Inter
national
J
our
nal
of
Electrical
and
Computer
Engineering
(IJECE)
V
ol.
11,
No.
2,
April
2021,
pp.
1761
1770
ISSN:
2088-8708,
DOI:
10.11591/ijece.v11i2.pp1761-1770
r
1761
DED
A:
An
algorithm
f
or
early
detection
of
topology
attacks
in
the
inter
net
of
things
J
alindar
Karande,
Sarang
J
oshi
Department
of
Computer
Engineering,
Pune
Institute
of
Computer
T
echnology
,
Sa
vitribai
Phule
Pune
Uni
v
ersity
,
Pune,
India
Article
Inf
o
Article
history:
Recei
v
ed
Jan
3,
2020
Re
vised
Jul
28,
2020
Accepted
Aug
28,
2020
K
eyw
ords:
Distrib
uted
algorithm
Early
detection
Internet
of
things
IoT
security
Predicti
v
e
detection
RPL
T
opology
attack
ABSTRA
CT
The
internet
of
things
(IoT)
is
used
in
domestic,
industrial
as
well
as
mission-critical
systems
including
homes,
transports,
po
wer
plants,
industrial
manuf
acturing
and
health-care
applications.
Security
of
data
gene
rated
by
such
systems
and
IoT
systems
itself
is
v
ery
critical
in
such
applications.
Early
detection
of
an
y
attack
tar
geting
IoT
system
is
necessary
to
mi
nimize
the
damage.
This
paper
re
vie
ws
sec
urity
attack
detec-
tion
methods
for
IoT
Infr
astructure
presented
in
the
state-of-the-art.
One
of
the
major
entry
points
for
attacks
in
IoT
system
is
topology
e
xploit
ation.
This
paper
proposes
a
distrib
uted
algorithm
for
early
detection
of
such
attacks
with
the
help
of
predicti
v
e
de-
scriptor
tables.
This
paper
also
presents
feature
selection
from
topology
control
pack
et
fields.
The
performance
of
the
proposed
algorithm
is
e
v
aluated
using
an
e
xtensi
v
e
simulation
carried
out
in
OMNeT++.
Performance
parameter
includes
accurac
y
and
time
required
for
detection.
Simulation
results
presented
in
this
paper
sho
w
that
the
proposed
algorithm
is
ef
fecti
v
e
in
detecting
attacks
ahead
in
time.
This
is
an
open
access
article
under
the
CC
BY
-SA
license
.
Corresponding
A
uthor:
Jalindar
Karande
Research
Scholar
,
Department
of
Computer
Engineering
Pune
Institute
of
Computer
T
echnology
,
Pune,
India
Email:
jalindar
.karande@ieee.or
g
1.
INTR
ODUCTION
The
internet
of
things
(IoT)
has
made
possible
seamless
communication
between
machines
and
hu-
man.
IoT
systems
ability
of
continuous
data
collection,
seamless
communication,
autonomous
decision
making
and
ability
to
control
the
ph
ysical
w
orld
by
implementing
decisions
changed
operational
paradigms
of
man
y
operational
systems.
IoT
made
it
possible
to
replace
human
in
man
y
critical
tasks.
Thes
e
resulted
in
minimizing
human
errors
and
increased
the
producti
vity
of
the
systems.
No
w
,
internet
of
things
(IoT)
has
become
a
vitally
important
application
in
e
v
ery
b
usiness
domain
including
b
ut
not
limited
to
smart
home,
smart
cit
y
,
smart
grid,
connected
cars,
connected
healthcare,
industrial
automation,
precision
f
arming,
smart
wearables,
retail
and
supply
chain
management.
Ev
en
in
the
CO
VID-19
outbreak
millions
of
population
are
lock
ed
do
wn
to
home
b
ut
IoT
systems
were
still
on
the
field.
IoT
systems
made
it
possible
to
k
eep
critical
infrastructures
functioning
through
remote
monitoring
and
controlling.
IoT
systems
were
e
xtensi
v
ely
used
during
CO
VID-19
outbreak
for
pandemic
management.
These
application
includes
the
use
of
smart
wearables
to
real-time
monitoring
of
health
data
as
well
as
compliance
with
home
quarantine,
real-time
data
collection
through
IoT
thermometers,
remote
instructions
and
application
of
IoT
enabled
robots
to
serv
e
patients
and
to
maintain
hospital
h
ygiene.
The
detailed
surv
e
y
of
IoT
applications
during
CO
VID-19
outbreak
is
presented
in
[1].
IoT
security
is
a
gro
wing
concern,
gi
v
en
that
v
ari
ous
critical
infrastructures
and
applications
are
di-
J
ournal
homepage:
http://ijece
.iaescor
e
.com
Evaluation Warning : The document was created with Spire.PDF for Python.
1762
r
ISSN:
2088-8708
rectly
connected
and
controlled
using
IoT
.
These
concerns
include
security
of
data,
connected
infrastructure,
human
as
well
as
IoT
infrastructure
itself.
The
s
ecurity
breach
of
the
IoT
system
may
lead
to
e
xploiting
crit-
ical
infrastructure
and
may
put
man
y
li
v
es
at
stak
e.
IoT
systems
are
connected
to
the
ph
ysical
w
orld
in
a
more
concrete
w
ay
than
con
v
entional
computer
systems.
This
mak
es
a
breach
of
security
of
IoT
system
more
catastrophic
in
nature.
This
raise
concerns
o
v
er
using
con
v
entional
security
algorithms
for
detection
of
secu-
rity
attacks
in
IoT
systems.
Security
requirement
s
of
IoT
embedded
into
critical
infrastructures
are
analysed
in
[2],
which
also
highlights
that
con
v
entional
internet
security
approaches
are
not
enough
to
address
the
secu-
rity
of
IoT
systems
used
for
management
of
critical
infrastructures.
Security
concerns
of
use
of
IoT
in
industrial
applications
are
highlighted
in
[3,
4].
Analysis
of
security
attack
detection
mechanisms
in
an
industrial
setting
is
presented
in
[5]
along
with
a
re
vie
w
of
dif
ferent
commercial
tools
a
v
ailable
for
attack
detection.
Authors
highlighted
the
need
for
more
focused
solutions
for
the
protection
of
industrial
IoT
systems.
The
not
only
breach
of
IoT
systems
security
leads
to
attack
on
IoT
systems
b
ut
compromised
IoT
de
vices
are
used
to
en-
able
lar
ge
scale
attacks
on
other
critical
infrastruct
ures.
A
detailed
assessment
of
such
IoT
enabled
atta
cks
is
presented
in
[6].
Man
y
techniques
ha
v
e
been
proposed
in
the
state-of-the-art
for
pre
v
enting
security
attacks
on
IoT
de
vices
which
includes
authentication
[7,
8],
Access
control
[9]
and
data
encryption
[10]
for
IoT
.
Although
se
v
eral
measures
ha
v
e
been
tak
en
to
pre
v
ent
security
attacks
on
IoT
systems,
the
lo
w
compute
po
wer
of
IoT
de
vices
still
mak
es
it
vulnerable
to
attacks.
The
vulnerability
assessment
of
consumer
IoT
de
vices
presented
in
[11]
sho
ws
that
around
10%
de
vices
are
prone
to
at
least
one
critical
risk
vulnerability
,
40%
de
vices
had
at
least
one
high-risk
vul
nerability
,
and
68%
de
vices
had
at
least
one
medium
risk
vulnerability
and
42%
de
vices
had
at
least
one
lo
w-risk
vulnerability
.
These
vulnerable
consumer
de
vices
analyzed
in
[11]
include
smart
TV
,
webcam
and
printers
from
a
wide
range
of
manuf
acturers.
These
highlights
that
security
attack
detection
is
of
critical
importance
e
v
en
though
pre
v
ention
mechanism
is
present
into
IoT
de
vices.
IoT
de
vices
use
RPL
protocol
for
b
uilding
netw
ork
topology
to
connect
to
the
Internet.
The
detailed
w
orking
of
the
RPL
protocol
is
presented
in
[12].
RPL
protocol
is
prone
to
be
e
xploited
and
becomes
an
entry
point
for
man
y
attacks
on
IoT
de
vices.
Security
of
RP
L
protocol
is
still
an
open
problem
[13].
The
resource-
constrained
nature
of
IoT
de
vices
,
the
possibility
of
bypass
of
pre
v
enti
v
e
mechanism
and
probable
catastrophic
loss
due
to
breach
of
security
of
IoT
de
vices
moti
v
ated
authors
to
design
of
an
algorithm
for
early
detection
of
such
security
attacks
without
putting
hea
vy
resource
load
on
indi
vidual
IoT
de
vice.
The
proposed
algorithm
is
distrib
uted
in
nature
and
will
run
in
tw
o
phases.
The
first
phase
in
v
olv
es
collecting
and
b
uilding
descripti
v
e
tables
locally
,
whereas
the
second
phase
in
v
olv
es
e
xchanging
descripti
v
e
tables
and
concluding
the
presence
of
an
attack
er
.
The
main
contrib
ution
of
this
paper
are
summarized
follo
ws:
This
paper
presents
a
comprehensi
v
e
re
vie
w
of
the-state-of-the-art
for
detection
of
IoT
security
attacks
This
paper
presents
the
selection
of
control
pack
et
parameters
for
attack
detection
This
paper
presents
a
distrib
uted
algorithm
for
early
detection
of
security
attacks
on
IoT
de
vices
This
paper
presents
a
performance
e
v
aluation
of
the
proposed
algorithm
in
early
detecting
attacks
Ne
xt
s
ection
presents
a
re
vie
w
of
the
state-of-the-art
for
security
attacks
and
countermeasures
on
the
IoT
system.
Section
4
proposes
distrib
uted
algorithm
for
early
detection
of
security
attacks
through
the
use
of
predicti
v
e
descriptor
tables.
Section
5
presents
the
result
analysis
to
assess
t
he
ef
fecti
v
eness
of
the
proposed
algorithm
for
the
early
detection
of
security
attacks.
2.
LITERA
TURE
REVIEW
Identifying
and
mitig
ating
attack
ers
from
the
netw
ork
ed
system
has
been
the
topic
of
importance.
Se
v
eral
methods
and
algorithms
ha
v
e
been
proposed
in
the
state
of
the
art
to
detect
specific
attacks.
Netw
ork
ed
systems
may
be
the
tar
get
of
multiple
attacks.
W
e
need
a
mechanism
to
inte
grate
se
v
eral
attack
detection
methods
into
a
single
frame
w
ork.
Standardised
frame
w
ork
for
such
detection
system
called
CIDF
[14]
is
presented
by
a
w
orking
group
created
by
D
ARP
A
no
w
called
intrusion
detection
w
orking
group
(ID
WG).
Snort
[15]
is
one
the
pro
v
en
open-source
attack
detection
tool,
b
ut
the
feasibility
of
deplo
ying
a
snort
system
in
IoT
nodes
is
ar
gued
in
[16]
due
to
resource
constraints
on
IoT
nodes.
Beha
viour
-based
analysis
of
vulnerabilities
of
the
drone-based
IoT
system
along
with
detection
of
vulnerability
using
Petri
net
is
presented
in
[17].
Attack
ers
e
xploit
vulnerabilities
in
IoT
de
vices
and
protocols
to
enter
into
the
IoT
netw
orks.
The
approach
based
on
the
modelling
relationship
between
vulnerabilities
as
a
graph
and
using
a
graph-theoretic
approach
for
detecting
attack
is
presented
in
[18].
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1761
–
1770
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
1763
V
arious
attacks
ag
ainst
RPL
protocol
ha
v
e
been
demonstrated
in
[19]
along
with
Lightweight
Heart-
beat
algorithm
to
detect
attack
ers.
The
proposed
algorithm
is
relying
only
on
IPsec
with
ESP
communication
and
in
man
y
cases,
IPsec
protocol
might
not
be
deplo
yed
in
IoT
nodes.
This
algorithm
is
also
creating
ad-
ditional
w
orkload
for
resource-constrained
IoT
nodes.
A
comprehensi
v
e
re
vie
w
of
security
challenges
in
IoT
topology
is
presented
in
[20].
This
paper
further
analysed
IoT
protocols,
including
RPL
and
6LoWP
AN
for
potential
security
weakness
along
with
the
need
for
further
research
in
IoT
topology
security
.
The
specification-
based
method
for
identifying
RPL
topology
attacks
on
the
IoT
system
is
presented
in
[21].
This
method
b
uilds
a
finite
state
machine
for
RPL
topology
operations.
T
opology
control
information
(DIO
pack
ets)
throughout
the
system
is
monitored
by
monitoring
nodes
and
information
within
these
pack
ets
is
used
for
state
transitions.
The
approach
presented
in
the
paper
is
ef
fecti
v
e
to
detect
more
comple
x
attack
scenarios
lik
e
multiple
and
collaborati
v
e
attacks.
W
ithin
the
multi-hop
IoT
system,
disco
v
ering
and
establishing
the
route
to
the
g
ate
w
ay
node
is
one
of
the
crucial
tasks.
The
ef
ficienc
y
of
this
task
leads
to
performance
impro
v
ement
in
the
o
v
erall
IoT
system.
This
task
is
e
x
ecuted
in
a
distrib
uted
manner
in
IoT
protocols
to
tak
e
care
of
runtime
link
f
ailures
or
ne
w
additions
of
nodes
in
the
system.
Unfortunately
,
this
crucial
distrib
uted
task
becomes
the
tar
get
of
the
attack.
Such
concern
of
security
of
route
disco
v
ery
has
been
presented
in
terms
of
MANET
[22],
which
applies
to
IoT
systems
also
as
it
shares
characteri
stics
lik
e
mobile
nodes
and
ad-hoc
nature
with
MANET
.
This
paper
also
highlights
pre
v
enting
such
attack
is
v
ery
costly
and
almost
impossible
in
gi
v
en
situations
and
more
focus
should
be
gi
v
en
on
detection
of
attack
than
pre
v
ention
of
it.
Intrusion
detection
in
IoT
trough
traf
fic
filtering
is
presented
in
[23].
This
w
ork
also
highlights
se
v-
eral
open
challenges
in
attack
detection
using
traf
fic
filtering
which
includes
comple
x
traf
fic
characterization,
dif
ficulties
in
preparing
the
blacklist
and
the
white
list
for
traf
fic
filtration,
traf
fic
sampling,
b
uilding
realistic
attack
models
and
the
impact
of
f
alse
positi
v
es.
Deep
pack
et
inspection
based
attack
detection
mechanism
is
presented
in
[24].
This
mechanism
mak
es
use
of
the
re
gular
e
xpression
in
terms
of
DF
A
to
represent
the
rule.
Representation
of
rules
in
re
gular
e
xpression
mak
es
it
easy
to
implement
in
the
hardw
are
through
field
pro-
grammable
g
ate
arrays
(FPGAs)
which
mak
e
it
f
aster
than
softw
are
approaches.
The
number
of
states
required
to
represent
all
possible
attack
signatures
is
v
ery
lar
ge
and
there
are
al
w
ays
chances
of
changing
the
signature
in
ne
w
attacks.
A
re
vie
w
of
machine
learning-based
approaches
for
enhancing
the
security
of
the
IoT
system
is
pre-
sented
in
[25].
These
approaches
include
authentication
based
on
a
prediction
of
communication
parameters,
machine
learning
algorithms
for
access
control,
secure
of
floading
and
machine
learning-based
attack
detec-
tion
methods.
This
paper
further
concluded
that
machine
learning
needs
intensi
v
e
computing
po
wer
and
high
communication
o
v
erhead.
Also,
the
need
for
a
lar
ge
amount
of
training
data
and
the
comple
x
feature
e
xtrac-
tion
process
mak
es
these
algorithms
unappealing
for
resource-constrained
de
vices.
Machine
learning-based
mechanism
usi
ng
inferencing
and
predicting
st
ates
of
t
he
system
is
presented
in
[26]
to
detect
anomal
ies
and
attacks
in
the
IoT
system.
Random
neural
netw
ork-based
approach
for
detection
of
attack
ers
in
IoT
systems
is
presented
in
[27].
This
approach
learns
anomalies
in
the
performance
of
the
system
using
the
random
neu-
ral
netw
ork
and
relates
it
to
the
f
ailure
of
IoT
node
or
attack
er’
s
presence.
A
g
ame
theory-based
approach
is
for
attack
detection
along
with
a
reputation
model
is
presented
in
[28],
which
is
capable
of
detecting
v
arious
attacks
on
IoT
systems.
Attack
detection
mechanism
s
are
traditionally
e
v
aluated
using
either
test
dataset
or
generating
attacks
manually
.
This
approach
gi
v
es
better
result
in
the
e
v
aluation
phase
b
ut
may
f
ail
to
detect
the
real
attack.
Genetic
programming-based
approach
for
generating
test
attacks
used
for
e
v
aluating
the
accurac
y
of
the
detection
mechanism
is
presented
in
[29].
Deep
learning-based
approach
for
attack
detection
in
IoT
is
presented
in
[30].
A
frame
w
ork
for
DDoS
attack
detection
in
IoT
systems
based
on
cosine
similarities
within
the
traf
fic
flo
w
is
presented
in
[31].
The
artificial
neural
netw
ork-based
architecture
for
detection
of
DDoS/DoS
attack
is
presented
in
[32].
This
architecture
mak
es
use
of
both
forw
ard
and
backw
ard
learning
mechanisms
to
train
and
identify
malicious
traf
fic.
Hidden
Mark
o
v
model-based
classifier
is
proposed
in
[33]
to
detect
anomalies
in
the
data,
which
is
used
to
alert
about
the
security
attack.
This
method
mak
es
use
of
multiple
kno
wledge
domains
lik
e
kno
wledge
of
the
ph
ysical
process
and
the
control
system
to
identify
the
attack.
This
approach
is
suitable
for
implementation
in
an
industrial
control
system,
b
ut
may
not
be
suitable
for
resource-constrained
IoT
systems.
Game
theory-based
approaches
for
detection
of
the
attack
er
mak
es
use
of
conflicting
goals
of
attack
ers
and
detection
engines.
A
tw
o-player
g
ame
theory-based
approach
where
the
attack
er
and
detection
engine
are
opponents
is
presented
in
[34]
to
collaborati
v
ely
detect
security
attacks
in
IoT
systems.
DED
A:
An
algorithm
for
early
detection
of
topolo
gy
attac
ks
in...
(J
alindar
Kar
ande)
Evaluation Warning : The document was created with Spire.PDF for Python.
1764
r
ISSN:
2088-8708
The
distrib
uted
attack
detection
method
with
monitoring
the
communication
pattern
of
nearby
nodes
and
identifies
suspicious
communication
is
presented
in
Kinesis
[35].
In
this
method,
identified
suspicious
communication
in
tw
o
hops
distance
is
reported
to
the
central
detection
node.
The
central
node
is
responsible
for
taking
the
final
decision
about
the
suspicious
node
and
notifies
its
decision
to
all
other
nodes
in
the
system.
This
mechanism
puts
additional
o
v
erhead
of
maintaining
the
tw
o-hop
communication
log.
The
ef
fecti
v
eness
of
the
mechanism
is
af
fected
by
f
abricated
communication
patterns
of
cooperati
v
e
attack
ers.
Another
distrib
uted
algorithm
for
detection
of
the
security
breach
called
v
ersion
number
attack
is
presented
in
[36].
This
method
of
attack
er
identification
mak
es
use
of
placement
of
additional
nodes
dedicated
to
monitoring
netw
ork
com-
munication,
which
results
in
the
higher
cost.
Further
,
this
approach
is
not
ef
fecti
v
e
in
the
case
of
multiples
cooperati
v
e
attack
ers
e
xploiting
the
IoT
system.
The
model
for
distrib
uted
detection
of
the
security
attack
on
the
IoT
system
is
presented
in
[37]
along
with
proof
of
the
concept
implementation,
which
mak
e
use
of
fog
computing
nodes
to
deplo
y
e
xtreme
learning
machine
based
mechanism
for
att
ack
detection
at
local.
Further
security
stat
e
information
collected
from
fog
computing
nodes
is
summarised
at
the
cloud
node
to
predict
the
future
course
of
action
of
the
attack
er
.
V
arious
security
attacks
o
v
er
RPL
based
IoT
netw
orks
ha
v
e
been
demonstrated
in
[38].
The
paper
also
e
v
aluated
v
arious
attack
detection
mechanism
in
dif
ferent
attack
scenario
lik
e
a
single
attack
er
,
multiple
attack
ers
and
collaborati
v
e
attacks.
This
paper
highlighted
the
need
for
designing
the
security
detection
mechanism
for
early
detection
of
attacks
and
include
capabilities
of
detecting
collaborati
v
e
attacks.
3.
IDENTIFICA
TION
OF
FEA
TUTES
FOR
A
TT
A
CK
DETECTION
Through
the
comparati
v
e
analysis
of
simulation
tools
used
for
IoT
research
presented
in
[39],
OM-
NeT++,
a
discrete
e
v
ent
simulation
tool,
is
used
for
the
simulation
study
.
Objecti
v
es
of
simulation
study
include
finding
out
ef
fecti
v
e
features
for
attack
detection,
the
ef
fecti
v
eness
of
predicting
v
alues
of
features
in
future
and
accurac
y
of
detection
mechanism.
The
IoT
system
is
simulated
in
OMNeT++
with
the
implementation
of
RPL
at
the
netw
ork
layer
and
IEEE
802.15.4
standard
at
the
ph
ysical
layer
and
MA
C
layer
.
Figure
1
sho
ws
a
comparison
of
the
total
number
of
data
pack
ets
recei
v
ed
at
the
g
ate
w
ay
node
with
time
during
v
arious
attacks
scenarios.
The
results
sho
w
that
there
is
a
drastic
increase
in
pack
et
loss
in
the
IoT
system
during
attacks.
V
ersion
number
attack
demonstrates
the
w
orst
performance
with
huge
pack
et
loss.
The
presented
results
sho
w
a
periodic
steep
increase
in
the
number
of
pack
ets
recei
v
ed
in
the
system
under
attack.
This
steep
increase
is
the
result
of
the
periodic
global
repair
of
netw
ork
topology
i.e
DOD
A
G
in
RPL
Protocol.
The
results
sho
w
that
de
viation
in
throughput
changes
is
a
good
feature
to
be
considered
for
attack
detection.
Figure
1
demonstrates
the
increase
in
the
DOD
A
G
v
ersion
number
during
dif
ferent
attack
scenarios.
Demon-
strated
a
high
increa
se
in
the
v
ersion
number
indicates
frequent
topology
reformat
ion
triggered
by
malicious
nodes.
This
result
moti
v
ates
to
use
the
rate
of
change
of
the
v
ersion
number
of
DOD
A
G
as
a
feature
to
identify
the
presence
of
the
v
ersion
number
attack
er
in
the
IoT
system.
(a)
(b)
Figure
1.
(a)
Throughput
changes
(b)
DOD
A
G
v
ersion
number
Figure
2
sho
ws
changes
in
the
rank
distrib
ution
o
v
er
time.
It
indicates
that
the
rank
v
alue
tends
to
shrink
in
rank
attack
and
tends
to
ele
v
ate
in
case
of
v
ersion
number
attack.
This
de
viation
in
the
distrib
ution
of
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1761
–
1770
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
1765
the
rank
is
a
good
feature
to
be
used
for
detecting
the
presence
of
an
attack
er
in
the
IoT
system.
Figure
2
also
sho
ws
the
dif
ference
between
the
standard
de
viation
of
the
rank
v
alue
of
DOD
A
G
for
dif
ferent
scenarios
o
v
er
time.
This
dif
ferences
will
be
useful
for
identifying
the
attack
er
in
RPL
based
IoT
netw
ork.
Figure
2
sho
ws
that
during
normal
operation,
the
rank
of
a
node
is
increasing
with
distance
from
the
g
ate
w
ay
node.
Whereas
in
case
of
attack
scenarios,
nodes
f
ar
from
g
ate
w
ay
node
also
tend
to
f
alsely
get
lo
wer
rank
v
alue.
Figure
3
sho
ws
de
viation
in
DIO
pack
ets
recei
v
ed
and
D
A
O
pack
ets
recei
v
ed
respecti
v
ely
in
dif
ferent
scenarios.
This
v
ariation
in
v
alues
of
dif
ferent
parameters
during
v
arious
attack
scenarios
will
be
useful
as
features
for
the
detection
of
the
attack
er
in
the
IoT
system.
(a)
(b)
(c)
(d)
Figure
2.
(a)
Distrib
ution
of
rank
(without
attack),
(b)
Distrib
ution
of
rank
(attack),
(c)
Standard
de
viation
of
Rank,
(d)
Rank
vs
distance
from
g
ate
w
ay
node
(a)
(b)
Figure
3.
(a)
DIO
pack
ets
recei
v
ed,
(b)
D
A
O
pack
ets
recei
v
ed
DED
A:
An
algorithm
for
early
detection
of
topolo
gy
attac
ks
in...
(J
alindar
Kar
ande)
Evaluation Warning : The document was created with Spire.PDF for Python.
1766
r
ISSN:
2088-8708
4.
PR
OPOSED
ALGORITHM
The
proposed
algorithm,
DED
A
for
topology
attack
detection
in
the
IoT
netw
ork
will
w
ork
by
the
placing
monitoring
nodes
in
addition
to
normal
nodes
for
attack
detection.
The
placement
of
additional
nodes
will
result
in
the
uninterrupted
operation
of
ordinary
nodes
by
a
v
oiding
computing
and
memory
o
v
erload
of
e
x
ecuting
detection
algorithms
on
them.
These
additional
monitoring
nodes
will
monitor
all
traf
fic
from
nearby
IoT
nodes
and
b
uild
a
descriptor
table
of
it.
This
descriptor
table
will
include
a
count
of
v
arious
control
pack
ets
transmitted
from
the
indi
vidual
node.
This
desc
riptor
table
will
also
include
information
about
netw
ork
pa-
rameters
lik
e
rank,
v
ersion
number
etc.
sent
in
control
pack
ets
by
indi
vidual
nodes.
Ev
ery
monitoring
node
is
preparing
a
partial
descriptor
table
and
also
making
a
log
of
changes
in
the
partial
descriptor
table.
This
log
of
changes
and
current
v
alues
is
used
to
predict
the
descriptor
table
early
in
time.
The
predicted
descriptor
table
is
shared
with
other
monitoring
nodes.
Ev
ery
monitoring
aggre
g
ate
v
alues
in
the
descriptor
table
recei
v
ed
from
other
monitoring
nodes.
This
aggre
g
ation
of
the
predict
ed
descriptor
table
will
gi
v
e
a
birds-e
ye
vie
w
of
the
current
state
of
the
system
to
e
v
ery
monitoring
node.
Monitoring
nodes
mak
e
use
of
the
predicted
descriptor
table
to
detect
the
presence
of
the
attack
er
and
identifies
which
node
is
the
attack
er
along
with
the
type
of
attack
being
launched.
Information
about
identified
attack
er
is
propag
ated
to
other
nodes
for
necessary
actions
and
precautions.
The
detailed
w
orking
of
the
algorithm
is
presented
in
Algorithm
1
and
Algorithm
2
Algorithm
1:
DED
A:
Distrib
uted
Early
Detection
Algorithm
Phase-I
RankT
uples
=
;
;
V
ersionNumberT
uples
=
;
;
DISRec
=
DIORec
=
D
A
ORec
=
D
A
O
ARec
=
;
;
localRouteT
able
=
;
;
n=
number
of
IoT
nodes
in
the
system;
i=0;
while
i
<
n
do
DISRec[i]
=
DIORec[i]
=
D
A
ORec[i]
=
D
A
O
ARec[i]
=
0;
end
listen
to
netw
ork
trafiic;
if
contr
ol
pac
k
et
then
source=source
address
from
base
pack
et;
dest=destination
address
from
base
pack
et;
if
DIO
pac
k
et
then
RankT
uples
=
RankT
uples
[
(source,
rank
in
DIO);
V
ersionNumberT
uples
=
V
ersionNumberT
uples
[
(source,
v
ersionNumber
in
DIO);
DIORec[getInde
x(source)]
++
;
else
if
D
A
O
pac
k
et
then
localRouteT
able
=
localRouteT
able
[
(source,
destination);
D
A
ORec[getInde
x(source)]
++
;
else
if
D
A
O
A
pac
k
et
then
localRouteT
able
=
localRouteT
able
[
(destination,source);
D
A
O
ARec[getInde
x(source)]
++
;
else
DISRec[getInde
x(source)]
++
;
end
end
end
end
PD
=
(RankT
uples,V
ersionNumberT
uples,DISRec,DIORec,D
A
ORec,D
A
O
ARec,localRouteT
able);
PPD
=
predict
using
timeseries
pattern(partial
descriptor
table);
broadcast
predicted
PPD;
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1761
–
1770
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
1767
Algorithm
2:
DED
A:
Distrib
uted
Early
Detection
Algorithm
Phase-II
Descriptor
T
able
=
;
;
i=0;
n=
number
of
PPD
recei
v
ed;
while
i
<
n
do
Descriptor
T
able
=
Descriptor
T
able
[
PPD[i];
end
i=0;
n=
number
of
nodes
in
the
IoT
system;
while
i
<
n
do
isAttack
er
=
Detect
using
machine
learning
model(Descriptor
T
able,
i);
if
isAttac
k
er
then
Announce
to
all
nearby
nodes;
else
Ignore;
end
end
4.1.
Mathematical
model
of
pr
oposed
system
Let
D
as
a
set
of
descriptor
tables
f
D
1
;
D
2
;
:
:
:
;
D
n
g
where
n
is
the
total
number
of
monitoring
nodes
and
D
i
,
the
descriptor
table
of
i
th
monitoring
node.
D
i
is
set
of
tuples
F
as
D
i
=
f
F
i
1
;
F
i
2
;
:
:
:
;
F
im
g
,
where
m
is
the
number
of
nodes
monitored
by
i
th
monitoring
node.
Let
F
ij
is
the
set
of
features
of
j
th
IoT
node
monitored
by
i
th
monitoring
node
as
F
ij
=
f
f
ij
1
;
f
ij
2
;
:
:
:
;
f
ij
l
g
,
where
l
is
j
F
ij
j
Let
L
is
the
set
of
log
tables
f
L
1
;
L
2
;
:
:
:
;
L
n
g
,
where
n
is
the
total
number
of
Monitoring
nodes
and
L
i
is
the
log
maint
ained
by
i
th
monitoring
node,
as
L
i
=
f
D
it
;
D
it
1
;
:
:
:
;
D
it
r
g
,
where
r
is
the
number
of
past
descriptor
tables
L
i
and
D
it
indicates
snapshot
of
D
i
at
time
t
.
Let
W
i
is
the
set
of
weights
of
i
th
descriptor
table
as
W
i
=
f
w
1
;
w
2
;
:
:
:
;
w
l
g
,
where
l
is
j
F
j
Let
P
D
as
the
set
of
predicted
descriptor
t
ables
f
P
D
1
;
P
D
2
;
:
:
:
;
P
D
n
g
,
where
n
is
the
t
otal
number
of
monitoring
nodes
and
P
D
i
as
the
predicted
descriptor
table
of
i
th
monitoring
node.
V
alue
of
the
predicted
features
f
at
time
t
is
calculated
using,
f
t
=
l
X
i
=0
n
X
j
=0
(
f
j
W
i
j
)
(1)
where
l
indicates
the
number
of
past
descriptor
tables
and
n
indicates
the
number
of
features.
5.
RESUL
T
AN
AL
YSIS
Accurac
y
of
predicting
rank
of
the
node
and
v
ersion
number
of
the
node
in
the
future
based
on
the
history
of
v
alues
present
into
the
descripti
v
e
table
is
presented
in
Figures
4(a)
and
(b)
(see
in
appendix)
respecti
v
ely
.
Features
other
than
the
rank
and
the
v
ersion
number
are
more
predictable
and
a
v
erage
accurac
y
of
prediction
is
sho
wn
in
Figure
4(c)
(see
in
appendix).
Results
also
sho
w
that
we
need
to
k
eep
the
history
of
descripti
v
e
tables
and
a
length
of
the
history
table
has
an
impact
on
the
accurac
y
of
prediction.
It
is
also
e
vident
that
k
eeping
a
v
ery
long
history
in
not
required
as
accurac
y
is
soon
coming
to
the
saturation
point.
W
e
need
to
use
predicted
features
for
early
detection
of
attacks
on
IoT
resources.
Figure
4(d)
(see
in
appendix)
sho
ws
the
accurac
y
of
detecting
the
attack
er
ahead
in
time.
The
accurac
y
of
predicting
long
ahead
is
less
and
tradeof
f
between
the
accurac
y
and
the
time
ahead
has
to
be
decided
in
the
deplo
yment
of
the
proposed
solution.
6.
CONCLUSION
The
proposed
algorithm
w
orks
in
tw
o
parts
in
parallel,
where
phase-I
b
uilds
local
descriptor
table
and
phase-II
b
uilds
a
global
descriptor
table
and
detects
the
presence
of
the
attack
er
.
The
predicted
local
descriptor
holds
future
v
alues
of
fields
present
in
the
control
pack
et.
This
use
of
future
v
alues
by
attack
detection
model
results
in
detection
of
attack
in
an
early
stage.
The
ef
fecti
v
eness
of
detecting
attack
early
is
pro
v
ed
through
the
e
xtensi
v
e
simulation
study
.
The
proposed
algorithm
will
be
v
ery
helpful
in
the
earl
y
detection
of
attacks
and
minimize
damage
in
IoT
systems.
Limitations
of
the
proposed
algorithm
include
the
additional
cost
of
putting
DED
A:
An
algorithm
for
early
detection
of
topolo
gy
attac
ks
in...
(J
alindar
Kar
ande)
Evaluation Warning : The document was created with Spire.PDF for Python.
1768
r
ISSN:
2088-8708
monitoring
nodes
and
incapable
of
detec
ting
unkno
wn
attacks.
Our
future
w
ork
includes
designing
a
predicti
v
e
algorithm
for
early
detection
of
collaborati
v
e
attacks
and
e
v
aluating
its
ef
fecti
v
eness.
APPENDIX
(a)
(b)
(c)
(d)
Figure
4.
Prediction
accurac
y
,
(a)
Rank
prediction
accurac
y
,
(b)
V
ersion
number
prediction
accurac
y
,
(c)
A
v
erage
prediction
accurac
y
of
all
features,
(d)
Accurac
y
of
attack
detection
REFERENCES
[1]
R.
P
.
Singh,
M.
Ja
v
aid,
A.
Haleem,
and
R.
Suman,
“Internet
of
things
(IoT)
applications
to
fight
ag
ainst
co
vid-19
pandemic,
”
Diabetes
and
Metabolic
Syndr
ome:
Clinical
Resear
c
h
and
Re
vie
ws
,
2020.
[2]
J.
Jenkins
and
M.
Burmester
,
“Runtime
inte
grity
for
c
yber
-ph
ysical
infrastructures,
”
International
Con-
fer
ence
on
Critical
Infr
astructur
e
Pr
otection
,
pp.
153-167,
2015.
[3]
A.
Sajid,
H.
Abbas,
and
K.
Saleem,
“Cloud-assisted
IoT
-based
scada
systems
securit
y:
A
re
vie
w
of
the
state
of
the
art
and
future
challenges,
”
IEEE
Access
,
v
ol.
4,
pp.
1375-1384,
2016.
[4]
Y
.
Cherdantse
v
a,
P
.
Burnap,
A.
Blyth,
P
.
Eden,
K.
Jones,
H.
Soulsby
,
and
K.
Stoddart,
“
A
re
vie
w
of
c
yber
security
risk
assessment
methods
for
scada
systems,
”
computer
s
and
security
,
v
ol.
56,
pp.
1-27,
2016.
[5]
J.
E.
Rubio,
C.
Alcaraz,
R.
Roman,
and
J.
Lopez,
“
Analysis
of
intrusion
detection
systems
in
industrial
ecosystems,
”
14th
International
Confer
ence
on
Security
and
Crypto
gr
aphy
(SECR
YPT
2017)
,
2017.
[6]
I.
Stellios,
P
.
K
otzanik
olaou,
M.
Psarakis,
C.
Alcaraz,
and
J.
Lopez,
“
A
surv
e
y
of
IoT
-enabled
c
yberat-
tacks:
Assessing
attack
paths
to
critical
infrastructures
and
services,
”
IEEE
Communications
Surve
ys
and
T
utorials
,
pp.
3453-3495,
2018.
[7]
A.-T
.
F
adi
and
D.
B.
Da
vid,
“Seamless
authentication:
F
or
IoT
-big
data
technologies
in
smart
industrial
application
systems,
”
IEEE
T
r
ansactions
on
Industrial
Informatics
,
2020.
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1761
–
1770
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
1769
[8]
Q.
Liu,
B.
Gong,
and
Z.
Ning,
“Research
on
clpkc-idpkc
cross-domain
identity
authenticat
ion
for
IoT
en
vironment,
”
Computer
Communications
,
2020.
[9]
Y
.
Y
ang,
X.
Zheng,
W
.
Guo,
X.
Liu,
and
V
.
Chang,
“Pri
v
ac
y-preserving
smart
IoT
-based
healthcare
big
data
storage
and
self-adapti
v
e
access
control
system,
”
Information
Sciences
,
v
ol.
479,
pp.
567-592,
2019.
[10]
S.
Rajesh,
V
.
P
aul,
V
.
G.
Menon,
and
M.
R.
Khosra
vi,
“
A
secure
and
ef
ficient
lightweight
symmetric
encryption
scheme
for
transfer
of
te
xt
files
between
embedded
IoT
de
vices,
”
Symmetry
,
v
ol.
11,
no.
2,
2019.
[11]
R.
W
il
liams,
E.
McM
ahon,
S.
Samtani,
M.
P
atton,
and
H.
Chen,
“Identifying
vulnerabilities
of
consumer
internet
of
things
(IoT)
de
vices:
A
scalable
approach,
”
IEEE
International
Confer
ence
on
Intellig
ence
and
Security
Informatics
(ISI)
,
pp.
179-181,
2017.
[12]
T
.
W
inter
,
P
.
Thubert,
A.
Brandt,
J.
W
.
Hui,
R.
K
else
y
,
P
.
Le
vis,
K.
Pister
,
R.
Struik,
J.-P
.
V
asseur
,
R.
K.
Ale
xander
,
“RPL:
IPv6
routing
protocol
for
lo
w-po
wer
and
lossy
netw
orks,
”
RFC
,
v
ol.
6550,
pp.
1-157,
2012.
[13]
A.
Arena,
P
.
Perazzo,
C.
V
allati,
G.
Dini,
and
G.
Anastasi,
“Ev
aluating
and
impro
ving
the
scalability
of
RPL
security
in
the
internet
of
things,
”
Computer
Communications
,
v
ol.
151,
pp.
119-132,
2020.
[14]
C.
ord
Kahn,
P
.
A.
Porras,
S.
Staniford-Chen,
and
B.
T
ung,
“
A
common
intrusion
detection
frame
w
ork,
”
J
ournal
of
Computer
Security
,
1998.
[15]
M.
Roesch,
“SNOR
T
:
Lightweight
intrusion
detection
for
netw
orks,
”
Lisa
,
v
ol.
99,
no.
1,
pp.
229-238,
1999.
[16]
A.
Sforzin,
F
.
G.
M
´
armol,
M.
Conti,
and
J.-M.
Bohli,
“RPIDS:
Ras
pberry
pi
ids-a
fruitful
intrusion
detec-
tion
system
for
IoT
,
”
2016
Intl
IEEE
Confer
ences
on
Ubiquitous
Intellig
ence
and
Computing
,
Advanced
and
T
rusted
Computing
,
Scalable
Computing
and
Communications,
Cloud
and
Big
Data
Computing
,
In-
ternet
of
P
eople
,
and
Smart
W
orld
Congr
ess
(UIC/A
TC/ScalCom/CBDCom/I
oP/SmartW
orld
,
pp.
440-448,
2016.
[17]
V
.
S
harma,
G.
Choudhary
,
Y
.
K
o,
and
I.
Y
ou,
“Beha
vior
and
vulnerability
assessment
of
drones-enabled
industrial
internet
of
things
(IIoT),
”
IEEE
Access
,
v
ol.
6,
pp.
43368-43383,
2018.
[18]
G.
Geor
ge
and
S.
M.
Thampi,
“
A
graph-based
security
frame
w
ork
for
securing
indust
rial
IoT
netw
orks
from
vulnerability
e
xploitations,
”
IEEE
Access
,
v
ol.
6,
pp.
43586-43601,
2018.
[19]
L.
W
allgren,
S.
Raza,
and
T
.
V
oigt,
“Routing
a
ttacks
and
countermeasures
in
the
RPL-based
internet
of
things,
”
International
J
ournal
of
Distrib
uted
Sensor
Networks
,
v
ol.
9,
no.
8,
2013.
[20]
D.
Airehrour
,
J.
Gutierrez,
and
S.
K.
Ray
,
“Secure
routing
for
Internet
of
things:
A
surv
e
y
,
”
J
ournal
of
Network
and
Computer
Applications
,
v
ol.
66,
pp.
198-213,
2016.
[21]
A.
Le,
J.
Loo,
Y
.
Luo,
and
A.
Lasebae,
“Specification-based
ids
for
securing
RPL
from
topology
attacks,
”
W
ir
eless
Days
(WD),
pp.
1-3,
2011.
[22]
M.
Burmester
and
B.
De
Medeiros,
“On
the
security
of
route
disco
v
ery
in
MANETs,
”
IEEE
T
r
ansactions
on
Mobile
Computing
,
v
ol.
8,
no.
9,
pp.
1180-1188,
2009.
[23]
W
.
Meng,
“Intrusion
detection
in
the
era
of
IoT
:
Building
trust
via
traf
fic
filtering
and
sampling,
”
Com-
puter
,
v
ol.
51,
no.
7,
pp.
36-43,
2018.
[24]
F
.
Y
u,
Z.
Chen,
Y
.
Diao,
T
.
Lakshman,
and
R.
H.
Katz,
“F
ast
and
memory-ef
ficient
re
gular
e
xpression
matching
for
deep
pack
et
inspection,
”
Pr
oceedings
of
the
2006
A
CM/IEEE
symposium
on
Ar
c
hitectur
e
for
networking
and
communications
systems,
pp.
93–102,
2006.
[25]
L.
Xiao,
X.
W
an,
X.
Lu,
Y
.
Zhang,
and
D.
W
u,
“IoT
security
techniques
based
on
machine
learning,
”
arXiv
pr
eprint
arXiv:1801.06275
,
2018.
[26]
N.
Muralidhar
,
C.
W
ang,
N.
Self,
M.
M
omtazpour
,
K.
Nakayama,
R.
Sharma,
and
N.
Ramakrishnan,
“ILLIAD:
Intelligent
in
v
ariant
and
anomaly
detection
in
c
yber
-ph
ysical
system
s,
”
A
CM
T
r
ansactions
on
Intellig
ent
Systems
and
T
ec
hnolo
gy
(TIST)
,
v
ol.
9,
no.
3,
pp.
1-20,
2018.
[27]
A.
Saeed,
A.
Ahmadinia,
A.
Ja
v
ed,
and
H.
Larijani,
“Intelligent
intrusion
detection
in
lo
w-po
wer
IoTs,
”
A
CM
T
r
ansactions
on
Internet
T
ec
hnolo
gy
,
v
ol.
16,
no.
4,
pp.
1-25,
2016.
[28]
H.
Sedjelmaci,
S.
m.
Senouci,
and
T
.
T
aleb,
“
An
accurate
security
g
ame
for
lo
w-resource
IoT
de
vices,
”
IEEE
T
r
ansactions
on
V
ehicular
T
ec
hnolo
gy
,
v
ol.
PP
,
no.
99,
pp.
1-1,
2017.
[29]
K.
Mrug
ala,
N.
T
uptuk,
and
S.
Hailes,
“Ev
olving
attack
ers
ag
ainst
wireless
sensor
netw
orks
using
genetic
programming,
”
IET
W
ir
eless
Sensor
Systems
,
v
ol.
7,
no.
4,
pp.
113-122,
2017.
[30]
A.
Abeshu
and
N.
Chilamkurti,
“Deep
learning:
the
frontier
for
distrib
uted
attack
detection
in
fog-to-
things
computing,
”
IEEE
Communications
Ma
gazine
,
v
ol.
56,
no.
2,
pp.
169-175,
2018.
DED
A:
An
algorithm
for
early
detection
of
topolo
gy
attac
ks
in...
(J
alindar
Kar
ande)
Evaluation Warning : The document was created with Spire.PDF for Python.
1770
r
ISSN:
2088-8708
[31]
S.
Sharmeen,
S.
Huda,
J.
H.
Aba
w
ajy
,
W
.
N.
Ismail,
and
M.
M.
Hassan,
“Mal
w
are
threats
and
detection
for
industrial
mobile-IoT
netw
orks,
”
IEEE
access
,
v
ol.
6,
pp.
15941-15957,
2018.
[32]
E.
Hodo,
X.
Bellek
ens,
A.
Hamilton,
P
.-L.
Dubouilh,
E.
Iork
yase,
C.
T
achtatzis,
and
R.
Atkinson,
“Threat
analysis
of
IoT
netw
orks
using
artificial
neural
netw
ork
intrusion
detection
system,
”
2016
International
Symposium
on
Networks,
Computer
s
and
Communications
(ISNCC),
pp.
1-6,
2016.
[33]
C.
Zhou,
S.
Huang,
N.
Xiong,
S.-H.
Y
ang,
H.
Li,
Y
.
Qin,
and
X.
Li,
“Design
and
analysis
of
multimodel-
based
anomaly
intrusion
detection
systems
in
industrial
process
automation,
”
IEEE
T
r
ansactions
on
Sys-
tems,
Man,
and
Cybernetics:
Systems
,
v
ol.
45,
no.
10,
pp.
1345-1360,
2015.
[34]
H.
W
u
and
W
.
W
ang,
“
A
g
ame
theory
based
collaborati
v
e
security
detection
method
for
Internet
of
things
systems,
”
IEEE
T
r
ansactions
on
Information
F
or
ensics
and
Security
,
v
ol
.
13,
no.
6,
pp.
1432-1445,
2018.
[35]
D.
Midi,
S.
Sultana,
and
E.
Bertino,
“
A
system
for
response
and
pre
v
ention
of
security
incidents
in
wireless
sensor
netw
orks,
”
A
CM
T
r
ansactions
on
Sensor
Networks
,
v
ol.
13,
no.
1,
pp.
1-38,
2016.
[36]
A.
Mayzaud,
R.
Badonnel,
and
I.
Chrisment,
“
A
distrib
uted
monitoring
strate
gy
for
detecting
v
ersion
num-
ber
attacks
in
RPL-based
netw
orks,
”
IEEE
T
r
ansactions
on
Network
and
Service
Mana
g
ement
,
v
ol.
14,
no.
2,
pp.
472-486,
2017.
[37]
S.
Praba
v
ath
y
,
K.
Sundarakantham,
and
S.
M.
Shalinie,
“Design
of
cogniti
v
e
fog
com
puting
for
intrusion
detection
in
Internet
of
things,
”
J
ournal
of
Communications
and
Networks
,
v
ol.
20,
no.
3,
pp.
291-298,
2018.
[38]
J.
Karande
and
S.
Joshi,
“Comprehensi
v
e
assess
ment
of
security
attack
detection
algorithms
in
Internet
of
things,
”
2018
F
ourth
International
Confer
ence
on
Computing
Communication
Contr
ol
and
A
utomation
(ICCUBEA),
pp.
1-6,
2018.
[39]
M.
Chern
yshe
v
,
Z.
Baig,
O.
Bello,
and
S.
Zeadally
,
“Internet
of
things
(IoT):
Research,
simulators,
and
testbeds,
”
IEEE
Internet
of
Things
J
ournal
,
v
ol.
5,
no.
3,
pp.
1637-1647,
2017.
BIOGRAPHIES
OF
A
UTHORS
J
alindar
Karande
recei
v
ed
Masters
in
Computer
Science
and
Engi
neering
and
Bachelors
in
Infor
-
mation
T
echnology
from
the
Uni
v
ersity
of
Pune.
He
is
currently
a
research
scholar
in
the
Department
of
Computer
Engineering,
Pune
Ins
titute
of
Computer
T
echnology
,
Sa
vitribai
Phule
Pune
Uni
v
ersity
,
Pune,
India.
He
is
Google
Cloud
Platform
Certified
Professional
Data
Engineer
and
also
holds
a
li-
cense
from
Databricks
for
Spark.
His
research
interests
include
IoT
,
Big
Data
and
Machine
Learning.
He
has
published
se
v
eral
papers
in
Journa
ls
and
reputed
conferences
in
these
areas.
He
is
af
filiated
with
IEEE
as
Graduate
Student
member
.
Pr
of
.
Sarang
J
oshi
recei
v
ed
a
PhD
in
Computer
Science
and
Engineering
from
Bharati
V
idyapeeth,
Pune,
India.
He
recei
v
ed
a
Masters
in
Computer
Engineering
and
a
Bachelors
in
Computer
Engi-
neering
from
Uni
v
ersity
of
Pune,
India.
He
is
currently
a
Professor
in
Department
of
Computer
Engineering,
Pune
Institute
of
Computer
T
echnology
,
Sa
vitribai
Phule
Pune
Uni
v
ersity
,
Pune,
In-
dia.
He
has
30
years
of
teaching
e
xperience.
His
research
interests
include
Algorithms,
Intelligence,
IoT
,
Big
Data
and
Machine
Learning.
He
has
guided
se
v
eral
research
scholars
and
published
se
v-
eral
papers
in
reputed
Journals
and
Conference
Proceedings
in
these
areas.
He
has
pre
viously
serv
ed
as
Chairman,
Board
of
Studies
of
Computer
Engineering
at
Sa
vitribai
Phule
Pune
Uni
v
ersity
.
He
has
authored
books
on
“Big
Data
Mining
-
Appli
cation
Perspecti
v
e”
and
“Design
and
Analysis
of
Algorithms”.
Int
J
Elec
&
Comp
Eng,
V
ol.
11,
No.
2,
April
2021
:
1761
–
1770
Evaluation Warning : The document was created with Spire.PDF for Python.