Scientific Reports,
Год журнала:
2024,
Номер
14(1)
Опубликована: Авг. 17, 2024
Federated
Learning
is
an
effective
solution
to
address
the
issues
of
data
isolation
and
privacy
leakage
in
machine
learning.
However,
ensuring
security
network
devices
architectures
deploying
federated
learning
remains
a
challenge
due
attacks.
This
paper
proposes
attention-based
Graph
Neural
Network
for
detecting
cross-level
cross-department
method
enables
collaborative
model
training
while
protecting
on
distributed
devices.
By
organizing
traffic
information
chronological
order
constructing
graph
structure
based
log
density,
enhances
accuracy
attack
detection.
The
introduction
attention
mechanism
construction
Attention
(FedGAT)
are
used
evaluate
interactivity
between
nodes
graph,
thereby
improving
precision
internal
interactions.
Experimental
results
demonstrate
that
our
achieves
comparable
robustness
traditional
detection
methods
prioritizing
protection
security.
Computer Networks,
Год журнала:
2021,
Номер
203, С. 108661 - 108661
Опубликована: Дек. 14, 2021
The
application
of
Machine
Learning
(ML)
techniques
to
the
well-known
intrusion
detection
systems
(IDS)
is
key
cope
with
increasingly
sophisticated
cybersecurity
attacks
through
an
effective
and
efficient
process.
In
context
Internet
Things
(IoT),
most
ML-enabled
IDS
approaches
use
centralized
where
IoT
devices
share
their
data
centers
for
further
analysis.
To
mitigate
privacy
concerns
associated
approaches,
in
recent
years
Federated
(FL)
has
attracted
a
significant
interest
different
sectors,
including
healthcare
transport
systems.
However,
development
FL-enabled
its
infancy,
still
requires
research
efforts
from
various
areas,
order
identify
main
challenges
deployment
real-world
scenarios.
this
direction,
our
work
evaluates
approach
based
on
multiclass
classifier
considering
distributions
scenario.
particular,
we
three
settings
that
are
obtained
by
partitioning
ToN_IoT
dataset
according
devices’
IP
address
types
attack.
Furthermore,
evaluate
impact
aggregation
functions
such
setting
using
IBMFL
framework
as
FL
implementation.
Additionally,
set
future
directions
existing
literature
analysis
evaluation
results.
IEEE Journal of Biomedical and Health Informatics,
Год журнала:
2022,
Номер
27(2), С. 778 - 789
Опубликована: Июнь 13, 2022
Recent
advances
in
electronic
devices
and
communication
infrastructure
have
revolutionized
the
traditional
healthcare
system
into
a
smart
by
using
internet
of
medical
things
(IoMT)
devices.
However,
due
to
centralized
training
approach
artificial
intelligence
(AI),
mobile
wearable
IoMT
raise
privacy
issues
concerning
information
communicated
between
hospitals
end-users.
The
conveyed
is
highly
confidential
can
be
exposed
adversaries.
In
this
regard,
federated
learning
(FL),
distributive
AI
paradigm,
has
opened
up
new
opportunities
for
preservation
without
accessing
data
participants.
Further,
FL
provides
end-users
as
only
gradients
are
shared
during
training.
For
these
specific
properties
FL,
paper,
we
present
privacy-related
IoMT.
Afterwards,
role
networks
introduce
some
advanced
architectures
incorporating
deep
reinforcement
(DRL),
digital
twin,
generative
adversarial
(GANs)
detecting
threats.
Moreover,
practical
end,
conclude
survey
discussing
open
research
challenges
while
future
systems.
Sensors,
Год журнала:
2022,
Номер
22(12), С. 4394 - 4394
Опубликована: Июнь 10, 2022
The
Internet
of
Things
(IoT)
revitalizes
the
world
with
tremendous
capabilities
and
potential
to
be
utilized
in
vehicular
networks.
Smart
Transport
Infrastructure
(STI)
era
depends
mainly
on
IoT.
Advanced
machine
learning
(ML)
techniques
are
being
used
strengthen
STI
smartness
further.
However,
some
decisions
very
challenging
due
vast
number
components
big
data
generated
from
STIs.
Computation
cost,
communication
overheads,
privacy
issues
significant
concerns
for
wide-scale
ML
adoption
within
STI.
These
can
addressed
using
Federated
Learning
(FL)
blockchain.
FL
address
preservation
handling
management
control.
Blockchain
is
a
distributed
ledger
that
store
while
providing
trust
integrity
assurance.
solution
add
more
security
This
survey
initially
explores
network
detail
sheds
light
blockchain
real-world
implementations.
Then,
applications
Vehicular
Ad
Hoc
Network
(VANET)
environment
perspectives
discussed
detail.
In
end,
paper
focuses
current
research
challenges
future
directions
related
integrating
Knowledge-Based Systems,
Год журнала:
2023,
Номер
274, С. 110658 - 110658
Опубликована: Май 22, 2023
Recent
developments
in
the
Internet
of
Things
(IoT)
and
various
communication
technologies
have
reshaped
numerous
application
areas.
Nowadays,
IoT
is
assimilated
into
medical
devices
equipment,
leading
to
progression
Medical
(IoMT).
Therefore,
IoMT-based
healthcare
applications
are
deployed
used
day-to-day
scenario.
Traditionally,
machine
learning
(ML)
models
use
centralized
data
compilation
that
impractical
pragmatic
frameworks
due
rising
privacy
security
issues.
Federated
Learning
(FL)
has
been
observed
as
a
developing
distributed
collective
paradigm,
most
appropriate
for
modern
framework,
manages
stakeholders
(e.g.,
patients,
hospitals,
laboratories,
etc.)
carry
out
training
without
actual
exchange
sensitive
data.
Consequently,
this
work,
authors
present
an
exhaustive
survey
on
FL-based
IoMT
smart
frameworks.
First,
introduced
devices,
their
types,
applications,
datasets,
framework
detail.
Subsequently,
concept
FL,
its
domains,
tools
develop
FL
discussed.
The
significant
contribution
deploying
secure
systems
presented
by
focusing
patents,
real-world
projects,
datasets.
A
comparison
techniques
with
other
schemes
ecosystem
also
presented.
Finally,
discussed
challenges
faced
potential
future
research
recommendations
deploy
IEEE Transactions on Industrial Informatics,
Год журнала:
2022,
Номер
19(1), С. 286 - 295
Опубликована: Март 7, 2022
Security
has
become
a
critical
issue
for
Industry
4.0
due
to
different
emerging
cyber-security
threats.Recently,
many
Deep
Learning
(DL)
approaches
have
focused
on
intrusion
detection.However,
such
often
require
sending
data
central
entity.This
in
turn
raises
concerns
related
privacy,
efficiency,
and
latency.Despite
the
huge
amount
of
generated
by
Internet
Things
(IoT)
devices
4.0,
it
is
difficult
get
labeled
data,
because
labeling
costly
time-consuming.This
poses
challenges
several
DL
approaches,
which
data.In
order
deal
with
these
issues,
new
should
be
adopted.This
paper
proposes
novel
federated
semi-supervised
learning
scheme,
that
takes
advantage
both
unlabeled
way.First,
an
AutoEncoder
(AE)
trained
each
device
(using
local/private
data)
learn
representative
low-dimensional
features.Then,
cloud
server
aggregates
models
into
global
AE
using
Federated
(FL).Finally,
composes
supervised
neural
network,
adding
fully
connected
layers
(FCN)
encoder
(the
first
part
AE)
trains
resulting
model
publicly
available
data.Extensive
case
studies
two
real-world
industrial
datasets
demonstrate
our
model:
(a)
ensures
no
local
private
exchanged;
(b)
detects
attacks
high
classification
performance,
(c)
works
even
when
only
few
amounts
are
available;
(d)
low
communication
overhead.
Information,
Год журнала:
2023,
Номер
14(1), С. 41 - 41
Опубликована: Янв. 9, 2023
Owing
to
the
prevalence
of
Internet
things
(IoT)
devices
connected
Internet,
number
IoT-based
attacks
has
been
growing
yearly.
The
existing
solutions
may
not
effectively
mitigate
IoT
attacks.
In
particular,
advanced
network-based
attack
detection
using
traditional
Intrusion
systems
are
challenging
when
network
environment
supports
as
well
protocols
and
uses
a
centralized
architecture
such
software
defined
(SDN).
this
paper,
we
propose
long
short-term
memory
(LSTM)
based
approach
detect
SDN
supported
intrusion
system
in
networks.
We
present
an
extensive
performance
evaluation
machine
learning
(ML)
deep
(DL)
model
two
SDNIoT-focused
datasets.
also
LSTM-based
for
effective
multiclass
classification
Our
proposed
shows
that
our
identifies
classifies
types
with
accuracy
0.971.
addition,
various
visualization
methods
shown
understand
dataset’s
characteristics
visualize
embedding
features.
