Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
9
, No
.
3
,
J
un
e
201
9
, pp.
1968~1
977
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v9
i
3
.
pp1968
-
19
77
1968
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Intellig
en
t
b
lack
hole d
etectio
n in
m
obile Ad
Hoc
n
etworks
Ya
se
r K
ha
m
ayseh,
Mun
eer
Bani Yassein
, Mai
Abu
-
Jaz
oh
Depa
rtment
o
f
C
om
pute
r
Scie
n
ce
,
Jordan
Univer
s
ity
for
Sci
enc
e
a
nd
Technol
og
y
,
Jordan
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Feb
9
, 2
01
8
Re
vised
Sep
1
4
, 2
01
8
Accepte
d
Oct
10
, 201
8
Secur
ity
is
a
cr
itical
and
ch
al
l
en
ging
issue
in
M
AN
ET
due
to
i
ts
open
-
na
ture
cha
ra
cteri
sti
cs
such
as:
m
obil
i
t
y,
wire
l
ess
comm
unic
at
ions,
se
l
f
-
orga
nizing
and
d
y
n
amic
to
polog
y
.
MA
NETs
are
comm
onl
y
the
t
arg
e
t
of
bla
ck
hol
e
at
t
ac
ks.
Th
ese
a
re
la
unc
hed
b
y
m
al
ic
ious
node
s
tha
t
joi
n
th
e
net
work
to
sabota
ge
and
dr
ai
n
i
t
of
it
s
reso
urc
es.
B
la
ck
hol
e
nodes
in
te
r
ce
p
t
exc
h
anged
dat
a
pac
k
et
s
and
sim
pl
y
drop
th
e
m
.
The
b
lack
ho
le
nod
e
uses
vul
ner
abilities
in
the
rou
ti
ng
pr
otoc
ol
of
MA
NE
TS
to
decla
r
e
it
s
el
f
as
th
e
c
losest
rel
a
y
node
to
an
y
desti
n
at
io
n.
Th
is
work
pr
oposed
two
de
tecti
on
p
rotoc
o
ls
base
d
on
the
col
l
ec
t
ed
datase
t,
namel
y
:
the
BDD
-
AODV
a
nd
H
y
brid
prot
ocol
s.
Bot
h
protoc
ols we
re
b
uil
t
on
top
of the
origi
nal
AO
DV
.
The
BDD
-
AO
D
V prot
ocol
depe
nds
on
the
fea
tur
es
col
l
ec
t
e
d
for
the
pre
ven
ti
on
and
de
te
c
tion
of
bla
ck
hole
atta
ck
tec
hnique
s.
On
th
e
othe
r
h
and,
the
H
y
br
id
pro
toc
ol
is
a
combina
ti
on
of
both
the
MI
-
AO
DV
and
the
proposed
BDD
-
AODV
protoc
ols.
Ex
te
n
sive
sim
ula
ti
on
expe
riments
wer
e
conduc
t
ed
to
e
val
ua
te
the
per
form
anc
e
of
the
proposed
a
l
gorit
hm
s.
S
imulat
ion
r
esult
s
show
tha
t
th
e
proposed
protocols
improved
th
e
detec
ti
on
and
pre
ven
ti
on
of
bla
ck
h
ole
nodes,
and
hence,
the
ne
twork
ac
hie
v
ed
a
highe
r
pac
ket
de
li
ver
y
rat
io
,
lower
droppe
d
pac
k
et
s
rat
io
,
and
lowe
r
over
hea
d
.
Ho
weve
r,
th
is
improvem
ent
le
d
to
a
slight
in
crea
se
in the
end
-
to
-
end
de
lay
.
Ke
yw
or
d
s
:
Aod
v
Bl
ack
h
ole
n
od
e
Detect
ion
Feat
ur
es
Hybr
i
d
Ma
nets
Rou
ti
ng
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Yaser K
ham
ayseh
,
Dep
a
rtm
ent o
f C
om
pu
te
r
Scie
nce,
Jo
r
da
n Un
i
ver
s
it
y fo
r Sci
ence
and Tec
hnolog
y
,
22110,
Irbid, J
orda
n.
Em
a
il
: y
aser@j
ust
.edu.
jo
1.
INTROD
U
CTION
A
m
ob
il
e ad
hoc n
et
work
(M
AN
E
T)
is a col
le
ct
ion
o
f
m
ob
i
le
d
evices t
hat
are conn
ect
e
d via wireless
li
nk
s;
these
de
vices
dynam
icall
y
con
sti
tute
a
tem
po
rar
y
ne
twork
without
a
need
f
or
a
central
m
anag
e
m
ent
or
netw
ork
in
fr
a
struct
ur
e.
Eac
h
m
ob
il
e
dev
ic
e
in
the
M
ANET
com
m
un
ic
at
es
directl
y
with
oth
e
r
m
ob
il
e
de
vices
within
it
s
rad
i
o
ra
ng
e
,
w
hile
it
can
com
m
u
nicat
e
with
no
des
that
are
lo
cat
ed
outsi
de
it
s
rad
io
range
thr
ough
interm
ediat
e
no
de
s.
Mo
reove
r,
al
l
the
node
s
m
us
t
cooper
at
e
in
for
wardin
g
the
pack
et
s
i
n
the
netw
ork
[1
]
-
[
3].
MANET
s
ha
ve
m
any
disti
nct
char
act
erist
ic
s:
d
ynam
ic
netw
ork
to
po
l
og
y,
no
in
fr
a
structu
re,
m
ulti
-
hop
routin
g,
li
m
i
ted
resou
rces,
a
nd
lim
it
ed
ba
ndwidt
h
[
4]
-
[
1
0
].
T
her
e
are
var
i
ou
s
ap
plic
at
ion
s
for
M
A
NETs
rangin
g
from
t
he
battl
efiel
d
to
cl
assr
oom
s
[
5]
,
[
1
1]
-
[
1
3
]
.
The
Ad
hoc
O
n
-
Dem
and
Dista
nce
Vect
or
(
AOD
V
)
is
a
reacti
ve
routin
g
protoc
ol
that
is
s
pe
ci
fical
ly
dev
el
op
e
d
f
or
MA
NETs
.
AOD
V
is
ba
sed
on
tw
o
m
echan
ism
s
fo
r
routin
g,
wh
ic
h
are:
r
ou
te
discov
e
ry
an
d
r
oute
m
ai
ntenan
ce
[
5],
[
8],
[
14
]
.
The
wi
des
pr
ea
d
us
e
of
MA
NET
s
is
bein
g
chall
en
ged
by
m
any
f
act
or
s,
i
n
pa
rtic
ular,
the
iss
ue
of
sec
ur
it
y.
N
od
e
s
are
ass
um
ed
to
trust
each
ot
he
r
an
d
to
w
ork
cl
os
el
y
with
each
ot
her
to
perform
the
ro
utin
g
operati
ons
.
T
his
trust
ass
umpti
on
vio
la
te
s
the
secur
it
y
pr
inci
ples
of
net
works
[3
],
[4
]
,
[14
]
.
Ther
e
a
re
m
an
y
ty
pes
of
at
tack
s
that
can
th
reate
n
MANET
s.
Bl
a
ck ho
le
att
acks
can gr
eat
ly
com
pr
omi
se the n
et
w
ork reso
ur
ces an
d degra
de
it
s p
er
form
ance [
6
].
In
t
he
se
at
ta
ck
s,
m
alici
ou
s
node
s
join
t
he
netw
ork,
cl
ai
m
that
they
ha
ve
the
best
path
to
destinat
io
n
nodes
durin
g
t
he
r
outi
ng
disc
ov
e
ry
op
e
rati
on, a
nd
finall
y, sim
ply
dro
p
al
l t
he dat
a p
ac
kets that
go th
rou
gh the
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
In
te
ll
igent
bla
c
k hole detec
ti
on in
mo
bile Ad
Ho
c
n
et
w
or
ks
(
Ya
ser
Kh
amay
seh
)
1969
Netw
orks
em
plo
y
seve
ral
m
echan
ism
s
t
o
protect
the
m
se
lves
fr
om
at
ta
cks.
Unf
or
t
un
at
el
y,
these
m
easur
es
do
not
gu
a
rant
ee
fu
ll
pr
otect
ion
f
ro
m
at
ta
c
ks
.
He
nce,
a
s
econd
li
ne
of
def
e
nce
that
ha
s
the
abili
ty
to
detect
e
m
erg
ing
vulne
rab
il
it
ie
s
i
s
need
e
d.
Ma
ny
syst
e
m
s
have
been
dev
el
oped
for
this
pur
pose,
m
os
t
no
ta
bly
i
ntr
us
io
n
detect
ion
syst
em
s
(I
DS
s
).
H
ow
e
ve
r,
t
hese
netw
or
k
-
base
d
IDSs
c
annot
deal
easi
ly
with
the
hu
ge
am
ou
nt
of
data
pa
ssing
thr
ough
the
net
work.
Ther
e
f
or
e,
a
f
ocused
featu
re
sel
ect
ion
strat
egy
f
or
decidin
g
on
th
e
m
os
t
i
m
po
rtant
featur
es
m
igh
t
b
e
ben
e
fici
al
.
Then
,
thes
e
sel
ect
ed
feature
s
can
be
fed
i
nto
a
le
arn
in
g
m
od
el
to
bu
il
d
a
n
eff
ect
ive
detect
ion
m
echan
ism
.
The
use
of
a
fo
c
us
e
d
datas
et
will
i
m
pr
ove
the
accuracy a
nd s
peed o
f dete
ct
ion
s
[3
]
,
[15]
,
[
16
].
Seve
ral
te
chn
i
qu
e
s
ha
ve
be
en
pro
posed
f
o
r
the
detect
ion
a
nd
the
preven
ti
on
of
s
uch
at
ta
cks
.
Gen
e
rall
y,
the
se
te
chn
i
ques
are
di
vid
e
d
int
o
tw
o
cat
e
gories
:
secur
i
ng
e
xisti
ng
prot
oc
ol
or
us
in
g
int
ru
si
on
detect
ion
syst
e
m
s
(I
DS)
[
8
]
.
P.
Ra
j
et
al
.
[
1
7
]
pr
opos
e
d
a
secur
e
AOD
V
r
ou
ti
ng
prot
oco
l
known
as
the
Detect
ion
,
P
re
ven
ti
on
a
nd
R
eact
ive
A
ODV
(
DP
R
A
ODV)
.
I
n
a
ddit
ion
to
the
basic
A
ODV
opera
ti
on
,
t
he
pro
po
se
d
prot
oc
ol
dev
el
op
e
d
introd
uced
the
dy
nam
ic
thresh
ol
d
value
.
Ac
cordin
g
t
o
t
he
pro
po
se
d
prot
oc
ol,
the
dynam
ic
threshold
value
i
s
cal
culat
ed
an
d
up
date
d
w
he
n
the
RR
EP
pa
cket
is
receive
d
from
the
repl
yi
ng
node
the
fir
st
tim
e.
The
m
a
i
n
draw
bac
k
of
this
te
chn
iq
ue
is
that
i
t
is
on
ly
base
d
on
a
hig
h
de
sti
na
ti
on
seq
uen
ce
num
ber
featu
re
to
diff
e
re
ntiat
e
a
m
al
ic
iou
s
node
from
a
no
rm
al
no
de
.
I
n
ad
diti
on
to
that,
the
netw
ork
over
hea
d
will
be
increase
d
du
e
to
the
exc
ha
ng
e
of
AL
A
RM
m
essages.
More
ov
e
r,
i
n
la
rg
e
netw
orks,
t
he
us
e
of
t
his
te
chn
i
qu
e
will
re
qu
i
re
a
lot
of
t
i
m
e
as
al
l
the
nodes
i
n
the
ne
twork
will
ha
ve
to
be
no
ti
fie
d.
Mi
str
y
et
al
.
[
18]
pro
posed
a
s
olut
ion
to
blac
k
ho
le
at
ta
ck
s
ba
sed
on
the
ori
gin
al
A
O
DV.
Their
appr
oach
dep
e
nd
s
m
ai
nly
on
m
od
ify
ing
th
e
functi
on
al
it
y
of
the
s
ource
node.
T
his
te
chn
i
qu
e
i
ncr
ea
ses
th
e
end
-
to
-
en
d del
ay
.
Kh
am
ay
seh
et
al
.
[
1
1
]
pro
pos
ed
a
MI
-
AOD
V
prot
oco
l,
whic
h
is
bu
il
t
on
top
of
t
he
or
igi
nal
AOD
V
protoc
ol.
Accord
i
ng
to
t
he
pr
opos
e
d
pr
oto
c
ol,
the
or
igi
nal
AODV
is
e
xte
nd
e
d
to
ha
ve
e
xtra
str
uctu
res,
wh
e
re
each
node
has
a
Tr
us
t
Table
con
ta
i
ning
th
e
ad
dr
esse
s
of
the
tru
st
node
s.
I
n
a
dd
it
io
n
to
that,
the
RR
EP
is
exten
ded
to
ha
ve
an
ad
dit
iona
l
fiel
d
(tru
st
f
ie
ld),
w
hich
is
us
e
d
to
ind
ic
at
e
if
the
rep
ly
ing
node
is
reli
able
or
no
t.
The
a
utho
rs
in
[
8
]
pro
pose
d
an
im
pr
ovem
ent
to
the
or
igi
nal
A
ODV
prot
oco
l
t
o
enh
a
nce
it
s
se
cur
it
y
against
blac
k
ho
le
at
ta
cks.
I
n
the
pro
pose
d
prot
oco
l,
ea
ch
node
i
n
the
MANET
has
two
ta
bles,
na
m
el
y
a
Su
s
pect
Ta
ble
an
d
a
Bl
ack
Table.
These
t
ables
a
re
desi
gn
e
d
to
c
onta
in
t
he
a
ddresse
s
of
t
he
inte
r
m
ediate
nodes
wh
ic
h
fa
il
ed
to
sen
d
the
data
thr
ough
them
.
In
ad
diti
on
to
that,
the
re
is
an
ad
diti
on
al
ackno
wled
ge
m
ent
pack
e
t. Kh
a
re
et
al
.
[
19
]
pro
pose
d
a
Secur
e A
d
-
hoc
O
n
-
De
m
and
Dista
nce
Vector
r
outi
ng
(S
A
ODV)
p
r
ot
oco
l
.
As
in
the
ori
gin
al
A
O
DV,
the
source
no
de
in
the
pro
po
s
ed
protoc
ol
trie
s
to
disc
ov
e
r
the
route
to
the
destinat
io
n
no
de
by
flo
od
i
ng
the
RR
EQ
m
essage
ac
ross
the
net
work.
T
he
ne
ig
hbouri
ng
node
s
recei
ve
the
RR
EQ
an
d
sen
d
the
RR
EP
ba
ck
to
the
sourc
e
node
if
they
hav
e
a
vali
d
route
in
their
r
ou
ti
ng
ta
ble
.
Acc
ordi
ng
to
the
S
A
ODV
protoc
ol,
th
e
s
ource
node
does
not
sel
ect
the
route
f
ro
m
the
first
re
ply.
T
he
m
ai
n
dr
aw
ba
ck
of
us
in
g
t
his tech
nique is t
hat it
incr
eases
the e
nd
-
to
-
en
d dela
y.
In
tr
us
i
on
Dete
ct
ion
Syst
em
s
(IDSs
)
ca
n
be
cl
assifi
ed
in
to
tw
o
m
ai
n
t
ypes:
an
om
aly
-
base
d
a
nd
sign
at
ur
e
-
base
d
(Mis
us
e
dete
ct
ion
m
od
el
)
[
20
]
.
A
n
IDS
de
pends
hea
vily
on
the
existe
nc
e
an
d
a
vai
la
bi
li
ty
of
good
a
nd
re
presentat
ive
dataset
s.
For
this
pur
po
se
,
in
19
99,
a
s
pecial
da
ta
set
,
known
as
the
KDDC
UP
’
99,
was
colle
ct
ed
from
real
network
trace
s
[
2
1
]
.
Feat
ur
e
r
edu
ct
io
n
is
us
ed
to
re
duce
the
com
plexity
and
com
pu
ta
ti
on
ti
m
e
of
the
m
odel
,
wh
e
re
us
i
ng
one
of
t
he
fe
at
ur
e
re
duct
io
n
te
chn
iq
ues
wi
ll
le
ad
to
an
in
crease
in
the
perf
or
m
ance
of
the
I
D
S.
I
n
[
20
]
,
W
a
’el
et
al
.
propo
sed
the
Ro
ugh
Set
Cl
assifi
ca
ti
on
Parall
el
G
eneti
c
Algorithm
(RSC
-
P
GA),
w
hich
is
a
hy
br
i
d
m
od
el
for
fe
at
ur
e
reducti
on.
This
m
od
el
w
as
te
ste
d
usi
ng
t
he
KDDCU
P’9
9
dataset
.
T
he
a
uthors
in
[2
2
]
e
m
plo
ys
r
ough
theo
ry
to
deal
with
va
gu
e
ne
ss
an
d
un
c
ertai
nty
in
decisi
on
syst
em
s.
This
too
l
i
s
us
e
d
to
ide
ntify
the
r
el
eva
nt
featur
e
s
f
ro
m
al
l
decisi
on
syst
e
m
featur
es
,
th
ereb
y
con
t
rib
uting
to
the
b
uildi
ng
of
an
ef
fici
ent
r
ule
set
f
or
cl
as
sific
at
ion
.
I
n
t
he
propose
d
m
od
el
,
the
featu
r
es
are
sel
ect
ed
acco
r
ding
to
the
ca
lc
ulati
on
of
both
t
he
discer
nib
il
it
y
m
at
rix
an
d
in
form
ation
gain
,
wh
e
re
the
cl
assifi
cat
ion
pro
du
ce
d
us
in
g
the
sel
ect
ed
featur
e
s
of
bot
h
m
et
ho
ds
gi
ve
s
bette
r
an
d
m
or
e
accurate
res
ults
than
t
he use
of
the sele
ct
ed fe
at
ur
es
us
i
ng ea
ch
m
et
ho
d
i
ndivid
ually
.
In
or
der
to
co
m
par
e
the
per
f
or
m
ances
of
di
ff
ere
nt
cl
assifi
ers
in
intrusi
on
detect
ion,
th
e
authors
in
[2
3
]
an
al
yse
d
t
he
C4.
5
a
nd
N
aï
ve
Ba
ye
s (
N
B) classi
fiers
usi
ng
th
e KDD
CUP’9
9
datase
t. Th
e a
uthors r
edu
ce
d
the
41
KDD
f
eat
ur
es
to
11
f
eat
ur
es
a
nd
7
featur
e
s
us
in
g
the
inf
or
m
at
ion
gain
m
easure
.
The
NB
a
nd
C4.
5
cl
assifi
ers
were
evaluated
on
the
dataset
us
ing
al
l
the
featur
es
,
the
sel
ect
ed
11
feat
ur
es
and
the
sel
ect
ed
7
featur
e
s.
T
he
r
esults
of
t
he
exp
e
rim
ents
sh
ow
e
d
that,
in
gen
e
ral,
the
overall
pe
rfor
m
ance
of
the
C
4.5
was
bette
r
tha
n
tha
t
of
t
he
NB
,
w
her
e
t
he
C4
.5
had
the
lo
west
error
rate
a
nd
the
hi
gh
e
st
cl
assifi
cat
ion
ac
cur
acy
.
On
the
ot
her
ha
n
d,
th
e
C4
.5
increase
d
t
he
t
ru
e
posit
ive
ra
te
an
d
ac
hieve
d
bette
r
detect
ion
rates.
Mo
r
eov
e
r,
the
res
ults
f
or
bo
t
h
cl
assifi
e
r
s
sho
wed
that
their
perform
a
nce
becam
e
m
or
e
ef
fici
ent
w
hen
the
i
nfo
rm
at
ion
gain
te
ch
ni
qu
e
was
use
d
t
o
se
le
ct
the
releva
nt
featu
res.
R
e
su
lt
s
show
t
hat
the
inf
or
m
at
ion
gai
n
is
a
su
c
cessf
ul
m
easur
e
for
sel
ect
ing
th
e m
os
t rele
va
nt fea
tur
es [
2
4]
,
[
25
].
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
3
,
June
2019
:
1968
-
1977
1970
Re
searche
rs
in
[
20
]
pro
pose
d
a
set
of
rel
evan
t
featu
res
that
co
ntribut
e
to
the
detec
ti
on
of
D
oS
at
ta
cks.
F
or
th
is
pur
po
se
,
th
ey
app
li
ed
the
N
SL
K
D
D
da
ta
set
to
the
C4.
5
al
gorith
m
with
a
6
-
fol
d
cr
oss
validat
io
n
us
i
ng
the
Or
a
nge
c
anv
a
s
data
m
ining
to
ol
to
a
na
ly
se
this
dataset
.
The
perfor
m
ance
was
m
e
asur
e
d
in
te
rm
s
of
the
accuracy
an
d
cl
assifi
cat
ion
ti
m
e.
The
obta
in
ed
res
ults
sho
wed
t
hat
the
in
crea
se
in
the
re
le
van
t
featur
e
s set
reduced
the classi
ficat
ion
ti
m
e a
nd inc
reased
th
e cla
ssific
at
ion acc
uracy
.
The
a
uthors
i
n
[
2
6
]
pro
pose
d
a
detect
io
n
te
chn
i
qu
e
f
or
s
el
ect
ed
black
ho
le
at
ta
cks
by
dep
loyi
ng
sever
al
IDS no
des
in M
AN
E
T. Th
e
I
DS
nodes
pe
r
form
an
an
ti
-
blac
k
hol
e functi
on that
est
i
m
at
es a su
sp
ic
ious
value
of
t
he
node
.
T
his
is
c
al
culat
ed
base
d
on
the
a
bnorm
al
diff
ere
nc
e
b
et
ween
the
RR
EP
and
RR
EQ
m
essages
tran
s
m
itted
from
th
e
node.
In
[
2
7
]
,
the
auth
or
s
pro
po
se
d
the
I
DAD
te
ch
niqu
e,
w
hich
is
ba
sed
on
bo
t
h
I
ntr
us
io
n
Detect
ion
a
nd
Anom
al
y
Dete
ct
ion
.
T
his
te
c
hn
i
qu
e
assum
es
that
the
act
iv
it
ie
s
of
any
node
in
the
netw
ork
is m
on
it
o
red
. I
n
add
it
io
n,
it
ass
um
es
that
the
a
no
m
al
ou
s
act
iv
it
ie
s
of
an
intr
uder
ca
n
be
i
dent
ifie
d
from
the
no
rm
al
act
ivit
ie
s
of
the
norm
al
n
od
e
s.
T
he
au
dit
data,
wh
ic
h
is
a
pr
e
-
colle
ct
ed
set
of
a
no
m
al
ous
act
ivit
ie
s,
sh
ou
ld
be
avail
able
to
the
IDA
D
syst
e
m
,
wh
ere
this
syst
e
m
com
par
es
ever
y
act
ivit
y
in
the
networ
k
with the
audit
data to
detect
t
he blac
k h
ole
node
.
Fr
om
the
pr
e
vi
ou
s works,
it
w
as
note
d
that
m
any
of
the p
r
opos
e
d
te
ch
nique
s
for
sec
ur
i
ng
an
existi
ng
protoc
ol
agai
nst
black
ho
le
a
tt
acks
la
cked
a
fu
ll
un
d
ersta
ndin
g
of
the
be
hav
i
our
of
the
netw
ork
wit
h
r
egards
to
black
hole
at
ta
cks.
More
ov
er,
m
os
t
of
the
ID
S
-
based
te
c
hn
i
qu
e
s
us
e
ei
ther
the
K
D
DC
UP
’
99
dataset
or
the
NS
L
K
D
D
dataset
.
T
he
la
tt
er
is
a
s
ubset
of
the
ori
gi
nal
K
DD
dataset
.
B
oth
dataset
s
a
r
e
out
-
dat
ed
an
d
wer
e
colle
ct
ed
for
wire
d
dom
ai
n
at
ta
cks.
A
ne
w
MAN
ET
-
ori
ented
dataset
with
a
fo
c
us
e
d
and
ef
fecti
ve
featur
e
sel
ect
ion
m
echan
ism
was
create
d
by
[
28
]
.
This
dataset
is
util
iz
ed
to
de
sign
tw
o
te
ch
niques
to
sec
ure
the
AOD
V protoc
ol ag
ai
ns
t
black ho
le
att
a
c
ks
.
This
paper
is
div
ide
d
into
4
sect
ion
s;
S
ect
ion
1
prese
nts
an
ov
e
r
view
of
MA
NET
S
a
nd
black
hole
at
ta
cks.
Mo
re
over
,
it
prese
nts
an
d
discusse
s
pr
e
vious
stu
dies
in
relat
ion
to
this
resea
rc
h.
Sect
io
n
3
des
cribes
the
pro
posed
s
chem
es:
the
B
DD
-
A
O
DV
pr
oto
c
ol,
an
d
Hy
br
i
d
prot
oco
l
.
The
re
su
lt
s
of
the
pro
posed
s
olu
ti
on
are
pr
ese
nted
in
Sect
io
n 4. Fi
nally
, S
ect
io
n 5 c
on
cl
ud
e
s thi
s work a
nd s
uggests s
om
e ideas for
futu
re
w
ork.
2.
PROP
OSE
D SCHE
MES
A
host
-
base
d
I
DS
was
us
e
d
by
the
no
des
in
the
MA
NET
,
with
eac
h
no
de
bei
ng
e
quip
ped
with
a
n
ID
S t
o
colle
ct
the audit
d
at
a.
Me
anwhil
e, an ano
m
al
y
-
based
d
et
ect
io
n
sc
hem
e w
as u
se
d for the
detect
ion
task
.
This
w
ork
pr
opose
d
tw
o
te
ch
niques
to
de
te
ct
and
pre
ven
t
bl
ack
hole
at
ta
c
ks
.
T
hese
te
ch
niques
use
the
dataset
dev
el
op
e
d
by [
28
]
.
The
g
ene
r
al
fr
am
ewo
r
k
c
on
si
der
e
d
in
t
hi
s
pa
per
is
s
hown
in
Fig
ur
e
1
.
The
detect
io
n
ph
a
se
is co
ns
ide
red in this
work.
Figure
1
.
Pro
pose
d
I
ntr
us
io
n Dete
ct
ion
M
od
el
To
e
xam
ine
the
abili
ty
of
t
he
dataset
in
[
28
]
to
detect
blac
k
ho
le
at
ta
c
ks
,
two
r
ou
ti
ng
protoc
ols
were
pro
po
se
d,
nam
el
y:
the
BDD
-
AOD
V
protoc
ol
an
d
t
he
Hyb
rid
protoc
ol,
w
her
e
these
prot
oco
ls
de
pe
nd
on
the
BDD
feat
ur
e
s.
The
f
ollo
wing
su
bse
ct
ions
de
m
on
strat
e
the
work
i
ng
m
ech
anism
s
of
the
BDD
-
A
O
DV
prot
oco
l
and the
Hyb
rid
pro
t
oco
l.
2.1. BD
D
-
AODV
p
r
otoc
ol
The
B
DD
-
AODV
pr
oto
c
ol
m
od
ifie
s
the
beh
a
viou
r
of
the
or
i
gin
al
A
ODV,
w
here
each
node
is
require
d
to h
a
ve
three
ta
bles:
a
Trust
Table, a
Bl
ack
Table, an
d
a
Co
un
t
T
able.
T
he
T
ru
st
Table
incl
ud
es
a
li
st
of
tr
us
te
d
node
s
in
the
net
w
ork,
wh
il
e
the
Bl
ack
Table
c
on
ta
in
s
a
li
st
of
the
no
des
m
a
rk
e
d
as
black
nodes
.
The
C
ount
Ta
ble
m
ai
ntains
t
he
f
ollo
wing
st
at
ist
ic
s
abo
ut
t
he
on
-
goin
g
ac
ti
viti
es
of
the
netw
ork
for
th
e
rep
l
y
node:
co
un
t of
RR
EP
from
this
node,
c
ount
of
m
axi
m
u
m
destinat
ion
seq
ue
nce
num
ber
fro
m
the
RR
EP
of
this
node,
a
nd
co
unt
of
t
he
lo
w
hop
c
ount
re
ply
from
the
RR
E
P
of
t
his
node
.
More
ov
e
r,
the
RR
EP
in
the
BDD
-
AOD
V
prot
oc
ol
has
tw
o
ext
r
a
fiel
ds
,
nam
el
y,
the
rep
ly
in
g
node
an
d
the
r
eply
ing
ho
p
co
un
t.
Fig
ur
e
2
de
picts
the BD
D
-
A
ODV
al
go
rithm
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
In
te
ll
igent
bla
c
k hole detec
ti
on in
mo
bile Ad
Ho
c
n
et
w
or
ks
(
Ya
ser
Kh
amay
seh
)
1971
H
a
ndle
RREQ
M
ess
a
g
e
H
a
ndle
RREP M
ess
a
g
e by So
urce
no
de
If
(
so
u
rce
n
o
d
e of
the RR
EQ
is n
o
t ex
ist in
T
T
)
Ins
ert
so
u
rce
no
d
e of
the RR
EQ
to
T
T
If
(
d
estin
atio
n
no
d
e of
the RR
EQ
is n
o
t in TT
)
Ins
ert
des
tin
atio
n
no
d
e o
f
the RR
EQ
to
TT
If
(
last n
o
d
e that
sen
d
s th
e RREQ i
s n
o
t in TT
)
Ins
ert
last no
d
e of
the RR
EQ
to
TT
If
(
Inter
m
ed
iate
no
d
e do
esn
’t
h
av
e valid
r
o
u
te to
d
estin
atio
n
)
Re
-
b
road
cast R
REQ
If
(
RR
EQ
is
r
ec
eiv
ed
by
d
estin
atio
n
no
d
e,
in
ter
m
ed
iate
has
a
v
alid
r
o
u
t or
m
ali
c
io
u
s n
o
d
e)
Un
icast RR
EP
to
t
h
e so
u
rce
n
o
d
e
If
(des
tin
atio
n
n
o
d
e
o
f
th
e
RR
EP
is
n
o
t
in
TT
)
Ins
ert
d
estin
atio
n
n
o
d
e
o
f
th
e
RR
EP
to
T
T
If
(
RR
E
P sen
d
er! = Rep
ly
in
g
no
d
e)
Ad
d
RREP sen
d
er
to
the TT
If
(
Rep
l
y
in
g
no
d
e is no
t in CT)
Ad
d
Rep
ly
in
g
no
d
e to th
e CT
Co
u
n
t of
RRER f
ro
m
the Rep
l
y
in
g
n
o
d
e =1
If
(
d
st seq
u
en
ce
nu
m
b
e
r
> 62
9
2
)
Max h
ig
h
seq
u
en
ce =1
Else
m
ax
hi
g
h
seq
u
en
ce =
0
If
(
1
<
r
ep
l
y
in
g
h
o
p
count <= 2)
Low repl
y
in
g
cou
n
t =1
Else
low repl
y
i
n
g
cou
n
t
=0
Else
I
f
(
Rep
l
y
in
g
no
d
e in CT)
+
+
Co
u
n
t o
f
RRER
If
(
d
estin
atio
n
seq
u
en
ce nu
m
b
er
>
6
2
9
2
)
Max h
ig
h
seq
u
en
ce
d
estin
atio
n
++
If
(
replying
ho
p
cou
n
t >
1
and
<=
2
)
Low
re
p
ly
in
g
ho
p
cou
n
t ho
p
++
If
(
reply
in
g
no
d
e exis
ts in
T
T)
Sen
d
data p
acket
Else I
f
(
replying
n
o
d
e exis
t in BT)
Discard R
R
EP
an
d
wait
an
o
th
er
RR
EP
Else I
f
(
Co
u
n
t of
RRER f
ro
m
the
Rep
ly
in
g
no
d
e =
m
a
x
hig
h
seq
u
en
ce
des
tin
atio
n
=
replying
h
o
cou
n
t and
Co
u
n
t of
RRER f
ro
m
th
e Rep
ly
in
g
n
o
d
e>=1
)
Ad
d
Rep
ly
in
g
no
d
e to th
e BT
If
(
reply
in
g
no
d
e exis
ts in
BT)
Disca
rd RR
EP
an
d
wait
an
o
th
er
RR
EP
Else sen
d
d
ata pac
k
et
H
a
ndle
RREP M
ess
a
g
e by Inter
m
e
dia
te
no
de
If
(
so
u
rce
n
o
d
e of
the RR
EP
is n
o
t exi
st in
T
T
)
Ins
e
rt
so
u
rce
n
o
d
e of
the RR
EP
to
T
T
If
(des
tin
atio
n
node o
f
the RR
EP
is
n
o
t exis
t in TT
)
Ins
ert
d
estin
ati
o
n
no
d
e o
f
the RR
EP
to
T
T
If
(RREP sen
d
er
!
= Rep
l
y
in
g
no
d
e)
Ad
d
RREP
sen
d
er
to
T
T
Figure
2. BD
D
-
AOD
V
al
go
rithm
2.
2.
H
yb
ri
d p
rot
oco
l
The
Hy
br
id
protoc
ol
com
bin
es
bo
t
h
BDD
-
AOD
V
an
d
MI
-
AOD
V
[
1
1
]
protoc
ols
to
de
rive
a
m
or
e
reali
sti
c
and
ef
fecti
ve
detect
ion
m
echan
ism
.
In
the
M
I
-
A
ODV
pr
oto
c
ol,
the
trust
fiel
d
is
us
ed
to
in
di
cat
e
if
the
re
plyi
ng
no
de
is
reli
a
ble
or
not.
It
is
gi
ve
n
a
value
of
ei
ther
0,
1
or
2.
I
n
the
Hyb
rid
prot
oco
l,
e
ach
node
is
pro
vid
e
d
with
two
ta
bles,
na
m
el
y,
a
Tru
st
Table
a
nd
a
Bl
ack
Ta
ble.
In
a
dd
it
i
on
to
that,
ever
y
s
ource
node
i
n
the
netw
ork
is
pro
vid
e
d
with
a
Co
un
t
Table
.
The
processes
f
or
the
ha
ndli
ng
of
the
RR
E
Q
i
n
the
BD
D
-
AOD
V
protoc
ol
an
d
Hybr
i
d
prot
oc
ol
are
sim
il
ar.
Figure
3
sho
ws
the
al
gorit
hm
fo
r
the
ha
nd
li
ng
of
the
RR
EQ
m
essages.
Figur
e
3.
Handli
ng of RRE
Q
Messages
in H
ybrid
Protoc
ol
Figure
4
s
hows
the alg
ori
thm
for
the
h
a
ndli
ng
of the RREP
by the
source
nod
e
.
H
a
ndle
RREP M
ess
a
g
e by So
urce
no
de
If
(
d
estin
atio
n
no
d
e of
the RR
EP
d
o
es n
o
t exis
t in th
e T
T)
Ins
e
rt
d
estin
atio
n
no
d
e o
f
th
e RREP to T
T
If
(
RR
EP
sen
d
er
!
= Rep
ly
in
g
no
d
e)
Ad
d
RREP
sen
d
er
to
the T
T
If
(
Rep
ly
in
g
no
d
e do
es n
o
t exis
t in CT)
{
Ad
d
Rep
ly
in
g
no
d
e to th
e CT
Co
u
n
t of
RRER fr
o
m
the Rep
l
y
in
g
no
d
e =1
If
(
d
estin
atio
n
se
q
u
en
ce nu
m
b
er
>
6
2
9
2
)
Max h
i
g
h
s
eq
u
en
ce
d
estin
atio
n
=1
Else
m
ax
hig
h
se
q
u
en
ce desti
n
atio
n
=0
If
(
replying
ho
p
cou
n
t >
1
and
<=
2
)
Low
r
ep
l
y
in
g
ho
p
count h
o
p
=1
}
Else
low repl
y
in
g
ho
p
cou
n
t hop =0
}
Else I
f
(
R
ep
ly
in
g
no
d
e exis
ts in
the CT)
{
Co
u
n
t of
RRER f
ro
m
the
Rep
ly
in
g
n
o
d
e ++
If
(
d
estin
atio
n
se
q
u
en
ce nu
m
b
er
>
6
2
9
2
)
m
ax
hig
h
seq
u
en
ce ++
If
(
replying
ho
p
cou
n
t >
1
and
<=
2
)
lo
w r
ep
l
y
in
g
ho
p
cou
n
t ++}
If
(
Tr
u
st f
ield
valu
e =
1
or 2)
If
(
Rep
ly
in
g
no
d
e do
es n
o
t exis
t in TT
)
Ins
ert
Rep
ly
in
g
no
d
e to TT
an
d
Send
data
p
acket
Else I
f
(
Tr
u
st f
ield
v
alu
e =
0
)
If
(
Co
u
n
t of
RR
ER f
ro
m
the R
ep
ly
in
g
no
d
e =
m
ax
hig
h
seq
u
en
ce desti
n
atio
n
=
replying
to co
u
n
t and
Co
u
n
t of
R
RER
f
ro
m
the Rep
ly
in
g
n
o
d
e >
= 1)
Ad
d
Rep
ly
in
g
no
d
e to
the
BT
Else I
f
(
Tr
u
st f
ield
v
alu
e equ
al
-
1)
If
(
Rep
ly
in
g
no
d
e do
es n
o
t in BT)
Ins
ert
R
ep
ly
in
g
n
o
d
e
to
BT
If
(
Rep
ly
in
g
no
d
e do
es n
o
t exis
t in BT)
S
sen
d
s th
e data
p
acket
Else Disca
rd RR
E
P and
wait
f
o
r
an
o
t
h
er
RR
EP}
Figure
4. Al
gorithm
f
or
t
he h
and
li
ng
of the
RR
EP b
y t
he
s
ource
node
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
3
,
June
2019
:
1968
-
1977
1972
3.
SIMULATI
O
N AND
A
NAL
YS
IS
OF R
ESULT
S
To
evalu
at
e
the
per
f
or
m
ance
of
the
pro
po
s
ed
protoc
ols,
s
i
m
ulati
on
exp
e
rim
ents
wer
e
cond
ucted
.
GloMoSi
m
s
i
m
ula
tor
was
use
d
to
eval
uate
the
per
f
orm
ance
of
f
our
di
fferent
prot
oc
ols,
nam
ely,
the
Hybr
i
d
protoc
ol,
BD
D
-
A
ODV
protoc
ol,
MI
-
A
ODV
protoc
ol,
an
d
ori
gin
al
A
O
DV
protoc
ol.
Ta
ble
1
su
m
m
arize
s
the
si
m
ulati
on
p
a
r
a
m
et
ers.
Table
1.
Sim
ul
at
ion
Pa
ram
et
e
rs
Para
m
e
ter
Valu
e
Para
m
eter
Valu
e
Si
m
u
lato
r
Glo
MoSi
m
2.0
3
Mini
m
u
m
no
d
es’ v
elo
city
0
.5
m
eter
/seco
n
d
Si
m
u
latio
n
du
ration
1
2
0
0
seco
n
d
s
Maxi
m
u
m
no
d
es’
v
elo
city
2
m
eter
/seco
n
d
Nu
m
b
e
r
o
f
no
d
es
2
0
,
2
5
,
3
0
,
3
5
,40
Rad
io
r
an
g
e
2
5
0
m
Mob
ility Mo
d
el
Ran
d
o
m
w
ay
p
o
in
t
Ban
d
wid
th
2
M
b
/s
Pau
se ti
m
e
0
,
1
0
Tr
af
f
ic T
y
p
e
CBR
Si
m
u
latio
n
ar
ea
1
0
0
0
m
* 1
0
0
0
m
CB
R p
acket si
ze
512
tes
Four
pe
rfor
m
ance
m
at
rices
wer
e
m
easur
ed
:
Packet
Deli
ve
ry
Ra
ti
o
(PD
R),
Dro
pp
e
d
P
ackets
Ra
ti
o,
Av
e
ra
ge
En
d
-
to
-
E
nd
Delay
a
nd
O
verhea
d.
Eigh
t
diff
e
re
nt
scenari
os
that
wer
e
sim
ulate
d
in
t
his
rese
arch.
These
sce
nar
i
os
wer
e
sim
ulated
with
differe
nt
par
am
et
ers,
wh
e
re
these
pa
ram
et
ers
had
a
direct
i
m
pact
on
t
he
black
ho
le
at
ta
ck.
T
hese
pa
ra
m
et
ers
are:
the
total
nu
m
ber
of
nodes,
t
he
num
ber
of
black
ho
le
nodes
,
and
th
e
pause
ti
m
e
(2
pause
ti
m
e
val
ues
wer
e
te
ste
d:
0
an
d
10.
A
pa
us
e
ti
m
e
of
0
in
dicat
es
a
hi
gh
m
ob
il
it
y,
wh
il
e
a
pause ti
m
e o
f
10 in
dicat
es a
slow m
ob
il
it
y).
3.1. Hig
h m
obi
li
ty
s
cen
ario
s
In
al
l
the
fo
ll
ow
i
ng
fi
gures,
the
blu
e
li
ne
re
pr
ese
nts
t
he
ori
gin
al
A
O
DV
pr
oto
c
ol,
the
re
d
li
ne
represe
nts
the
MI
-
A
O
DV
pro
tocol,
the
gree
n
li
ne
re
presen
ts
the
BDD
-
A
ODV
protoc
ol,
and
t
he
ora
nge
li
ne
represe
nts the Hy
br
i
d protoc
ol
.
Figure
5
s
how
s
an
im
pr
ovem
ent
in
the
Pac
kets
Deli
ve
ry
Ra
ti
o
(PDR)
f
or
al
l
the
4
protoc
ols
f
or
a
netw
ork
t
hat
w
as
bei
ng
at
ta
ck
ed
by
1
a
nd
2
black
hole
(s
).
The
obta
ined
r
esults
s
how
t
ha
t
the
Hybr
i
d
prot
oco
l
in
the
netw
ork
,
wh
ic
h
was
at
ta
cked
by
one
black
ho
le
no
de
,
ha
d
the
hi
ghest
pack
et
deli
ver
y
rati
o.
As
sh
ow
n
in
Figure
5,
t
he
PD
R
increas
ed
w
hen
the
num
ber
of
no
de
s
increase
d.
As
a
resu
lt
of
inc
reasin
g
the
nu
m
ber
of
norm
al
no
des
in
the
net
wor
k,
the
source
node
had
a
bette
r
chan
ce
of
rece
iving
t
he
RR
EP
m
essages
fro
m
the
norm
al
and
re
li
able
nodes.
I
t
was
note
d
that
the
e
xistence
of
a
sec
ond
blac
k
hole
node
ca
us
e
d
a
sli
gh
t
decr
ease
in
t
he
p
ac
ket d
el
i
very
r
at
io.
(a)
(b)
Figure
5
.
Pac
ke
t D
el
ivery Rat
io w
it
h Pa
us
e
Ti
m
e 0
,
(a
) 1 B
la
ck
H
ole,
(
b)
2
Bl
ack
Holes
Figure
6
sho
w
s
that
the
BD
D
-
A
ODV
a
nd
H
ybrid
proto
cols
con
t
rib
uted
to
a
decr
ea
se
in
t
he
num
ber
of
dro
pped
p
ac
kets co
m
par
e
d t
o
the
or
i
gin
al
AOD
V
a
nd
M
I
-
A
ODV pr
ot
oc
ols for a
netw
ork
t
hat w
a
s att
acked
by 1 an
d 2 bla
c
k ho
le
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
In
te
ll
igent
bla
c
k hole detec
ti
on in
mo
bile Ad
Ho
c
n
et
w
or
ks
(
Ya
ser
Kh
amay
seh
)
1973
(a)
(b)
Figure
6
.
Pac
k
et
D
r
op Rat
io
with Pa
us
e
Tim
e 0
,
(a)
1
Bl
a
ck Ho
le
,
(b) 2
Bl
ack Hole
As
sho
wn
i
n
F
igure
6,
t
he
nu
m
ber
of
dro
pp
ed
pac
kets
dec
reased
as
t
he
num
ber
of
node
s
increase
d
from
20
to
40.
W
it
hin
this
i
nt
erv
al
,
eac
h
s
ource
no
de
w
as
surr
ounded
by
m
or
e
no
rm
al
neig
hbo
rs,
w
hi
le
the
nu
m
ber
of
blac
k
hole
node
s
in
the
netw
ork
w
as
fixe
d
at
2.
T
her
e
fore,
th
e
s
ource
no
des
in the
netw
ork
ha
d
the
chan
ce
to
rec
ei
ve
m
or
e
al
te
rn
at
ive
act
ive
routes
to
the
destinat
io
n
fro
m
the
norm
al
an
d
reli
able
nodes
.
The
a
gr
eem
en
t
betwee
n
t
he
res
ults
of
the
dro
pp
e
d
pac
ke
ts
an
d
pack
et
delivery
rati
o
cou
l
d
be
obse
rv
e
d.
More
ov
e
r, i
ncrea
sing t
he n
umber
of
black h
ol
es led to
inc
re
ase in t
he nu
m
ber o
f dro
pp
e
d packets
.
Figure
7
s
how
s
the
end
-
to
-
e
nd
delay
resu
lt
s
for
the
Hyb
r
id,
BD
D
-
AOD
V,
MI
-
A
ODV
and
orgi
nal
AOD
V
proto
c
ols
w
hen
the
ne
twork
was
at
ta
cked
by
1
an
d
2
blac
k
hole
s.
O
n
the
oth
e
r
hand,
Fig
ure
7
sho
ws
the end
-
to
-
e
nd
delay
r
es
ults w
hen the
netw
ork was
at
ta
cke
d by tw
o black
hole
no
des.
Acc
ordin
g
to
the
re
su
lt
s
in
both
fig
ur
es
,
the
Hy
br
i
d
a
nd
B
DD
-
AOD
V
prot
oc
ols
in
creased
t
he
e
nd
-
to
-
e
nd
delay
com
par
ed
to
both
the
MI
-
A
O
DV
an
d
ori
gi
nal
A
O
DV
prot
oco
ls.
For
a
netw
ork
at
ta
cke
d
by
on
e
blac
k
hole
,
the
or
i
gin
al
AOD
V
achieve
d
the
be
st
end
-
to
-
e
nd
delay
resu
lt
,
w
hile
the
Hybr
i
d
prot
oco
l
ach
ei
ved
the
hi
gh
est
end
-
to
-
e
nd
delay
resu
lt
.
(a)
(b)
Figure
7
.
Dela
y wit
h
Pa
us
e
T
i
m
e 0
. (
a
) 1 Bl
ack
Ho
le
,
(b) 2
Bl
ack
Ho
le
As
s
how
n
in
Figure
7,
t
he
en
d
-
to
-
e
nd
delay
incr
ease
d for al
l t
he
prot
oco
ls
wh
e
n
t
he
num
ber
of no
de
s
increase
d.
This
increas
e
in
th
e
num
ber
of
node
s
decr
ease
d
the
c
ha
nces
of
the
destinat
io
n
node
bec
om
i
ng
th
e
neig
hbour
of
the
source
node
.
Ther
e
f
or
e,
th
e
pack
et
s
we
re
transm
i
tt
ed
over
seve
ral
hops
in
order
to
rea
ch
th
e
destinat
io
n
no
de.
Mo
re
ov
e
r,
the
Hyb
rid
and
B
DD
-
A
O
DV
protoc
ols
wer
e
re
quire
d
to
pe
rfo
rm
m
or
e
op
e
rati
ons
to
detect
and
av
oi
d
the
black
hole
node,
a
nd
hen
ce
,
increas
ed
the
delay
.
A
si
m
il
ar
patter
n
w
a
s
ob
ta
ine
d
i
n
the
case
of
a
net
w
ork
bein
g
at
ta
c
ked
by
t
wo
bla
ck
ho
le
nodes
.
The
ori
gin
al
AOD
V
ac
hiev
ed
the
best
en
d
-
to
-
e
nd
delay
com
par
ed
t
o
the
oth
e
r
three
prot
ocol
s,
wh
il
e
the
H
ybrid
protoc
ol
achieve
d
the
hi
gh
est
end
-
to
-
en
d del
ay
r
esult.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
3
,
June
2019
:
1968
-
1977
1974
As
sho
wn
i
n
F
i
gure
8(a),
t
he
ov
e
rh
ea
d
inc
r
eased
as
the
num
ber
of
node
s
increase
d
f
rom
20
to
40
nodes
.
T
his
co
uld
be
inte
rpre
te
d
to
m
ean
th
at
the
increa
se
in
the
num
ber
of
node
s
in
t
he
netw
ork
le
d
to
an
increase
in
the
nu
m
ber
of
c
ontr
ol
pack
et
s
(
e.g
.
RR
EQ
a
nd
RR
EP)
that
wer
e
exc
hange
d
thr
ough
the
net
work.
The
or
i
gin
al
AOD
V
ac
hiev
ed
a
sig
nifica
ntly
higher
ov
erh
ea
d
a
gain
st
al
l
the
oth
er
protoc
ols
du
e
to
the
existe
nce
of th
e b
la
ck
hole a
nd the
lack
of a
ny d
et
ect
io
n
a
nd
pr
e
ve
ntion m
echan
ism
.
(a)
(b)
Figure
8
.
O
verhead
w
it
h Pa
use
Tim
e 0
,
(a) 1
Bl
ack
Ho
le
,
(b
) 2 Bla
ck H
ole
3.2.
Low
M
obi
li
ty
Scen
arios
The
perf
or
m
ance
of
the
te
s
te
d
pr
oto
c
ols
was
e
valuate
d
unde
r
t
he
as
su
m
ption
of
l
ow
m
ob
il
it
y.
As
show
n
in
F
igure
9,
f
or
networks
at
ta
cked
by
on
e
blac
k
ho
le
an
d
tw
o
bl
ack
ho
le
no
de
s,
the
Hyb
rid
a
nd
the
BDD
-
A
O
DV
protoc
ols
outp
erfor
m
ed
bo
t
h
the
MI
-
AOD
V
and
the
or
igi
na
l
AO
D
V
pr
oto
cols
in
te
rm
s
of
th
e
pack
et
delive
ry r
at
io.
(a)
(b)
Figure
9
.
Pac
ke
t De
li
ver
y R
at
io w
it
h Pa
us
e
Ti
m
e 1
0,
(a
) 1
Bl
ack Hole,
(b) 2 Bl
ack
H
oles
Figure
9(
a
)
s
hows
the
pac
ke
t
delivery
rati
o
res
ults
f
or
a
netw
ork
t
hat
was
bein
g
at
ta
cked
by
on
e
black
no
de.
T
he
pack
et
delive
ry
rati
o
increas
ed
w
hen
the
num
ber
of
no
de
s
in
the
netw
ork
increase
d
f
r
om
20
to
40
nodes.
W
it
hin
this
int
erv
al
,
inc
reasi
ng
the
nu
m
ber
of
node
s
in
the
netw
ork
le
d
to
a
decr
eas
e
in
the
chan
ce
s
of
t
he
black
hole
node
s
to
obta
in
RR
EQ
m
essages.
Th
us,
the
cha
nces
of
th
e
black
hole
nodes
dro
pp
i
ng
data
pack
et
s
al
so
de
creased
.
M
oreov
e
r,
inc
reasing
the
num
ber
of
norm
al
nodes
i
n
t
he
networ
k
inc
rease
d
the
possible
nu
m
ber
of
node
neig
hbou
rs,
hen
ce
i
ncr
easi
ng
it
s
c
han
ce
s
of
getti
ng
t
o
the
destinat
ion
node
(s)
su
cces
sfu
ll
y.
Fi
gure
9(b)
sho
ws
tha
t
the
pac
ket
de
li
var
y
rati
o
i
nc
reased
w
he
n
the
nu
m
ber
of
node
s
increase
d
from
20
t
o
40
no
de
s.
T
he
a
ddit
ion
of
a
sec
ond
black
hole
ha
d
a
ne
gative
eff
ect
on
the
pack
et
delivery
rati
o.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
In
te
ll
igent
bla
c
k hole detec
ti
on in
mo
bile Ad
Ho
c
n
et
w
or
ks
(
Ya
ser
Kh
amay
seh
)
1975
Figure
10
s
hows
that
t
he
droppe
d
pack
et
s
rati
o
for
al
l
the
protoc
ols
de
creased
as
t
he
num
ber
of
nodes
inc
rease
d
from
20
to
40
node
s.
A
gai
n,
ad
ding
a
se
cond
blac
k
hol
e
increase
d
the
nu
m
ber
of
droppe
d
pack
et
s
f
or
bo
t
h
cases.
From
Figure
9
to
Fi
gure
10,
the
a
gr
eem
ent
betw
een
the
dro
pp
e
d
pac
kets
resu
l
ts
and
pack
et
delive
r
y
rati
o
resu
lt
s
can
be
obser
ve
d,
w
her
e
the
pr
oto
c
ol
w
hic
h
ha
d
a
hi
gh
dr
opped
pa
ckets
rat
io
wa
s
the pr
oto
c
ol w
i
th a lo
w deli
ve
ry r
at
io.
(a)
(b)
Figure
10
.
Dro
pp
e
d P
ackets
Ra
ti
o
with
Pause
Tim
e 1
0,
(a) 1
Bl
ac
k Ho
le
,
(b) 2 Bl
ack
Ho
le
s
Figure
11
s
ho
ws
t
he
e
nd
-
to
-
end
delay
res
ul
ts
for
a
net
work
bei
ng
at
ta
cked
by
one
an
d
tw
o
blac
k
ho
le
no
des.
T
he
Hybr
i
d
an
d
BDD
-
A
O
DV
protoc
ols
w
ork
ed
pro
per
ly
wi
th
a
pau
se
ti
m
e
of
10,
w
he
re
these
two p
ro
t
oco
ls i
ncr
ease
d
t
he
e
nd
-
to
-
en
d dela
y com
par
ed
t
o t
he
MI
-
A
ODV
an
d or
i
gin
al
A
ODV pr
otoc
ols
.
(a)
(b)
Figure
11
.
Del
ay
w
it
h
Pa
us
e
Ti
m
e 1
0,
(a
) 1
Bl
ack Hole,
(b) 2 Bl
ack
H
oles
As
s
how
n
in
F
igure
11(a
),
t
he
end
-
to
-
en
d
de
la
y
increased
for
al
l
the
prot
oco
ls
w
hen
t
he
nu
m
ber
of
nodes
in
crease
d
f
ro
m
20
to
40
nodes
.
T
he
Hybr
i
d
pr
otoc
ol
achieve
d
t
he
hi
gh
e
st
en
d
-
to
-
e
nd
delay
resu
lt
s
com
par
ed
to t
he
o
the
r
three
protoc
ols,
w
hile t
he
or
i
gin
al
A
ODV
ac
hieve
d t
he
lowest e
nd
-
to
-
e
nd d
el
ay
re
su
lt
s.
This
diff
e
ren
c
e
in
the
res
ults
was
due
to
t
he
diff
e
ren
c
e
in
the
pac
kets
delivery
rati
o
ob
ta
ine
d
from
eac
h
protoc
ol,
w
here
increasin
g
th
e
pack
et
s
deliv
ery
rati
o
requi
red
m
or
e
tim
e,
thereb
y
increa
sing
the
en
d
-
to
-
en
d
delay
r
es
ults.
Fo
r
a
net
work
at
ta
cked
by
two
black
hole
nodes
,
the
or
i
gin
al
A
O
DV
s
howe
d
the
bes
t
end
-
to
-
e
nd
delay
com
par
ed
with
the
ot
he
r
protoc
ols.
T
he
Hyb
rid
protoco
l
s
howe
d
the
highest
en
d
-
to
-
e
nd
delay
r
esults.
Figure
11
(b)
s
hows
the
e
nd
-
to
-
e
nd
delay
r
esults
for
a
ne
twork
bein
g
a
tt
acked
by
tw
o
blac
k
hole
nodes
.
An
y
inc
rease
i
n
the
P
DR
res
ulted
in
a
n
inc
rease
in
the
e
nd
-
to
-
en
d
delay
.
As
the
Hy
bri
d
pr
oto
c
ol
ha
d
the
highest P
DR, h
ence it
had
the
highest e
nd
-
to
-
end d
el
ay
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
3
,
June
2019
:
1968
-
1977
1976
Fo
r
a
pa
us
e
ti
m
e
of
10,
t
he
Hybr
i
d
a
nd
B
DD
-
A
O
DV
pr
oto
c
ols
outpe
rfor
m
ed
the
MI
-
AOD
V
an
d
the
or
i
gin
al
A
ODV
protoc
ols
w
it
h
resp
ect
to
the
over
hea
d.
F
igure
12
sh
ows
the
ov
e
rh
ea
d
resu
lt
s
for
net
works
bein
g
at
ta
cke
d by one
black
hole an
d
t
wo b
la
ck ho
le
s
.
(a)
(b)
Figure
12
.
O
ve
rh
ea
d wit
h Pau
se Tim
e 1
0,
(a) 1
Bl
ac
k Ho
le
,
(b) 2 Bl
ack
Ho
le
s
As
s
how
n
by
Figure
12,
the
ov
e
r
head
i
ncrea
sed
as
th
e
num
ber
of
node
s
increase
d
f
r
om
20
to
40
nodes
.
This
in
dicat
ed
that
in
creasin
g
the
num
ber
of
node
s
in
the
netw
ork
will
le
ad
to
an
inc
rease
in
the
nu
m
ber
of
c
ontrol
pac
kets
(e
.
g.
RR
E
Q
an
d
RR
EP)
exc
ha
nged
t
hro
ugh
th
e
netw
ork.
Fig
ur
e
12(
b)
s
ho
ws
that
the
ov
e
rhead
r
esult
s
for
the
or
i
gin
al
AOD
V
was
sig
nific
antly
hig
he
r
than
the
over
hea
d
for
the
oth
e
r
three
protoc
ols.
T
his
was
due
to
th
e
i
m
pact
of
bl
ack
hole
no
de
s
in
the
ori
gina
l
AODV
a
nd
the
abse
nce
of
any
detect
ion
an
d
pr
e
ve
ntion
m
e
chan
ism
.
Dec
r
easi
ng
the
im
pact
of
blac
k
ho
le
in
the
net
work
le
d
t
o
a
dec
r
ease
in
the n
et
wor
k overhea
d.
4.
CONCL
US
I
O
N AND F
UT
U
RE W
ORK
A
com
m
on
thr
eat
to
MANET
s
is
black
hole
at
ta
cks.
This
pa
per
was
buil
t
on
to
p
of
the
w
ork
do
ne
in
[
28
]
to
en
han
c
e
the
secur
it
y
le
vel
in
MAN
ETS.
T
o
f
ully
util
iz
e
the
BDD
dataset
,
a
BDD
-
A
O
DV
protoc
ol
was
pro
posed
,
w
hich
de
pe
nds
on
the
feat
ur
es
of
t
he
B
DD
dataset
t
o
buil
d
it
s
pr
e
ve
ntion
a
nd
det
ect
ion
m
echan
ism
s.
The
B
DD
-
AODV
pr
oto
c
ol
m
od
ifie
s
the
be
hav
i
our
of
th
e
ori
gin
al
A
O
DV,
m
aking
it
m
or
e
secur
e
a
gainst
black
ho
le
at
ta
cks,
wh
e
re
it
checks
the
reli
a
bili
ty
of
the
node
t
hat
sen
ds
the
RR
EP
m
e
ssage.
The
Hyb
rid
pr
oto
c
ol
wa
s
cre
at
ed
by
c
om
bin
in
g
the
MI
-
A
ODV
a
nd
BD
D
-
AOD
V
prot
oco
ls
,
incl
ud
i
ng
al
l
the
featur
e
s
of
th
e
BDD
datase
t.
In
oth
e
r
w
ords
,
the
w
orking
m
echan
ism
of
the
MI
-
A
ODV
prot
oco
l
was
com
bin
ed
with
the
w
orki
ng
m
echan
ism
of
the
BD
D
-
AODV
prot
oco
l
i
n
or
der
t
o
cre
at
e
a
Hyb
rid
prot
oco
l
.
As
in
the
BD
D
-
AOD
V
prot
oco
l,
eac
h
no
de
in
the
MAN
ET
has
a
Tr
us
t
Table
,
a
Bl
ack
Table
,
a
nd
a
Count
Table
.
Sim
ulatio
n
re
su
lt
s
sho
wed
t
hat
the
BDD
-
A
O
DV
a
nd
the
Hy
br
i
d
prot
oco
ls
re
duc
ed
the
im
pact
of
black
ho
le
at
ta
cks
a
nd
ou
t
perform
ed
both
the
or
iginal
AOD
V
and
MI
-
A
O
D
V
protoc
ols
in
te
r
m
s
of
the
PD
R,
dro
pp
e
d pac
kets r
at
io,
and
ove
rh
ea
d, w
hile t
he
end
-
to
-
e
nd del
ay
w
as m
ai
nt
ai
ned
i
n
s
om
e i
nter
vals.
The
co
op
e
rati
ve
and
sel
ect
ive
black
ho
le
no
des
we
re
no
t
c
on
si
der
e
d
in
th
is
wo
r
k.
T
he
coope
rati
ve
black
hole
pr
ob
le
m
occurs
w
hen
m
or
e
than
one
black
ho
le
c
oope
r
at
es
tog
et
her
.
O
n
the
ot
he
r
ha
nd
,
the
sel
ect
ive b
l
ack hole
at
ta
ck
s
sel
ect
a
set
of
d
at
a p
ackets
to b
e dro
pped
, d
el
ivere
d
to th
e d
est
inati
on
nod
e
, o
r
m
od
ifie
d.
It
is
pro
po
se
d
th
at
the
be
ha
viour
of
co
operati
ve
a
nd
sel
ect
ive
bl
ack
hole
s
be
st
ud
ie
d
a
nd
t
he
BDD
dataset
be
e
xpan
de
d
to
inc
lud
e
the
relev
ant
featu
res
t
hat
co
ntribute
to
the
detect
ion
of
the
se
at
ta
cks.
More
ov
e
r,
it
is
pro
po
se
d
that t
he
BD
D
-
AOD
V
an
d
Hyb
rid
protoc
ols
be
en
han
ce
d
with
ce
rtai
n
m
echan
ism
s
to
so
lve
these
tw
o
pro
blem
s.
In
a
ddit
ion
,
the
ex
per
im
ental
resu
lt
s
s
how
ed
that
the
e
nd
-
to
-
e
nd
delay
in
th
e
Hybr
i
d
a
nd
BDD
-
A
O
DV
pr
oto
c
ols
wa
s
hi
gh
e
r
tha
n
t
he
delay
in
the
or
iginal
A
O
DV.
The
refor
e
,
th
ere
is
a
need f
or furt
he
r
im
pr
ov
em
ents in or
de
r
to
r
e
du
ce
the
delay
values
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
In
te
ll
igent
bla
c
k hole detec
ti
on in
mo
bile Ad
Ho
c
n
et
w
or
ks
(
Ya
ser
Kh
amay
seh
)
1977
REFERE
NCE
S
[1]
Ochola
E
.
O
.
an
d
El
off
M
.
M
.
,
“
A Re
vie
w o
f
Bl
a
ck
Hole
a
tt
a
ck
o
n
AO
DV
Routi
ng
in
MA
NET
,
”
I
SSA
,
2011
.
[2]
T.
Kaur
and
A.
Singh,
“
Perform
anc
e
Ev
al
u
at
ion
of
MA
NET
with
Bla
ck
Hole
A
tt
a
ck
Us
ing
Routi
ng
Protocol
s
,
”
Inte
rnational
Jo
urnal
of Engi
ne
e
ring R
ese
arch
a
nd
Applications
(
IJE
RA
)
,
v
ol
/
issue:
3
(
4
)
,
2013
.
[3]
S.
Kum
ar,
et
al
.
,
“
An
Empiric
al
Crit
ique
of
On
-
Dem
and
Routi
ng
Protocol
s
a
ga
i
nst
Rushing
Atta
ck
in
MA
NET
,
”
Inte
rnational
Jo
urnal
of El
e
ct
ri
c
al
and
Comput
er
Engi
n
ee
ring
(
IJE
CE)
,
v
ol
/
issue:
5
(
5
)
,
pp
.
1102
-
1
110,
2015
.
[4]
Aart
i
and
S
.
T
y
agi,
“
Stud
y
of
MA
NET:
Chara
c
te
rist
ic
s,
Chal
le
ng
es,
Ap
p
li
c
at
ion
and
Secur
ity
At
ta
ck
s
,
”
In
te
rnational
Jo
urnal
of Adv
an
c
ed
R
ese
arch
in Com
pute
r Sc
i
en
ce
and
Sof
tware Engi
ne
ering
,
v
ol
/i
ss
ue:
3
(
5
)
,
201
3.
[5]
R
.
M
.
Desai
,
e
t
al
.
,
“
Routi
ng
Protocol
s
for
Mobi
le
Ad
Hoc
N
et
w
ork
-
A
Surve
y
a
nd
Anal
y
s
is
,
”
In
donesian
Journa
l
of
E
le
c
tric
al
En
gine
ering
and
C
ompu
te
r Sc
i
ence
,
v
ol
/i
ss
ue:
7
(
3
)
,
pp.
795
-
801
,
20
17.
[6]
A
.
P
.
Murdan
an
d
A
.
Bhowon,
“
Mobile
Ad
Hoc
Networks
in
Presenc
e
of
Bl
ac
k
Hole
Attack
,
”
In
donesian
Journal
of
E
le
c
tric
al
En
gine
ering
and
C
omputer
Scienc
e
,
v
ol
/i
ss
ue:
7
(2)
,
pp.
577
-
582
,
20
17.
[7]
Go
y
al,
et
a
l
.
,
“
MA
NET:
Vulner
ab
il
i
ti
es,
Chal
le
ng
es,
At
ta
cks,
Applicat
ion
,
”
Inte
rnat
i
onal
Journal
of
Computati
onal Engi
ne
ering
&
Manage
ment
(
IJ
CEM)
,
v
ol.
11
,
2
011.
[8]
B.
Nawafl
eh
,
e
t
al
.
,
“
Im
prove
d
AO
DV
Proto
col
to
Det
ec
t
and
Avoid
Black
Hole
Nodes
in
MA
NETs
,
”
Master
Th
esis
i
n
Com
pute
r
Sci
en
ce
,
Dept
.
of
Com
p.
Sci.,
JU
ST
Uni
v.
,
2012.
[9]
Baa
da
che,
e
t
al
.
,
“
Avoiding
Bl
ac
k
hol
e
and
C
ooper
ative
Bl
ack
hole
At
ta
ck
s
in
W
ire
l
ess
Ad
hoc
Networks
,
”
Inte
rnational
Jo
urnal
of
Comput
er
Scienc
e
and
I
nformation
Se
cu
rity
(
IJCSI
S)
,
v
ol
/i
ss
ue:
7
(
1
)
,
201
0.
[10]
Saini
,
et
al
.
,
“
Com
par
ison
bet
wee
n
Vari
ous
Bl
a
ck
Hole
De
te
c
tion
Te
chn
ique
s
i
n
MA
NET
,
”
Na
ti
onal
Conf
ere
n
ce
on
Computati
on
al
Instrum
ent
at
i
on
,
2010
.
[11]
Y.
Kham
a
y
seh,
et
al
.
,
“
A
New
Protocol
for
Dete
c
ti
ng
Black
Hole
Nodes
in
A
d
Hoc
Network
s
,
”
Inte
rnationa
l
Journal
of
Comm
unic
ati
on
Net
w
orks and
Information
S
ec
urit
y
(
IJCNIS)
,
v
ol
/i
ss
u
e:
3
(
1
)
,
2011
.
[12]
Bat
tula,
et
a
l
.
,
“
Mixed
signal
base
d
tr
ansm
ission
in
m
ixe
d
lay
ers
for
h
ig
h
throug
hput
w
ire
l
ess
net
work
,
”
El
ixir Int
ernati
o
nal
Journal
,
201
1.
[13]
Ta
ne
ja
,
et
al
.
,
“
A
Surve
y
of
Ro
uti
ng
Protocol
s
i
n
Mobile
Ad
Hoc
Networks
,
”
Int
ernati
onal
Journ
al
of
Innov
at
ion
Manage
ment
an
d
Technol
og
y
,
v
ol
/i
ss
ue:
1
(
3
)
,
20
10.
[14]
Singh,
et
al
.
,
“
S
ec
uri
t
y
Iss
ues
And
Li
nk
Expi
rati
on
In
Secur
e
Routi
ng
Protocol
s
In
Mane
t:
A
Revi
ew
,
”
N
e
twork
,
v
ol
/
issue:
3
(
7
)
,
2
014.
[15]
Li
u,
e
t
al
.
,
“
To
ward
Inte
gra
t
ing
Feat
ure
Selection
Algorit
hm
s
f
or
Cla
ss
ifi
ca
t
ion
and
Cluste
ring
,
”
Knowle
dge
an
d
Data
Engi
n
ee
rin
g,
IE
EE Tr
ansacti
ons
,
v
ol
/i
ss
ue:
17
(
4
)
,
2005.
[16]
John,
et al
.
,
“
Irre
le
van
t
Fea
ture
s
a
nd
The Subs
et
S
el
e
ct
ion
Problem
,
”
ICML
,
v
ol
.
94
,
1994
.
[17]
Raj
,
e
t
al
.
,
“
DP
RAO
DV
:
A
Dy
namic
Le
arn
ing
Sy
st
em
aga
inst
Bla
ckho
le
Att
a
ck
in
AO
DV
b
ase
d
MA
NET
,
”
Inte
rnational
Jo
urnal
of
Comput
er
Scienc
e
Iss
ue
s (
IJCSI
)
,
v
ol.
2
,
2009
.
[18]
Mistr
y
,
et
al
.
,
“
Im
proving
AO
D
V
Protocol
Aga
inst
Attacks
,
”
I
nte
rnational
Mu
lt
i
Con
fe
renc
e
o
f
Engi
n
ee
rs
and
Computer
Scient
ists (
IME
CS)
,
v
ol.
2
,
2010
.
[19]
Khara
,
et
al
.
,
“
Secur
ity
in
Rou
ti
ng
Proto
col
to
Avoid
Thr
eat
o
f
Bl
ac
k
Hole
Atta
ck
in
MA
N
ET
,
”
In
te
rnation
al
journal
of
Elec
tric
al
,
E
lectronic
s
&
C
o
mm
unic
ation E
ng
in
ee
ring
,
v
ol.
2
,
2012.
[20]
Bhoria
,
et
al
.
,
“
Dete
rm
ini
ng
Feat
ur
e
Set
of
DO
S
Atta
cks
,
”
Inte
rnational
Jo
urnal
of
Adv
an
ce
d
Re
search
i
n
Computer
Scien
ce
and
Sof
tware Engi
ne
ering
,
v
ol
.
3
,
2013
.
[21]
KD
DCUP
,
1999
.
ht
tp:
/
/kdd.
i
cs.
u
ci
.
edu/
da
ta
b
ase
s/
kddcup99/
kddcu
p99.
html
.
[22]
Riss
ino,
et
al
.
,
“
Rough
Set
T
heor
y
–
Fund
amenta
l
Con
ce
pts
,
Princi
pal
s
,
Da
t
a
Ext
ra
ction,
an
d
Applic
a
ti
ons
,
”
Data
Mini
ng
an
d
Knowle
dg
e
Di
scov
ery
in Re
al
Life
Appl
i
cat
ion
s
,
2009.
[23]
Upendra
,
“
An
E
ffic
i
ent
Fe
at
ur
e
Reduc
ti
on
Com
par
ison
of
M
ac
h
ine
Le
a
rning
Al
gorit
hm
s
for
Int
rusion
Dete
ct
io
n
S
y
stem
,
”
Int
ernati
onal Journal of
Eme
rging
Tr
ends
&
Technol
ogy
in
Computer
Sci
en
ce
,
v
ol
.
2
,
2013.
[24]
M
.
Tav
allaee,
e
t
al
.
,
“
A
det
a
iled
an
aly
sis
of
t
he
KD
D
CUP
99
data
se
t
,”
P
roce
edi
ngs
o
f
t
he
Se
cond
IEEE
int
ernati
ona
l
co
nfe
renc
e
on
C
omputati
onal
in
te
lligen
ce
for
sec
urit
y
and
d
efense
applicat
io
ns
(
CISDA'09
)
.
IEE
E
Press, Pis
cat
away
,
N
J, USA
,
pp
.
53
-
58
,
20
09.
[25]
J.
M
chugh,
“
Te
s
ti
ng
Intrusion
D
et
e
ct
ion
S
y
stem
s:
A
Crit
ique
of
the
1998
and
1
999
DA
RP
A
Int
rusion
Dete
c
ti
on
S
y
stem
Eva
lu
at
i
ons
as
Perform
e
d
b
y
Li
n
col
n
L
a
bora
tor
y
,
”
ACM
Tr
ansacti
ons
on
Information
and
Syste
m
Sec
urit
y
,
v
ol.
3
,
2000
.
[26]
M.
Y.
Su,
“Pre
vent
ion
of
Selective
Bla
ck
Hole
Atta
cks
on
Mobile
Ad
Hoc
Networks
throug
h
I
ntrusion
Dete
c
tion
S
y
stems
,
”
Comp
ute
r Comm
unic
a
ti
ons
,
v
ol
/i
ss
ue:
34
(
1
)
,
2011
.
[27]
Alem,
et
al
.
,
“
Preve
nt
ing
Bl
ac
k
Hole
Att
ac
k
in
Mobile
Ad
-
hoc
N
et
works
Us
ing
Anom
aly
De
te
c
t
ion
,
”
IE
EE
,
v
ol
.
3,
2010.
[28]
M
.
B
.
Yass
ei
n
,
et
a
l
.
,
“
Feat
ur
e
Select
ion
for
Bla
ck
Hol
e
Att
ac
ks
,
”
Journal
Of
Unive
rs
al
C
omputer
Scienc
e
,
v
ol
/i
ss
ue:
22
(
4
)
,
pp.
521
-
536
,
20
16.
Evaluation Warning : The document was created with Spire.PDF for Python.