CETEF : A Comprehensive and Efficient Trust Evaluation Framework for Distributed Networks

Abstract

In order to defend the distributed networks from malicious behaviors, many trust evaluation models have been proposed by the researchers . However, most of the existing trust models do not consider the uncertainty in distributed network environment adequately. In addition, current trust models do not carry out a fine-grained analysis on the recommendations and cannot filter the malicious recommendations effectively. Therefore, current trust evaluation model cannot resist the malicious behaviors effectively. To attack these challenges, a comprehensive and efficient distributed trust evaluation framework is proposed in this paper to improve the evaluation accuracy. There are three types of trust in our trust evaluation framework: direct trust, recommendation trust and indirect trust. Firstly, the direct trust is calculated based on an improved subjective logic model. Then, the recommendation trust is calculated based on the recommendations from nodes who had interactions with both the subject node and the object node. Thirdly, upon indirect recommendations clustering, a public opinion is generated to represent the indirect trust of the node. Lastly, the three types of trust are fussed to calculate the final trust of the given node. The experimental analysis shows the effectiveness of our trust evaluation framework.