Assessment Of Trust In Vehicular Social Networks

The current transportation system is going through a revolution with the rapid development of autonomous vehicle technologies and wireless communications.Information trustworthiness in vehicular network plays a vital role in facilitating data sharing among vehicles,to achieve better driving safety and convenience.Data can be generated from the sensors on either vehicles and the roadside infrastructure.By utilizing the data shared from other vehicles or roadside infrastructures,a vehicle is able to broaden its sensing range and has a better perception about its surrounding environment.As data are collected by individual vehicles,equipped with various types of sensors and different data processing algorithms,the trustworthiness of the data cannot be guaranteed.Existing research on trustworthy vehicular systems mainly focuses on its security aspect,i.e.,to guarantee the system’s confidentiality,integrity,and authentication.There is not adequate study on the trustworthiness of the data and/or the data generators.To tackle the information trustworthiness issue in connected vehicular systems,we consider the data exchanging between vehicles as a kind of social interaction between them,and model a vehicular network as a vehicular social network.Within the network,vehicles are connected if they exchanged data from which the interacting vehicles can derive the trustworthiness of each other,based on the trustworthiness of the shared data.With the vehicular social network,we design the OpinionsWalk algorithm to accurately assess the trustworthiness of vehicles,leveraging the three-valued subjective logic trust model.By investigating the process of data exchange between vehicles,we design a mechanism to conduct objective trust assessment,which is proven to be more accurate than subjective trust assessment.Finally,we treat vehicle in groups/communities and propose the solution to intra-and intercommunity trust assessments.Simulation results demonstrate the accuracy and efficiency in trust assessment in vehicular social networks.Evaluation of the proposed solution is divided into two parts.We first evaluate the performance of the OW algorithm to a few benchmark algorithms,including EigenTrust(ET),TrustRank(TR),MoleTrust(MT),and TidalTrust(TT).Because these algorithms are all designed for online social networks,we adopt two real-world OSN datasets,Advogato and Pretty Good Privacy(PGP),to make a fair comparision between OW and the benchmark algorithms.Then,we simulate a vehicular nework where vehicles exchanging data are able to conduct trustworthi-ness assessment of each other.The accuracy of the proposed dynamic trustworthiness assessment mechanism is evaluated in the simulations.