Inter
national
J
our
nal
of
Electrical
and
Computer
Engineering
(IJECE)
V
ol.
11,
No.
2,
April
2021,
pp.
1375
1387
ISSN:
2088-8708,
DOI:
10.11591/ijece.v11i2.pp1375-1387
r
1375
A
trust-based
authentication
framew
ork
f
or
security
of
WP
AN
using
netw
ork
slicing
Sazia
P
ar
vin
1
,
Amjad
Gawanmeh
2
,
Sitalakshmi
V
enkatraman
3
,
Ali
Alwadi
4
,
J
amal
N.
Al-Karaki
5
,
P
aul
D
.
Y
oo
6
1,3
Melbourne
Polytechnic,
V
ictoria,
Australia
2
Department
of
Electrical
Engineering,
Colle
ge
of
Engineering
and
IT
,
Uni
v
ersity
of
Dubai,
Dubai,
United
Arab
Emirates
4
School
of
Engineering,
Auckland
Uni
v
ersity
of
T
echnology
Auckland,
Ne
w
Zealand
5
Department
of
Information
Security
Engineering
T
echnology
,
Ab
u
Dhabi
Polytechnic,
Ab
u
Dhabi,
United
Arab
Emirates
5
Department
of
Computer
Engineering,
The
Hashemite
Uni
v
ersity
,
Zarqa-Jordan
6
Department
of
Computer
Science
and
Information
Systems,
Birkbeck
Colle
ge,
Uni
v
ersity
of
London,
London,
United
Kingdom
Article
Inf
o
Article
history:
Recei
v
ed
Apr
25,
2020
Re
vised
Aug
7,
2020
Accepted
Aug
18,
2020
K
eyw
ords:
Authentication
Security
Substantiality
T
rust
WP
AN
ABSTRA
CT
Ne
w
technologies
and
their
seamless
wireless
interconnecti
vity
bring
along
man
y
chal-
lenges
including
security
and
pri
v
ac
y
issues
that
require
immediate
attention.
W
ireless
personal
area
netw
orks
(WP
ANs)
are
characterized
by
limited
ener
gy
res
ources
and
computing
po
wer
which
call
for
lightweight
security
mec
hanisms
in
these
netw
orks
as
a
mandatory
requirement.
In
this
paper
,
a
lightweight
trust-based
frame
w
ork
for
node
authentication
in
WP
AN
is
proposed.
Our
main
obj
ecti
v
e
is
to
minimise
the
ef
fort
in
distinguishing
v
alid
requests
of
trustw
orth
y
nodes
from
in
v
alid
requests
of
mali-
cious
nodes
that
can
result
in
netw
ork
compromises.
W
e
achie
v
e
this
through
netw
ork
slicing
whic
h
di
vides
the
netw
ork
into
primary
and
secondary
virtual
net
w
orks.
The
proposed
frame
w
ork
has
three-fold
benefits.
Firstly
,
it
authenticates
nodes’
requests
based
on
a
no
v
el
method
of
trust
v
alue
calculation.
Secondly
,
the
frame
w
ork
maintains
ener
gy
ef
ficienc
y
while
authenticating
nodes’
requests
to
access
WP
AN
resources.
Fi-
nally
,
the
frame
w
ork
pro
vides
a
solution
for
the
biasing
problem
that
can
arise
due
to
une
xpected
beha
viour
of
malicious
users
in
WP
ANs.
The
frame
w
ork
ef
ficac
y
is
illustrated
by
using
a
case
study
to
sho
w
ho
w
it
can
accurately
capture
trust
relations
among
nodes
while
pre
v
enting
malicious
beha
vior
.
This
is
an
open
access
article
under
the
CC
BY
-SA
license
.
Corresponding
A
uthor:
Amjad
Ga
w
anmeh
Colle
ge
of
Engineering
and
IT
Uni
v
ersity
of
Dubai,
Dubai,
U
AE
Email:
amjad.g
a
w
anmeh@ud.ac.ae
1.
INTR
ODUCTION
Recent
rapid
adv
ancements
in
wireless
communications,
sensor
technologies,
and
the
Internet
of
things
(IoT)
ha
v
e
resulted
in
the
e
v
olution
of
wireless
body
area
netw
ork
(WB
AN)
and
wireless
personal
area
netw
orks
(WP
AN)
into
an
interdisciplinary
netw
orking
paradigm,
connecting
people
with
v
arious
real-time
applications
including
heal
thcare,
sports,
entertainment
and
social
media
[1–5].
WP
AN
is
a
v
ery
common
and
pre
v
ailing
technology
that
is
used
to
monitor
and
collect
data
from
a
ph
ysical
en
vironment
through
sensors
that
can
be
attached
to
the
person,
implanted,
or
deplo
yed
around
the
person’
s
confined
area.
While
a
WB
AN
interconnects
independent
nodes
(e.g.
sensors
and
actuators)
that
are
present
in,
on,
or
around
a
human
body
,
a
J
ournal
homepage:
http://ijece
.iaescor
e
.com
Evaluation Warning : The document was created with Spire.PDF for Python.
1376
r
ISSN:
2088-8708
WP
AN
is
a
connected
netw
ork
of
de
vices
centered
on
an
indi
vidual’
s
personal
w
orkspace
.
Due
to
their
inherent
limitations
in
memory
and
po
wer
,
the
y
are
vulnerable
to
malicious
attacks
since
current
security
solutions
are
not
suitable
to
protect
each
node
as
well
as
dynamic
topology
of
distrib
uted
data
processing
in
WSN
[6,
7].
W
eak
security
in
WSN
can
ha
v
e
an
adv
erse
impact
that
could
e
v
en
result
in
life-threatening
condi-
tions
in
healthcare
[6].
In
addition,
dif
ferent
types
of
short
range
wireless
netw
orks
are
constantly
e
xpanding
with
v
arious
de
vices
attached
as
demonstrated
in
[8].
The
v
arious
types
of
WP
ANs
applications
raises
se
v-
eral
issues
related
to
preserving
data
confidentiality
and
pri
v
ac
y
[9].
W
ith
the
increased
use
of
WP
ANs
in
v
arious
applications
that
can
impact
both
personal
and
professional
life,
it
is
important
to
b
uild
a
lightweight
trust-based
authentication
frame
w
ork
for
ensuring
their
security
[10,
11].
W
ith
this
aim,
the
focus
of
this
research
is
to
propose
a
security
frame
w
ork
to
safe
guard
WP
AN
with
a
simplified
and
lightweight
trust
man-
agement
mechanism.
Se
v
eral
studies
ha
v
e
demonstrated
the
feasibility
of
security
attacks
on
these
types
of
netw
orks
[12],
thus,
WP
AN
is
considered
as
a
threat
by
indi
viduals
due
to
the
e
xisting
vulnerabilities
in
pro-
tecting
their
personal
data
and
pri
v
ac
y
[13].
T
raditional
authentication
and
trust
management
methods
of
WP
AN
are
inef
fecti
v
e
and
the
current
security
situation
w
arrants
an
ef
ficient
lightweight
security
method
with
minimum
o
v
erhead.
Such
a
solution
must
address
se
v
eral
security
concerns
in
WP
ANs:
i)
a
malicious
node
in
WP
AN
could
ab
use
its
resources
with
selfish
beha
vior;
ii)
a
malicious
node
treated
as
a
primary
user
could
stop
secondary
nodes
from
using
the
netw
ork;
and
iii)
malicious
nodes
could
create
biased
nodes
with
traditional
trust
management
schemes
as
their
true
trust
v
alues
are
not
kno
wn.
While
trust
in
WP
AN
has
been
discussed
in
the
current
state
of
the
art
[14],
the
method
applied
are
not
suitable
for
WP
AN.
Firstly
,
most
trust-based
schemes
are
applied
only
for
routing
problem
in
WP
ANs
and
not
for
pre
v
ention
of
access
to
untrustw
orth
y
node.
Secondly
,
due
to
the
dif
ficulty
in
calculating
the
true
trust
v
alues
for
biased
nodes
for
authentication
in
WP
AN
conte
xt,
there
is
paucity
of
research
t
o
address
this
problem.
Finally
,
to
the
best
of
our
kno
wledge,
no
solution
for
authentication
in
WP
AN
using
trust
management
scheme
e
xists
to
enhance
security
in
WP
ANs
using
netw
ork
slicing,
where
nodes
belong
to
either
a
primary
or
a
secondary
virtual
group.
Se
v
eral
methods
were
proposed
to
address
security
in
wireless
application
oriented
netw
orks,
includ-
ing
the
w
ork
reported
in
[8],
which
addressed
the
issue
of
detecting
impersonation
attack
in
wireless
W
iFi.
Recently
,
time-bound
lightweight
frame
w
orks
that
are
services-based
in
processing
security
policies
are
stud-
ied
[15].
Some
security
approaches
are
tool-based
and
focus
on
the
netw
ork
routing
protocols
as
well
as
authentication
and
encryption
techniques
[16].
Existing
WP
AN
security
methods
are
predominantly
limited
in
solving
a
specific
problem
[17,
13].
Man
y
ne
w
models
and
metrics
for
trust
e
v
aluation
proposed
are
not
generic
and
consider
specific
location-bas
ed
pri
v
ac
y
,
backw
ard
broadcasting
or
tunnelling
issues
[18,
19].
Such
schemes
require
h
ybrid
security
architectures
that
use
dif
ferent
algorithms
for
dif
ferent
problems.
In
pre
vious
w
ork,
Sazia
et
al.
[20]
proposed
trust-based
authentication
mechanism
in
cogniti
v
e
radio
netw
orks
(CRNs)
to
secure
the
spectrum
sharing
process.
Ho
we
v
er
,
it
did
not
address
trust-biasing
problem.
Subsequently
,
Sazia
et
al.
[21]
proposed
a
community-based
grouping
technique
to
ensure
the
trust-based
authentication
pro-
cess
in
CRNs,
and
in
[22]
secondary
node
authentication
i
n
WB
AN
w
as
introduced.
Ho
we
v
er
,
it
lack
ed
the
proof
of
trust
e
v
aluation
for
aut
hentication.
T
rust
is
important
as
it
is
a
measure
of
a
node’
s
reliability
and
security
within
the
netw
ork
throughout
its
communication
with
other
nodes.
Other
w
orks
ha
v
e
limited
focus
of
trust
in
healthcare
applications
for
quality
of
data
deli
v
ery
and
netw
ork
operations
[23].
This
paper
aims
to
address
the
issue
of
node
authentication
in
WP
ANs
in
general.
It
propos
es
a
no
v
el
lightweight
method
t
h
a
t
can
enhance
the
security
of
WP
AN
taking
into
consideration
the
topology
and
nature
of
the
netw
ork
paradigm.
Existing
solutions
are
resource
e
xtensi
v
e
while
WP
AN
do
not
ha
v
e
computational
po
wer
.
Further
,
some
nodes
or
implanted
de
vices
are
dif
ficult
to
be
maintained
with
limited
po
wer
supply
,
and
WP
ANs
ha
v
e
dynamic
and
mobile
characteristics.
So
f
ar
,
WP
AN
can
only
support
lo
w
security
require-
ment
applications,
such
as
athletes’
wearabl
es,
and
critical
applications,
such
as
continuous
heart
monitoring
de
vices.
These
form
our
research
moti
v
ation
in
proposing
a
reliable
li
ghtweight
security
mechanism
for
node
authentication
in
WP
ANs.
2.
NODE
A
UTHENTICA
TION
IN
WP
ANS
B
ASED
ON
TR
UST
In
WP
AN/WB
AN,
the
sensors
and
de
vices
might
be
wearable,
implanted
or
used
for
personal
mon-
itoring.
Such
sensors
and
monitoring
de
vices
are
usually
designed
with
minimal
comple
xity
and
lo
w
po
wer
consumption
in
order
to
maximize
battery
life.
Ho
we
v
er
,
e
xisting
security
primiti
v
e
operations,
such
as
hash-
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1375
–
1387
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
1377
ing
and
encryption,
require
po
werful
processing
and
high
resources
making
them
unsuitable
for
these
sensors
and
de
vices.
Therefore,
in
this
paper
,
we
propose
a
lightweight
frame
w
ork
for
trust
management
and
authenti-
cation
of
nodes
in
WP
AN.
The
frame
w
ork
treats
the
WP
AN
as
cogniti
v
e
radio
enabled
wireless
netw
ork,
and
di
vides
nodes
into
tw
o
types,
primary
user
(PU)
and
secondary
user
(SU).
PU
is
a
trusted
node
in
WP
AN
that
is
ph
ysically
secure,
and
SU
is
a
node
that
cogniti
v
ely
operates
through
the
WP
AN
without
causing
harmful
interference
to
the
PU.
In
addition,
there
is
a
de
vice
that
al
lo
ws
other
wireless
de
vices
to
connect
to
a
wired
or
wireless
netw
ork
using
W
iFi,
or
related
standards.
In
addition,
Further
,
our
frame
w
ork
consists
tw
o
units
called
base
stations,
the
primary
user
base
station
(PUBS)
and
secondary
user
base
station
(SUBS).
2.1.
System
model
and
ar
chitectur
e
Figure
1
depicts
the
proposed
trust
based
authent
ication
WP
AN
architecture
by
dynamically
di
viding
nodes
into
tw
o
groups:
primary
and
secondary
.
This
architecture
fits
well
for
WP
AN
due
to
the
mobility
of
the
netw
ork
and
the
dynamics
of
its
topology
that
allo
ws
nodes
to
be
part
of
the
netw
ork
based
on
se
v
eral
aspects,
including
time,
and
location.
Figure
1.
System
architecture
for
the
trust-based
authentication
frame
w
ork
for
WP
AN/WB
AN
The
k
e
y
components
of
the
frame
w
ork
include
the
(Base
station)
which
acts
as
a
point
of
c
o
nt
act
between
primary
nodes
as
well
as
secondary
nodes.
The
architecture
contains
on
primary
user
base
station
(PUBS)
that
connects
primary
nodes,
and
a
secondary
user
bas
e
station
(SUBS)
that
connects
secondary
nodes.
In
addition,
the
certificate
aut
hority
(CA)
is
used
to
maintain
trust
v
alues
for
all
nodes
as
well
as
for
updating
the
trust
repository
whene
v
er
requested
from
the
PUBS
or
SUBS.
In
addition,
it
plays
a
role
in
the
authentication
process.
Finally
,
T
rust
repository
acts
as
a
database
for
nodes’
trust
v
alues.
Since
the
trust
v
alue
is
publicly
kno
wn,
its
inte
grity
is
achie
v
ed
storing
a
pri
v
ate
trust
v
alue
for
e
v
ery
node
locally
.
The
pri
v
ate
trust
v
alue
is
only
visible
to
the
CA
so
that
it
can
be
v
alidated
ag
ainst
an
y
malicious
modification.
2.2.
Operation
of
trust-based
authentication
framew
ork
This
section
sho
ws
ho
w
nodes
can
join
the
WP
AN
as
secondary
nodes
using
the
proposed
archi
tecture.
Figure
2
sho
ws
the
steps,
where
the
node
request
is
made
by
the
node
to
the
WP
AN.
The
request
is
processed
into
the
BS,
which
i
n
turn
calculates
a
trust
v
alue
for
the
node,
and
communicate
it
with
the
CA.
The
CA
first
checks
for
biasing,
and
if
detected,
the
biasing
algorithm
is
used
to
handle
this
situation.
Otherwise,
the
authentication
decision
is
made
based
on
the
calculated
trust
v
alue.
The
decision
is
communicated,
and
the
trust
v
alue
is
updated
if
necessary
.
The
first
step
of
the
propos
ed
method
is
to
perform
trust
calculation
for
candidate
nodes.
When
a
node
w
ants
to
use
the
netw
ork
resources
i.e.
as
a
secondary
node,
the
SUBS
calculates
its
corresponding
trust
v
alue.
this
is
achie
v
ed
by
asking
all
member
nodes
of
the
SUBS
to
assign
a
trust
v
alue
to
the
candidate
node.
Once
e
v
ery
member
communicates
back
it
s
recommended
v
alue,
based
on
history
of
the
node,
the
SUBS
aggre
g
ates
the
recei
v
ed
trust
v
alues
and
obtains
one
for
the
candidate
node.
If
the
node
is
deemed
trustw
orth
y
,
then
it
is
either
authenticated
by
the
SUBS
or
the
request
is
forw
arded
to
the
PUBS.
In
this
case,
the
PUBS
A
trust-based
authentication
fr
ame
work
for
security
of
WP
AN
using
network
slicing
(Sazia
P
arvin)
Evaluation Warning : The document was created with Spire.PDF for Python.
1378
r
ISSN:
2088-8708
also
calculates
the
candidate
node’
s
trust
v
alue
to
decide
i
ts
eligibili
ty
to
access
the
required
resources.
This
is
done
through
the
trust
based
authentication
frame
w
ork
process
as
sho
wn
in
Figure
2.
An
y
calculated
trust
v
alue
of
the
candidate
node
must
be
communicated
to
the
CA
to
process
it.
Once
a
trust
v
alue
is
calculated,
authentication
process
is
started.
Authentication
is
performed
bas
ed
on
trust
v
alue
and
a
predefined
threshold.
While
processing
trust
v
alue
through
the
CA,
a
biasing
check
is
performed,
i
f
the
calculated
trust
v
alue
is
found
biased,
then
SUBS
handle
this
situation
by
marking
nodes
with
improper
trust
v
alue
is
biased.
T
rust
calculation,
authentication,
and
biasing
are
discussed
in
more
detail
in
the
follo
wing
tw
o
sections.
N
od
e
R
eq
u
es
t
B
S
T
ru
s
t
C
alcu
la
t
ion
T
ru
s
t
p
r
ocessed
in
C
A
B
ias
D
e
t
ect
ion
Alg
or
it
h
m
B
ias R
esolu
t
ion
Alg
or
it
h
m
Acc
ep
t
N
od
e
Au
t
h
en
t
ic
a
t
ion
D
e
cisio
n
R
ej
ect
N
od
e
Figure
2.
T
rust
based
authentication
frame
w
ork
3.
TR
UST
EV
ALU
A
TION
In
a
real
life
scenario,
lar
ge
number
of
connected
de
vices,
such
as
wearable
de
vices,
smart
de
vices,
WBSN
nodes,
de
vices
with
tags,
etc
get
connected
dynamically
making
trust
calculation
becomes
a
continuous
process.
W
e
define
trust
in
WP
AN
as
a
de
gree
of
reliabi
lity
and
security
that
can
be
ensured
when
a
SU
requests
to
access
an
y
PU’
s
node.
In
this
conte
xt,
trust
is
calculated
for
authenticating
a
SU’
s
request
and
to
determine
whether
the
SU
can
be
allo
wed
to
use
a
PU’
s
free
spectrum
or
other
netw
ork
resources
without
causing
an
y
security
breach
or
malicious
attack
in
the
WP
AN.
In
our
proposed
frame
w
ork,
we
calculate
the
candidate
SU
node’
s
trust
comprehensi
v
ely
using
the
direct
trust,
indirect
trust
and
inte
grated
trust
e
v
aluation
models,
which
are
described
ne
xt.
Our
proposal
dif
fers
from
the
methods
adopted
in
literature
[24,
25]
as
it
is
uniquely
modeled
for
WP
AN.
Figure
3
sho
ws
a
scenario
for
trust
e
v
aluation
models
in
our
proposed
frame
w
ork
by
depicting
the
dif
ferent
trust
relationships
between
a
candidate
SU
node
and
other
member
PU
nodes.
The
trust
v
alues
calculated
using
three
trust
e
v
aluat
ion
models,
namely
direct
trust,
indirect
trust
and
inte
grated
trust,
with
trust
v
alues
are
in
the
range
of
0
(complete
distrust)
and
1
(complete
trust).
In
the
ne
xt
subsections,
we
e
xplain
the
process
adopted
in
calculating
the
candidate
node’
s
trust
v
alue
using
v
arious
scenarios
of
the
member
node
relationships
established.
N
o
d
e
1
N
o
d
e
4
N
o
d
e
3
N
o
d
e
2
D
i
re
ct
t
ru
s
t
D
i
re
c
t
t
ru
s
t
D
i
re
ct
t
ru
st
I
n
d
i
re
c
t
t
ru
st
D
i
r
e
ct
t
ru
s
t
C
andida
t
e
Node
Figure
3.
A
scenario
for
trust
e
v
aluation
models
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1375
–
1387
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
1379
3.1.
Dir
ect
trust
v
alue
calculation
In
this
model,
we
e
v
aluate
the
direct
trust
of
a
candidate
node
by
proposing
a
computation
method
for
determining
a
quantitati
v
e
direct
trust
v
alue.
The
direct
trust
v
al
u
e
of
a
node
is
calculated
by
other
member
nodes
if
a
direct
relationship
has
been
established
between
them
via
some
past
interaction
and
cooperation
e
xperiences.
Let
us
denote
the
direct
trust
v
alue
as
T
D
ir
ecttr
ust
.
In
Figure
3,
node
2
represents
the
candidate
SU
node,
and
it
has
established
direct
communication
links
with
both
node
1
and
node
4.
Hence,
node
1
and
node
4
will
use
the
direct
trus
t
e
v
aluation
model
based
on
the
past
communication
link
to
calculate
the
candidate
node’
s
trust
v
alue.
A
candidate
node’
s
direct
trust
v
alue
is
determined
based
on
the
member
nodes
that
ha
v
e
established
a
direct
communication
link
or
cooperation.
In
a
dynamic
WP
AN,
the
relationships
between
nodes
k
eep
chang-
ing.
Hence,
the
trust
v
alues
determined
by
the
member
nodes
ha
v
e
to
be
constantly
monitored
by
the
base
station
according
to
the
users’
multi-attrib
ute
trust
v
alues.
A
historic
record
of
information
about
all
the
past
communication
links
established
are
maintained
in
a
table
of
cooperation
records
as
sho
wn
in
T
able
1.
The
candidate
node’
s
trust
v
alue
is
calculated
based
on
its
attrib
ute
v
alues.
In
our
proposed
approach,
each
at-
trib
ute
has
three
v
alues
associated,
namely
the
number
of
successes
(
S
i
;
i
=
1
;
2
::::n
)
,
the
number
of
f
ailures
(
F
i
;
i
=
1
;
2
:::n
)
and
the
amount
of
cooperation
(
C
i
;
i
=
1
;
2
;
::::n
)
.
W
e
assume
equal
v
alues
assigned
between
the
users
for
cooperati
v
e/non-cooperati
v
e
beha
vior
during
their
interaction
process.
T
able
1.
Cooperation
record
table
Attrib
utes
Success
F
ailure
Cooperation
Sum
A
1
S
1
F
1
C
1
A
2
S
2
F
2
C
2
....
....
....
....
A
n
S
n
F
n
C
n
W
e
define
the
cooperation
sum
used
in
T
able
1
as
the
total
sum
of
the
number
of
successes
and
f
ail
ures
of
a
candidate
node’
s
request
or
e
v
ent
and
is
represented
by
C
i
=
S
i
+
F
i
;
i
=
1
;
2
;
:::n
.
The
trust
v
alue
for
attrib
ute
A
i
can
be
computed
based
on
the
v
alues
in
T
able
1
and
then
can
be
used
to
calculate
the
candidate
node’
s
o
v
erall
t
rust
v
alue
(denoted
by
T
D
ir
e
ct
tr
ust
)
with
n
attrib
utes
A
i
;
i
=
1
;
2
;
::n
by
combining
the
trust
v
alues
for
each
attrib
ute
T
A
i
as
follo
ws:
T
D
ir
ec
t
tr
ust
=
n
P
i
=1
T
A
i
n
P
i
=1
T
A
i
+
n
P
i
=1
(1
T
A
i
)
;
w
her
e
T
A
i
=
S
i
C
i
;
(1)
3.2.
Indir
ect
trust
v
alue
calculation
In
order
to
e
v
aluate
the
indirect
trust
of
a
candidate
node
in
the
absence
of
an
y
direct
communicati
on
link
wi
th
another
node,
we
consider
the
recommendations
from
the
neighboring
member
nodes.
The
past
interactions
between
the
surrounding
member
nodes
and
the
candidate
node
is
used
to
establish
the
indirect
trust
v
alue
of
the
candidate
node.
In
this
model,
each
member
node
that
does
not
ha
v
e
an
y
past
communication
link
or
cooperation
e
xperience
with
the
candidate
node
will
then
mak
e
a
request
to
other
neighboring
member
nodes
for
pro
viding
recommendations
about
the
candidate
node.
By
this
method,
the
candidate
node’
s
trust
is
calculated
indirectly
Let
us
denote
such
an
indirect
trust
v
alue
as
T
I
ndir
ec
t
tr
ust
.
As
sho
wn
in
Figure
3,
node
3
has
no
direct
communicati
on
link
with
the
candidate
node
(node
2).
Ho
we
v
er
,
node
3
has
a
direct
communication
with
node
1
and
node
4.
Hence,
node
3
can
f
acilitate
a
request
t
o
node
1
and
node
4
t
o
mak
e
a
recommendation
about
the
candidate
node.
Therefore,
the
indirect
trust
e
v
aluation
model
will
be
adopted
for
node
3
to
compute
the
candidate
node’
s
trust
v
alue.
According
to
the
second
scenario
of
our
trust
e
v
aluation
model,
where
the
member
node
does
not
ha
v
e
direct
communication
link
or
cooperation
established
with
the
candida
te
node,
an
indirect
trust
v
alue
is
calcu-
lated.
In
this
case,
the
member
node
mak
es
a
request
to
other
neighboring
member
nodes
for
recommendations
about
the
candidate
node.
W
e
di
vide
these
surrounding
nodes
into
three
dif
ferent
types,
namely
reliable
nodes
(trustw
orth
y
nodes),
unkno
wn
nodes
and
unreliable
nodes.
Reliable
nodes
are
those
surrounding
nodes
that
ha
v
e
pro
vided
trust
w
orth
y
recommendations
in
the
past.
Let
(
T
r
el
i
abl
e
)
denote
the
trust
v
alue
recommended
A
trust-based
authentication
fr
ame
work
for
security
of
WP
AN
using
network
slicing
(Sazia
P
arvin)
Evaluation Warning : The document was created with Spire.PDF for Python.
1380
r
ISSN:
2088-8708
by
such
reliable
nodes.
Unkno
wn
nodes
are
t
hose
surrounding
nodes
from
whom
the
member
node
has
not
solicited
recommendations
in
the
past.
Let
(
T
unk
now
n
)
represent
the
trust
v
alue
recommended
by
thes
e
un-
kno
wn
nodes.
Based
on
history
of
e
v
ents,
if
a
neighboring
node
has
pro
vided
incorrect
recommendations
in
the
past,
it
is
considered
as
an
unreliable
node.
Let
(
T
unr
el
iabl
e
)
denote
the
trus
t
v
alue
recommended
by
unreliable
nodes.
In
the
Indirect
T
rust
v
alue
calculation,
the
member
node
considers
the
recommendations
from
reliable
and
unreliable
nodes
only
and
not
from
the
unreliable
nodes.
In
addition,
a
weight
f
actor
which
represents
the
de
gree
of
utility/importance
based
on
the
e
v
ents
that
were
monitored
and
quantified
are
gi
v
en
to
distinguish
between
the
recommendations
from
relia
ble
and
unkno
wn
nodes.
It
is
logical
to
e
xpect
a
higher
weight
f
actor
assigned
to
reliable
nodes’
recommendations
as
that
of
unkno
wn
nodes.
The
trust
v
alues
of
both
T
r
el
iabl
e
and
T
unk
now
n
are
required
to
be
first
calculated
in
order
to
determine
the
candidate
node’
s
Indirect
T
rust
v
alue.
The
process
of
calculating
trust
v
alue
from
reliable
n
odes
consists
of
tw
o
steps:
i)
the
member
node
retrie
v
es
the
trust
v
alues
s
tored
locally
from
the
reliable
nodes
the
y
ha
v
e
directly
e
v
aluated;
and
ii)
the
member
node
recei
v
es
trust
v
alues
from
reliable
nodes
who
ha
v
e
pro
vided
the
recommendations
about
the
candidate
node’
s
trust
v
alue.
The
member
node
combines
all
these
trust
v
alues
and
computes
T
r
el
iabl
e
using
2.
T
r
el
iabl
e
=
n
r
eliable
P
i
=1
T
M
R
i
T
R
C
i
n
r
el
ia
bl
e
(2)
where
T
M
R
i
denotes
the
trust
v
alue
from
member
node
(M)
to
i
th
reliable
node
(R);
T
R
C
i
denotes
the
trust
v
alue
from
reliable
node
(R)
to
candidate
node
(C)
and
n
r
el
iabl
e
denotes
the
total
number
of
reliable
nodes.
Ne
xt,
the
trust
v
alues
from
unkno
wn
nodes
T
unk
now
n
are
calculated
using
3.
T
unk
now
n
=
n
unk
now
n
P
k
=1
T
U
C
k
n
unk
now
n
(3)
where
T
U
C
k
denotes
the
trust
v
alue
from
k
th
unkno
wn
node
to
candidate
node
and
n
unk
now
n
denotes
the
total
number
of
unkno
wn
nodes.
Let
us
denote
a
weight
W
r
el
iabl
e
to
be
allocat
ed
to
the
reliable
nodes’
trust
v
alue
T
r
el
iabl
e
and
a
weight
W
unk
now
n
to
be
a
ssigned
to
unkno
wn
nodes’
trust
v
alue
T
unk
now
n
.
W
e
use
the
approach
gi
v
en
in
[Reference
thesis]
to
allocate
the
weights,
W
r
el
ia
bl
e
and
W
unk
now
n
.
W
e
adopt
t
he
traditional
weighted
approach
to
calculate
the
candidate
node’
s
o
v
erall
Indirect
T
rust
v
alue
as
follo
ws,
where
W
r
el
iabl
e
+
W
unk
now
n
=
1
and
W
r
el
iabl
e
;
W
unk
now
n
2
[0
;
1]
:
T
I
ndir
ectt
r
ust
=
T
r
el
iabl
e
W
r
el
iabl
e
+
T
unk
now
n
W
unk
now
n
(4)
3.3.
Integrated
trust
v
alue
calculation
It
is
typical
that
WP
AN
e
xhibits
dynamic
situations
from
time
to
time,
establishing
direct
as
well
as
indirect
communication
links
bet
ween
the
nodes.
An
inte
grated
trust
e
v
aluation
model
is
proposed
here
for
situations
where
both
direct
and
the
indirect
trust
v
alues
ha
v
e
bee
n
computed.
In
this
model,
we
combine
both
direct
trust
and
indirect
trust
v
alues
from
member
nodes
in
order
to
compute
the
candidate
node’
s
o
v
erall
trust
v
alue
indirectly
.
Let
us
denote
the
inte
grated
trust
v
alue
as
T
ig
.
In
Figure
3,
both
node
1
and
node
4
has
established
both
direct
and
indirect
communication
links
with
the
candidate
node
(node
2).
Therefore,
node
1
and
node
4
will
adopt
the
inte
grated
trust
e
v
a
luation
model
to
combine
the
direct
trust
as
well
as
Indirect
T
rust
v
alues
computed
from
both
links
for
determining
the
candidate
node’
s
o
v
erall
trust
v
alue.
In
the
third
scenario
of
trust
e
v
aluation
model
of
our
proposed
frame
w
ork,
the
Inte
grated
T
rust
v
alue
of
a
candidate
node
is
calculated
for
situations
when
the
candidate
node
has
established
both
direct
and
indirect
communication
links
with
member
nodes.
In
this
case
based
on
the
requirements
of
a
certain
task,
each
member
node
assigns
dif
ferent
weights
using
the
weighted
approach.
Let
us
denote
W
dt
as
the
weight
for
the
direct
trust
v
alue
and
W
it
as
the
weight
for
the
indirect
trust
v
alue.
Then,
the
candidate
node’
s
Inte
grated
T
rust
v
alue
is
calculated
by
the
follo
wing
equation,
where
W
dt
+
W
it
=
1
and
W
dt
;
W
it
2
[0
;
1]
T
ig
=
W
dt
T
dt
+
W
it
T
it
(5)
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1375
–
1387
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
1381
These
calculations
determine
if
a
cooperation
between
nodes
ha
v
e
tak
en
place
and
each
member
node
updates
the
Inte
grated
T
rust
v
alue
of
its
cooperation
node
(candidate
node)
accordingly
.
The
SUBS
can
no
w
mak
e
a
decision
on
an
SU’
s
request
to
authenticate
access
to
PU’
s
netw
ork
based
on
the
abo
v
e
calculated
trust
v
alue.
While
the
abo
v
e
trust
e
v
aluation
model
w
orks
for
e
xisting
SU
in
the
netw
ork,
it
does
not
cater
to
situations
when
a
ne
w
user
who
has
securely
joined
a
secondary
netw
ork
of
the
WP
AN
mak
es
a
request
to
access
the
PU’
s
netw
ork,
This
is
because
the
trust
v
alue
may
not
be
a
v
ailable
with
the
PUBS
or
other
PUs.
Hence,
we
propose
a
bootstrapping
method
to
address
such
a
ne
w
SU
case
scenario
in
the
ne
xt
subsection.
3.4.
Bootstrapping
of
trust
v
alue
calculation
W
e
propose
a
boot
strapping
method
for
the
PUBS
to
determine
the
trust
v
alue
for
a
ne
wly
joined
SU
under
tw
o
main
methods:
i)
T
riangular
T
rust
v
alue
calculation;
and
ii)
Reference
T
rust
v
alue
calculation.
The
algorithm
for
bootstrapping
a
ne
w
SU’
s
trust
v
alue
is
pro
vided
in
the
follo
wing
three
steps:
A
ne
wly
joined
SU
requests
the
SUBS
to
access
the
PU’
s
free
spectrum.
This
request
is
subjecti
v
e
to
the
trust
v
alue
determined
by
the
SUBS
and
is
based
on:
i)
its
joining
process
to
the
secondary
netw
ork;
and
ii)
its
beha
viour
pattern
observ
ed
for
a
certain
period
of
time.
On
successful
completi
on
of
this
authentication
process,
the
SUBS
sends
the
request
to
the
PUBS
for
further
action.
When
the
PUBS
recei
v
es
the
request
from
the
SUBS
for
the
SU
to
access
the
primary
netw
ork,
the
authentication
is
based
on
the
SU’
s
trust
v
alue.
At
this
stage,
s
ince
the
SU
node
has
no
trust
v
alue
with
either
the
PUBS
or
the
other
PUs,
triangular
trust
v
alue
calculation
method
is
in
v
ok
ed,
which
is
based
on
the
trust
relationship
between
the
PUBS
and
SUBS.
There
are
tw
o
pathw
ays:
i)
a
trust
relationship
e
xists
between
PUBS
and
SUBS-SU’
s
trust
v
alue
is
based
on
the
triangular
trust
v
alue
calculation
method;
or
ii)
no
trust
relationship
e
xists
between
the
SUBS
and
PUBS-trustw
orth
y
member
nodes
in
the
secondary
netw
ork
are
used
by
the
PUBS
and
the
SU’
s
trust
v
alue
is
based
on
reference
trust
v
alue
calculation
method.
W
e
pro
vide
a
detailed
algorithm
of
trust
v
alue
calculations
under
both
triangular
trust
and
reference
trust
methods
in
the
follo
wing
subsections.
3.4.1.
T
riangular
trust
v
alue
calculation
This
bootstrapping
method
is
used
when
some
trust
relationship
e
xists
between
the
SUBS
and
PUBS.
The
triangular
trust
calculation
method
is
used
here
to
determine
t
he
trust
v
alue
of
a
ne
w
SU
when
it
has
a
relationship
with
the
SUBS
b
ut
not
with
the
PUBS.
Let
us
ass
ume
the
trust
relationship
between
PUBS
!
SUBS;
SUBS
!
SU1
for
calculating
the
trust
v
alue
of
the
SU
node
(SU1),
and
denote
T
P
U
B
S
;S
U
1
as
the
trust
v
alue
of
the
PUBS
for
the
SU
node
(SU1).
Assuming
the
trust
v
alue
between
PUBS
and
SUBS
to
be
0.8,
and
the
trust
v
alue
between
SUBS
and
SU1
to
be
0.7,
we
ha
v
e
T
P
U
B
S
;S
U
B
S
=
0
:
8
and
T
S
U
B
S
;S
U
1
=
0
:
7
,
we
can
calculate
the
T
riangular
T
rust
v
alue
between
the
PUBS
and
SU1
as
follo
ws:
T
P
U
B
S
;S
U
1
=
T
P
U
B
S
;S
U
B
S
T
S
U
B
S
;S
U
1
=
0
:
8
0
:
7
=
0
:
56
3.4.2.
Refer
ence
trust
v
alue
calculation
This
bootstrapping
method
is
used
in
cases
where
there
is
no
relationship
between
PUBS
and
SUBS.
The
reference
trust
v
alue
calculation
for
a
ne
wly
joined
SU
is
based
on
the
recomme
n
da
tions
from
other
users.
The
dif
ference
here
from
the
indirect
trust
v
alue
calculation
method
is
that
the
SUs
and
not
the
PUs
request
for
recommendations
from
the
PUBS.
Here,
v
arious
trust
v
al
ues
recei
v
ed
are
collected
and
the
a
v
erage
of
these
v
alues
a
re
computed
to
determine
the
o
v
erall
trust
v
alue
for
the
ne
w
SU.
As
sho
wn
in
Figure
4,
let
us
assume
the
trust
relationship
between
dif
ferent
nodes
such
that
PUBS
trusts
SU2
and
SU3
with
a
v
alue
of
0.5
and
0.6,
respecti
v
ely
,
and
SU1
who
is
a
ne
wly
joined
SU
has
a
trust
v
alue
of
0.5
with
SU2
and
0.4
with
SU3.
I
n
this
case,
the
a
v
erage
trust
v
alue
between
the
PUBS
and
SU1
is
determined
as
follo
ws:
T
P
U
B
S
;S
U
1
=
(
T
P
U
B
S
;S
U
2
T
S
U
2
;S
U
1
+
T
P
U
B
S
;S
U
3
T
S
U
3
;S
U
1
))
=
2
=
(0
:
5
0
:
5
+
0
:
6
0
:
4)
=
2
=
0
:
249
=
)
0
:
25
Finally
,
the
PUBS
decides
to
declare
whether
the
SU
node
(SU1)
is
trustw
orth
y
or
not
based
on
whether
the
abo
v
e
computed
trust
v
alue
(either
through
T
riangular
T
rust
relationship
or
Reference
T
rust
rela-
tionship)
is
greater
than
the
predefined
threshold
v
alue
(
T
thr
eshol
d
)
or
not.
Hence,
the
predefined
threshold
v
alue
plays
a
major
role
in
determining
the
minimum
le
v
el
of
trust
required
by
the
PUBS.
A
trust-based
authentication
fr
ame
work
for
security
of
WP
AN
using
network
slicing
(Sazia
P
arvin)
Evaluation Warning : The document was created with Spire.PDF for Python.
1382
r
ISSN:
2088-8708
PU
BS
S
U
3
S
U
1
S
U
2
0
.
5
0
.
5
0
.
6
0
.
4
0
.
2
5
Figure
4.
A
ne
w
SU
node
within
a
reference
trust
relationship
4.
A
UTHENTICA
TING
AND
BIASING
PR
OBLEM
RESOLUTION
If
a
node
is
eligible
to
access
the
netw
ork
resources,
then,
it
will
authenticated
by
the
BS.
This
is
done
by
ha
ving
the
SUBS
compares
the
computed
trust
v
alue,
T
,
ag
ainst
the
predefined
threshold,
(
T
thr
eshol
d
)
.
The
candidate
node
is
authenticated
only
if
its
trust
v
alue
is
greater
than
the
required
threshold,
otherwise
it
request
is
rejected.
On
the
other
hand,
the
SUBS
may
forw
ard
the
request
to
the
PUBS,
which
also
obtains
the
node’
s
trust
v
alue
and
v
alidates
it
ag
ainst
the
authentication
threshold.
The
node
is
authenticated
only
if
it
has
trust
v
alue
higher
than
the
threshold.
There
are
se
v
eral
adv
antages
of
adopting
this
mechanism.
First,
the
netw
ork
can
support
tw
o
le
v
els
of
authentication.
Second,
certain
nodes
might
be
authenticated
for
certain
resources
b
ut
not
for
others,
and
finally
,
the
threshold
of
authentication
can
be
adjusted
in
order
to
relax
or
strict
access
based
on
ho
w
critical
the
application
is.
Biasing
may
occur
only
at
the
SUBS,
therefore,
it
is
processed
by
the
SUBS
by
asking
the
CA
to
pro-
vide
the
current
candidate
node’
s
trust
v
alue,
if
there
is
one.
The
SUBS
also
obtain
a
trust
recommended
v
alues
from
member
nodes.
The
CA
v
alidates
the
candidate
node’
s
trust
v
alue
it
has
through
the
public
and
pri
v
ate
repositories,
and
communicates
a
v
alidated
one
to
the
SUBS.
The
SUBS
compares
the
computed
recei
v
ed
v
alue
from
CA
with
the
calculated
one
through
members.
If
there
is
a
dif
ference,
then
the
SUBS
mark
the
candidate
node’
s
trust
v
alue
as
biased.
SUBS
then
check
the
v
alue
communicated
from
members
node
before
and
after
the
biasing
occurs.
The
SUBS
then
reports
nodes
that
pro
vide
dif
ferent
trust
v
alues
are
‘malicious
node’
or
‘biased
by
a
malicious
node’.
This
information
is
communicated
to
all
of
SUBS
member
nodes.
T
dt
=
n
P
i
=1
T
A
i
n
P
i
=1
T
A
i
+
n
P
i
=1
(1
T
A
i
)
=
0
:
57
+
0
:
71
+
0
:
42
(0
:
57
+
0
:
71
+
0
:
42)
+
(1
0
:
57)
+
(1
0
:
71)
+
(1
0
:
42)
=
0
:
56
(6)
T
r
el
iab
l
e
=
n
r
eliable
P
i
=1
T
M
R
i
T
R
C
i
n
r
el
iabl
e
=
0
:
45
0
:
5
+
0
:
6
0
:
45
+
0
:
5
0
:
7
3
=
0
:
28
(7)
5.
PERFORMANCE
EV
ALU
A
TION
W
e
assume
that
none
of
the
nodes
is
biased.
The
input
parameters
for
the
tw
o
netw
orks
are
the
number
candidate
nodes,
the
number
of
nodes
in
the
netw
ork,
and
the
threshold
for
trust.
The
first
netw
ork
acts
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1375
–
1387
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
1383
as
secondary
netw
ork
and
has
5
nodes.
The
second
netw
ork
acts
as
a
primary
netw
ork
and
also
has
5
member
nodes
and
one
base
station.
W
e
also
assume
that
each
has
3
criteria
for
trust.
W
e
also
assume
that
a
threshold
trust
v
alue
of
0.6
is
initially
used.
T
rust
v
alue
calculation
is
conducted
by
members
in
the
tw
o
netw
orks
using
direct
trust,
and
then
communicated
with
the
base
station.
The
final
trust
v
alue
is
them
calculated
by
the
base
station
and
sent
to
the
CA.
The
final
trust
v
alue
is
then
used
to
check
for
biasing
mak
e
a
decision
about
user
authentication.
In
this
section
we
demonstrate
authentication
base
d
on
the
trust
v
alue,
where
we
consider
three
cases
for
node
v
alidation
in
both
netw
orks,
in
the
first
we
address
node
authentication
is
granted
in
the
secondary
netw
ork,
then
node
authentication
is
granted
in
the
primary
netw
ork,
and
finally
,
node
authentication
is
denied.
F
or
authentication
in
the
secondary
netw
ork,
we
assume
that
node
2
is
t
he
candidate
node
in
the
secondary
netw
ork,
and
that
the
rest
of
nodes
will
pro
vide
their
trust
recommendation
about
node
2
to
the
secondary
netw
ork
base
station.
It
is
also
assumed
that
nodes
are
f
air
and
not
biased.
W
e
conducted
simulation
by
generating
100
repetiti
v
e
cases
for
the
success
and
f
ailure
for
each
time
period
for
three
criteria,
where
trust
v
alue
for
the
candidate
node
is
assigned
using
equations
1
and
6.
The
trust
v
alues
of
the
candidate
node
recommended
from
the
four
nodes
in
the
secondary
netw
ork
for
dif
ferent
time
periods
are
sho
wn
in
T
able
2.
Therefore,
the
a
v
erage
trust
v
alue
of
the
candidate
node
for
e
v
ery
criteria
can
be
calculated.
F
or
instance
for
the
first
criteria
it
is
equal
to
:
0
:
81+0
:
32+0
:
76+0
:
99
4
=
0
:
72
.
Then,
the
final
trust
v
alue
of
the
candidate
node
calculated
from
the
using
the
three
dif
ferent
criteria
as
:
0
:
72+0
:
67+0
:
45
3
=
0
:
61
.
T
able
2.
Recommended
trust
v
alue
calculated
from
nodes
in
the
secondary
netw
ork
using
three
criteria
Candidate
Node
Node
number
in
‘Netw
ork
1’
Criteria
1
Criteria
2
Criteria
3
4*Node
2
1
0.818
0.996
0.645
3
0.321
0.045
0.998
4
0.763
0.316
0.959
5
0.996
0.855
0.487
In
the
ne
xt
step,
the
computed
trust
v
alue
is
compared
with
the
trust
threshold
and
the
decision
is
made
based
on
this
comparison,
as
sho
wn
in
T
able
3.
Since
he
computed
final
trust
v
alue
of
0.61
is
lar
ger
than
the
threshold
v
alue
of
0.6,
the
SUBS
authentica
tes
the
candidate
node
to
accept
its
request
to
use
the
netw
ork
resources
or
forw
ards
its
request
to
the
secondary
netw
ork
as
sho
wn
in
T
able
3.
T
able
3.
Authentication
of
candidate
node
in
the
secondary
netw
ork
Candidate
node
T
rust
V
alue
of
Candidate
node
Final
trust
v
alue
Threshold
v
alue
threshold
?
Decision
Criteria
1
Criteria
2
Criteria
3
3*
Node
2
0.72
0.67
0.45
0.61
0.6
Y
es
Request
is
Netw
ork
1’
authenticated
or
forw
arded
to
‘
‘Netw
ork
2’
F
or
authentication
in
the
primary
netw
ork,
we
demonstrate
an
authentication
request
for
a
candi
d
a
te
node
for
the
resources
in
the
primary
netw
ork.
W
e
assume
that
the
pri
mary
netw
ork
has
fi
v
e
no
de
s
and
three
criteria,
where
node
5
is
assumed
to
be
the
candidate
node.
The
member
nodes
of
the
prim
ary
netw
ork
calculate
the
trust
v
alue
for
node
5
using
direct
trust
from
dif
ferent
time
periods
using
three
criteria
as
illustrated
in
T
able
4.
The
a
v
erage
trust
v
alue
of
the
candidate
node
for
each
criteria
is
calculated,
for
instance,
for
criteria
1
it
is
0
:
59+0
:
95+0
:
76+0
:
93
4
=
0
:
81
.
The
base
station
calculates
the
final
trust
v
alue
using
:
0
:
81+0
:
55+0
:
77
3
=
0
:
71
,
which
is
then
communicates
it
to
the
CA.
This
v
alue
is
also
used
compared
with
the
threshold
of
0.6,
and
as
a
result,
the
candidate
node
is
granted
access
to
the
primary
netw
ork
as
illustrated
in
T
able
5.
In
this
e
xample
we
sho
w
request
rejected
case,
where
the
same
steps
in
the
first
case
are
perform
ed,
where
node
number
2
in
the
secondary
netw
ork
is
the
candidate
node.
T
able
5
sho
ws
the
final
calculated
trust
v
alue
of
0
:
3
is
less
than
the
threshold
v
alue,
therefore,
the
request
is
not
authenticated
in
the
secondary
netw
ork.
In
other
w
ords,
the
candidate
node
is
not
considered
v
alid
and
in
its
o
wn
netw
ork,
therefore,
the
base
station
of
the
secondary
netw
ork
does
not
forw
ard
the
request
to
access
the
primary
netw
ork
resources.
Ne
xt,
we
consider
the
biased
nodes
case
where
a
number
of
biased
nodes
tak
es
part
in
the
authentication
process
of
certain
nodes.
This
e
xample
will
sho
w
ho
w
biasing
is
detected,
and
ho
w
the
base
station
resolv
es
this
iss
ue
along
with
the
CA.
W
e
assume
that
the
secondary
netw
ork
has
5
nodes,
and
that
node
3
is
the
candidate
node.
W
e
will
also
assume
A
trust-based
authentication
fr
ame
work
for
security
of
WP
AN
using
network
slicing
(Sazia
P
arvin)
Evaluation Warning : The document was created with Spire.PDF for Python.
1384
r
ISSN:
2088-8708
that
some
member
nodes
will
intentionally
assign
biased
trust
v
alues
for
the
candidate
node.
Once
biasing
is
detected,
resolution
is
performed
by
the
netw
ork
as
demonstrated.
T
able
4.
Recommended
trust
v
alue
calculated
from
nodes
using
three
criteria
in
primary
netw
ork
Candidate
Node
Node
number
in
‘Netw
ork
2’
Criteria
1
Criteria
2
Criteria
3
4*Node
5
1
0.599
0.996
0.645
3
0.960
0.045
0.998
4
0.765
0.316
0.959
5
0.931
0.855
0.487
T
able
5.
Recommended
trust
v
alue
calculated
from
nodes
using
three
criteria
in
primary
and
secondary
netw
orks
Candidate
node
T
rust
V
alue
of
Candidate
node
Final
trust
v
alue
Threshold
v
alue
threshold
?
Decision
Criteria
1
Criteria
2
Criteria
3
3*
Node
2
in
0.81
0.55
0.77
0.71
0.6
Y
es
Request
is
‘Netw
ork
1’
authenticated
Node
4
in
0.21
0.33
0.47
0.33
0.6
No
Request
is
‘Netw
ork
1’
Declined
The
candidate
node’
s
trust
v
alue
is
calculated
by
the
member
nodes
in
‘Netw
ork
1’,
where
all
member
nodes
are
f
air
.
F
or
criteria
1,
four
nodes
in
‘Netw
ork
1’
assign
a
trus
t
v
alue
to
the
candidate
node
accordingly
,
as
sho
wn
in
T
able
6
sho
ws
trust
v
alue
for
candidate
node
3
before
biasing.
The
final
trust
v
alue
for
the
candidate
node
as
calculated
by
the
base
station
in
the
secondary
netw
ork
is
0
:
8056+0
:
7842+0
:
5032
3
=
0
:
6976
.
This
v
alue
is
communicated
to
the
CA,
which
stores
it
as
the
same
for
pri
v
ate
v
alue
and
public
v
alue
for
each
node
in
the
database
before
biasing.
Assuming
that
member
node
5
is
biased,
and
that
it
will
assign
f
alse
trust
v
alues
for
the
candidate
node,
the
base
station
calculates
the
final
trust
v
alue
for
node
3
in
netw
ork
1
bec
omes
0.69,
0.93,
and
0.57
for
the
three
criteria
with
final
v
alue
of
0.80,
which
is
dif
ferent
from
the
trust
v
alue
computed
before
biasing
tak
es
place.
Once
this
trust
v
alue
is
communicated
with
the
CA,
it
will
be
detected
as
biased
due
to
lar
ge
dif
ference.
In
order
to
resolv
e
this
problem,
the
CA
identifies
all
the
trust
v
alues
for
all
member
nodes
in
both
netw
orks
in
the
repository
and
forw
ards
the
information
to
the
base
station
in
the
secondary
netw
ork
where
biasing
w
as
initiated.
This
e
xample
sho
ws
ho
w
the
biased
v
alue
is
detected
in
the
secondary
netw
ork.
Once
a
biasing
is
detected,
the
case
station
checks
for
the
trust
v
alue
assigned
by
its
nodes
for
all
criteria
as
illustrated
in
T
able
7,
which
sho
ws
that
node
5
w
as
biased
with
its
trust
assignment.
Therefore,
the
secondary
netw
ork
base
st
ation
communicates
this
information
to
all
nodes
in
the
netw
ork
and
e
xcludes
the
biased
node
from
further
contrib
utions
in
the
trust
calculation.
T
able
6.
T
rust
v
alue
calculated
in
secondary
netw
ork
using
three
criteria
Candidate
Node
number
Criteria
1
Criteria
2
Criteria
3
Node
in
‘Netw
ork
1’
4*Node
2
1
0.701
0.998
0.555
2
0.983
0.787
0.582
4
0.973
0.983
0.182
5
0.584
0.376
0.725
T
able
7.
T
rust
v
alue
calculated
for
biasing
detection
in
secondary
netw
ork
for
dif
ferent
criteria
Criteria
Criteria
Node
number
T
rust
v
alue
before
biasing
T
rust
v
alue
after
biasing
Similar
Decision
4*1
Node
1
0.7016
0.7016
Y
es
Not
biased
Node
2
0.9833
0.9833
Y
es
Not
biased
Node
4
0.9737
0.9737
Y
es
Not
biased
Node
5
0.5843
0.9999
No
Biased
4*2
Node
1
0.9982
0.9982
Y
es
Not
biased
Node
2
0.7866
0.7866
Y
es
Not
biased
Node
4
0.9833
0.9833
Y
es
Not
biased
Node
5
0.3760
0.9999
No
Biased
4*3
Node
1
0.5554
0.5554
Y
es
Not
biased
Node
2
0.5826
0.5826
Y
es
Not
biased
Node
4
0.1822
0.1822
Y
es
Not
biased
Node
5
0.7253
0.9999
No
Biased
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1375
–
1387
Evaluation Warning : The document was created with Spire.PDF for Python.