In the challenging and dynamic context of Edge Computing, we propose a distributed trust-based approach, which includes a trust model and a nodes pooling algorithm, both validated through simulation, where edge servers collaborate to manage task offloading requests. Our model analyzes the past behavior of edge domains by considering both the resources they have shared and requested from other domains, as well as the characteristics of the tasks involved. In addition, we characterize the trust relationships established between edge servers. As we explain in our work, such trust relationship is directly connected to the measured QoS of task offloading. These two aspects (i.e., historical behavior and trust) are then combined into a unified metric, termed Collaborative Edge Computing Trust (CEC-Trust), which serves to assess the benefits (in terms of QoS) of joining a pool of edge domains for collaborative task offloading. To this end, we designed a fully decentralized procedure, guided by the CEC-Trust measure to form pools of edge domains. A set of simulation results are reported and prove that, by the designed procedure and along with the exploitation of the CEC-Trust measure, the overall performance of the edge servers in the collaborative scenario are notably improved.
Messina et al. (Tue,) studied this question.