Reputation and Trust Models with Data Quality Metrics for Improving Autonomous Vehicles Traffic Security and Safety

In this paper, we develop, implement, test, and analyze a novel technique that allows to improve the security and safety of intersections crossing by the group of autonomous vehicles. The proposed approach is based on augmenting the trust and reputation models with data quality evaluation and using it for initial trust value assignment. This technique allows increasing accuracy and recall rates in detecting agents that might supply incorrect data and facilitating their removal from the agent group consideration. To evaluate the proposed method, we performed the simulation study of the autonomous vehicle traffic control through an intersection. The conducted experiments showed that the employment of data quality metrics improves detecting autonomous vehicles and other agents that might transmit incorrect data.