Human
faces
have
been
widely
adopted
in
many
applications
and
systems
requiring
a
high-security
standard.
Although
face
authentication
is
deemed
to
be
mature
nowadays,
existing
works
demonstrated
not
only
the
privacy
leakage
of
facial
information
but
also
success
spoofing
attacks
on
biometrics.
The
critical
reason
behind
this
failure
liveness
detection
This
work
advances
most
biometric-based
user
schemes
by
exploring
dynamic
biometrics
(human
activities)
rather
than
traditional
static
faces).
Inspired
observations
from
psychology,
we
propose
mmFaceID
leverage
humans'
activities
when
performing
word
reading
for
achieving
robust,
highly
accurate,
effective
via
mmWave
sensing.
By
addressing
series
technical
challenges
capturing
micro-level
muscle
movements
using
sensor,
build
neural
network
reconstruct
estimated
expression
parameters.
Then,
unique
features
can
extracted
enable
robust
regardless
relative
distances
orientations.
We
conduct
comprehensive
experiments
23
participants
evaluate
terms
distances/orientations,
length
lists,
occlusion,
language
backgrounds,
demonstrating
an
accuracy
94.7%.
extend
our
evaluation
real
IoT
scenario.
speaking
commends,
average
reach
up
92.28%.
IEEE Internet of Things Journal,
Год журнала:
2024,
Номер
11(12), С. 22668 - 22683
Опубликована: Март 28, 2024
The
popularity
of
smart
home
devices
has
led
to
an
increase
in
security
incidents
happening
homes.
A
key
measure
avoid
such
is
authenticate
users
before
they
can
interact
with
devices.
However,
current
methods
often
require
additional
hardware.
This
paper
proposes,
a
gesture-based
authentication
system,
effective
method
built
on
top
the
voice
interfaces
already
available
these
devices,
without
adding
new
uses
gesture
processing
pipeline
that
identifies
Doppler-existing
frames
and
detects
Direction
Arrival
Reflection
low
SNR
environments
at
longer
distances.
Furthermore,
regarding
nature
authentication,
this
system
also
supports
detecting
user
liveness,
preventing
replay
synthesis
attacks
from
remote
attackers.
evaluation
shows
high
accuracy
False
Accept
Rate
(FAR)
0.08%
Reject
(FRR)
3.10%
for
within
1.5m
device.
2022 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW),
Год журнала:
2024,
Номер
unknown, С. 565 - 572
Опубликована: Март 16, 2024
Amidst
the
evolving
landscape
where
cyber
and
physical
realms
seamlessly
intertwine,
smishing
(SMS
phishing)
emerges
as
a
so-phisticated
security
threat.
This
paper
introduces
novel
cross-device
solution
for
detecting
attacks
in
augmented
reality
(AR)
environments.
Our
method
uses
AR
glasses
to
enhance
user
perception
interaction
by
analyzing
images
of
URLs
displayed
across
various
devices,
including
smartphones,
laptops,
digital
displays.
approach
aligns
with
"Seamless
Reality"
concept,
integrating
space
perception,
cognition,
interaction.
We
detail
development
prototype
system
present
findings
from
study
that
evaluates
its
effectiveness
diverse
range
settings.
results
underscore
potential
technologies
detection,
contributing
safer
between
virtual
real-world
elements.
The
also
delves
into
aspects
display
systems,
interface
design,
integration
haptic
feedback,
offering
insights
broader
implications
seamless
experiences.
Virtual
Reality
(VR)
technology,
extensively
utilized
in
gaming,
social
networking,
and
online
collaboration,
has
raised
significant
security
concerns
due
to
the
array
of
sensors
integrated
into
VR
headsets.
This
paper
discusses
several
our
ongoing
research
that
explore
sensor
vulnerabilities
within
headsets
proposes
appropriate
mitigation
strategies.
Specifically,
we
focus
on
three
types
embedded
headsets:
unrestricted
motion
sensors,
optical
eye-tracking
sensors.
Our
investigation
outlines
potential
attacks
exploiting
these
vulnerabilities,
which
could
result
privacy
leakage
malicious
signal
injections.
Furthermore,
detail
design
implementation
effective
countermeasures
defend
against
threats.
ACM Transactions on Multimedia Computing Communications and Applications,
Год журнала:
2024,
Номер
unknown
Опубликована: Сен. 9, 2024
With
metaverse
attracting
increasing
attention
from
both
academic
and
industry,
the
application
of
virtual
reality
(VR)
has
extended
beyond
3D
immersive
viewing/gaming
to
a
broader
range
areas,
such
as
banking,
shopping,
tourism,
education,
etc.,
which
involves
growing
amount
sensitive
private
user
data
into
VR
systems.
However,
with
current
password-based
authentication
schemes
in
mainstream
devices,
studies
demonstrate
that
side-channel
attacks
can
pose
severe
threat
privacy.
To
mitigate
threat,
we
propose
novel
Panoramic-image-based
system,
i.e.,
Pivot
,
defend
against
attacks,
yet
maintain
high
usability.
Specifically,
design
an
image-random-pivoting-based
interaction
mechanism
assist
users
quickly
securely
selecting
memorable
points
interest
panoramic
image.
Then
image
region
segmentation
algorithm
is
designed
automatically
scatter
regions
form
customized
graphic
password
for
user,
could
ensure
sufficiently
large
space
also
reduce
near-region
point
misclicks.
Afterward,
indexes
are
used
generate
hashed
authentication.
Both
theoretical
security
analysis
extensive
secure
user-friendly
practice.
Human
faces
have
been
widely
adopted
in
many
applications
and
systems
requiring
a
high-security
standard.
Although
face
authentication
is
deemed
to
be
mature
nowadays,
existing
works
demonstrated
not
only
the
privacy
leakage
of
facial
information
but
also
success
spoofing
attacks
on
biometrics.
The
critical
reason
behind
this
failure
liveness
detection
This
work
advances
most
biometric-based
user
schemes
by
exploring
dynamic
biometrics
(human
activities)
rather
than
traditional
static
faces).
Inspired
observations
from
psychology,
we
propose
mmFaceID
leverage
humans'
activities
when
performing
word
reading
for
achieving
robust,
highly
accurate,
effective
via
mmWave
sensing.
By
addressing
series
technical
challenges
capturing
micro-level
muscle
movements
using
sensor,
build
neural
network
reconstruct
estimated
expression
parameters.
Then,
unique
features
can
extracted
enable
robust
regardless
relative
distances
orientations.
We
conduct
comprehensive
experiments
23
participants
evaluate
terms
distances/orientations,
length
lists,
occlusion,
language
backgrounds,
demonstrating
an
accuracy
94.7%.
extend
our
evaluation
real
IoT
scenario.
speaking
commends,
average
reach
up
92.28%.
