
-
En
d
dela
y
perform
ance
sh
ows
that
TMC
-
Ge
netic
node
recovery
al
gorithm
is
per
form
ed
bette
r
than
the
TMS
an
d
TMC
al
gorithm
in
te
rm
s o
f
m
illi
sec
onds
.
6.3.2
. Num
ber
of h
op
s
The
nu
m
ber
of
ho
ps
is
def
i
ne
d
as
t
he
nu
m
ber
of
interm
ediat
e
li
nk
s
or
nodes
betwee
n
the
s
ource
node
an
d
de
sti
nation
node
pe
r
route.
T
he
num
ber
of
ho
ps
si
m
ulati
on
com
par
ison
sho
wn
in
Fi
gure
5.
TMC
and
TMC
-
Ge
ne
ti
c
no
de
recovery
al
gorithm
hav
e
ta
ke
n
f
ewer
hops
c
om
par
ed
to
TMS
routin
g
al
gorithm
beca
use
in
TM
S
routin
g
al
gor
it
h
m
so
m
e
ro
uting
m
ay
un
de
rgo
back
forth
pro
pag
at
io
n
to
fetch
the
ne
xt
no
de
in
the
netw
ork
.
The
refo
re,
T
MS
routin
g
al
gorithm
m
a
y
visit
an
al
read
y
visit
ed
no
de
i
n
the
routin
g
process
wh
e
reas
TMC
and
TMC
-
Ge
ne
ti
c
no
de
recovery
al
gorith
m
do
es
no
t
visit
nodes
twic
e
in
the
netw
ork.
Hen
c
e
in
TMS
routin
g
al
gorithm
a
route
m
a
y
hav
e
rep
eat
ed
node
s
in
the
netw
ork,
a
nd
al
so
TMS
al
gorith
m
m
ay
cro
ss
the
total
nu
m
ber
of avai
la
ble nod
e
s in
the syste
m
.
TMC
al
go
rith
m
nu
m
ber
of
hops
fall
s
with
in
total
nu
m
be
r
avail
able
node
s
in
the
netw
ork,
a
nd
this
will
gu
a
ran
te
e
that
qu
al
it
y
co
m
m
un
ic
at
ion
c
hannel
f
ro
m
source
t
o
destina
ti
on
,
inc
reases
the
num
ber
of
al
ive
nodes
in
the
ne
twork
th
us
inc
r
eases
netw
ork
li
fetim
e,
increa
se
resid
ual
ene
rg
y
of
nodes
a
nd
the
refo
r
e
en
erg
y
eff
ic
ie
nt t
han the TMS
alg
or
it
hm
.
Figure
4.
En
d
-
to
-
E
nd
delay
p
e
rfor
m
ance
of T
MC
,
TMS, a
nd TM
C
-
GA
Figure
5. N
umber
of
Hops
Pe
rfor
m
ance
of T
MC
,
TMS, a
nd TM
C
-
GA
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.
5
,
Oct
ober
20
19
:
3
7
5
9
-
3
7
7
1
3768
6.3.3
.
T
otal
en
ergy co
nsum
p
tion
Energy
c
on
s
um
pt
ion
is
def
i
ned
as
s
pe
nd
i
ng
e
ner
gy
by
t
he
node
in
te
rm
s
of
no
de
gen
e
rati
on,
data
transm
issi
on
,
and
route
dis
cov
e
ry.
T
he
total
energy
con
s
um
ption
de
fines
that
tot
al
wasted
ene
rg
y
f
or
deliveri
ng
the p
acket
s
f
ro
m
t
he
source nod
e
to
the
destinat
ion
no
de
in
the
route
Ov
e
rall
energy
co
ns
um
pt
ion
si
m
ulati
on
co
m
par
ison
is
sh
ow
n
in
Fig
ure
6.
TMC
a
nd
TMC
-
Ge
ne
ti
c
no
de
rec
ov
e
ry
al
gorith
m
hav
e
consum
ed
le
ss
energy
com
par
ed
to
TMC
and
TM
S
r
ou
t
ing
al
gorit
hm
because
in
T
MS
routin
g
al
gorithm
so
m
e
ro
utin
g
m
ay
un
derg
o
ba
ck
an
d
f
or
t
h
pro
pag
at
io
n
to
fetch
ne
xt
f
orward
no
de
in
the
net
work.
H
ence
i
n
TMS
r
ou
ti
ng
a
lgorit
hm
a
ro
ut
e
m
ay
hav
e
r
epeate
d
node
s
in
syst
em
wh
ereas
TMC
a
nd
TMC
-
Ge
netic
no
de
recovery
al
gor
it
h
m
do
es
not
visit
nodes
twi
ce
in
the
netw
ork.
T
he
num
ber
of
hops
incr
eases
in
the
ne
tw
or
k
wh
ic
h
le
ads
to
the
total
ener
gy
consum
ption
increase
s.
T
MS
al
go
rithm
is
le
ss
ener
gy
eff
ic
ie
nt
com
par
ed
t
o
TMC
and
TM
C
-
Ge
netic
node
recovery
al
gorithm
.
Figu
r
e
6
sho
ws
the
Pe
rfor
m
ance
energy
consum
ed
sh
ows
that TM
S c
ons
um
ed
higher
th
an
the
TMC
al
gorithm
an
d T
MC
-
Genet
ic
node
r
ec
overy a
lgorit
hm
.
6.3.4
.
Residu
al
e
nerg
y
Re
sidu
al
ene
rgy
of
al
l
the
no
de
s
in
the
network
de
fines
that
total
avail
able
energy
of
al
l
node
s
in
the
netw
ork
ov
e
r
a
per
i
od
ic
ti
m
e.
This
is
the
t
otal
rem
a
ining
e
nergy
in
the
ne
twork
.
S
om
e
ro
utin
g
netw
ork
will
hav
e
Re
si
du
al
Energy
Lim
its
(REL)
wh
e
r
e
so
m
e
ro
utin
g
netw
ork
will
no
t
ha
ve
a
REL
con
ce
pt.
TMS
al
gorithm
will
not
hav
e
RE
L
co
nce
pt
wh
e
reas
t
he
TMC
al
gorithm
has
REL.
REL
is
cal
culat
ed
by
Energy
consum
ption
by
n
odes
.
Re
sidu
al
ene
r
gy
si
m
ulati
on
com
par
iso
n
is
sh
own
in
Figure
7.
TM
C
-
Ge
netic
node
recove
ry
al
gorithm
has
m
or
e
resid
ual
energy
c
om
par
ed
t
o
TMC
a
nd
TMS
r
outi
ng
al
gorithm
bec
ause
i
n
TM
S
r
ou
ti
ng
al
gorithm
has
back
a
nd
forth
pr
opa
gatio
n.
The
num
ber
of
hops
increa
se
s
in
the
netw
ork
w
hic
h
le
ads
to
the
t
otal
energy
c
on
s
um
ption
in
creases.
TMS
al
gorithm
is
less
ene
rg
y
e
ff
ic
ie
nt
com
par
ed
to
TMC
an
d
TMC
-
Gen
et
ic
node
recovery
al
gor
it
h
m
.
TMC
-
G
eneti
c
node
re
cov
e
ry
al
gorit
hm
reco
ve
rs
t
he
nodes
t
o
m
ake
th
e
netw
ork
m
or
e
energy
ef
fici
ent,
inc
reases
t
he
netw
ork
li
f
e
an
d
inc
rease
s
the
nu
m
ber
of
al
ive
no
des
in
the
syst
e
m
to
m
a
ke
avail
able
f
or
data
com
m
un
ic
at
ion
from
so
ur
ce
to
destinat
ion.
Figure
7
s
ho
ws
the
Perfo
rm
ance
Re
sidu
al
E
ne
r
gy
s
hows
t
hat
TMC
-
Ge
netic
node
rec
ov
e
ry
al
gorithm
has
high
resi
du
al
energy
than
t
he
TMS
a
nd TMC
ro
uting al
gorithm
.
Figure
6. Per
f
orm
ance en
er
gy
c
ons
um
ed
of
TMC
,
TMS, a
nd TM
C
-
GA
Fig
ure
7.
Per
f
orm
ance Resi
dual
En
er
gy
of T
MC
,
TMS, a
nd TM
C
-
GA
4.3.5
.
R
ou
tin
g
o
verhe
ad
Rou
ti
ng
ov
e
r
he
ad
is
de
fine
d
as
the
total
r
outi
ng
t
raffic
se
nt
in
bits/
s
ec
by
al
l
the
avai
la
ble
sens
or
nodes
in
t
he
r
ou
ti
ng
netw
or
k
durin
g
sim
ulati
on
.
I
n
the
si
m
ula
ti
on
c
om
par
iso
n
the
de
fi
ned
TMS,
TM
C
an
d
TMC
-
ge
netic
node
recove
ry
al
gorithm
sends
a
set
of
c
ontr
ol
pac
kets
to
fi
nd
r
ou
te
s
.
Ro
ut
ing
ov
e
r
head
us
e
d
to
m
easur
e
the
e
f
f
ic
ie
ncy
of
the
route.
Higher
t
he
r
outi
ng
over
head
m
or
e
was
ta
ge
of
ba
ndwi
dth
.
TMS
al
go
rithm
us
es
m
or
e b
a
ndwi
dth
i
n
te
rm
s
of r
outi
ng d
is
cov
e
ry
a
nd
up
dating
in
the c
orres
pondin
g
r
ou
ti
ng
ta
bles.
Figure 8
sh
ows
the
Per
f
or
m
ance
Rou
ti
ng
Ov
e
r
head
s
hows
that
the
TM
C
and
TM
C
-
Ge
netic
node
recovery
al
gorithm
has
le
ss
rou
ti
ng
ov
e
r
head tha
n
the
TMS
r
ou
t
ing
al
gorithm
.
Evaluation Warning : The document was created with Spire.PDF for Python.