Abstract:

Vehicular ad hoc networks(VANETs) are an important part of intelligent transportation systems and can improve traffic safety and efficiency through the dissemination of vehicle messages.However,the dissemination of false emergency messages by malicious vehicles poses a serious threat to the normal operation of VANETs and traffic safety.To address the low accuracy of message trust evaluation in existing trust management schemes for VANETs with a high proportion of malicious vehicles,a machine learning-assisted trust evaluation scheme is proposed which optimizes the existing trust evaluation algorithm by introducing a random forest model to assist roadside units in analyzing emergency messages and outputting the prediction probability of messages being true.Based on the prediction probability output by the random forest model,a switchable caching mechanism is designed,and a trust value query algorithm is designed to balance the conflict between query efficiency and storage space overhead of roadside units in the existing scheme.Meanwhile,the prediction probability is used as a reference factor in the trust evaluation algorithm to obtain a higher message evaluation accuracy.Finally,the proposed scheme is compared with the existing scheme,and experimental results show that the message trust evaluation accuracy of the proposed scheme is improved by approximately 6.2%~21.9% and that the proposed scheme exhibits good robustness under several proportions of malicious vehicles.

Key words: vehicular ad-hoc networks, emergency message, machine learning, smart contract, trust evaluation