An autonomous dynamic trust management system with uncertainty analysis

In order to help users select a reliable service in the open and complicated network, an autonomous dynamic trust management system is put forward based on collaborative recommendation mechanism and uncertainty analysis of satisfaction data. Any user in the network can store and update autonomously his trust table which contains direct interaction experience, find easily the reliable referrers according to collaborative recommendation, and collect valuable trust data and calculate the satisfaction and its reliability based on uncertainty analysis. Then the direct, recommended and self-recommended trusts can be calculated and finally be used to compute the global trust which will help the user to make a trust decision. A complicated network environment which includes quite a part of malicious nodes is built for investigating the robustness and effectiveness of the trust system. The experimental results show that the quality of service selection can be improved with the increase of interaction; the system can keep stable even though there are 50% malicious nodes initially in the network or a large number of malicious nodes join the network after stability; the system can resist the collusion attack more effectively and has lower computation and storage overhead than contrastive models.