TELKOM
NIKA
, Vol. 13, No. 4, Dece
mb
er 201
5, pp. 1242
~1
250
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i4.3104
1242
Re
cei
v
ed Se
ptem
ber 16, 2015; Revi
se
d No
vem
ber
4, 2015; Acce
pted No
vem
b
er 17, 201
5
Influence of Sensor Nodes on the Invulnerability of Tree
Network
Lifeng Jian
g
1,2,a*
, Fengming Zhang
1,b
, Renn
ong Ya
ng
3,c
,
Kun Xu
2,d
1
Institute of Equipm
ent mana
geme
n
t and sa
fet
y
Eng
i
ne
eri
n
g, Air F
o
rce Engin
eeri
ng U
n
iv
ersit
y
,
Xi’
an
710
05
1, Chin
a
2
Air F
o
rce Aviation Un
iversit
y
,
Chan
gch
un, 1
300
22, Ch
ina
3
Aerona
utics a
nd Astrona
utic
s Engin
eeri
ng
Coll
eg
e, Ai
r F
o
rce Engi
ne
erin
g Univ
ersit
y
,
Xi
’an 7
1
0
038, Ch
ina
e-mail: re
dish
3
737
@16
3
.com
a
, zfm@163.com
b
, y
r
n@1
63.c
o
m
c
, x
k
@163.com
d
A
b
st
r
a
ct
In the transfor
m
ation
proc
es
s from
the com
p
lex system
of great indu
strial
era to the informatio
n
era, the co
mpo
nent h
a
vin
g
se
nsin
g functio
n
plays
an i
m
port
ant rol
e
in th
e
evol
ution
of co
mp
lex syste
m
.
To
abstract the com
p
lex system
of “tree”
struct
ure as “t
ree”
network, To abstract
the components
include the
compo
nent
ha
ving s
ens
ing
functio
n
as
n
o
d
e
s, an
d h
o
w
the s
ensor
n
o
d
e
s in
the
n
e
tw
ork affect n
e
tw
ork
invul
ner
abi
lity is studie
d
qu
antitat
ive
l
y in
this pap
er. F
i
rstly, the experi
m
ent
al pro
g
r
am for netw
o
rk
invul
ner
abi
lity i
s
desi
g
n
ed; s
e
con
d
ly, the
i
ndic
a
tors
for
me
asuri
n
g
the
netw
o
rk inv
u
l
nera
b
il
ity an
d
the
importa
nce of nod
e are pr
op
osed; t
hen, the
invul
nera
b
il
ity exper
iments ar
e carrie
d
out i
n
tw
o conditio
n
s-
w
i
th or w
i
thout sensor no
des
in “tree
”
n
e
tw
ork; fi
nally the
experi
m
e
n
tal
data are statis
tically an
aly
z
e
d
.
Results s
how
that after the
ad
ditio
n
of sens
or
nod
es,
the i
n
v
u
ln
erab
ility of “t
ree
”
n
e
tw
ork is pro
m
ote
d
w
h
e
n
subj
ect to ra
n
d
o
m
attack an
d p
a
rticul
ar att
a
ck. T
he
r
e
se
arch res
u
lts
ar
e of ref
e
renc
e
sign
ifica
n
ce f
o
r
improving self-
i
nvulner
ability
in th
e transfor
m
ation proc
es
s from com
p
lex system
to
infor
m
ation. T
h
e
exper
imenta
l
p
r
ogra
m
an
d th
e rel
e
va
nt con
c
lusio
n
s o
b
tai
n
ed by
exp
e
ri
ment in
this
pap
er hav
e certa
i
n
i
n
no
va
ti
on
.
Ke
y
w
ords
: inv
u
lnerability, com
p
lex system
, sensor
Copy
right
©
2015 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Whe
n
the
co
mplex sy
ste
m
in la
rge
in
dustri
a
l e
r
a
make
s th
e transfo
rmatio
n
to meet the
need
s of th
e i
n
formatio
n e
r
a, to enh
an
ce
the ability
of
the syste
m
to
obtain info
rm
ation is one
of
the important purposes of
the tran
sform
a
tion. If the sensor
with the ability to obtain informati
on
can
be
used
as co
mpo
n
e
n
t to be i
n
te
grated
into
t
he
compl
e
x
system, it
wil
l
have a
po
si
tive
impact o
n
in
formation tra
n
sformation.
Ho
weve
r,
su
ch
comp
one
nt can
not on
ly enhan
ce t
he
ability to obtain information of the sy
stem, but al
so
have impact
on the i
n
vulnerability of sy
stem
by cha
ngin
g
the way of informatio
n interacti
on
betwe
en the
comp
o
nents i
n
the
system. Beca
u
s
e
the invulnerability of system directly det
ermines th
e viability of the
system
i
n
a
specifi
c
external
environ
ment,
it ha
s g
r
e
a
t sig
n
ifica
n
ce
to st
udy it.
Curre
n
tly, so
me a
c
hieve
m
ents have
been
made in th
e resea
r
ch of system invuln
erability. And
in parti
cula
r, the re
sea
r
ch
method by t
he
abs
trac
tion from complex s
y
s
t
em to complex net
w
o
rk
and the s
t
udy in invulnerability has been
proved effe
ctive. Howeve
r, the
rese
arch
es on the rel
a
tionship
bet
wee
n
the co
mpone
nt and
the
invulnerability of system
are not
much.
Based
on the existing
rese
arches, experimental program
in the i
n
vulnerability of
complex
system is
designed from the perspec
tive of net
work,
the
indicators for mea
s
uri
ng t
he net
wo
rk i
n
vulnerability and th
e imp
o
rtan
ce
of no
de a
r
e p
r
o
p
o
s
ed
,
the invulnera
b
ility experiments are
ca
rri
ed out in two
condition
s-wi
th or without sen
s
o
r
nod
es in
“tree”
network, and based on the experimental
dat
a, the invulnerability
relationshi
p
between
sen
s
o
r
a
nd
the complex
system
is
analyzed.
Th
e re
sult
s of
Experime
n
t sh
ow th
at
the
components
of the sy
stem can affect
the invulnerability of
the system, the research from t
h
is
point is a imp
o
rtant supple
m
ent
to existing re
sea
r
ch
method
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Influence of Sensor Node
s
on the
Invul
n
erabilit
y of Tree Net
w
ork
(L
ifeng Jia
ng)
1243
2. State o
f
th
e Art
At this
stage,
the
re
sea
r
ch on
invuln
erability
of the
system
is m
a
inly reflecte
d in
the
following three as
pec
t
s
:
(1)
Re
sea
r
ch
on how to
build net
work model of co
mplex syste
m
[1]-[4], the network
model ha
s a
cha
r
a
c
teri
stic of small-worl
d netwo
rk
m
o
del [5] or scal
e-fre
e
network model [6].
(2) A
c
cordin
g to the re
sea
r
ch requi
reme
nts to
desi
gn expe
rimental
pro
g
ram
of
invulnerability and
getting
variou
s type
s of d
a
ta
requi
red
for the
re
sea
r
ch
of inv
u
lnerability from
the experiment [7]-[9].
(3)
Research on the measure of invulnerabilit
y [10]-[12], especi
ally
the research for
large
scale complex syste
m
has ma
de i
m
porta
nt achi
evements [13
]
-[15].
The existing
results
are mainly studie
d
the
topolo
g
y structu
r
e
on the
imp
a
c
t of th
e
invulnerability, however the research on the re
lationship between compon
ent and invulnerabil
i
ty
is not mu
ch, but this is the
key point of the pap
er.
3. Method
3.1. Experimental Progr
am for In
v
u
ln
erabilit
y
Based on Ran
d
o
m
Attack Str
a
tegy
Ran
dom
atta
ck st
rategy
m
ean
s to
ra
nd
omly
sele
ct several
nod
es
that co
mpo
s
e
network
with eq
ual p
r
obability a
s
a
ttack ta
rget
s
whe
n
attacki
ng the n
e
two
r
k. A n
e
two
r
k with N nod
e
s
is
subj
ect to
ra
ndom
attack
by ten time
s
in the ex
p
e
ri
ment. Figu
re
1(a
)
sho
w
s a
flowcha
r
t of
the
experim
ental prog
ram.
(a)
Ran
dom a
ttack expe
rim
ent
(b) Pa
rticul
ar
attack exp
e
ri
ment
Figure 1. Flowchart of exp
e
rime
nt
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 124
2 – 1250
1244
3.2. Experi
mental Program for
Net
w
ork
Inv
u
lnerabilit
y
Based
on Particular Attack
Strateg
y
Particul
ar
attack
strategy
mean
s to
sele
ct m
o
re i
m
porta
nt net
work
nod
es
as
prio
r
targets wh
en
attacki
ng th
e network. T
he key of this st
rategy lie
s in h
o
w to
o
r
de
r the n
o
d
e
s in
terms
of their importa
nce t
o
determine t
he order
of
p
a
rticul
ar atta
ck a
c
cordi
ng t
o
a stan
da
rd
.
Figure 1(b
)
shows the flowcha
r
t
of the experime
n
tal p
r
otocol.
3.3. Preparation of the Ex
periment
3.3.1. Build Net
w
o
r
k M
o
del
It is suppo
se
d that there
are 30
0 nod
e
s
after
the co
mplex syste
m
of “tree”
structu
r
e i
s
abstracte
d
a
s
a
n
e
two
r
k mod
e
l, an
d
these
nod
e
s
a
r
e
divide
d into
two
categori
e
s:
on
e is
comm
and
-issue no
de, of
whi
c
h the
function i
s
to issue
b
ehavior i
n
st
ruction
s
to t
heir
corre
s
p
ondin
g
actio
n
no
d
e
s, 50 totally
(den
oted by
C); the
othe
r is
actio
n
n
ode, 25
0 totally
(den
oted by A), of which the function i
s
to ex
ecute the instructio
n
s
from co
mm
and-i
s
sue n
o
d
e
.
The comma
n
d
-issu
e
no
de
s of ea
ch lev
e
l can
exch
a
nge info
rmati
on between
uppe
r an
d lo
wer
levels. An
d al
l the
actio
n
n
ode
s of
ea
ch
level
can only
exchang
e informatio
n with
correspon
din
g
comm
and
-issue no
de
s, an
d ultimately can e
s
tabli
s
h
a network mo
del of “tree
”
structu
r
e. Fig
u
r
e
2(a
)
sh
ows two-dime
nsio
nal view of the net
wo
rk m
o
del of “tree
”
c
o
mplex
sy
ste
m
.
(a) “Tr
ee”
n
e
twor
k
(b)
“Tree” n
e
twork with a
d
d
i
tion of sen
s
o
r
node
s
Figure 2. Two
-
dime
nsi
onal
view of netwo
rk mo
del
The network
mod
e
l cont
aining
sen
s
o
r
n
ode
s ca
n
be
e
s
tablished ba
sed
on “tree
”
netwo
rk mo
d
e
l. Throug
h study in
the
p
o
sition
s
and f
unctio
n
s of
real compl
e
x system se
nsors,
the intro
d
u
c
tion of
sen
s
o
r
node
s
can
de
rive the follo
wing
assu
mp
tions:
sen
s
o
r
node
s b
e
lon
g
to
comm
and
-issue no
de
s wit
h
in the
sub
-
n
e
twork;
se
n
s
or n
ode
s
can
monitor
all the no
de
s wit
h
in
the su
b-network except fo
r comma
nd-i
s
sue n
ode
s; the
se
nsor n
ode
s
of
the uppe
r comm
and
level can m
o
nitor all
the
n
ode
s of l
o
wer levels
that a
r
e subje
c
t to t
he
control of
comm
and
-issue
node
s withi
n
the same
sub
-
network; se
n
s
or n
ode
s onl
y transmit the
colle
cted inf
o
rmatio
n to the
comm
and
-issue nod
es fro
m
the same
sub-n
e
two
r
k.
Finally “tre
e”
netwo
rk m
o
d
e
l with se
nso
r
nod
es
can
be built, of which the
2D
view is
s
h
ow
n
in
F
i
gu
r
e
2(
b
)
.
3.3.2 Indicators for Measur
ing Net
w
o
r
k In
v
u
lnerabilit
y
and Nodes Importance Lev
e
l
The in
dicators for mea
s
u
r
i
ng net
wo
rk invulnerability used in
the
pape
r a
r
e th
e rel
a
tive
size and n
e
twork
releva
nce
.
Network rele
vance: a
me
asu
r
e of
the
rele
va
nce o
f
each n
ode
in the
net
work.
Th
e
formula is
as
follows
:
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Influence of Sensor Node
s
on the
Invul
n
erabilit
y of Tree Net
w
ork
(L
ifeng Jia
ng)
1245
1
1/
2
V
C
NN
(
1
)
is the numbe
r of the node
s that
are not reachabl
e in n
e
twork;
is the scale o
f
the network.
Relative si
ze:
the ratio of the num
ber
o
f
network no
des in th
e m
a
ximum su
b-netwo
rk
after the attack to the numb
e
r of netwo
rk
node
s that are not subj
ect
to attack.
3.3.3. Dete
r
m
ination of the Attac
k
Ti
mes and th
e Order o
f
the
Net
w
o
r
k
No
des
(1)
Determina
t
ion of the attack times
Random
attack needs to
be conducted
by
10
times in
t
he i
n
vulnerability experi
ment of
two n
e
two
r
k
model
s
base
d
on
rand
om
attack st
rateg
y
, and th
e p
r
o
babilitie
s of
b
e
ing
attacked
for
the node a
r
e
10%, 20%, 30%, 40%, 50%, 60%,
70%, 80%, 90% and 100% respectively.
In the invulnerability experi
m
ent
based
on par
ti
cular attack
strategy
, the “tree” network
without a
dditi
on of
sen
s
o
r
node
s atta
cks
5 no
de
s
of the net
work on
e time,
with 10 tim
e
s of
particula
r attack in th
e exp
e
rime
nt; the “tree” n
e
two
r
k with additio
n
of sen
s
or
no
des atta
cks 1
0
node
s on
e time, with 10 times of pa
rticular attack in
the experim
e
n
t.
(2) O
r
d
e
r of the network n
ode
s in term
s of importan
c
e
The no
de
s of
two net
work
model
s a
r
e o
r
de
red
acco
rding to the d
egre
e
centrality levels.
For
th
e “tre
e” netwo
rk witho
u
t
additio
n
of
sen
s
o
r
node
s, only the
first 50
nod
es in
the im
porta
nce
are li
sted,
be
cau
s
e
the ex
perim
ent i
s
carri
ed
out 10
times
and
o
n
ly 5 no
de
s
are
attacked
one
time. Table
1
sho
w
s the
o
r
der of the
“tre
e”
network n
o
des a
c
cordi
n
g to the
sum
values of the
in
-
degree an
d
out-de
g
ree o
f
degre
e
ce
n
t
rality. For t
he “tree
”
net
work
with ad
di
tion of sen
s
or
node
s, only the first 10
0 n
ode
s in the importa
nce
are listed, be
ca
use the
exp
e
r
iment is
ca
rried
out 10 time
s
and o
n
ly 10
node
s a
r
e att
a
cked
one ti
me
. Table
2
sho
w
s the o
r
der of the
“tree”
netwo
rk no
de
s
with ad
ditio
n
of sen
s
o
r
n
ode
s a
c
cordi
ng to the
su
m value
s
of t
he in
-de
g
re
e
and
out-de
g
ree of
degre
e
ce
ntrality.
Table 1. Orde
r of “tree
”
net
work no
de
s in terms of de
gree
cent
ralit
y
Node
Out-d
egree
In-degr
ee
C19, C20,
C21 a
nd C22
9.000
9.000
C1, C2,
C3, C
4
,
C5, C6,
C7,
C
8
, C9, C
10, C1
1
,
C12, C
13,
C14, C15,
C16,
C17 and C
1
8
8.000
8.000
C23, C24, C25,
C26, C27, C28
,
C29, C30, C3
1, C32, C33,
C34, C35, C36,
C37, C38, C39
,
C40, C41, C4
2, C43, C44,
C45, C46,
C47,
C48, C49 a
nd C
50
6.000
6.000
Table 2. Orde
r of degree centra
lity of the “tree
”
network
with additi
on of sen
s
o
r
node
s
Net
w
ork nod
e
In-degr
ee
Out-d
egree
S1
349 1
S4
132 2
S3 and S2
104
2
S10
62 5
S9
62 3
S5, S6, S7 and S
8
48
3
S21 and S22
34
4
S19 and S20
27
4
S11, S12, S13, S
14, S15, S16, S1
7 and S18
20
4
C19, C20,
C21 a
nd C22
13
9
C11,
C12, C13, C14,
C15, C16, C17
and
C
1
8
12
8
C5, C6, C7
, C8,
C9 and C1
0
11
8
C2, C3 and
C4
10
8
C1
9 8
C23,
C24, C25, C26,
C27, C28, C29,
C30, C31, C32,
C33, C34, C35,
C36,
C37,
C38, C39, C40,
C41, C42, C43,
C44, C45, C46,
C47, C48, C49
and
C50
11 6
S23, S24, S25, S
26, S27, S28, S2
9, S30, S31, S32
,
S33, S34, S35,
S36,
S37, S38, S39, S
40, S41, S42, S4
3, S44, S45, S46
,
S47, S48, S49
and
S50
5 6
V
N
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 124
2 – 1250
1246
3.4. Experiment
3.4.1. Rando
m Atta
ck Ex
periment
Ran
dom atta
ck expe
rime
n
t
can be divided into
the random atta
ck experiment
of “tree”
netwo
rk a
nd
random
atta
ck expe
riment
o
f
“tree
”
netwo
rk with
the
ad
dition of
sensor
nod
es.
Th
e
pro
c
ed
ures o
f
two
experi
m
ental
part
s
are
sim
ila
r
and
co
nsi
s
te
nt with
the
p
r
ocess sho
w
n in
Figure 1(a
)
. T
he experi
m
en
tal data of “tree” net
wo
rk a
fter ten times of rando
m attack are
sho
w
n
in
Table
3. T
h
e
e
x
p
e
r
i
m
e
n
t
a
l
d
a
t
a
of
“tree”
n
e
two
r
k with the addi
tion of sensor node
s
a
f
t
e
r
t
e
n
times o
f
rand
om attack ar
e shown in
T
able 4
.
Table 3. Statistics of invuln
erability indi
cators
fo
r “tre
e
”
netwo
rk in random atta
ck experime
n
t
Attack times
Relative size
Netw
or
k relevance
1 0.413
0.2342
2 0.4567
0.3681
3 0.1067
0.0378
4 0.1933
0.1166
5 0.0333
0.0197
6 0.0567
0.0275
7 0.04
0.0227
8 0.0233
0.0172
9 0.01
0.0076
10 0
0
Table 4. Statistics of invuln
erab
ility indi
cators fo
r “tre
e
”
netwo
rk wit
h
addition of
sen
s
o
r
nod
es in
rand
om attack experi
m
ent
Attack times
Relative size
Netw
or
k relevance
1 0.8857
1
2 0.7886
1
3 0.6971
1
4 0.3086
0.3984
5 0.1971
0.2832
6 0.1314
0.1543
7 0.3143
1
8 0.0457
0.1337
9 0.0229
0.069
10 0
0
3.4.2. Partic
ular Attac
k
Experiment
Particular
attack experiment can be
d
i
vided
in
to
th
e particular
attack e
x
periment o
f
“tree” network and particular attack experiment of
“
t
ree” network with the add
ition o
f
sensor
nodes. The
procedures o
f
two e
x
perimental par
ts
are consistent with the
p
r
ocess sho
w
n in
Figure 1(b).
The experimental da
ta of “tree” net
work after ten
times of partic
u
lar attack a
r
e
shown in Ta
ble 5
.
Table 5. Statistics of invulnerab
ility indi
cators for “tree” network
in particul
ar attack experi
m
ent
Attack times
Relative size
Netw
or
k relevance
1 0.3
0.1924
2 0.14
0.0607
3 0.06
0.0317
4 0.06
0.0171
5 0.02
0.0100
6 0.02
0.0083
7 0.02
0.0064
8 0.02
0.0045
9 0.02
0.0023
10 0
0
The experimental da
ta “tree” network with
the ad
dition o
f
sensor nodes after ten
times o
f
particular attack
are shown in
Table
6.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Influence of Sensor Node
s
on the
Invul
n
erabilit
y of Tree Net
w
ork
(L
ifeng Jia
ng)
1247
Table 6
.
Statistics o
f
invu
lnerability ind
i
ca
tors for
“tr
ee” network
with add
ition
of
sensor nodes in par
ticula
r attack expe
riment
Attack times
Relative size
Netw
or
k relevance
1 0,9714
1
2 0.9429
1
3 0.4
1
4 0.0686
0.0180
5 0.02
0.0123
6 0.02
0.0117
7 0.02
0.0111
8 0.0171
0.0083
9 0.0171
0.0045
10 0
0
4. Results a
nd Analy
s
is
4.1. Statistic
a
l Analy
s
is o
f
Ran
dom Attack Exp
e
ri
ment
According to
the definitio
n relative siz
e
,
it can be known that
th
e value can
be used
to measure
the damage
degree of the main
network after being attacked randomly;
th
e
value closer to 1 indicate
s that the main net
work is subject to low-level da
mage, and th
e
network in
vulnerability is
b
e
tter
.
Figure 3 sho
w
s that
in th
e face o
f
eq
ual probability o
f
a rando
m attack,
the
damage
exten
t
of
the
“tree” network after the
additi
on o
f
sensor nodes is lower than before, when
subject to
a
random atta
ck, wh
ich suggests the
in
troduction
of
sensor nodes can alleviate
the da
mage
of
the
networ
k
thus
impro
v
ing
the
ne
twork invulne
r
ability.
F
i
g
u
r
e
3. Co
mparative an
alysis
in
the
r
e
lative
size o
f
“
t
ree” ne
twork after
bein
g
attack
randomly before and a
f
ter the
addi
tion
of sensor no
des
F
i
g
u
r
e
4. Co
mparative an
alysis
in
the
network
rele
vance o
f
“
t
re
e” network a
f
ter be
ing a
t
ta
ck
randomly before and a
f
ter the
addition
of sensor no
des
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 124
2 – 1250
1248
After the ne
twork is rand
omly a
ttacke
d,
some nodes are directly eliminated, and the
other nodes become iso
l
ated ones.
These nodes
and the network have lost contact in
informa
t
ion
,
so the nodes have been isolated out
side the main network. Network relevance
is used in the paper to measure the number
of isolated nodes in the network, the large
number of is
olated
nodes means h
i
gh
er degree o
f
the ne
twork
damage.
When the network is subject
to
the
random
attacks of differ
ent
probabilities, the
network relevance value
of “tree” network
with addition o
f
sensor nodes
is higher than
before as shown in Figur
e 4. This ind
i
cates
that
th
e addition
of
sensor node
s enhances the
survivab
ility of the
networ
k
.
4.2. Statistical Anal
y
s
is o
f
Particular Attac
k
Exper
i
ment
Figure 5 sho
w
s that when subject
to
a particular
attack, the
d
a
mage exten
t
o
f
the
“tree” networ
k
after
the a
ddition
of se
nsor
nodes is far lower than before in the
first
th
ree
at
t
a
cks,
whic
h ex
hibit
s
good inv
u
lnerabilit
y
accord
ing t
o
t
he definition of
relat
i
v
e
size. From
the four
th a
t
tack, the damage degrees of two ma
in
networks
are
similar, ind
i
cating that the
two ha
ve si
m
ilar
invu
lnera
b
ili
ty.
F
i
g
u
r
e
5. Co
mparative an
alysis
in
the
r
e
la
ti
ve
s
i
z
e
o
f
“
t
ree” ne
twork
after
bein
g
attack par
tic
u
larly be
fore
and after
the
addition
of s
ensor nodes
F
i
g
u
r
e
6. Co
mparative an
alysis
in
the
net
work relev
ance of “
t
ree” network after
being a
ttack
particularly b
e
fore and
a
f
ter the
additio
n
of sensor n
odes
Figure 6 shows that when subject to a particular attack, th
e network
with the
addition o
f
sensor node
s is less prone to gener
ate isolated
nodes com
pared with that
without sensor nodes in
the firs
t
three particu
lar attacks, thus s
howing good invulnerability.
From the
fo
urth attack,
the
probabilities to genera
t
e isola
t
ed
n
odes
of two main
ne
tworks
are similar
,
indicating
tha
t
th
e
two
have similar in
vu
lnerability.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Influence of Sensor Node
s
on the
Invul
n
erabilit
y of Tree Net
w
ork
(L
ifeng Jia
ng)
1249
5. Discussio
n
The followin
g conclusions can be drawn from the
experimen
t
s
above based
on the
analysis o
f
th
e exper
imental da
ta:
(
1
)
After th
e addition o
f
sensor nodes, the
in
vulnerabili
ty of “tree” network is
promoted wh
en subject
to
random atta
ck, ma
inly e
m
bodied in
the decreased damage le
vel
of
the ma
in
network and the
decreased probabili
ty o
f
ne
twork node
to ge
nerate isola
t
ed
nodes.
(
2
)
In the “
t
ree” network with h
i
erarchical
characteristics, the higher
le
vel of th
e
sensor node
s means m
o
re nodes e
s
tablish th
e
information
exchange relationship with
them;
the hig
her in-degree and out-degree of valu
e
s
of the degree centrality
indicate better
invulnerab
ility promo
t
ion
when subject
to
random a
ttack, and
vice versa.
(
3
)
After th
e addition o
f
sensor nodes, the in
vulnerabili
ty
of “tree” network in
particular attack is improved to some
exten
t
,
but
from the
fou
r
th attack, the addition
o
f
sensor node
to the network has little
effect
on in
vulnerability. This
is mai
n
ly because the
importan
t
22
sensor nodes (S1-S22)
tha
t
cause
the change in the
information e
x
chan
ge
relationship
of network node has b
een destroy
ed after the first three
attacks, and the
remaining 28 sensor no
des have low out-degree and in-deg
ree values due to the bottom
loc
a
tion
in
th
e network
,
wi
th
li
tt
le
influe
nce on
the n
e
twork in
vuln
erability.
(
4
)
After
the
addi
tion o
f
sensor nodes to “
t
r
ee” n
e
twork,
the i
m
proving
fu
nction
in
invulnerab
ility in
particular attack is
not so good
as tha
t
in r
andom attac
k
. Although
the
addition o
f
sensor nodes
to the ne
twork can
chang
e the exchan
ge relationsh
i
p between the
informa
t
ion b
e
tween nodes, the sensor node may bec
ome particular target of attack, as the
out-degree and in-degree values o
f
centrality
are high. O
n
ce these sensor nodes are
attacked,
th
e new infor
m
ation e
x
ch
ange rela
tio
n
ship ceases to exist,
and then
th
e
promotion
of
invulnerab
ility
will be no
t
effec
t
ive.
(
5
)
The add
ition o
f
sensor nodes t
o
the network can
chan
ge the infor
m
ation
exchange relationship between network
nodes and has impact on
the netwo
rk
invulnerab
ili
ty.
This indic
a
tes tha
t
in
the info
rma
t
iz
ation o
f
the complex s
y
ste
m
,
the netwo
rk
invulnerab
ility can be i
m
proved by optimiz
ing
the sensor
node information excha
nge
relationship
with
the
othe
r nodes in
th
e network.
6. Conclusio
n
s
The rel
a
tion
ship between
sen
s
o
r
an
d the invul
nerabil
i
ty of comple
x system is
studied in
this pa
per, th
e re
sults
sh
o
w
that addi
n
g
ne
w ty
pes
of comp
one
n
t
s to the sy
stem will h
a
ve an
impact on its
invulnerability. The results
of the re
search will help to furt
her
research on how
to
improve the i
n
vulnerability of complex sy
stem.
Ackn
o
w
l
e
dg
ement
Duri
ng the writing pap
ers, t
he study was of gre
a
t help Dr. Ying
Zhang, to expre
ss m
y
gratitude.
Referen
ces
[1]
Nasiruzz
ama
n
ABM, Pota
H
R
. Com
p
le
x N
e
t
w
o
r
k F
r
am
e
w
o
r
k B
a
sed
C
o
mpar
ative St
ud
y
of Po
w
e
r
Grid Central
i
t
y
Measures.
Internati
ona
l Jour
nal of Electr
ica
l
and Co
mpute
r
Engine
eri
n
g
. 201
3; 3(4)
:
543-
552
[2]
Zhong PY, Sh
uai B, Chen G
T
. Model and
simulati
on o
n
cascad
i
ng fai
l
u
r
e survivab
ilit
y
of hazard
ous
materials tra
n
s
portatio
n
n
e
t
w
ork un
der terr
orist attack.
Applic
atio
n Res
earch
of Co
mputers
. 20
13
;
30(1): 10
7-1
1
0
[3]
Lv JG, Guo JF
, Liu YY, Z
h
a
n
g
W
,
Allen
J. Appro
a
ch
es of
influ
enc
e ma
xi
mizatio
n
in s
o
c
i
al
net
w
o
rk
s
w
i
t
h
pos
itive a
nd ne
gativ
e opi
nitio
n
s.
Dyna
. 201
5; 90(4): 40
7-41
5
[4]
Lazár Iva
n
, Hu
sár Jozef. Veri
fication
of seq
uent
i
a
l p
a
ttern
s in pro
ductio
n
using
informat
i
on e
n
trop
y.
T
ehnick
i Vjesn
i
k
. 2013; 20(
4): 669-
676
[5]
W
a
tts DJ, Strogatz SH. C
o
ll
e
c
tive D
y
n
a
mics
of Small-
w
o
rl
d
Net
w
orks.
Na
tu
re
. 19
98;
393
(668
4): 44
0-
442
[6]
Barab
a
si AL, A
l
bert R. Emerg
enc
y of Scal
in
g in R
and
om
Net
w
orks.
Science
. 19
99; 28
6
(
543
9): 509-
511
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 124
2 – 1250
1250
[7]
Baba
ei M, Ghassemi
eh H, J
a
lil
i M. Casca
din
g
fail
ure tol
e
ranc
e of mod
u
lar sma
ll-
w
o
rl
d net
w
o
rks
.
IEEE Transaction on Circuits
and System
s-II: Express Briefs
. 2011; 58(
8): 527-
531
[8]
Cetink
a
y
a EK,
Bro
y
les D, D
and
ekar A. Mode
lli
ng comm
unic
a
tion
net
w
o
rk chal
len
ges
for future
intern
et resi
li
ence, s
u
rviva
b
ilit
y, a
n
d
d
i
srupt
i
on to
ler
ance: A
sim
u
lati
on-b
a
se
d
ap
proac
h.
T
e
leco
mmunic
a
tion Syste
m
s
.
201
1; 52(2): 75
1-76
6
[9]
Albert R, Jeo
ng H, Bara
ba
si AL. Erro
r a
nd attack toler
ance of com
p
l
e
x
net
w
o
rks.
Nature
.2
00
0;
406:3
78-
382
[10]
T
i
an CG, Lu
XY, C
h
u
LS,
Don
g
T
,
Li D
X
. Mu
lti-Obj
e
ctive T
r
ansmissi
on
Net
w
ork
Pl
ann
ing
w
i
t
h
Consi
der
atio
n
of Po
w
e
r
Grid
Vuln
erab
ilit
y
a
nd Win
d
Po
w
e
r Accommod
a
ti
on.
Jo
urna
l of
Engi
neer
in
g
Scienc
e an
d T
e
chn
o
lo
gy Rev
i
ew
. 2013; 6(3)
:30-34
[11]
W
u
J, Barahona M,
T
an YJ.
Spectral mea
s
ure of structural
robust
ness in compl
e
x net
w
o
rks.
IEEE
T
r
ansactio
n
s o
n
Systems Ma
n and Cy
ber
ne
tics Part A.
2011; 41(6): 12
44-
125
2
[12]
Liu Y
N
, T
ang H, Zhao GF, Xi
ao YP,
Xu
C. Net
w
ork Inv
u
ln
erab
ilit
y Asses
s
ment T
e
chnol
og
y b
a
se
d
o
n
the ENI.
T
E
LKOMNIKA Indon
esia
n Journ
a
l o
f
Electrical Eng
i
ne
erin
g.
201
3; 11(9): 489
6-4
903
[13]
Estrada E,
Hat
ano
N, B
enzi
M. T
he ph
ysics
of comm
unic
a
bilit
y
in
comp
le
x
net
w
o
rks.
Phys Rep
.20
12;
514(
3): 89-1
1
9
[14]
Shan
g YL. Loc
al natur
al co
nn
ectivit
y
i
n
com
p
le
x n
e
t
w
orks.
Chin P
h
ys Lett. 2011;
2
8
(6): 0
689
03
[15]
W
u
J,
T
an SY,
T
an YJ, Deng
HZ
. Anal
y
s
is of
Invuln
erab
ili
t
y
in com
p
le
x
net
w
o
rks b
a
se
d on n
a
tur
a
l
conn
ectivit
y
.
Complex System
s and Complexity Scienc
e
. 201
4: 11(1): 77
-86
Evaluation Warning : The document was created with Spire.PDF for Python.