Research On Distributed Resource Allocation Strategy On The Internet Of Vehicles

Automated,connected,and electric vehicles have become widely recognized as development trends and strategic directions for vehicle technology.The evolution and revolution of vehicle technology will reduce the cost of travel and change the way of travel,thus bringing profound changes to people’s daily life.The internet of vehicles is a new industrial form that deeply integrates automobiles,energy,information,transportation,and traffic management industries.It is also an important application of high-tech such as 5G,artificial intelligence,and clean energy in industries such as automobiles and transportation.However,the resource allocation on the internet of vehicles becomes complicated and challenging since the internet of vehicles has a hierarchical data processing architecture,where are multiple distributed and autonomous entities with built-in data storage,computing,and communication support,providing diversified services such as dynamic environment perception,connected autonomous driving,and coordinated charging of electric vehicles.This research provides novel insights on improving the design and analysis of the internet of vehicles,especially from the perspective of resource allocation.The research objective of this dissertation is to design,analyze,and evaluate a hierarchical framework with distributed resource allocation for the internet of vehicles based on game theory,considering the characteristics of automated,connected,and electric vehicles and the scenario of multiple service operators serving multitudes of users at the same time.This dissertation is interdisciplinary by nature,and encompasses contents from economics,optimization,game theory,communications,networking,intelligent transportation system,and smart grid.The main research results of this dissertation are as follows:1.For the problem of communication resource allocation in MEC-assisted vehicle onroad analysis,we propose the pricing strategies of smart mobility service providers and the data downloading strategies of vehicle users based on the multi-leader-follower game,which effectively improves the utilities of smart mobility service providers and vehicle users.Vehicle users can receive multiple real-time data files from several smart mobility service providers and combine these files and local observations to extend sensing range,reinforce and validate local observations.However,as the amount of downloaded data increases,the download speed will be very slow due to the limitation of network capacity.We assume a data downloading market between smart mobility service providers and VUs in offering and purchasing files to support vehicle on-road analysis.Each smart mobility service provider and each vehicle user are assumed to compete with their peers to maximize their own profits.We optimize the pricing strategies of smart mobility service providers and the data downloading strategies of vehicle users in the scenarios of the single service provider and multiple service providers,to balance the reliability of the fusion result and network delay.Numerical simulation results verify the theoretical analysis and show that the proposed strategy maximizes the utility of smart mobility service providers and vehicle users.2.For the problem of electric vehicle public charging recourse allocation,we use the hierarchical game and matching theory to analyze the charging time optimization for electric vehicles,the matching between electric vehicles and charging stations,and the pricing mechanism for charging stations under the time-based billing model,so as to effectively improves the efficiency of the entire public charging system.Electric vehicle public charging is important to meet the exploding charging demand and to address the range anxiety issue.We focus on the connected electric vehicle public charging market with heterogeneous charging stations under the time-based billing model.We develop a hierarchical game to jointly consider the charging time optimization for electric vehicles,the matching between electric vehicles and charging stations,and the pricing mechanism for charging stations.The optimal charging time strategies are analyzed for electric vehicles,and a many-to-one matching algorithm is applied to solve the matching between electric vehicles and charging stations.Besides,a block coordinate descent based algorithm is applied for each charging station to solve the pricing problem.Simulation results show that the proposed schemes can achieve the performance improvement of the charging system.3.For the incentive mechanism design towards model training and data analytics related to autonomous driving,we propose an incentive mechanism for the federated learning and analytics system with multiple tasks based on federated optimization and multi-leader-follower game,which improves the efficiency of resource allocation for data owners and system utility.Autonomous driving is essentially a complex large-scale multi-task learning problem that requires collaborative learning of model weights or statistics from decentralized vehicle driving data.Using federated learning and analysis in autonomous driving can keep data in the vehicles,thereby reducing data collection costs,improving communication efficiency,and protecting data privacy and security.However,the performance of task models relies heavily on the participation of data owners.Furthermore,for a multi-task system,data owners need to allocate limited computation resources efficiently among different local tasks.In the proposed mechanism,each task owner decides its reward rate to incentivize data owners to participate in federated learning or analytics.Based on the reward rates of tasks,each data owner determines the accuracy level of local tasks’ solutions.Simulation results show that our proposed mechanism can achieve the performance improvement of the whole system.