Electronics,
Год журнала:
2024,
Номер
14(1), С. 24 - 24
Опубликована: Дек. 25, 2024
This
study
presents
a
predictive
maintenance
system
designed
for
industrial
Internet
of
Things
(IoT)
environments,
focusing
on
resource
efficiency
and
adaptability.
The
utilizes
Nicla
Sense
ME
sensors,
Raspberry
Pi-based
concentrator
real-time
monitoring,
Long
Short-Term
Memory
(LSTM)
machine-learning
model
analysis.
Notably,
the
LSTM
algorithm
is
an
example
how
system’s
sandbox
environment
can
be
used,
allowing
external
users
to
easily
integrate
custom
models
without
altering
core
platform.
In
laboratory,
achieved
Root
Mean
Squared
Error
(RMSE)
0.0156,
with
high
accuracy
across
all
detecting
intentional
anomalies
99.81%
rate.
real-world
phase,
maintained
robust
performance,
sensors
recording
maximum
Absolute
(MAE)
0.1821,
R-squared
value
0.8898,
Percentage
(MAPE)
0.72%,
demonstrating
precision
even
in
presence
environmental
interferences.
Additionally,
architecture
supports
scalability,
accommodating
up
64
sensor
nodes
compromising
performance.
enhances
platform’s
versatility,
enabling
customization
diverse
applications.
results
highlight
significant
benefits
contexts,
including
reduced
downtime,
optimized
use,
improved
operational
efficiency.
These
findings
underscore
potential
integrating
Artificial
Intelligence
(AI)
driven
into
constrained
offering
reliable
solution
dynamic,
operations.
Sensors,
Год журнала:
2024,
Номер
24(15), С. 5022 - 5022
Опубликована: Авг. 3, 2024
The
number
of
connected
devices
or
Internet
Things
(IoT)
has
rapidly
increased.
According
to
the
latest
available
statistics,
in
2023,
there
were
approximately
17.2
billion
IoT
devices;
this
is
expected
reach
25.4
by
2030
and
grow
year
over
for
foreseeable
future.
share,
collect,
exchange
data
via
internet,
wireless
networks,
other
networks
with
one
another.
interconnection
technology
improves
facilitates
people's
lives
but,
at
same
time,
poses
a
real
threat
their
security.
Denial-of-Service
(DoS)
Distributed
(DDoS)
attacks
are
considered
most
common
threatening
that
strike
devices'
These
be
an
increasing
trend,
it
will
major
challenge
reduce
risk,
especially
In
context,
paper
presents
improved
framework
(SDN-ML-IoT)
works
as
Intrusion
Prevention
Detection
System
(IDPS)
could
help
detect
DDoS
more
efficiency
mitigate
them
time.
This
SDN-ML-IoT
uses
Machine
Learning
(ML)
method
Software-Defined
Networking
(SDN)
environment
order
protect
smart
home
from
attacks.
We
employed
ML
based
on
Random
Forest
(RF),
Logistic
Regression
(LR),
k-Nearest
Neighbors
(kNN),
Naive
Bayes
(NB)
One-versus-Rest
(OvR)
strategy
then
compared
our
work
related
works.
Based
performance
metrics,
such
confusion
matrix,
training
prediction
accuracy,
Area
Under
Receiver
Operating
Characteristic
curve
(AUC-ROC),
was
established
SDN-ML-IoT,
when
applied
RF,
outperforms
algorithms,
well
similar
approaches
work.
It
had
impressive
accuracy
99.99%,
less
than
3
s.
conducted
comparative
analysis
various
models
algorithms
used
results
indicated
proposed
approach
others,
showcasing
its
effectiveness
both
detecting
mitigating
within
SDNs.
these
promising
results,
we
have
opted
deploy
SDN.
implementation
ensures
safeguarding
homes
against
network
traffic.
Scientific Reports,
Год журнала:
2025,
Номер
15(1)
Опубликована: Март 25, 2025
Cybersecurity
professionals
depend
on
multi-layered
techniques
to
find
even
the
most
minor
anomalies
that
can
point
possible
attacks
given
complexity
of
network
data.
Modern
threat
environment
concerns
include
feature
representation,
scalability,
and
flexibility
demand
for
improved
techniques.
This
work
presents
Multi-Layer
Deep
Autoencoder
(M-LDAE),
especially
tailored
cross-layer
IoT
detection,
solve
these
difficulties.
Specifically
designed
based
Internet
Things
(IoT)
attack
Multi-Layered
Auto
Encoder
(M-LDAE)
is
introduced
in
present
research
overcome
challenges.
With
use
deep
autoencoders
hierarchical
simplification
capabilities,
M-LDAE
able
extract
latent
representations
contain
both
global
local
attributes.
technology
effectively
safeguards
against
various
cyber
threats,
including
Man-in-the-Middle
at
layer
Distributed
Denial
Service
(DDoS)
transport
networks.
To
improve
detection
adapt
emerging
methods,
system
employs
learning
algorithms
such
as
RNNs,
GNNs,
TCNs.
proves
new
vectors,
enhance
accuracy,
reduce
false
positives
through
extensive
simulations,
using
benchmark
datasets
real-world
scenarios.
A
paradigm
presented
this
paper,
which
provides
a
flexible
robust
solution
complete
cybersecurity
across
different
domains
thereby
improves
field
identification.
Advances in computational intelligence and robotics book series,
Год журнала:
2025,
Номер
unknown, С. 125 - 154
Опубликована: Март 7, 2025
In
order
to
detect
denial-of-service
(DoS)
and
distributed
denial
of
service
(DDoS)
intrusions
on
the
organization's
e-healthcare
data
warehouse
infrastructure,
authors
this
study
proposed
a
computing
framework
that
combines
federated
learning
system
based
blockchain
technology.
A
Message
Queuing
Telemetry
Transport
(MQTT)
broker
gathers
from
an
IoT
node
sends
it
platform
for
analysis.
As
creation
several
new
technologies
applications,
has
created
opportunities
in
age
cloud
communication.
Due
increasing
use
technologies,
computer
networks
have
had
serious
security
concerns,
there
are
vulnerabilities
as
well.
DoS
DDoS
attacks
servers
may
compromise
general
stability,
efficacy
services,
real-time
information
federation.
This
provided
efficient
MQTT
approach
secure
cyberattacks
presented
state-of-the-art
defenses
against
DoS/DDoS
digital
healthcare
ecosystem.
Security and Privacy,
Год журнала:
2024,
Номер
7(6)
Опубликована: Май 21, 2024
Abstract
With
the
rapid
proliferation
of
insecure
Internet
Things
(IoT)
devices,
security
Internet‐based
applications
and
networks
has
become
a
prominent
concern.
One
most
significant
threats
encountered
in
IoT
environments
is
Distributed
Denial
Service
(DDoS)
attack.
This
attack
can
severely
disrupt
critical
services
prevent
smart
devices
from
functioning
normally,
leading
to
severe
consequences
for
businesses
individuals.
It
aims
overwhelm
victims'
resources,
websites,
other
by
flooding
them
with
massive
packets,
making
inaccessible
legitimate
users.
Researchers
have
developed
multiple
detection
schemes
detect
DDoS
attacks.
As
technology
advances
facilitating
factors
increased,
it
challenge
identify
such
powerful
attacks
real‐time.
In
this
paper,
we
propose
novel
distributed
scheme
network
traffic‐based
deploying
Kafka
Streams
processing
framework
named
Kafka‐Shield.
The
Kafka‐Shield
comprises
two
stages:
design
deployment.
Firstly,
designed
on
Hadoop
cluster
employing
highly
scalable
H2O.ai
machine
learning
platform.
Secondly,
portable,
scalable,
deployed
framework.
To
analyze
incoming
traffic
data
categorize
into
nine
target
classes
real
time.
Additionally,
stores
each
flow
input
features
predicted
outcome
File
System
(HDFS).
enables
development
new
models
or
updating
current
ones.
validate
effectiveness
Kafka‐Shield,
performed
analysis
using
various
configured
scenarios.
experimental
results
affirm
Kafka‐Shield's
remarkable
efficiency
detecting
rate
over
99%
process
0.928
million
traces
nearly
3.027
s.
