Research On Reputation Model Based On Blockchain Technology For Internet Of Vehicles

With the industrialization of the intelligent automobile industry,satellite communication and other technologies,as well as the emergence of concepts like intelligent driving,autonomous vehicle,et al.,The Internet of Vehicles(IoV)is in a thriving stage of development.Meanwhile,as more and more vehicles are connected to the network,security concerns arise as well.Although the IoV is a typical example of the Internet of Things(IoT),unlike other traditional Internet of Things scenarios,the IoV has unique characteristics such as high node mobility,rapid changes in network topology,multiple and heterogeneous devices and communication channels,imbalanced and limited device resources,large network scale but continuous changes in network strength et al.Therefore,traditional network security solutions cannot be directly applied to IoV scenarios.While researchers have proposed solutions based on identity authentication and reputation evaluation,but due to issues like the centralized architectures,data privacy concerns,and single points of failure,make those traditional solutions difficult to implement.In recent years,the academic and industrial circles have conducted extensive research on reputation-aware and blockchain-based technology for the scenario of the IoV in both theory and practice ways.However,as the blockchain technology itself is still a new and immature technology,there are still several secure challenges and problems in its performance,such as:Blockchain scalability and security,blockchain-based reputation definition,quantification and update schemes,cold start-up issues,etc.The breakthroughs in these key issues are of great significance for the development of blockchain.This thesis focuses on several problems existing in the application of blockchain technology in the scenario of IoV.Selecting three typical scenarios of IoV and from the perspective of reputation,research is carried out from three aspects:designing secure,lightweight,and efficient blockchain consensus algorithms for ride sharing scenarios,designing solutions for existing security issues based on blockchain technology in data sharing scenarios,and federated learning scenarios,respectively.Corresponding new methods and models are proposed.The main work and innovation points of this thesis are as follows.(1)Aiming at the problem of poor efficiency and scalability,limited security defense coverage,and lack of incentives in present reputationbased blockchain consensus,a consensus algorithm called MPoR(Modified Proof of Reputation)based on adaptive stochastic filtering and multi-feature reputation evaluation optimization was proposed.This method uses the unique hash value of a unique identification of vehicle nodes as a quantitative criterion for stochastic filtering,significantly reducing the number of consensus members while achieving adaptive control of the size of consensus group.At the same time,in order to resist malicious attacks and eliminate the shortcomings of existing single reputation evaluation standards,a multi feature-based reputation evaluation algorithm was designed,which optimized the consensus mining group and the security during the consensus process has been improved,in terms of resistance and scope.Theoretical analysis and plenty simulation experiments show that MPoR can effectively improve consensus efficiency and resist latent and collusive attacks.(2)In view of the present security challenges that exist in data interaction and reputation evaluation and management,this thesis proposes an effective secure reputation model for IoV based on blockchain,called Double-Layer Blockchain-based Reputation Evaluation&Management Model(DBREMM).Based on this model,a complete reputation update scheme is proposed.Firstly,a reputation management model based on two parallel blockchains work collaboratively is designed,which is used to record events and reputation values respectively.Then,according to the typical model of secure event verification in data sharing scenario,a complete set of reputation evaluation and updating scheme is designed,including:1)Direct trust calculation based on multi-factor Bayesian inference,which reduces the observation error in the process of trust calculation and improves the reliability of evaluation.2)Indirect trust calculation based on historical accumulated reputation with attenuation factor,and security reputation fusion scheme based on number threshold with fluctuation factor,thus reducing the possibility of attacks such as collusive attacks and false information injection.Theoretical analysis and extensive simulation experiments reflect DBREMM’s security algorithm effectiveness,accuracy and stronger ability to resist several attacks.(3)To address the cold start and reputation quantification issues in current federated learning scenarios for IoV,a Blockchain-based Hybrid Reputation-aware Federated Learning Model(BHRFL)is proposed.In BHRFL,two consortium blockchain,called Node Reputation Blockchain(NRBC)and Learning Reputation Blockchain(LRBC),are used cooperatively,which record the accumulated reputation of vehicles during long-term activities in the network,and the reputation generated according to training conditions during federated learning tasks,respectively.Based on the parallel chain model,the entire operation process of the BHRFL framework was implemented through smart contracts,and the performance of BHRFL was evaluated in the Ethereum development environment.In addition,a brand-new reputation evaluation scheme is designed to solve the cold start issue through vehicle historical reputation,and to solve the node selection problem in Non-IID(Non-Independent and Identically Distributed)scenarios by integrating vehicle historical reputation and task contribution.Experimental results show that the proposed algorithm has better performance compared to existing reputation-aware federated learning models for efficiency,security,and latency.