Resource Allocation Methods In Trusted Vehicular Networks

With the development of modern communication technology and artificial intelligence,the Internet of Vehicles(Io V)has been realizing efficient perception,intelligent analysis,and safe sharing of information among people,vehicles,and roads.Whether data can be shared trusted matters the efficiency and safety of the transportation system.Constructing a trusted data sharing environment is beneficial to form a benign data sharing and utilization cycle with big data and artificial intelligence(AI)technology,and also beneficial to incorporate with the transportation and grid system to provide threedimensional services.However,the existing Io V suffers many security issues,such as mutual untrust between entities,onboard data authentication difficulty,onboard training data privacy leakage,and vehicle-to-grid(V2G)trading process insecurity.Blockchain has an advantage in building trust between untrust entities at a low cost so that can be utilized to construct trusted vehicular networks.This dissertation firstly designs a consortium chain-based distributed trust management architecture,which can assist the following research content.Introducing blockchain in the resource-constrained Io V will cause extra computation and communication overhead,so this dissertation investigates the resource allocation problem under public chain-based trusted data sharing scenario,consortium chain-based trusted application training scenario,and consortium chain-based energy trading scenario.The main contributions are as follows:1)To prompt mutual trust between vehicles,a consortium chain-based distributed vehicular trust management architecture is designed.Under this architecture,ratings and trust value of vehicles can be shared trusted in the Io V,and can also realize distributed operation and centralized supervision.Then,to enhance randomness and security during the consensus node selection process,the architecture associates the ratings collection amount of roadside unit with the consensus node selected probability.Finally,theoretical analysis and simulation results verify the proposed architecture has strong security properties.2)To solve the problem of low vehicle participation caused by the competitive consensus mechanism,in the public chain-based trusted data sharing scenario,a computation offloading chosen strategy is designed.Firstly,a cooperative mining strategy is designed,in which vehicles can rent computation resources from edge servers and form coalitions to participate in the consensus mechanism.Secondly,a coalition formation algorithm is designed to improve system utility and profit rate by optimizing the computation offloading choice of vehicles.Finally,simulation results verify the effectiveness of the proposed algorithm in improving vehicle profit rate and system utility.3)To solve the problem of low federated learning training efficiency caused by the dynamic and time-varying environment,in the consortium chain-based trusted AI training scenario,a computation frequency,wireless channel,and computation offloading chosen strategy is designed.Firstly,the machine learning training latency is quantified under the generalized architecture of federated learning combined with consortium chain.Secondly,to allow the vehicles to select the CPU frequency and choose wireless channel and computation offloading choice autonomously in the dynamic and time-varying vehicular network,a distributed deep reinforcement learning-based algorithm is designed to reduce the training latency.Finally,simulation results verify the effectiveness of the proposed algorithm in reducing the training latency in the dynamic time-varying environment.4)The vehicle transaction fee and other factors can influence the consensus security and efficiency,in the consortium chain-based energy trading scenario,a transaction fee and recruit ratio adjustment strategy is designed.Firstly,a three-stage energy demand negotiation protocol is designed to enhance the flexibility of the energy matching process,and combined with smart contract and consortium chain to ensure the self-enforcement of the protocol.Secondly,to enhance the consensus security,edge servers can recruit idle vehicles to participate in the consensus process.A robust Stackelberg algorithm is designed under consideration of vehicle unstable state.The proposed algorithm can adjust the transaction fee of the consumer vehicle and recruit ratio of edge servers under consideration of balancing network security and block propagation latency to maximize their respective utility.Finally,theoretical analysis shows the proposed energy matching mechanism has strong security properties,and simulation results verify the effectiveness of the proposed algorithm in balancing consensus security and efficiency.