TELKOM
NIKA
, Vol.11, No
.2, Februa
ry 2013, pp. 81
9
~
82
6
ISSN: 2302-4
046
819
Re
cei
v
ed Au
gust 7, 201
2; Re
vised Decem
ber
28, 20
12; Accepted
Jan
uary 13, 2
013
An IOT Security Risk Autonomic Assessmen
t Algorithm
Ruijuan Zhe
ng*
1
, Mingchuan Zhan
g
1
, Qingtao
Wu
1
, Chunlei Yang
1
, Wang
y
a
ng Wei
1
,
Dan Zh
ang
1
, Zhengc
hao
Ma
1
1
Electronic & In
formation En
gi
neer
ing C
o
ll
eg
e Hen
an U
n
ive
r
sit
y
of Scie
nce
and T
e
chnol
o
g
y
Luo
ya
n
g
,
Chin
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
:rj
w
o@
16
3.co
m
A
b
st
r
a
ct
In term
s of I
n
ternet of Things (IOT) sys
tem
wi
th the possi
bil
i
ty criterio
n of fu
zz
i
ness and
rand
o
m
ness s
e
curity risk, w
e
qu
alitativ
ely
ana
ly
z
e
the
security risk level
of IOT s
e
curity scene
by
descri
b
in
g ge
n
e
rali
z
a
tio
n
metrics the potenti
a
l i
m
p
a
ct
and l
i
keli
ho
od of oc
currenc
e of every ma
jor thre
a
t
scenar
ios. On this basis,
we propos
ed se
lf-assess
me
nt
alg
o
rith
m of IOT
security ri
sk, adopting three
-
di
me
nsio
nal n
o
rmal clo
ud mode
l integr
ate
d
consid
erat
i
o
n of risk indicators, researc
h
in
g the multi-
rul
e
ma
pp
ing r
e
lati
onsh
i
p b
e
tw
ee
n the q
ual
itativ
e inp
u
t of
safe
ty indicat
o
rs a
nd the
qua
ntitative reas
on
in
g of
self-assess
me
nt. F
i
nally
, w
e
buil
d
se
curity ri
sk assessment
simul
a
tio
n
pl
at
form, and verif
y
the validity and
accuracy of the
algor
ith
m
in th
e pre
m
is
e of substanti
a
tin
g
the risk leve
l and
the safety criterion d
o
m
ai
n.
Ke
y
w
ords
:
N
o
rmal Cl
ou
d Mode
l, IOT
, Security
Risk, Autonom
i
c Assess
m
e
nt
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
As an ultra
-
l
a
rge
-
scale n
e
twork, it brings
e
norm
o
u
s
ch
alleng
es to its secu
rity and
cre
d
ibility because of comp
osition of co
mplex geom
e
t
ry, the application of the non-d
e
termi
n
istic
and
ru
nning
fuzzi
ne
ss of
IOT. Moreov
er, the
het
e
r
ogen
eity of t
he te
rminal
a
nd
sub
net m
o
re
brings
maximum technic
a
l
diffic
u
lty to sec
u
rity of
cro
s
s-d
o
mai
n
an
d
acro
ss subn
et. Therefore,
in
the Internet
of Things
se
curity re
se
arch, it
is necessary to co
mpre
hen
sivel
y
consi
der t
he
dynamic
cha
nge of re
sou
r
ce
s an
d the
many ty
pes of abnormal
status, and
overall take into
ac
cou
n
t
sy
st
em se
cu
rit
y
mech
ani
sm
s an
d st
rategie
s
time-sharin
g sequ
ence. Auton
o
mic
comp
uting
ha
s be
en
reg
a
rded a
s
a ne
w effectiv
e
way to achieve
system
auto
nomy an
d so
lve
the probl
em
of system se
curity
pe
rformance de
clin
e based on a
syst
em of internal a
nd external
cha
nge
s in
d
e
mand
auto
n
o
mou
s
ly adj
u
s
ting th
e
software a
nd
hardwa
r
e
re
so
urce
s to im
pro
v
e
servi
c
e p
e
rfo
r
man
c
e. How it autonomo
u
sly co
nv
erg
e
s, und
ersta
nds a
nd a
ssesse
s
to ma
ny
se
curity fa
ctors affectin
g
IOT sy
ste
m
se
cu
rity and
com
p
let
e
s the
fine
mea
s
ureme
n
t of
autonomi
c
se
curity in dynamic chan
gef
ul compl
e
x environm
ent, is a key pre
r
equi
site for the
autonomi
c
se
curity of the Internet of Thi
ngs
system.
Because IOT
self-security
research i
s
still in
its infancy, its direct
ly related literatures
now a
r
e relat
i
vely less. But the existing res
earch
shows a goo
d
pros
pe
ct and trend of ra
pid
developm
ent. These have
provide
d
a th
eoreti
c
al
refe
rence an
d technical directio
n for u
s
to lea
r
n
from the
se
cu
rity of the co
mputer
netwo
rk a
nd the
au
tonomic
se
cu
rity mech
ani
sms an
d ways of
system, reali
z
e the d
epth
fusion
autono
my cha
r
a
c
te
ri
stics an
d IOT
se
curity an
d
see
k
the tren
ds
and
co
re e
ssence of Thi
n
gs. Th
e curre
n
t re
sea
r
ch
more fo
cu
se
s on the
re
sea
r
ch
an
d an
al
ysis
of the
risk
asse
ssm
ent m
o
del. A dyna
m
i
c trust m
odel
ba
sed
on
re
putation
and
risk a
s
se
ssm
ent
[1] synthe
size
s dyn
a
mi
sm
and
ri
sk of th
e tru
s
t d
egre
e
evalu
a
tion f
o
r th
e p
r
obl
e
m
that the
tru
s
ted
netwo
rk can
not
effectivel
y
deal with of
malici
o
u
s
n
o
de atta
cks. In
tegrated
fu
zzy logic an
d
re
al-
time risk assessment of Petri nets [2] and risk
assessment of the fusion of fuzzy theory and BP
neural netwo
rk [3] have constructe
d theoreti
c
al
mo
del of their o
w
n. Mean
whi
l
e, assessme
nt
strategi
es an
d metho
d
s
h
a
ve also ma
de some
pro
g
re
ss.
Do
cu
ment [4] refe
ren
c
ing
imm
une
dang
er theo
ry has propo
sed n
e
two
r
k intrusio
n ri
sk dete
c
tion
and qu
antitative assessm
ent
method
s usi
n
g the antibod
y density. Docume
nt [5
] propo
sed a net
work ri
sk assessment met
hod
based on
cl
oud mo
del.
A quantization, co
ding
and
contro
l
scheme i
s
p
r
esented
un
der
comm
uni
cati
on con
s
traint
s [6], a hie
r
archi
c
al m
o
del of survi
v
al situation
a
l aware
n
e
s
s is
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 2, Februa
ry 2013 : 819 – 826
820
prop
osed in [7]. And Integrating
cur
r
e
n
t
st
udy status, the releva
nt theor
ie
s an
d strategie
s
are n
o
t
enou
gh matu
re, as a
re
su
lt of autonom
y problem
s i
n
com
puter
n
e
tworks, the resea
r
ch in the
field of syste
m
se
cu
rity and IOT secu
rity are
in th
e exploration
and the initi
a
l stage. Sel
f
-
asse
ssm
ent pro
c
e
s
s
of se
curity risk
i
s
prop
osed
i
n
t
h
is
pap
er, fo
cu
sing
on
the
problem
of I
O
T
self-a
sse
s
sm
ent, com
b
inin
g with hi
gh p
r
omi
s
cuou
s a
nd hete
r
og
en
eou
s charact
e
risti
cs
of IO
T,
adoptin
g the
three
-
dime
nsi
onal no
rmal
clou
d mod
e
l to re
sea
r
ch th
e self-asse
ssment algo
rith
m
of system
se
curity ri
sk, a
nd jud
g
ing
gl
obal
m
u
lti-val
ued d
epe
nde
ncy cha
r
a
c
te
ristics
betwe
en
possibility crit
erion and security risks.
2. Possibilit
y Criterion
Clou
d model
[8] which i
s
a model of qual
itat
ive and qu
antitative con
v
ersio
n
empl
oying a
natural la
ngu
age expressi
on by Acade
mician Li Deyi
is able to the unce
r
tainty conve
r
t between
qualitative
co
nce
p
t an
d it
s
quantitat
ive
repre
s
e
n
tation
a
natural la
n
guag
e. It ha
s
been
ap
plied
in
data mini
ng,
intelligent
co
ntrol, fuzzy e
v
aluation,
etc. In the vario
u
s b
r
a
n
che
s
of the natu
r
a
l
sci
en
ce
s and
social
sci
en
ce
s,
the pervasivene
ss of the nor
mal
distributio
n and the normal
membe
r
ship
function tog
e
ther h
a
s lai
d
the found
atio
n for the univ
e
rsalit
y theory of the normal
clou
d mod
e
l
[9]. One-di
mensi
onal
n
o
rmal
clou
d
model (X, Y
)
co
nsi
s
ts
of particular
cl
oud
gene
rato
r, g
enerates
qu
antit
ative co
nversi
on of t
he co
n
c
ept ,
embodi
es random
ne
ss
and
fuzzi
ne
ss
of the concept b
y
the
expecte
d value Ex, entropy En an
d hyper
entro
py He. Becau
s
e
of its good
m
a
thematical n
a
ture, the n
o
rmal clo
ud mo
del is u
s
e
d
to
indicat
e
a la
rge n
u
mbe
r
o
f
uncertain p
h
e
nomen
on [10
]
in natural science and
so
cial scie
nce. At prese
n
t, the norm
a
l clo
u
d
model h
a
s
be
come
the mo
st wid
e
cl
oud
model.
Th
e
curve
expression i
s
rep
r
e
s
ented a
s
sho
w
n
belo
w
.
22
exp[
(
)
/
2
(
'
)
]
xn
yx
E
E
In view of th
e unive
rsality [9] of the
n
o
r
mal
clo
ud
m
odel,
com
b
in
ed
with th
e f
eature
of
se
curity indi
cator of IOT system, on the basi
s
of
IOT system wit
h
the possibi
lity criterion of
fuzzi
ne
ss a
n
d
ran
domn
e
ss secu
rity risk, the pot
enti
a
l impa
ct an
d likelih
ood
of occurren
ce of
every major threat
scenari
os w
ill be described, eval
uated and fine measured.
And the level
o
f
se
curity ri
sk and syste
m
toleran
c
e d
egre
e
with the heteroge
neou
s IOT secu
rity sce
n
e
of
increme
n
tal d
eployment ch
ara
c
teri
stics
will be qu
alitative analyzed.
Figure 1. IOT archite
c
tu
re
IOT stru
ctu
r
e
s
an "Interne
t of Things"
of co
verage
of all things i
n
the wo
rld
utilizing
these
technol
ogie
s
su
ch
a
s
RFID,
wi
rel
e
ss d
a
ta
com
m
unication
o
n
the
ba
sis o
f
the Internet
in
the comp
uter. In
this network, go
od
s (p
rodu
cts)
can comm
uni
cate
with each ot
her with
out the
need for hu
man interven
tion. Its esse
nce is to
achieve autom
atically i
denti
f
ication of items
(produ
cts)
an
d the
interco
nne
ction
and
sh
ari
ng
of i
n
formatio
n th
roug
h
com
p
u
t
er Inte
rnet
b
y
usin
g radi
o freque
ncy ide
n
t
ification (RFID) te
chnol
og
y.
Things architecture is shown in Figu
re 1.
As a multi
-
source h
e
tero
gene
ou
s fusi
on net
work,
IOT Secu
rity Crite
r
ion
su
bject to
con
s
trai
nts
of the a
r
chitect
u
re
se
cu
rity e
l
ements
a
n
d
s
e
c
u
r
i
ty thr
e
ats
.
N
e
tw
or
k la
ye
r
o
f
IO
T
has
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
An IOT Security Risk Auton
o
m
i
c Asse
ssm
ent Algorithm
(Ruijuan
g Zheng
)
821
the sa
me
se
curity p
r
obl
e
m
s
with sen
s
or net
wo
rks, mobile
com
m
unication n
e
tworks a
nd
the
Internet. But there i
s
a g
r
e
a
ter differen
c
e between IO
T high
-level
and lo
w-l
e
vel
with traditio
n
a
l
netwo
rk
se
cu
rity
. Perce
p
tio
n
lay
e
r
has
its o
w
n
u
n
iqu
e
nature du
e to the differen
c
e of p
e
rce
p
tion
equipm
ent a
nd gathe
r wa
y. Application
layer presen
ts a different
se
curity attri
butes b
a
sed
on
different sce
n
a
rio
whi
c
h is t
he co
ntent fe
ature
s
, wo
rk
environ
ment, operator m
a
n
ageme
n
t of the
appli
c
ation
-
oriented. Ri
sk
crite
r
ion i
n
flu
enci
ng
IOT
system se
cu
rit
y
will be extracted to
cle
a
r
l
y
state the
com
p
reh
e
n
s
ive re
stri
cti
on facto
r
of IOT security, this
paper based o
n
Internet of T
h
in
g
s
hiera
r
chi
c
al d
i
vision.
The info
rmati
on of pe
rcept
ion Laye
r
is t
o
go
throug
h
these
pro
c
e
ss flo
w
s, su
ch as th
e
informatio
n p
e
rception,
acquisitio
n
, agg
regatio
n,
fusi
on, tran
smi
s
sion, sto
r
ag
e, mining, d
e
ci
si
on-
makin
g
and
control, etc. Theref
ore, th
e perce
ption
crite
r
ion
(
pc
)
affecting the
layer security is
from several
asp
e
ct
s , su
ch as
se
cu
rity of awa
r
e n
o
d
e
s
(
pc
1
),
re
so
urce con
s
trai
nt of percepti
o
n
and the
co
nverge
nce poi
nt(
pc
2
), the
se
curity of the informatio
n coll
ection
(
pc
3
), the priva
c
y of
the
informatio
n transmi
ssion (
pc
4
) to preve
n
t these pot
ential se
cu
rity problem
s, su
ch as n
o
d
e
camo
uflage
(
pc
11
), ad
dition
of no
de
s en
e
r
gy con
s
um
ption (
pc
12
), sig
nal
le
aka
ge a
nd
inte
rferen
ce
(
pc
21
), i
n
form
ation tamp
eri
ng (
pc
22
)
,
the p
e
r
c
e
ived
da
ma
g
e
of h
a
r
dware/software
(
pc
31
), n
on-
authori
z
e
d
u
s
e
(
pc
32
), p
e
r
ce
ption d
a
ta
dest
r
u
c
tion
(
pc
41
), p
e
rce
p
tion data t
heft (
pc
42
), e
t
c.
Network
layer
is
be
v
i
ewe
d
as
th
e
core data f
o
rwarding le
vel of Thing
s
. The n
e
twork
credibility and security (
nc
1
), se
cu
rity of data a
nd
privacy
(
nc
2
), and
relia
bility (
nc
3
) of ro
u
t
ing
proto
c
ol
s are
simultane
ou
sly taken int
o
account by
Netwo
r
k
Cri
t
erion (
NC
). These
proble
m
s
inclu
de o
c
cu
pied tra
n
smission
ban
dwid
th (
nc
11
),
rapi
d sp
rea
d
of secu
rity threat
s (
nc
12
), me
ssage
to steal (
nc
21
), message ta
mperi
ng (
nc
22
), messag
e d
e
stru
ction
(
nc
23
), protocol d
e
stru
ction
(
nc
31
),
sho
r
tenin
g
th
e network lifetime (
nc
32
), too long d
e
l
a
y (
nc
33
), h
u
ge ene
rgy
consumption
(
nc
34
)
cau
s
e
d
by Floodin
g
/ LEACH / PEGASIS / SPIN routing proto
c
ol
s.
Acco
rdi
ng to different ap
plicatio
ns an
d manag
eme
n
t mecha
n
ism in applicat
ion layer,
appli
c
ation Criterio
n
(
AC
) is re
stri
cted
by the se
rvice indu
stry
(
ac
1
), a
c
ce
s
s
c
ontrol
(
ac
2
),
informatio
n st
orag
e (
ac
3
), a
nd man
age
m
ent mod
e
ls
(
ac
4
),
inclu
d
ing multi-a
s
pe
ct conte
n
t
su
ch as
the type of
servi
c
e
(
ac
11
), se
rvice
obj
ect
(
ac
12
), privacy
p
r
ote
c
tion (
ac
13
), au
thenticatio
n of
hetero
gen
eo
us network (
ac
21
), remote signing ide
n
tification of appl
ication termi
n
al (
ac
22
), attac
k
of virus / hacker / mal
w
a
r
e (
ac
31
), illegal use of 3G terminal (
ac
32
), the intern
al authenti
c
at
ion
(
ac
41
), man
a
g
e
ment co
ntra
ct (
ac
42
), etc.
The
se
cu
rity crite
r
io
n of
perce
ption
laye
r, net
wo
rk laye
r a
n
d
appli
c
ation
layer i
s
integrate
d
, an
d its scop
e of
application a
nd the in
flue
n
c
ing fa
ctors a
r
e fused, exte
nded to fo
rm a
dist
rib
u
t
i
on st
ruct
u
r
e of
I
O
T se
curit
y
criterion, a
s
sh
o
w
n in Figu
re
2.
Figure 2. A distributio
n stru
cture
of
Thi
n
g
s
se
cu
rit
y
crit
erion
Multi-dime
nsi
onal, multi-l
a
yered
se
cu
rity crit
e
r
ion fo
r the
se
cu
rity of the Inte
rnet of
Thing
s
ha
s the effect of varying deg
rees in Fi
gure 2 to form
the set of attributes
of the
evaluation of
the se
cu
rity
risks of the Int
e
rnet of Thi
n
gs. Acco
rd
ing
l
y, three-tier
secu
rity crite
r
i
on
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 2, Februa
ry 2013 : 819 – 826
822
that may affect IOT se
cu
rit
y
is analyzed
,
from
whi
c
h
extract
s
seve
ral attrib
utes
influen
cing th
e
degree
high
e
s
t num
be
r of
IOT security
risk, to
fo
rm
t
he key crite
r
ion set
(K
ey
Crite
r
ion,
KC
) of
I
O
T
s
e
c
u
rit
y
as
se
ssm
ent
.
Mean
while,
it
s se
curit
y
crit
erion
lev
e
l (
C
rit
e
rio
n
G
r
ad
e,
CG
) i
s
divided
into the following grades
:
12
{,
,
.
.
.
,
}
,
R
C
G
cg
cg
cg
R
Z
, acco
rdi
ng to
the
bad
or g
ood
deg
ree
o
f
each ind
e
x of
KC
in the a
c
tual wo
rk pro
c
e
ss. F
o
r
diffe
rent i
ndicato
rs
en
counte
r
ed security ri
sk
probability in
heter
ogeneous IOT
environment, Integer
R
po
ssesses a
variety o
f
options. Ba
sed
on the
pri
n
ci
ple of
one
-di
m
ensi
onal
cl
oud m
odel,
singl
e
crite
r
io
n is de
scrib
e
d
by m
u
lti-le
vel
s
e
c
u
rity. Security c
r
iterion
C
kc
formed ca
n be expre
ssed as
KC
CK
C
C
G
.
3 Self-as
ses
sment Alg
o
rithm
3.1 Gener
a
lized
Risk Lev
e
ls
Qualitative re
pre
s
entatio
n of Things
saf
e
ty ri
sk is d
e
t
e
rmine
d
by
t
he se
cu
rit
y
crit
erio
n
C
KC
mapp
ed
by the key
cri
t
erion
KC
an
d se
cu
rity crit
erion
grade
CG
of third q
u
a
rter. T
herefo
r
e,
in the begi
nni
ng of esta
blishing the eval
uation mo
del,
we a
r
gu
es th
at the divided
rule of level
of
risk (Level of
Risk
,
L
R
i) is
the similar
wi
th the division
grad
e of its secu
rity crite
r
ion grade
CG
,
whi
c
h
P
gra
de i
s
divide
d
into mai
n
ly
based
on th
e differe
nce
betwe
en the
oriente
d
spe
c
ific
appli
c
ation a
nd the a
c
curacy re
qui
rem
ents of a
s
se
ssment. Thu
s
,
12
{
,
,
.
..,
}
,
RP
L
l
r
lr
lr
P
Z
. In
view of th
e
above, o
ne-d
i
mensi
onal
n
o
rmal
cl
oud
C
KC
generate
d
by the
aforemention
ed
key
crit
e
r
ion i
s
r
e
lat
ed t
o
t
h
e se
cu
rit
y
risk a
s
se
ssm
ent pro
c
e
ss by employing reve
rse
clou
d
gene
rato
r ge
nerali
z
a
b
ility to the risk lev
e
l to est
ablish the security risk a
s
sessment clou
d model
that depen
ds
on the multi-l
e
vel and mult
i-dime
nsi
onal
criteri
on.
1
2
12
12
{
,
,
...,
,
,
,
...
,
,
,
...,
}
M
NT
KC
pc
pc
pc
nc
nc
nc
ac
ac
ac
is the
criteri
on d
o
main
of
M
+
N
+
T
,
M
,
N
,
T
Z
, in
which security
crite
r
ion
is no
t releva
nt to
each oth
e
r.
L
R
is
qualitative comme
nt i
n
KC
. The element
1
2
12
12
{
,
,
.
..
,
,
,
,
.
.
.
,
,
,
..
.,
}
M
NT
p
c
p
c
p
cn
c
n
c
n
c
a
c
a
c
a
c
in
KC
for
K
C
of
L
R
is
expre
s
sed a
s
a rando
m nu
mber
with sT
able tren
d.
12
1
2
1
2
1
2
12
12
:
[
0
,
1
]
,
{
,
,
...,
,
,
,
...
,
,
,
...,
}
,
{
,
,
...,
,
,
,
...
,
,
,
...,
}
KC
M
N
T
MN
T
K
C
K
C
pc
pc
pc
nc
nc
nc
ac
ac
ac
K
C
pc
pc
pc
nc
nc
nc
ac
ac
ac
Then the
normal clo
ud of
M
+
N
+
T
dime
nsio
n ri
sk
asse
ssm
ent ca
n be de
scri
b
ed by the
following 3(
M
+
N
+
T
) digital feature
s
.
11
1
2
2
2
11
1
2
2
2
11
1
2
2
2
((
,
,
),
(
,
,
)
,
.
..
,
(
,
,
)
,
(
,
,
)
,
(
,
,
)
,
..
.,
(
,
,
)
,
(
,
,
)
,
(
,
,
)
,
..
.,
(
,
,
)
)
pc
pc
pc
pc
pc
pc
pc
M
p
c
M
pc
M
nc
nc
nc
nc
nc
nc
nc
N
n
c
N
nc
N
ac
ac
ac
ac
ac
ac
ac
T
a
c
T
ac
T
Ex
En
H
e
Ex
En
H
e
Ex
En
He
Ex
En
H
e
Ex
En
He
Ex
En
He
Ex
En
H
e
Ex
En
He
Ex
En
H
e
3.2 Algorith
m
Procedure
The initial mappe
d rela
tionshi
p bet
wee
n
the secu
rity criteri
on
KC
and
the risk
asse
ssm
ent i
s
given by th
e as
se
ssmen
t
algorithm. T
he detaile
d reasonin
g
rule
s between th
em
need to
be a
nalyze
d
u
s
in
g the a
s
soci
ation rul
e
s
b
e
twee
n the
multi-dime
nsi
onal the
se
curity
crite
r
ion
cl
ou
d an
d IOT
ri
sk level
cl
oud.
Here, the
pot
ential relation
ship
that exi
s
ts b
e
twe
en th
em
is mainly de
scribe
d by the co
rrel
a
tio
n
rule
s between key security criteri
o
n
KC
of
M
+
N
+
T
dimen
s
ion a
n
d
R
-dim
en
sio
nal level of risk
L
R
. The fusi
on of security
criterio
n set
and level of risk
has formed th
e correl
ation rule set
I
of IOT s
e
c
u
rity risk
. Then
1
2
12
12
1
2
{
,
,
.
.
.
,
,
,
,
..
.
,
,
,
..
.,
;
,
,
.
..
,
}
,
,
,
,
RM
N
T
P
I
K
C
L
p
c
p
c
p
c
n
cn
c
n
c
a
ca
c
a
c
l
r
l
r
l
r
M
N
T
P
Z
, in which
V
I
is stated a
s
eleme
n
t value in
I
,
{
,
,
;
}
,
{
1
,
2
,
...
}
,
{
1
,
2
,
.
..
}
,
{
1
,
2
,
...
}
,
{
1
,
2
,
...
}
mn
t
p
Ip
c
n
c
a
c
l
r
VV
V
V
V
m
M
n
N
t
T
p
P
.
As you
see,
both
KC
and
L
R
are
sub
s
et
of
I
, and
R
KC
L
.
C
KC
and
C
LR
, respec
tively,
is multi-dime
nsio
nal a
nd
one-dime
nsio
nal no
rm
al
cloud fo
rmed
by the afore
m
entione
d 3.
1.
()
()
()
(
)
(
)
(
)
(
)
(
)
(
)
{(
,
,
)
}
{(
,
,
)
}
{(
,
,
)
}
,
{
1
,
mm
m
n
n
n
t
t
t
rr
r
pc
p
c
pc
nc
nc
nc
ac
ac
ac
K
C
c
g
cg
cg
C
E
xE
n
H
e
E
xE
n
H
e
E
xE
n
H
e
m
2
,
...,
},
{
1
,
2
,
...,
},
{
1
,
2
,
...,
},
{
1
,
2
,
...,
}
M
nN
t
T
r
R
,
(,
,
)
,
{
1
,
2
,
.
.
.
}
Rp
p
p
Ll
r
l
r
l
r
CE
x
E
n
H
e
p
P
.
Any eleme
n
t value
s
in
I
V
are
expresse
d by
,
im
n
t
p
v
i
pc
nc
ac
lr
to si
m
p
lify above
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
An IOT Security Risk Auton
o
m
i
c Asse
ssm
ent Algorithm
(Ruijuan
g Zheng
)
823
expre
ssi
on. Then the expression fo
rm o
f
correlation rules is
()
(
)
(
)
nt
ma
b
c
p
cV
n
c
V
a
c
V
p
d
lr
V
,
,,
,
mn
t
p
a
b
c
pc
nc
ac
d
l
r
.
A reasonin
g
idea of mu
lti-con
d
ition
and mu
lti-rul
e
is forme
d
betwee
n
o
u
r multi-
dimen
s
ion
a
l
se
curity
crite
r
ion
and
one
-dime
n
si
onal
level of ri
sk. The rea
s
on
ing ante
c
e
d
e
n
t
()
(
)
(
)
()
(
)
(
)
(
)
(
)
(
)
(,
,
;
,
,
;
,
,
)
mn
t
m
n
t
m
n
t
r
pc
nc
ac
pc
n
c
ac
pc
nc
ac
K
C
cg
C
E
xE
xE
xE
n
E
n
E
n
H
e
H
e
H
e
and co
nse
que
nt
(,
,
)
Rp
p
p
Ll
r
l
r
l
r
CE
x
E
n
H
e
are sha
ped
based on the
correlation rule
I
of securi
ty risk. For every item
multi-dime
nsi
onal se
cu
rit
y
crite
r
ion
()
(
)
(
)
nt
ma
b
c
pc
V
n
c
V
ac
V
, sp
ecifi
c
algorith
m
pro
c
e
ss of au
tonomic a
s
se
ssment is a
s
follows.
Step1
Determine their respective rul
e
s,
if
(
)
()
()
()
3(
)
3
mm
m
m
p
c
pc
pc
p
c
ma
Ex
E
n
p
c
V
E
x
E
n
()
(
)
()
(
)
3(
)
3
nn
n
n
n
nc
n
c
nc
n
c
b
Ex
E
n
n
c
V
E
x
E
n
()
()
()
()
3(
)
3
tt
t
t
t
ac
a
c
ac
ac
c
E
xE
n
a
c
V
E
x
E
n
Then the rule
()
(
)
(
)
nt
p
ma
b
c
d
pc
V
n
c
V
a
c
V
l
r
V
is dire
ctly activated to step
4
, or el
se to step 2;
Step2
Calcul
ating re
spe
c
ti
vely correspo
nding ri
sk level sup
port
(
(
)(
)(
)
)
nt
p
ma
b
c
d
Sp
c
V
n
c
V
a
c
V
l
r
V
And
((
)
(
)
(
)
)
nt
p
ma
b
c
d
Sp
c
V
n
c
V
a
c
V
l
r
V
Els
e
if
(
(
)(
)(
)
)
nt
p
ma
b
c
d
Sp
c
V
n
c
V
a
c
V
l
r
V
Els
e
if
(
(
)(
)(
)
)
nt
p
ma
b
c
d
Sp
c
V
n
c
V
a
c
V
l
r
V
Deg
r
ee of co
nfiden
ce
(
(
)(
)(
)
)
nt
p
ma
b
c
d
Cp
c
V
n
c
V
a
c
V
l
r
V
And
((
)
(
)
(
)
)
nt
p
ma
b
c
d
Cp
c
V
n
c
V
a
c
V
l
r
V
Els
e
if
((
)
(
)
(
)
)
nt
p
ma
b
c
d
Cp
c
V
n
c
V
a
c
V
l
r
V
Els
e
if
((
)
(
)
(
)
)
nt
p
ma
b
c
d
Cp
c
V
n
c
V
a
c
V
l
r
V
;
Step3
In the
same
con
d
ition eleme
n
ts, the rule
(m
ax
)
(
m
a
x
)
(
m
ax
)
(
m
a
x
)
(
m
ax
)
(
m
a
x)
(m
a
x
)
(
m
a
x
)
()
(
)
(
)
nt
p
ma
b
c
d
pc
V
n
c
V
ac
V
l
r
V
corre
s
p
ond
ed
by the maximum of the produ
ct of
S
and
C
is
ac
tivated.
Step4
Th
e d
edu
ction re
su
lt
(m
ax
)
d
V
of risk valu
e is output a
c
cordi
ng to correspon
ding
rule. And
the su
ppo
rt a
nd confide
n
ce co
efficient
of co
rre
sp
on
ding rule a
r
e
adaptively ad
justed. T
hen i
t
is
input to
ope
ration
process of
Step3,
thus, th
e
au
tonomic de
d
u
ction
of
re
aso
n
ing
rul
e
is
ac
compli
sh
ed
.
Note: rea
s
o
n
i
ng pro
c
e
s
s of other crite
r
i
ons
is exe
c
ut
ed by compl
y
ing with thinking of
Step3, Step4 and Step5.
4. Simulation Experimen
t
To verify self-asse
ssm
ent effect of the pr
opo
se
d se
cu
rit
y
risk s
e
lf
-a
ssessme
n
t algorithm
to netwo
rk st
atus d
a
ta det
ermin
ed by th
e ma
ss
dat
a
of IOT, a sim
u
lation exp
e
ri
ment platform is
built in this p
aper a
nd its
data set
s
are
trai
ned, incl
u
d
ing pe
rcepti
on port sca
n
n
ing sample
s set,
packet
s
steal
sampl
e
set a
nd intern
al au
thenticat
io
n data set. By w
h
ich, a multi-dimen
s
ion
a
l risk
asse
ssm
ent norm
a
l clo
u
d
is gene
rate
d, on this ba
sis, a
c
tual o
peratio
n pe
rforma
nce of self-
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 2, Februa
ry 2013 : 819 – 826
824
rea
s
oni
ng ru
le is analyzed. The sim
u
lation ex
pe
riment platfo
rm of IOT autonomi
c
ri
sk
asse
ssm
ent is sh
own in Figure 3.
Figure 3. Simulation Experi
m
ent platform
of IOT Autonomic Ri
sk Asse
ssm
ent
Acco
rdi
ng to simulatio
n
pla
tform in Figure 5,
M
+
N
+
T
dimensi
onal
se
curity criterio
n of the
theory do
mai
n
1
2
12
12
{
,
,
.
..
,
,
,
,
..
.
,
,
,
.
.
.
,
}
M
NT
K
C
pc
pc
pc
nc
nc
nc
ac
ac
ac
is em
bodi
e
d
as
12
4
{,
,
}
K
C
pc
n
c
ac
.
Analogo
usly,
both level of
risk
L
R
a
nd g
r
ade
of se
cu
rity criterio
n
CG
of IOT are
divided into
P
grad
e. Here,
P
is em
bod
ied a
s
P
=
7
, that
is
1
2
34
56
7
{,
,
,
,
,
,
}
{
,
,
,
R
C
G
L
l
r
l
r
l
r
l
r
l
r
l
r
l
r
w
or
s
t
w
o
r
s
e
ba
d
,,
,
}
me
diu
m
go
od
be
tte
r
b
e
st
. Based on the above inf
o
rmatio
n em
bodie
d
, the specifi
c
divisio
n
of
three ki
nd
s of possibility se
curity criterio
n and level of
risk is shown
in Table 1.
Table 1. Division of se
curit
y
criterio
n an
d risk level
Comment
pc
1
nc
2
ac
4
L’
R
best [0.0,0.1)
[0.0,0.1)
[0.9,1.0)
[0.0,0.1)
better [0.1,0.2)
[0.1,0.2
) [0.8,0.9)
[0.1,0.2)
good
[0.2,0.4)
[0.2,0.4)
[0.6,0.8)
[0.2,0.4)
medium [0.4,0.6)
[0.4,0.6)
[0.4,0.6)
[0.4,0.6)
bad [0.6,0.8)
[0.6,0.8)
[0.2,0.4)
[0.6,0.8)
w
o
rse
[0.8,0.9)
[0.8,0.9)
[0.1,0.2)
[0.8,0.9)
w
o
rst
[0.9,1.0]
[0.9,1.0]
[0.0,0.1)
[0.9,1.0]
Based
on
the
gen
erali
z
e
d
se
curity
crite
r
ion,
the
evalu
a
tion exp
e
cta
t
ion of the
ge
nerate
d
three
-
dime
nsi
onal ri
sk
a
s
sessment
no
rmal clou
d set
K
CL
R
C
b
a
sed
on fo
rwa
r
d
cl
oud
g
enerator is
sho
w
n
in T
a
b
l
e 2
ado
pting
the p
r
opo
se
d
self
-a
sse
ssm
ent p
r
ocess. I
t
assum
e
s th
at
ac
4
co
uld
b
e
set to any value. Thu
s
, three-di
men
s
ion
a
l self-a
sse
s
sment clou
d is gained, sho
w
n in Figu
re
4.
Multi-group
12
4
{,
,
}
K
Cp
c
n
c
a
c
is train
ed
according
to
the
algo
rithm p
r
o
c
e
s
s an
d
sema
ntic co
mbination
of
theory
evalua
tion. Part
of trainin
g
sam
p
l
e
s i
s
sh
own i
n
Ta
ble
2.
M
ulti-
combi
nation
relation
shi
p
s betwee
n
th
ree
-
dime
nsi
o
nal sa
mple
and a
s
sessment re
sult
are
synthe
sized, inclu
d
ing sev
e
ral
type
s su
ch as
one
-di
m
ensi
onal
co
rre
sp
ondi
ng,
two-di
men
s
io
nal
corre
s
p
ondin
g
, and thre
e-dim
e
n
s
iona
l corre
s
po
nd
ing, one
-dim
ensi
onal e
r
ror, and two-
dimen
s
ion
a
l error, thre
e-di
mensi
onal e
r
ror.
The ge
nerate
d
self-asse
ssment erro
rs o
f
se
curity risk are
sho
w
n i
n
Figu
re 5 b
a
s
ed
on
result of t
r
ai
ning
sam
p
le.
It ca
n b
e
see
n
that th
e e
rro
r
between
asse
ssment result
and
theoreti
c
al
e
v
aluation
predictio
n valu
e s
houl
d b
e
less or e
q
ual 0.0
015,
and
evaluati
o
n
con
c
lu
sio
n
betwee
n
them
is basi
c
ally con
s
i
s
t
ent, which
can satisfy the evaluation accu
ra
cy
requi
rem
ents in the pre
m
ise of two-dim
e
n
s
iona
l corre
s
p
ond
ing, and th
ree-di
men
s
ion
a
l
corre
s
p
ondin
g
, one-dimen
s
ion
a
l erro
r, and two-dim
ensi
onal e
r
ro
r. Only wh
en
the situation
of
one-dime
nsio
n corre
s
po
ndi
ng a
n
d
thre
e-dimen
s
ion
a
l
error ap
pea
rs sim
u
ltane
ou
sly, it will
cau
s
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
An IOT Security Risk Auton
o
m
i
c Asse
ssm
ent Algorithm
(Ruijuan
g Zheng
)
825
error rel
a
tive larger be
ca
use
sampl
e
i
nput exi
s
ts the m
obility o
f
ope
ration
a
nd
sele
ction
of
sup
port an
d confiden
ce. It will achieve 0
.
003 in sp
ecifi
c
ca
se.
Figure 4. Self-asse
s
sme
n
t clou
d for
goo
d
Table 2. Part
of training sa
mples
Sample Input
Assessment Result
pc
1
nc
2
ac
4
Theor
y evaluatio
n
Error value
0.0 0.09
0.92
best
0.001
0.13 0.15
0.85
better
0.0014
0.28 0.33
0.77
good
0.0015
0.58 0.48
0.45
medium
0.0012
0.75 0.69
0.28
bad
0.0014
0.88 0.85
0.17
w
o
rse
0.0015
0.92 0.98
0.08
w
o
rst
0.0013
0.79 0.88
0.58
bad
0.0015
0.25 0.44
0.55
good
0.016
0.60 0.57
0.64
medium
0.0030
0.12 0.55
0.7
better
0.0028
Figur
e 5. The
erro
r of IOT se
curity
ri
sk
s
e
lf-as
s
e
s
sme
n
t
5. Conclusio
n
An autono
mi
c characte
ri
stic is give
n to IO
T aimin
g
at system
feature a
n
d
se
curity
informatio
n of
IOT an
d u
n
certainty, unp
redictio
n
an
d f
u
zzine
s
s of it
s
cha
nge. F
o
cu
sing
on
sel
f
-
asse
ssm
ent of security risk, t
he self-a
sse
ssm
ent alg
o
rithm of
IOT security risk based on three-
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 2, Februa
ry 2013 : 819 – 826
826
dimen
s
ion
a
l
norm
a
l cl
oud
wa
s
studie
d
ba
sed
on th
e dynami
c
fu
sion
re
sult
of heteroge
neo
us
se
curity fa
ctors.
We
st
rive to ma
ke
a
bre
a
kt
h
r
o
u
g
h
in th
e research
of aut
onomi
c
se
cu
rity
mech
ani
sm
of heteroge
n
eou
s securit
y
of IOT. It
provid
es a
pplication se
rvice se
cu
rity
of
ensurin
g IOT in uncertain e
n
vironm
ent for new
solutio
n
and thin
kin
g
.
Referen
ces
[1]
Z
hou Q, Yu J. D
y
namic T
r
ust Mode
l Base
d o
n
Re
putatio
n a
nd Risk Ass
e
s
s
ment
in T
r
usted N
e
t
w
ork
.
Co
mp
uter Appl
icatio
n Rese
ar
ch
. 2010; 2
7
(1
1): 4211-
42
14.
[2]
Lia
o
ND, Li F
,
Song Y.
Research o
n
rea
l
-time netw
o
rk security risk assess
me
nt and forecast
.
Procee
din
g
of
201
0 Internati
o
nal C
onfer
ence
on Inte
ll
ige
n
t Comp
utation T
e
chn
o
lo
g
y
an
d
Automatio
n
.
201
0; (3): 84-8
7
.
[3]
Hu CJ, Lv CM.
Method of ris
k
assess
ment
base
d
on c
l
as
sifi
ed sec
u
rity
protectio
n
an
d
fu
zz
y
n
eur
al
netw
o
rk
. Proce
edi
ng
of th
e 2
010
Asi
a
-Pacifi
c
Co
nf
erenc
e
on W
e
arab
le
Comp
uting
S
ystems. 201
0
;
379-
382.
[4]
Cai Z
Y
, Z
h
en
g
LP, Z
h
u
SF
.
Quantitative
a
ssessm
ent
of
Net
w
ork
intrus
i
on r
i
sk Bas
e
d
on
immu
ne
antib
od
y co
nce
n
tration.
Hi
gh T
e
chno
logy Le
tters
. 2010; 20(
10): 102
7-1
03.
[5]
Liu
Y, L
YH. A
n
ev
alu
a
tio
n
m
ode
l for
net
w
o
r
k
risk
base
d
on
clo
ud t
heor
y.
Co
mp
uter Si
mulati
on.
2
0
10;
27(1
0
): 95-9
8
.
[6]
Liu QQ. Coordinated Motion Control of
Autonom
ous
and S
e
miautonom
ous Mobile Agents
.
T
E
LKOMNIKA Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2012; 1
0
(8).
[7]
Z
hao JH, Z
h
o
u
Y, Shuo L
X
. A
Situatio
n A
w
ar
eness M
ode
l o
f
S
y
stem Survi
v
abil
i
t
y
B
a
se
d
on Var
i
ab
l
e
Fuzzy
Set.
T
E
LKOMNIKA Indon
esia
n Jour
nal
of Electric
al
Engin
eeri
n
g
. 2
012; 10(
8).
[8]
Li DY, Du Y.
Uncertai
nty artificial int
e
ll
ige
n
ce
, Beijin
g: Defen
s
e Industr
y
Pr
e
ss, 2005.
[9]
Li DY, Liu CY.
Stud
y
o
n
the univ
e
rsal
it
y
of
the norma
l clo
ud mod
e
l.
Chi
nese En
gin
eer
i
ng Scie
nce
.
200
4; 6(8): 28-
34.
[10]
Zhang
HB, Pei QQ, Ma JF. A
n
algorithm
for
sensing insider threat bas
ed
on cloud m
o
del.
Chin
es
e
Journ
a
l of Co
mputers
. 200
9; 3
2
(4): 784-
79
2.
Evaluation Warning : The document was created with Spire.PDF for Python.