A probabilistic trust model for cloud services using Bayesian networks

Abstract

With the industry’s fast adoption of cloud computing, it’s been hard to keep trust between cloud service providers and people who use their services. In fact, trust management in cloud computing has become very challenging given the urgent need for cloud service requesters to choose efficient, trustworthy, and non-risky services. One of the most important factors that can be considered in the applicant’s trust or distrust of a service is the different quality of services related to the service. Therefore, approaches are needed to assess the trustworthiness of cloud services with respect to their quality of service (QoS). Given the uncertainty that exists for cloud services, it is more realistic to model their QoS parameters as random variables and also consider different dependencies between them. In this paper, a new trust model for cloud services is proposed using a Bayesian network, which is a probabilistic graphical model that can be used as one of the best methods to control uncertainty. Using Bayesian networks makes it possible to infer more accurate QoS values, which leads to the selection of highly trustworthy services by several cloud service requesters. To evaluate the proposed trust model, experimental results are performed using a real-world cloud service quality feedback, collected by the $$CloudArmor$$ project. The results of the experiments show that the proposed trust model is highly accurate and significantly reduces the estimation error.