Research On Intelligent Resource Allocation In Internet Of Vehicles Based On Edge Computing

With the development of society and technology,increasing vehicles and complicated traffic conditions are leading to the lack of global information and the time-varying character of Internet of Vehicles(IoVs),which makes it challenging to perform effective task arrangement and resource allocation in IoVs.Considering the characteristics of vehicles in different driving scenarios,we propose task scheduling and resource allocation schemes to dynamically match network resources and vehicular applications,which provides users with high-quality services.First,when vehicles move straightly,we propose a hierarchical resource allocation model,in which a base station integrating with several RSUs cooperatively orchestrate edge computing and content caching to improve IoV performance.Moreover,we formulate a joint optimization problem by analyzing MNO's revenue and users' quality of experience,and develop a mobility-aware resource allocation schemes by investigating deep reinforcement learning.Then,when vehicles arrive at a crossroad,we propose an edge cooperative resource allocation model,which enables multi-RSU offloading to utilize unoccupied resources in IoVs,aiming to minimize network delay.Then we develop an online optimization framework by leveraging Lyapunov optimization theory,and prove that it can achieve near-optimal performance.Leveraging strong cognition of Artificial Intelligence(AI),we put forward an imitation learning enabled branch-and-bound solution to speed up the problem solving process with a few training examples.Lastly,experimental results based on real-world traffic data demonstrate that our proposed methods are suitable for intelligent transportation systems in 5G era.In addition,our proposed schemes greatly promote IoV performance and outperform other methods from various aspects.