Résumé:
With the growth of the Internet of Things (IoT), the majority of apps run on mobile devices
are increasingly reliant on artificial intelligence (AI) algorithms, such as approaches for
automatic learning, image/video processing, augmented reality, online gaming, and the longawaited
arrival of metaverse applications. These applications need a large amount of resources
in order to perform extremely complicated calculations, mobile terminals face significant limits
in terms of computation and storage capacity, in addition to a limited battery life. As a
solution, computation / data offloading has been presented as a possible resource allocation
strategy that comprises transferring intensive jobs and massive volumes of data to be performed
by the mobile edge computing (MEC).
However, the lack of confidence between mobile terminals and MEC servers is seen as a
serious security and confidentiality problem to the computation/data offloading approach’s
deployment. Indeed, before engaging in any interaction, each entity of the system must verify
and confirm the reliability and the security of the other entity. Blockchain has recently been
identified promising solution for increasing the security of MEC systems. Furthermore, the
blockchain can provide high levels of security and trust for MEC by depending on a vast
community’s distributed verification of the authenticity of MEC system communicators via
highly trustworthy consensus protocols.
Motivated by the previous discussion, we investigate in this work a trust management system
to ensure more secure and reliable interactions in the MEC domain for task offloading.
Our system is built on blockchain technology and consists primarily of two parts: a bidirectional
trust management mechanism based on a reputation metric and an incentive approach
that uses game theory to ensure more efficient block mining. We used a Stackelberg-type game
to model the incentive problem in verification with asymmetric information in which the players
are: a leader who acts first, followed by several followers. Our Stackelberg game is being
played between edge devices.
The leader begins by announcing its strategy, after which the followers compete and respond
with their best strategy that maximizes the leader’s approach. After receiving all of the
followers’ replies, the leader optimizes its final strategy. In the last part we tried to show the
relationship between the utility of the players (leader/follower) and the other parameters, and
we obtained logical results.
