Artificial intelligence based trust management for vehicular networks using blockchain

Abstract

Vehicular network is a network that vehicles and roadside units (RSUs) communicate with each other, in order to improve traffic safety and efficiency, with a blockchain-based trust management system inside calculating the trust value for each vehicle and saving them in a public blockchain to ensure the credibility of messages transferred inside. However, in the original vehicular network model, the vehicle predicts the events only depending on the distance between the message sender and the reported event, and the only usage of the trust value is to show the credibility of a vehicle. This paper is planning to implement a machine-learning algorithm to support the model. In addition, since the trust value can be considered as an aggregation of the history of a vehicle, this paper is also planning to take the trust value into consideration. In this paper, the modification will be made during the event prediction process. A machine-learning algorithm with several features will be applied here to improve the accuracy and the trust value will be considered as a positively related factor. This will increase the cost of malicious behaviors and against spoofing attacks and badmouthing attacks. What's more, A double-chain structure will be introduced to fight against the compromised RSUs attack.