IEEE Consumer Electronics Magazine - January/February 2024 - 15

TABLE 1. Comparison among related trust proposals.
Work
Technique Dataset
15
16
17
Bayesian
networks
Agentbased
+ BC
BC
18
Weighted
averages
+data
fusion
19
20
None
None
None
Network
level
No
No
No
Synthetic No
Entropy Synthetic No
RPL
+subjective
logic
21
22
23
24
This
proposal
Machine
learning
Fuzzy logic
IBE +BC
Agentbased
+ BC
Relative
entropy +
BC
None
Real +
synthetic
Real
None
None
Real +
synthetic
Yes
No
No
No
No
Yes
Application
level
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Architecture
Distributed
Distributed
Distributed
Decentralized
Distributed
Distributed
Centralized
Distributed
Distributed
Distributed
Distributed
Resourceconstraint
awareness
No
Yes
No
Yes
No
Yes
No
Yes
No
No
Yes
Dynamism
awareness
Yes
Yes
No
Yes
No
No
No
No
No
No
Yes
Offers
trust
metric
No
No
No
Yes
Yes
No
No
Yes
No
Yes
Yes
3)
whether the approach considers networklevel
and/or application-level characteristics;
4) which type of architecture is used;
5) whether the approach is resource-constraintaware
and/or dynamism-aware; and
6) if the approach offers trust values.
Most of the works consider applicationlevel
characteristics, and only a couple (including
our approach) take into consideration
the network level. Out of the 12 presented
works, ten consider a distributed architecture,
six are resource-constraint-aware, five are
dynamism-aware, and five actually present a
trust metric.
The foregoing works emphasize the importance
and relevance of building trust-based
approaches to provide security in IoT devices
communication. In this article, besides presenting
a trust model that combines BC and
information theory techniques, the key contribution
of our work is the double perspective
of both application level and network level.
Hence, our approach provides a more comprehensive
trust metric that can deal with the
particularities of consumer IoT devices.
January/February 2024
TRUST MODELING
To model trust for IoT, we need to know
which information is necessary to compose a
trust metric. First, a receiver IoT device needs to
build an initial trust to enable communication,
since it does not know the sender IoT device previously.
Given the initial trust, the receiver
should dynamically adjust its trust in the sender
accordingly to the network behavior of the
sender (more trust if the behavior is as estimated,
less trust otherwise). Then, the trust
value should not last endlessly, but instead, be
decreased over time, given that, after the last
contact, the receiver does not know if the sender
was compromised or if it presents any unexpected
(potentially malicious) behavior.
The essential features for trust are the guarantee
of the uniqueness, veracity, and authenticity
of the device's identity. Therefore, a device must
know if a communicating device is indeed the
device it supposes it is. This issue raises the need
for an IoT device identification mechanism that
provides assurance of unique identities as well as
protection against tampering, duplication, forgery,
and impersonation. A possible strategy to do
so is to use the encryption scheme IBE. IBE, as its
15
IEEE Consumer Electronics Magazine - January/February 2024

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