IEEE Consumer Electronics Magazine - January/February 2024 - 18

Special Section on Security and Privacy-Aware Emerging Computing
threshold. When trust value falls below the
threshold, devices will need to return to the first
case of trust establishment, i.e., devices will
need to obtain a minimum trust from BC again.
In our model, we consider a proportional temporal
decay as described in the following equation:
C3 ¼ TRji d
(3)
where d is the decay factor.
The following equation presents the proposed
composition with all three components
together:
TRji ¼ C1 þC2C3:
(4)
Regarding the four aspects that we assume
for the IoT trust concept in the " Introduction "
section, in our proposal, we incorporate identity
assurance with the use of IBE for devices identification,
network behavior is addressed by using
the Low Level component, the data integrity is
incorporated by assuming the successful communications
reported by the confirmations of
the transactions in BC, and the respective protection
of devices through communications that are
only allowed for licit devices.
MODEL VERIFICATION
An experimental evaluation of our trust analytic
model was developed to analyze the behavior
of the model over time. As the model
verification evolves, each component of (4) is
computed. For instance, a device i generates
traffic to a device j according to some distribution.
This means, in each time interval t (e.g.,
one second), a value of bytes per t is generated
according to such a distribution. In the same
way, device j also sends traffic to device i
according to another distribution, but changes
its behavior during a certain interval by sending
traffic according to a different distribution. What
we aim to demonstrate is the dynamism of the
trust metric as the traffic behavior changes and
verify our model with traffic patterns based on
known distributions (not yet realistic).
Figure 2 illustrates the traffic generation of
two devices according to a Poisson distribution
with parameter ¼ 10 Bps. During the interval
between the instants 400 and 600 of the experiment,
the traffic of device j changes its pattern
18
FIGURE 2. Generated traffic over time.
to another distribution, namely, a Pareto with
parameter a ¼ 10Bps. Therefore, the trust metric
is computed throughout this behavior, taking
into consideration (4).
The trust metric over time is shown in
Figure 3. The trust values vary over time, following
the variation in traffic behavior (observed in
Figure 2). More specifically, it reduces when the
traffic pattern changes (between instants 400
and 600), and recovers when the pattern goes
back to the expected behavior. The green and
red lines represent the thresholds set by each
device to decide on the minimum trust value
accepted to establish a communication. Threshold
values are key to our approach and must be
properly set according to the application the
devices are associated with, and possibly tuned
overtime to cover application changes. Hence,
the high level of our approach makes available
application-level information that includes the
selected threshold set by the application in
question, which can accept lower trust values
(less restrict) or only higher trust values (more
restrict). In this example, the device i stops
trusting in device j as soon as it starts perceiving
the anomalous behavior of j. Conversely, device
FIGURE 3. Calculated trust over time considering
synthetic traffic.
IEEE Consumer Electronics Magazine
IEEE Consumer Electronics Magazine - January/February 2024

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - January/February 2024
