Inter
national
J
our
nal
of
Recongurable
and
Embedded
Systems
(IJRES)
V
ol.
13,
No.
3,
No
v
ember
2024,
pp.
748
∼
757
ISSN:
2089-4864,
DOI:
10.11591/ijres.v13.i3.pp748-757
❒
748
Impr
o
ving
the
perf
ormance
of
IoT
de
vices
that
use
W
i-Fi
Ali
Ahmed
Razzaq,
K
unjam
Nageswara
Rao
Department
of
Computer
Science
and
Systems
Engineering,
Colle
ge
of
Engineering,
Andhra
Uni
v
ersity
,
V
isakhapatnam,
India
Article
Inf
o
Article
history:
Recei
v
ed
No
v
10,
2023
Re
vised
May
26,
2024
Accepted
Jul
3,
2024
K
eyw
ords:
Identity
management
system
Internet
of
things
Machine
learning
Po
wer
consumption
Quality
of
service
ABSTRA
CT
Pro
viding
quality
service
to
users
of
the
internet
of
things
(IoT)
entails
address-
ing
tw
o
cruci
al
aspects:
one
related
to
security
and
the
other
concerning
the
limited
resources
of
IoT
de
vices.
W
e
will
f
ace
a
challenge
while
using
time-
sensiti
v
e
applications
within
a
netw
ork
that
utilizes
a
high-performance
W
i-Fi
technology
with
e
xceeding
ener
gy
consumption.
Due
to
this
research
challenge,
we
propose
a
ne
w
algorithm,
IoT
-quali
ty
of
service
(QoS),
designed
to
achie
v
e
a
true
balance
between
enhancing
the
security
aspects
of
IoT
de
vices
and
im-
pro
ving
netw
ork-hardw
are
performance.
Thus,
the
algorit
hm
ef
ciently
man-
ages
the
limited
ener
gy
resources
by
monitoring
ener
gy
le
v
els,
communication
quality
,
and
queuing
delay
at
access
points.
This
is
accomplished
by
utilizing
a
streamlined
identity
management
system
capable
of
achie
ving
authentication
and
access
authorization
with
reduced
loading
for
IoT
de
vices.
The
research
h
y-
pothesis
underwent
v
alidation
through
a
comparati
v
e
analysis
of
its
performance
ag
ainst
the
con
v
entional
model
of
a
W
i-Fi-based
IoT
de
vice.
This
e
v
aluation
w
as
conducted
utilizing
the
NS3
simulator
and
w
as
based
on
a
predeterm
ined
set
of
parameters
inuencing
the
e
xamined
perfor
mance
metrics,
including
po
wer
consumption,
throughput,
delay
,
and
response
time.
The
ndings
e
xposed
the
superiority
of
the
proposed
algorithm.
This
is
an
open
access
article
under
the
CC
BY
-SA
license
.
Corresponding
A
uthor:
Ali
Ahmed
Razzaq
Department
of
Computer
Science
and
Systems
Engineering,
Colle
ge
of
Engineering,
Andhra
Uni
v
ersity
V
isakhapatnam
530003,
Andhra
Pradesh,
India
Email:
taif
ali607@gmail.com
1.
INTR
ODUCTION
The
quality
of
service
(QoS)
is
a
parameter
that
assesses
the
o
v
erall
performance
of
a
service,
par
-
ticularly
the
performance
observ
ed
(e
xperienced)
by
service
users.
In
light
of
the
e
xtensi
v
e
utilization
and
implementation
of
internet
of
things
(IoT)
services
in
our
day-to-day
acti
vities,
it
becomes
essential
to
lo
wer
the
e
xpenses
associated
with
IoT
de
vices,
all
while
ensuring
that
the
le
v
el
of
pro
vided
QoS
remains
uncom-
promised.
Also,
Under
the
concept
of
the
IoT
,
there
are
countless
de
vices
with
dif
ferent
characteristics
and
capabilities.
So
to
impro
v
e
the
QoS
associated
with
IoT
,
tw
o
k
e
y
elements
should
be
ensured,
namely:
netw
ork
security
to
achie
v
e
pri
v
ac
y
and
the
security
of
netw
ork
resources,
and
the
ef
cient
administration
and
allocation
of
netw
ork
resources.
Prominennt
research
papers
in
IoT
de
vices
and
QoS
can
be
found
in
[1]-[6].
The
concept
of
de
vice
isolation
is
crucial,
as
it
prohibits
direct
access
from
the
internet,
ensuring
pre
v
ention
of
unauthorized
access
and
pri
v
ac
y
violations.
Both
authentication
and
access
authorization
pose
challenges
for
the
IoT
,
gi
v
en
its
di
v
er
gence
from
traditional
internet
components,
wherein
IoT
de
vices
are
predominantly
purpose-specic
and
typically
ha
v
e
limited
resources.
The
identity
m
anagement
(IdM)
system
pro
vides
both
authentication
and
access
authorization
for
internet
users
(user
identity
information
management)
This
system
consists
of
four
components,
as
illustrated
in
Figure
1,
namely
entities
(users
or
de
vices),
identiers
(entity
Identities),
identier
pro
vider
(IdP),
and
service
pro
vider
(SP).
J
ournal
homepage:
http://ijr
es.iaescor
e
.com
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Recongurable
&
Embedded
Syst
ISSN:
2089-4864
❒
749
Figure
1.
Components
of
the
IdM
system
Ef
forts
and
research
ha
v
e
been
directed
to
w
ards
proposing
se
v
eral
approaches
that
could
le
v
erage
IdM
with
IoT
,
specically
in
the
realms
of
authenti
cation
and
access
authorization
schemes.
A
proposed
authentication
and
access
control
frame
w
ork
for
IoT
de
vices
considered
de
vices
as
nal
entities
in
the
internet
architecture,
communicating
through
unique
IPv6
addresses.
It
utilized
the
OpenID
protocol
for
authentication
and
the
role-based
access
control
(RB
A
C)
protocol
for
access
authorization.
Ho
we
v
er
,
the
proposal
did
not
address
single
sign-on
(SSO)
is
sues
and
did
not
highlight
an
y
results
that
could
v
alidate
the
suggestion
carried
out
by
Liu
et
al
.
[7].
In
Chibelushi
et
al
.
[8]
studied
an
IdM
system
for
IoT
in
the
healthcare
conte
xt,
b
ut
it
f
ailed
to
pro
vide
secure
communication,
lea
ving
IoT
de
vices
accessible
directly
from
the
internet.
Later
,
Leo
et
al
.
[9]
utilized
web
services
between
the
internet
and
IoT
to
ensure
condentiality
and
security
of
transmitted
information.
This
study
,
ho
we
v
er
,
is
not
considered
to
be
capable
of
securing
end-to-end
security
interaction
between
the
internet
and
IoT
,
secure
communication
channels,
or
e
v
en
a
SSO
service.
Ho
we
v
er
,
W
itk
o
vski
et
al
.
[10]
suggested
inte
grating
IdM
with
IoT
to
maintain
SSO
and
data
encryption
between
communicating
parties.
Ho
we
v
er
,
the
study
did
not
pro
vide
an
y
results
related
to
po
wer
consumption,
especially
considering
that
the
pro
vision
of
SSO
is
based
on
encryption
k
e
ys.
Recenctly
,
Santos
et
al
.
[11]
introduced
the
unied
federated
lightweight
authentication
of
things
(FLA
T)
authenticati
on
protocol,
combining
symmetric
encryption
systems
and
embedded
certicates,
bypassing
the
principles
of
asymmetric/symmetric
encryption
used
in
traditional
federated
IdM
systems.
Y
et,
it
did
not
tak
e
into
account
access
authorization
processes
and
service
disco
v
ery
.
Other
studies
about
using
IoT
with
IdM
are
found
in
[12]-[17].
This
research
aims
at
le
v
eraging
articial
intelligence
(AI)
techniques
to
accomplish
this
task,
gi
v
en
their
notable
presence
in
addressing
the
chal
lenges
posed
by
the
upcoming
generations
in
the
eld
of
wireless
communications.
In
this
research,
we
use
IoT
de
vices
emplo
ying
W
i-Fi
technology
for
netw
ork
connecti
vity
.
This
choice
is
based
on
the
widespread
use
of
wireless
local
area
netw
orks
(WLAN)
in
the
unlicensed
spectrum,
highlighting
the
increasing
comple
xity
in
wireless
netw
orks.
Quality
of
communication
standards
in
the
IoT
must
ensure
stability
and
accurac
y
for
the
utilized
learning
technology
.
Sindjoung
and
Minet
[18]
distinguished
between
tw
o
types
of
communication
quality
standards,
one
is
hardw
are-dependent
and
the
other
is
softw
are-
dependent.
Hardw
are-dependent
standards
directly
collect
data
from
the
de
vices
without
preprocessing
and
include
indicators
such
as
recei
v
ed
signal
strength
indicator
(RSSI),
link
quality
indicator
(LQI),
and
signal-
to-noise
ratio
(SNR).
The
precision
pro
vided
by
hardw
are-dependent
standards
is
insuf
cient
for
tw
o
main
reasons.
Firstly
,
only
successfully
transmitted
pack
ets
are
considered,
and
secondly
,
the
e
v
aluation
does
not
tak
e
into
account
the
entire
pack
et
b
ut
only
its
initial
symbols.
Ho
we
v
er
,
acquiring
its
v
alues
requires
undertaking
computational
operations,
namely:
pack
et
deli
v
ery
ratio,
the
required
number
of
pack
et
transmissions,
and
the
de
gree.
As
attaining
performance
quality
for
the
IoT
netw
ork
is
the
primary
objecti
v
e,
it
is
necessary
to
con-
sider
the
allocat
ed
resources
and
attempt
t
o
utilize
them
optimally
based
on
the
outcomes
of
machine
learning
algorithms
(i.e.,
channel
state).
Channel
access
is
often
congested
simultaneously
,
particularly
when
a
lar
ge
Impr
o
ving
the
performance
of
IoT
de
vices
that
use
W
i-F
i
(Ali
Ahmed
Razzaq)
Evaluation Warning : The document was created with Spire.PDF for Python.
750
❒
ISSN:
2089-4864
number
of
de
vices
connect
to
the
same
wireless
channel
at
the
same
time.
Consequently
,
the
channel
becomes
o
v
erloaded,
So
Ma
et
al
.
[19]
proposed
a
deep
learning-based
channel
allocation
algorithm,
applying
time
of
w
ait
(T
oW)
for
selecting
communication
channels
in
massi
v
e
cogniti
v
e
IoT
netw
orks.
Their
results
demon-
strated
signicant
impro
v
ement
in
interference
det
ection
compared
to
traditional
methods
not
relying
on
deep
learning.
Ener
gy
allocation
and
inte
rference
management
are
crucial
aspects
af
fect
ing
IoT
netw
orks.
F
or
this
reason
L
yngg
aard
[20]
proposed
a
dynamic
system
for
interference
detection
and
ener
gy
all
ocation,
based
on
the
interference
le
v
el
in
radio
channels.
The
y
applied
the
channel
state
information
(CSI)
algorithm
to
predict
transmission
ener
gy
le
v
els
(based
on
CSI).
Considering
that
man
y
IoT
de
vices
are
small-sized
with
limited
bat-
tery
capacity
,
intel
ligent
management
and
allocation
of
this
scarce
resource
are
essential.
Hence,
Zeki
´
c-Su
ˇ
sac
et
al
.
[21]
suggested
an
AI
based
ener
gy
management
system
for
smart
cities
relying
on
IoT
.
Neural
netw
orks,
decision
trees,
and
random
learning
methods
were
emplo
yed
to
predict
ener
gy
consumption
in
those
cities,
demonstrating
impro
v
ed
ener
gy
consumption
predictions
compared
to
non-AI-based
approaches.
In
Becv
ar
et
al
.
paper’
s
[22],
it
w
as
found
that
predicting
channel
quality
using
machine
learning,
le
v
eraging
netw
ork
correlations,
pro
v
ed
ef
cient
in
reducing
o
v
erall
e
xpenses
compared
to
the
traditional
pilot-
based
approach,
e
xceeding
90%.
It’
s
note
w
orth
y
that
the
study
netw
ork
includes
a
lar
ge
number
of
nodes
that
communicate
with
each
other
.
Y
et,
another
study
T
orres-Alv
arado
et
al
.
[23]
emphasized
the
importance
of
adopting
machine
learning
algorithms
to
predict
channel
quality
(lo
w
or
high)
for
IoT
netw
orks,
where
authentication
processes
are
af
fected
by
noise
and
radiation
(associated
with
channel
quality),
especially
when
implemented
in
hardw
are
(such
as
cogniti
v
e
radio
de
vices).
According
to
their
e
xperiments,
the
random
forest
algorithm
achie
v
ed
the
highest
classication
accurac
y
of
95.54%.
In
our
presented
research,
we
seek
to
preserv
e
service
quality
with
both
its
k
e
y
elements
via
a
tw
o
fold
strate
gy
.
Firstly
,
we
utilize
a
modied
IdM
to
enhance
c
ybersecurity
.
Secondly
,
we
adopt
AI
techniques
to
predict
communication
quality
.
This
is
coupled
with
monitoring
ener
gy
le
v
els
and
queuing
delays
at
access
points
to
ef
ciently
manage
the
ener
gy
re
source
in
IoT
de
vices.
This
will
be
achie
v
ed
without
adv
ersely
impacting
latenc
y
,
recognizing
it
as
a
critical
criterion
for
time-sensiti
v
e
applications.
This
paper
is
structured
as
follo
ws:
the
proposed
algorithm
presented
in
section
2,
In
section
3,
discussion
and
results.
Finally
,
section
4
concludes
our
paper
.
2.
THE
PR
OPOSED
ALGORITHM
In
this
section,
it
is
essential
to
re
vie
w
the
k
e
y
points
upon
which
our
research
proposal
is
based,
aiming
to
achie
v
e
the
research
goal,
before
delving
into
the
detailed
operational
mechanisms
(as
outlined
in
the
o
wchart
re
vie
w).
The
foundational
aspects
of
the
w
ork
are
di
vided
into
tw
o
parts
according
to
its
objecti
v
e.
2.1.
Fundamentals
of
r
esour
ce
management
mechanism
In
this
research,
we
rely
on
se
v
eral
k
e
y
points
to
accomplish
our
w
ork.
Due
to
the
limited
r
esources
of
IoT
de
vices,
ef
fecti
v
e
resource
management
translates
to
enhancing
the
quality
of
service
pro
vided
to
netw
ork
users.
In
our
proposed
algorithm,
the
focus
is
directed
to
w
ards
the
limited
ener
gy
resource
and
ho
w
to
ef
-
ciently
utilize
it
while
emplo
ying
W
i-Fi
as
a
means
of
data
e
xchange.
This
includes
considering
the
potential
delays
introduced
by
ener
gy-sa
ving
measures
and
mitig
ating
t
h
e
ir
impact
on
time-sensiti
v
e
applications.
In
the
follo
wing
ar
gument,
we
will
re
vie
w
the
mathematical
models
emplo
yed
to
check
both
parameters.
More-
o
v
er
,
we
will
identify
the
machine
learning
algorithm
that
will
contrib
ute
to
enhancing
ener
gy
ef
cienc
y
by
encouraging
the
wireless
card
to
enter
a
sleep
mode
when
the
channel
quality
is
poor
.
2.1.1.
P
o
wer
consumption
model
The
proposed
approach
depends
on
predicting
connection
quality
while
concurrently
monitoring
net-
w
ork
load
dir
ected
to
w
ards
the
acti
v
ated
IoT
de
vice
in
po
wer
-sa
ving
mode
(i.e.,
when
it
is
in
sleep
mode)
f
acilitated
by
W
i-Fi.
Ho
we
v
er
,
such
prediction
and
monitoring
are
contingent
on
the
de
vice
remaining
po
wer
capacity
.
It
is
crucial
to
emphasize
that
the
algorithm
necessitates
dependence
on
a
mathematical
model
to
compute
the
wireless
card’
s
po
wer
consumption,
as
dened
by
(1)
[24].
P
av
g
=
P
T
x
∗
T
T
x
+
P
R
x
∗
T
R
x
+
P
I
∗
T
I
+
P
S
∗
T
S
T
(1)
As
it
is
re
v
ealed
by
the
former
equation,
the
lifetime
of
the
wireless
card
stays
in
each
of
its
operating
modes
(transmitting
T
T
x
),
recei
ving
T
R
x
,
idle
T
I
,
and
sleeping
T
S
)
is
multiplied
by
the
basic
po
wer
consump-
tion
v
alue
of
the
mode
(transmitting
mode
P
T
x
,
recei
ving
mode
P
R
x
,
idle
mode
P
I
,
and
Sleeping
mode
P
S
),
Int
J
Recongurable
&
Embedded
Syst,
V
ol.
13,
No.
3,
No
v
ember
2024:
748–757
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Recongurable
&
Embedded
Syst
ISSN:
2089-4864
❒
751
gi
v
es
a
simplied
model
for
calculating
consumption,
noting
that
the
symbol
T
indicates
the
simulation
time
(the
sum
of
the
presence
times
in
the
operating
modes).
2.1.2.
A
v
erage
delay
The
proposal
did
not
o
v
erlook
the
nature
of
the
transmitted
data
,
taking
into
consideration
the
e
xi
s-
tence
of
tw
o
types
or
classications
of
data,
one
of
which
requires
calculating
the
delay
standard,
gi
v
en
that
it
is
time-sensiti
v
e
(as
in
critical
industrial
applications).
Therefore,
the
a
v
erage
delay
e
xperienced
by
a
data
pack
et
destined
for
an
IoT
de
vice
depends
on
the
po
wer
-sa
ving
mode
pro
vided
by
W
i-Fi
communication
technology
,
with
consideration
of
tw
o
f
actors,
one
of
which
is
the
probability
of
the
pack
et
arri
ving
while
the
wireless
card
of
the
IoT
de
vice
is
in
sleep
mode
P
r
sl
eep
,
and
this
leads
to
a
delay
in
the
queue
of
the
access
point,
consider
ing
that
notication
of
its
e
xistence
(in
order
to
be
reco
v
ered
by
the
de
vice)
will
be
made
only
at
the
be
ginning
of
the
ne
xt
beacon
period.
Not
only
that,
b
ut
there
is
another
w
aiting(delay)
that
the
stored
pack
et
suf
fers
from,
with
the
be
ginning
of
the
beacon
period,
namely
the
serving
time
of
the
pack
ets
that
precede
it
in
the
queue
of
the
access
point
¯
d
|
(
sl
eep
n
)
.
Another
f
actor
contrib
uting
to
the
calculation
of
the
a
v
erage
del
ay
criterion
is
the
a
v
erage
reco
v
ery
time
of
pack
ets
stored
at
the
access
point
by
the
IoT
de
vice,
after
w
aking
up
d
av
g
.
Based
on
the
abo
v
e,
the
a
v
erage
is
gi
v
en
according
to
(2)
(see
[25],
[26]):
D
S
I
n
=
P
r
sl
eep
∗
¯
d
|
sl
eep
n
+
d
av
g
(2)
2.1.3.
Communication
quality
pr
ediction
Machine
learning
enables
syste
ms
to
of
fer
dynamically
learning
and
enhance
performance
wit
hout
being
e
xplicitly
programmed.
There
e
xist
both
linear
and
non-linear
models
for
machine
learning
techniques
(see
[12]).
The
random
forest
classier
w
as
used
as
part
of
the
proposed
algorithm
for
assessing
the
quality
of
netw
ork
communication,
based
on
tw
o
standards.
These
are:
the
RSSI
by
the
IoT
de
vice,
which
is
a
simple
hardw
are
standard
that
can
pro
vide
an
accurate
and
f
ast
estimate
of
the
quality
of
communication
(see
[22]).
The
RSSI
a
v
erage
of
an
IoT
de
vice
retrie
ving
data
pack
ets
from
the
access
point
(AP)
(which
is
numbered
n
pack
ets
during
the
beacon
period)
is
gi
v
en
according
to
(4)
(see
[18]):
R
S
S
I
av
g
=
P
n
i
=1
R
S
S
I
i
n
(3)
Furthermore,
there
is
the
standard
called
pack
et
deli
v
ery
ratio
(PDR),
which
is
a
softw
are
standard
equal
to
the
ratio
of
the
number
of
pack
ets
successfully
recei
v
ed
by
an
IoT
de
vice
(successful
receipt
necessarily
means
the
recipient
sending
an
acki
notication
to
the
s
ender
(which
is
the
access
point
here))
to
the
number
of
pack
ets.
P
ack
et
j
sent
by
the
access
point
a
t
the
be
ginning
of
each
beacon
period
and
is
gi
v
en
according
to
the
relationship
as
in
(4)
(see
[18]):
P
D
R
=
P
n
i
=1
ack
i
P
m
j
=1
pack
et
j
(4)
2.2.
Fundamentals
of
security-r
elated
operational
mechanism
The
research
objecti
v
e
is
to
achie
v
e
service
quality
in
IoT
netw
orks,
and
true
service
quality
cannot
be
attained
without
considering
the
security
aspect
of
the
netw
ork.
The
proposed
algorithm
relies
on
the
concept
of
IdM
system
to
e
x
ecute
authentication
and
access
authorization
operations,
yet
the
adopted
system
is
a
modied
one.
2.2.1.
The
used
identity
management
system
The
modied
IdM
system
adopted
in
the
proposed
algorithm
depends
basically
on
tw
o
fundamental
points.
Firstly:
using
conte
xtual
parameters
that
distinguish
the
user
(such
as
its
identity
,
role,
acti
vities,
location,
whether
ph
ysical
(global
positioning
system
(GPS))
or
virtual
(internet
protocol
(IP)
address)
and
the
social
netw
orks
i
t
utilizes,
in
addition
to
the
type
of
data
that
determines
the
sensiti
vity
of
the
data),
within
the
user
identiers
,
in
which
the
y
will
participate
in
the
access
control
mechanism.
Secondly:
encryption
of
the
transmitted
data
at
tw
o
le
v
els
using
tw
o
encryption
k
e
ys
k
e
y
encryption
k
e
y
(KEK)
(encrypts
the
content
of
messages
e
xchanged
during
the
session),
and
MKK
(
k
e
y
encryption
KEK),
where
the
ANSI
X.9.17
standard
is
used
to
manage
the
distrib
ution
of
k
e
ys
(see
[10]).
In
accordance
with
what
ha
v
e
been
discussed,
an
IoT
de
vice
should
implement
tw
o
programmed
modules
(see
[8]):
i)
the
conte
xt
unit,
composed
of
tw
o
sub-module
s.
One
sub-module
deals
with
identiers
and
utilizes
them
within
an
algorithm
that
lters
response
content
to
serv
e
the
Impr
o
ving
the
performance
of
IoT
de
vices
that
use
W
i-F
i
(Ali
Ahmed
Razzaq)
Evaluation Warning : The document was created with Spire.PDF for Python.
752
❒
ISSN:
2089-4864
request.
The
other
sub-m
od
ul
e
focuses
on
constructing
identiers
based
on
requests
from
users
or
IoT
de
vice
users;
and
ii)
the
pri
v
ac
y
unit,
responsible
for
sending
requests
and
recei
ving
responses
subject
to
authentication
and
authorization
processes
through
dedicated
serv
ers
e
xternal
to
the
IoT
netw
ork
(the
crucial
point
here
is
the
of
oading
of
pri
v
ac
y
polic
y
b
urdens
from
the
IoT
de
vice,
contrary
to
the
study),
in
addition
to
the
required
encryption
and
decryption
operations.
2.3.
The
pr
oposed
algorithm
inter
net
of
things-quality
of
ser
vice
The
ener
gy
of
the
IoT
de
vice
is
considered
a
vital
and
crucial
resource
in
the
netw
ork.
It
should
not
be
compromised,
as
preserving
it
without
ne
glecting
service
quality
is
essential
to
meet
the
users’
e
xpectations.
Therefore,
we
proposed
the
IoT
-QoS
algorithm,
which
operates
as
follo
ws:
-
The
algorithm
is
in
v
ok
ed
at
the
be
ginning
of
each
Beacon
Frame
guidance
period
when
the
wireless
card
of
the
IoT
de
vice
w
ak
es
up
and
recei
v
es
the
guidance
frame.
The
de
vice
utilizes
a
po
wer
-sa
ving
mode
supported
by
W
i-Fi
technology
.
-
The
algorithm
rst
looks
at
the
de
vice’
s
battery
po
wer
as
the
thres
hold
for
decision-making
in
maintaining
service
quality
.
Predicting
poor
connection
quality
is
done
using
the
random
for
est
classier
,
relying
on
the
RSSI
and
PDR
metrics
or
a
drop
in
the
IoT
de
vice’
s
ener
gy
le
v
el
(
P
I
oT
)
belo
w
the
threshold
(
P
T
hr
e
).
In
such
cases,
the
W
i-Fi
radio
is
turned
of
f
(transitioning
the
wireless
card
to
sleep
mode),
and
entering
sleep
mode
for
the
longest
possible
period
(
S
l
eep
max
)
helps
e
xtend
the
de
vice’
s
operational
lifespan
due
to
lo
w
po
wer
consumption
in
this
state.
Furthermore,
there
is
a
need
for
pack
et
aggre
g
ation
for
uplink
data.
The
wireless
card
transitions
from
sleep
mode
to
an
acti
v
e
state
when
data
pack
ets
are
a
v
ailable
in
the
transmission
queue.
F
ailure
to
aggre
g
ate
data
w
ould
result
in
transmission
operations
at
a
lo
w
transfer
rate
(due
to
poor
channel
conditions),
increasing
po
wer
consumption.
-
If
the
prediction
indicates
good
connection
quality
and
the
de
vice’
s
ener
gy
le
v
el
is
higher
or
equal
to
the
threshold
le
v
el
(
P
T
hr
e
),
and
there
are
no
stored
pack
ets
a
w
aiting
transmission,
the
card
is
put
into
sleep
mode
for
a
duration
equal
to
twice
the
pre
vious
sleep
peri
od.
This
is
because
data
mo
v
ement
on
the
internet
occurs
in
the
form
of
b
ursts
(see
[27]).
-
If
the
IoT
de
vice
possesses
stored
data
pack
ets
at
the
access
point,
it
wil
l
perform
the
required
netw
ork
transmission
and
reception
operations.
Here,
a
distinction
is
made
based
on
whether
the
stored
data
is
a
data
request
(e
xternal
query)
accepted
in
terms
of
authentication
and
authorization
(subject
to
the
modied
IdM
system).
In
this
case,
an
algorithm
is
in
v
ok
ed
to
pro
vide
pack
et
fragmentation
to
reduce
the
amount
of
data
to
be
sent
according
to
the
request
conte
xt,
relying
on
conte
xtual
information
used
to
isolate
users.
Y
et,
if
the
transferred
data
is
time-sensiti
v
e,
the
a
v
erage
delay
of
the
retrie
v
ed
pack
ets
m
ust
be
calculated,
and
the
duration
of
the
wireless
card’
s
stay
in
sleep
mode
is
reduced
to
the
minimum
v
alue
assigned
if
the
imposed
time
constraint
is
not
met.
This
condition
serv
es
as
a
real
constraint
for
an
y
delay
caused
by
both
data
aggre
g
ation
and
entering
sleep
mode
operation.
It
is
w
orth
noting
that
the
netw
ork
transmission
and
reception
operations
for
the
mentioned
data
types
are
conducted
according
to
a
higher
access
priority
to
the
wireless
medium,
compared
to
traditional
IoT
user
data.
This
is
achie
v
ed
by
granting
IoT
de
vices
applying
the
proposed
algorithm
a
shorter
back-of
f
time
than
those
not
using
it
(traditional
de
vices).
The
latter
applies
the
binary
e
xponential
increase
for
back-of
f
time
(distrib
uted
coordination
function
(DCF)
access
pattern),
while
our
algorithm
relies
on
linear
increase.
The
proposed
algorithm
IoT
-QoS
diagram
sho
wn
in
Figure
2.
3.
DISCUSSION
AND
RESUL
TS
In
this
section,
we
e
xplained
the
results
by
using
the
NS3
simulator
[28]
to
complete
the
e
v
aluation
process
according
to
the
parameters
sho
wn
in
T
able
1,
noting
that
the
training
data
set
is
collected
from
the
trace
le,
which
is
used
to
train
the
random
forest
classier
,
which
in
turn
generates
a
training
model
that
is
used
to
complete
the
classication
process.
According
to
Figure
3,
de
vices
using
the
proposed
IoT
-QoS
algorithm
were
able
to
achie
v
e
lo
wer
ener
gy
consumption
than
their
counterparts
relying
solely
on
the
s
tandard
po
wer
-
sa
ving
mode
(IoT),
with
an
inc
rease
in
both
the
primary
sleep
interv
al
and
the
number
of
netw
ork
users.
This
is
a
direct
result
of
the
algorithm
monitoring
three
parameters:
ener
gy
le
v
els,
the
presence
of
stored
pack
ets
at
the
access
point,
and
connection
quality
.
The
algorithm
utilizes
these
parameters
to
achie
v
e
ener
gy
sa
vings
in
de
vices.
Additionally
,
it
pri
oritizes
access
to
the
medium,
reducing
the
wireless
card’
s
idle
time
during
netw
ork
contention.
This
positi
v
ely
af
fects
the
de
vice’
s
battery
ener
gy
.
Int
J
Recongurable
&
Embedded
Syst,
V
ol.
13,
No.
3,
No
v
ember
2024:
748–757
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Recongurable
&
Embedded
Syst
ISSN:
2089-4864
❒
753
Figure
2.
The
proposed
algorithm
IoT
-QoS
diagram
T
able
1.
The
parameters
used
in
the
simulator
(NS3)
P
arameter
V
alue
P
T
hr
e
0.5
mJ
T
ransmitting
po
wer
1400
mW
Recei
ving
Po
wer
900
mW
Idle
po
wer
700
mW
Sleeping
po
wer
60
mW
C
W
min
32
C
W
max
1024
PSM
timeout
25
msec
Max
sleep
interv
al
1,000
msec
Simulation
time
200
Sec
Impr
o
ving
the
performance
of
IoT
de
vices
that
use
W
i-F
i
(Ali
Ahmed
Razzaq)
Evaluation Warning : The document was created with Spire.PDF for Python.
754
❒
ISSN:
2089-4864
Figure
3.
Ener
gy
consumption
as
a
function
of
increasing
IoT
de
vices
and
primary
sleep
duration
(insert
link
image)
Also,
the
p
r
op
os
ed
algorithm
distinguishes
between
tw
o
types
of
data,
one
of
these
w
as
subjected
to
modied
authentication
processes.
The
critical
netw
ork
standard
for
this
data
type
is
response
time,
considering
that
the
additional
security
system
load
will
af
fect
data
transfer
time.
According
to
our
study
,
de
vices
imple-
menting
the
proposed
algorithm
achie
v
ed
lo
wer
response
times
than
those
not
using
it,
e
v
en
with
increased
netw
ork
data
traf
c,
as
depicted
in
Figure
4.
This
can
be
justied
as
the
priority
access
scheme,
which
plays
a
crucial
role
in
f
aster
access
to
the
wirel
ess
medium.
Moreo
v
er
,
the
algorithm
for
pro
viding
pack
et
width
had
a
signicant
impact
on
adjusting
the
transmitted
information
in
response
to
the
request
conte
xt.
Figure
4.
Response
time
as
a
function
of
increasing
netw
ork
load
Hence,
the
proposed
algorithm
managed
to
mitig
ate
the
impact
of
increasing
the
number
of
de
vices
in
the
wireless
netw
ork
on
latenc
y
,
a
crucial
performance
metric
for
time-sensiti
v
e
applications.
This
data
type,
distinguished
as
the
second
type
by
the
IoT
-QoS
algorithm,
is
illustrated
in
Figure
5,
where
t
he
superiority
of
the
proposal
becomes
e
vident,
especially
under
netw
ork
congestion.
This
action
can
be
considered
a
natural
outcome
of
the
IoT
-QoS
algorithm’
s
ability
to
indirectly
reduce
congestion
le
v
els
compared
to
IoT
.
This
reduc-
tion
is
achie
v
ed
by
predicting
the
wireless
link
state
through
random
forest,
which
forces
the
wireless
card
into
a
sleep
state
when
de
vices
are
not
in
proximity
to
the
access
point.
In
such
cases,
the
card
can
enter
transmis-
sion
and
reception
operations
at
a
higher
rate
with
the
be
ginning
of
each
beacon
frame
period.
This
ef
fecti
v
ely
reduces
both
contention
time
for
the
wireless
medium
and
the
time
required
for
reception
operations,
resulting
in
f
aster
data
retrie
v
al.
Int
J
Recongurable
&
Embedded
Syst,
V
ol.
13,
No.
3,
No
v
ember
2024:
748–757
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Recongurable
&
Embedded
Syst
ISSN:
2089-4864
❒
755
Figure
5.
Delay
as
a
function
of
increasing
the
number
of
IoT
de
vices
The
results
depicted
in
Figure
6
indicate
that
the
beha
vior
of
the
proposed
IoT
-QoS
algorithm
has
an
impact
on
the
round
trip
time
(R
TT)
between
an
IoT
de
vice
and
the
serv
er
,
especially
in
the
presence
of
a
small
number
of
de
vices
in
the
netw
ork.
This
led
to
a
decrease
in
producti
vity
compared
to
traditional
de
vices.
Ho
we
v
er
,
the
performance
of
the
IoT
-QoS
algorithm
surpasses
that
of
traditional
de
vices
under
netw
ork
con-
gestion.
This
impro
v
ement
can
be
attrib
uted
to
the
algorithm’
s
ability
t
o
reduce
contention
time
for
the
wireless
medium
through
the
linear
access
scheme.
It
allo
ws
for
transmission
operations
with
impro
v
ed
channel
con-
ditions,
a
v
oiding
the
need
for
retransmission
and
enabling
data
transfer
at
a
higher
rate.
Naturally
,
putting
the
wireless
card
to
sleep
under
poor
channel
conditions
reduces
congestion
le
v
els,
and
impacting
congestion
in
one
w
ay
or
another
.
Figure
6.
Throughput
as
a
function
of
increasing
the
number
of
IoT
de
vices
4.
CONCLUSION
In
light
of
the
ndings,
it
is
e
vident
that
the
IoT
-QoS
algorithm
has
demonstrated
a
capacity
to
enhance
the
performance
of
IoT
de
vices
utilizing
W
i-Fi
as
their
communication
medium
,
all
while
ensuring
the
pi
v
otal
aspect
of
security
is
not
compromised.
Consequently
,
a
strong
recommendation
emer
ges
for
the
adoption
of
the
proposed
IoT
-QoS
algorithm,
particularly
for
IoT
de
vices
characterized
by
constrained
resources,
notably
in
po
wer
.
The
algorithm
sho
wcases
its
ef
cac
y
by
successfully
upholding
QoS
standards
while
notably
diminish-
ing
po
wer
consumption
within
the
wireless
cards
of
IoT
de
vices
operating
in
congested
netw
ork
en
vironments,
thereby
outperforming
traditional
de
vices
in
similar
conditions.
In
the
future,
we
seek
to
enhance
the
security
Impr
o
ving
the
performance
of
IoT
de
vices
that
use
W
i-F
i
(Ali
Ahmed
Razzaq)
Evaluation Warning : The document was created with Spire.PDF for Python.
756
❒
ISSN:
2089-4864
aspect
of
the
IoT
netw
ork.
Se
v
eral
research
directions
may
be
e
xplored
in
this
re
g
ard,
including
the
adoption
of
blockchain
technology
for
authentication
processes.
Addit
ionally
,
there
is
a
gro
wing
interest
in
le
v
eraging
AI
and
softw
are
dened
netw
orks
to
detect
attacks
and
mak
e
informed
decisions
to
mitig
ate
their
impact.
REFERENCES
[1]
D.
Bandyopadh
yay
and
J.
Sen,
“Internet
of
things:
applications
and
challenges
in
technology
and
standardization,
”
W
ir
eless
P
er
sonal
Communications
,
v
ol.
58,
no.
1,
pp.
49–69,
May
2011,
doi:
10.1007/s11277-011-0288-5.
[2]
L.
Math
y
,
C.
Edw
ards,
and
D.
Hutchison,
“Princi
ples
of
QoS
in
group
communications,
”
T
elecommunication
Systems
,
v
ol.
11,
pp.
59–84,
1999,
doi:
10.1023/A:1019132914996.
[3]
M
.
Singh
and
G.
Baranw
al,
“Quality
of
service
(QoS)
in
internet
of
things,
”
in
2018
3r
d
International
Confer
ence
On
Internet
of
Things:
Smart
Inno
vation
and
Usa
g
es
(IoT
-SIU)
,
IEEE,
Feb
.
2018,
pp.
1–6.
doi:
10.1109/IoT
-SIU.2018.8519862.
[4]
R
.
Duan,
X.
Chen,
and
T
.
Xing,
“
A
QoS
archit
ecture
for
IO
T
,
”
in
2011
International
Confer
ence
on
Internet
of
Things
and
4th
International
Confer
ence
on
Cyber
,
Physical
and
Social
Computing
,
IEEE,
Oct.
2011,
pp.
717–720.
doi:
10.1109/iThings/CPSCom.2011.125.
[5]
M.
S.
Bernard,
T
.
Pei,
and
K.
Nasser
,
“QoS
strate
gies
for
wireless
multimedia
sensor
netw
orks
in
the
conte
xt
of
IoT
at
the
MA
C
layer
,
application
layer
,
and
cross-layer
algorithms,
”
J
ournal
of
Computer
Networks
and
Communications
,
v
ol.
2019,
pp.
1–33,
Dec.
2019,
doi:
10.1155/2019/9651915.
[6]
S.
P
a
w
ar
et
al
.,
“Security
and
QoS
(quality
of
service)
related
current
challenges
in
IoT
,
”
International
J
ournal
of
Electr
onics
and
Communication
Engineering
,
v
ol.
10,
no.
4,
pp.
9–20,
Apr
.
2023,
doi:
10.14445/23488549/IJECE-V10I4P102.
[7]
J.
Liu,
Y
.
Xi
ao,
and
C.
L.
P
.
Chen,
“
Authentication
and
access
control
in
the
internet
of
things,
”
in
2012
32nd
International
Confer
-
ence
on
Distrib
uted
Computing
Systems
W
orkshops
,
IEEE,
Jun.
2012,
pp.
588–592.
doi:
10.1109/ICDCSW
.2012.23.
[8]
C.
Chibelushi,
A.
Eardle
y
,
and
A.
Arabo,
“Identity
management
in
the
internet
of
things:
the
role
of
MANETs
for
healthcare
applications,
”
Computer
Science
and
Information
T
ec
hnolo
gy
,
v
ol.
1,
no.
2,
pp.
73–81,
Sep.
2013,
doi:
10.13189/csit.2013.010201.
[9]
M.
Leo,
F
.
Battisti,
M.
Carli,
and
A.
Neri,
“
A
federated
architecture
approach
for
internet
of
things
security
,
”
in
2014
Eur
o
Med
T
elco
Confer
ence
(EMTC)
,
IEEE,
No
v
.
2014,
pp.
1–5.
doi:
10.1109/EMTC.2014.6996632.
[10]
A.
W
itk
o
vski,
A.
Santin,
V
.
Abreu,
and
J.
Maryno
wski,
“
An
IdM
and
k
e
y-based
authentication
method
for
pro
viding
single
sign-
on
in
IoT
,
”
in
2015
IEEE
Global
Communications
Confer
ence
(GLOBECOM)
,
IEEE,
Dec.
2015,
pp.
1–6.
doi:
10.1109/GLO-
COM.2015.7417597.
[11]
M.
L.
B.
A.
Santos,
J.
C.
Carneiro,
A.
M.
R.
Franco,
F
.
A.
T
eix
eira,
M.
A.
A.
Henriques,
and
L.
B.
Oli
v
eira,
“FLA
T
:
federated
lightwei
ght
authentication
for
the
internet
of
things,
”
Ad
Hoc
Networks
,
v
ol.
107,
p.
102253,
Oc
t.
2020,
doi:
10.1016/j.adhoc.2020.102253.
[12]
M.
J.
M.
Al-Saadi
and
M.
Ilyas,
“Identity
management
approach
in
internet
of
things
(IoT),
”
in
2020
4th
International
Sym-
posium
on
Multidisciplinary
Studies
and
Inno
vative
T
ec
hnolo
gies
(ISMSIT)
,
IEEE,
Oct.
2020,
pp.
1–6.
doi:
10.1109/ISM-
SIT50672.2020.9254470.
[13]
M.
A.
Bouras,
B.
Xia,
A.
O.
Ab
uassba,
H.
Ning,
and
Q.
Lu,
“IoT
-CCA
C:
a
blockchain-based
consortium
capability
access
control
approach
for
IoT
,
”
P
eerJ
Computer
Science
,
v
ol.
7,
p.
e455,
Apr
.
2021,
doi:
10.7717/peerj-cs.455.
[14]
V
.
Dehal
w
ar
,
M.
L.
K
olhe,
S.
Deoli,
and
M.
K.
Jhariya,
“Blockchain-based
t
rust
management
and
authentication
of
de
vices
in
smart
grid,
”
Cleaner
Engineering
and
T
ec
hnolo
gy
,
v
ol.
8,
p.
100481,
Jun.
2022,
doi:
10.1016/j.clet.2022.100481.
[15]
X.
Zhu
and
Y
.
Badr
,
“Identity
management
systems
for
the
internet
of
things:
a
surv
e
y
to
w
ards
blockchain
solutions,
”
Sensor
s
,
v
ol.
18,
no.
12,
p.
4215,
Dec.
2018,
doi:
10.3390/s18124215.
[16]
K.
M.
Sadique,
R.
Rahmani,
and
P
.
Johannesson,
“DIdM-EIoTD:
distrib
uted
identity
management
for
edge
internet
of
things
(IoT)
de
vices,
”
Sensor
s
,
v
ol.
23,
no.
8,
p.
4046,
Apr
.
2023,
doi:
10.3390/s23084046.
[17]
J.
Chen,
Y
.
Liu,
and
Y
.
Chai,
“
An
identity
management
frame
w
ork
for
internet
of
things,
”
in
2015
IEEE
12th
International
Confer
-
ence
on
e-Business
Engineering
,
IEEE,
Oct.
2015,
pp.
360–364.
doi:
10.1109/ICEBE.2015.67.
[18]
M.
L.
F
.
Sindjoung
and
P
.
Minet,
“W
ireless
link
quality
prediction
in
IoT
netw
orks,
”
in
2019
8th
International
Confer
-
ence
on
P
erformance
Evaluation
and
Modeling
in
W
ir
ed
and
W
ir
eless
Networks
(PEMWN)
,
IEEE,
No
v
.
2019,
pp.
1–6.
doi:
10.23919/PEMWN47208.2019.8986920.
[19]
J.
Ma,
T
.
Nag
atsuma,
S.-J.
Kim,
and
M.
Hase
g
a
w
a,
“
A
machine-learning-based
channel
assignment
algori
thm
for
IoT
,
”
in
2019
International
Confer
ence
on
Articial
Intellig
ence
in
Information
and
Communication
(ICAIIC)
,
IEEE,
Feb
.
2019,
pp.
1–6.
doi:
10.1109/ICAIIC.2019.8669028.
[20]
P
.
L
yngg
aard,
“Using
machine
learning
for
adapti
v
e
interference
suppression
in
wireless
sensor
netw
orks,
”
IEEE
Sensor
s
J
ournal
,
v
ol.
18,
no.
21,
pp.
8820–8826,
No
v
.
2018,
doi:
10.1109/JSEN.2018.2867068.
[21]
M.
Zeki
´
c-Su
ˇ
sac,
S.
Mitro
vi
´
c,
and
A.
Has,
“Machine
learning
based
system
for
managing
ener
gy
ef
cienc
y
of
public
sector
as
an
approach
to
w
ards
smart
cities,
”
International
J
ournal
of
Information
Mana
g
ement
,
v
ol.
58,
p.
102074,
Jun.
2021,
doi:
10.1016/j.ijinfomgt.2020.102074.
[22]
Z.
Becv
ar
,
D.
Gesbert,
P
.
Mach,
and
M.
Najla,
“Machine
learning-based
channel
quality
prediction
in
6G
mobile
netw
orks,
”
IEEE
Communications
Ma
gazine
,
v
ol.
61,
no.
7,
pp.
106–112,
Jul.
2023,
doi:
10.1109/MCOM.001.2200305.
[23]
A.
T
orres-Alv
arado,
L.
A.
Morales-Rosales,
and
I.
Algredo-Badillo,
“
A
channel-quality
classication
analysis
for
IoT
communica-
tion
based
on
machine
learning,
”
in
2021
Me
xican
International
Confer
ence
on
Computer
Science
(ENC)
,
IEEE,
Aug.
2021,
pp.
1–4.
doi:
10.1109/ENC53357.2021.9534815.
[24]
J.
Li,
J.
Xiao,
J.
W
.-K.
Hong,
and
R.
Boutaba,
“FSM-based
W
i-Fi
po
wer
estimation
method
for
smart
de
vices,
”
in
2015
IFIP/IEEE
In-
ternational
Symposium
on
Inte
gr
ated
Network
Mana
g
ement
(IM)
,
IEEE,
May
2015,
pp.
147–155.
doi:
10.1109/INM.2015.7140287.
[25]
D.
Jung,
R.
Kim,
and
H.
Lim,
“Po
wer
-sa
ving
strate
gy
for
balancing
ener
gy
and
delay
performance
in
WLANs,
”
Computer
Commu-
nications
,
v
ol.
50,
pp.
3–9,
Sep.
2014,
doi:
10.1016/j.comcom.2014.02.005.
[26]
R.
Zheng,
J.
Hou,
and
L.
Sha,
“Performance
analysis
of
the
IEEE
802.11
po
wer
sa
ving
mode,
”
in
Pr
oc.
CNDS
,
2004,
pp.
1–9.
Int
J
Recongurable
&
Embedded
Syst,
V
ol.
13,
No.
3,
No
v
ember
2024:
748–757
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Recongurable
&
Embedded
Syst
ISSN:
2089-4864
❒
757
[27]
Y
.
Xiao,
P
.
Sa
v
olainen,
A.
Karppanen,
M.
Siekkinen,
and
A.
Yl
¨
a-J
¨
a
¨
aski,
“Practica
l
po
wer
modeling
of
data
transmi
ssion
o
v
er
802.11g
for
wireless
applications,
”
in
Pr
oceedings
of
the
1st
International
Confer
ence
on
Ener
gy-Ef
cient
Computing
and
Network-
ing
,
Ne
w
Y
ork,
NY
,
USA:
A
CM,
Apr
.
2010,
pp.
75–84.
doi:
10.1145/1791314.1791326.
[28]
K.
W
ehrle,
M.
G
¨
unes
¸
,
and
J.
Gross,
Eds.,
Modeling
and
tools
for
network
simulation
.
Berlin,
Heidelber
g:
Springer
Berlin
Heidel-
ber
g,
2010.
doi:
10.1007/978-3-642-12331-3.
BIOGRAPHIES
OF
A
UTHORS
Ali
Ahmed
Razzaq
had
a
master’
s
de
gree
in
computer
netw
ork
engineering
from
Andhra
Uni
v
ersity
.
No
w
I
am
a
research
scholar
at
Andhra
Uni
v
ersity
in
the
IoT
specialty
for
the
purpose
of
obtaining
a
Ph.D.
He
ha
v
e
se
v
eral
skills
in
the
eld
of
articial
i
ntelligence
and
its
programming
in
the
eld
of
netw
orks
and
design
websites
by
frame
w
ork
django
in
p
ython.
He
currently
w
ork
in
the
eld
of
air
na
vig
ation
and
its
systems
as
a
data
entry
for
a
viat
ion
transit
(FDO)
in
the
Iraqi
Air
T
raf
c
Management
Center
in
the
Area
Control
Center
(A
CC)
a
nd
no
w
an
air
traf
c
controller
at
Baghdad
International
Airport
(Iraq).
He
can
be
contacted
at
email:taif
ali607@gmail.com.
Pr
of
.
K
unjam
Nageswara
Rao
is
a
Professor
in
Department
of
Computer
Science
and
Systems
Engineering
at
Andhra
Uni
v
ersity
Colle
ge
of
Engineering.
He
has
more
than
24
years
of
teaching
e
xperience.
He
has
published
3
patents
and
more
than
50
research
papers
so
f
ar
in
v
ari-
ous
highly
reputed
international
journals.
His
research
interest
includes
cloud
computing,
wireless
netw
orks,
sensor
netw
orks,
IoT
,
bioinformatics,
medical
image
processing,
netw
ork
security
,
data
mining
and
data
analyticss.
He
can
be
contacted
at
email:
kunjamnag@gmail.com.
Impr
o
ving
the
performance
of
IoT
de
vices
that
use
W
i-F
i
(Ali
Ahmed
Razzaq)
Evaluation Warning : The document was created with Spire.PDF for Python.