Research On Video Service-based Resource Allocation In The Internet Of Vehicles

With the rapid development of the automotive industry and mobile communication technology,video services in the Internet of Vehicles(Io V)will play a more important role in entertainment,road safety,and other aspects.Due to the dynamic characteristics of wireless channel and the limited wireless spectrum resources,computational resources,and storage resources,it is a big challenge to provide high-quality,low-latency,and low-jitter video services for vehicles.This article studies the allocation of spectrum resources,computational resources,and storage resources based on video services in the Internet of Vehicles to maximize the user's video experience quality.The specific works are as follows:(1)This paper proposes a resource allocation scheme based on live video transcoding and transmission in the Internet of Vehicles.Different from the existing studies that only consider the video bitrate when modeling video streams,this article considers both the bitrate and the video type,since different videos require different computing resources.The study aims to maximize the video quality of all vehicles and reduce latency,and quality jitter,by jointly optimizing vehicle scheduling,bitrate selection,computational and spectrum resource allocation.Because the wireless channels and available resources in Io V have Markov characteristics,the above joint optimization problem is modeled as a Markov Decision Process(MDP).To solve the MDP,the study utilizes the "Soft Actor-Critic" algorithm that is based on maximum entropy framework,to solve this MDP problem.Finally,extensive simulation results based on the dataset of the real world show that compared with other reinforcement learning(RL)algorithms,the proposed algorithm accesses more excellent performance in terms of learning speed,exploration ability and stability.(2)This paper proposes a video transmission scheme based on scalable video coding(SVC)technology in the Internet of Vehicles.Scalable Video Coding(SVC)technology can encode a video stream into multiple video layers with different bitrates,and adaptively transmit videos with different bitrates to match the dynamic variance of vehicle channels and diverse requirements,thereby improving the user experience.The proposed scheme utilizes the storage function of the roadside unit(RSU).By transmitting high popularity videos to the RSUs,it can relieve the transmission burden of the backhaul links,save spectrum resources,and reduce latency.The joint optimization problem of vehicle scheduling,video layer selection and spectrum resource allocation is modeled as MDP.To solve this MDP,an actor-critic deep reinforcement learning algorithm is employed,which can deal with the action space with multi-dimensional continuous actions mixed with discrete actions.Simulation results show that the proposed algorithm can effectively improve the video experience quality of vehicle users while reducing the delay,and the proposed algorithm has significantly excellent performance in terms of learning speed,compared with the policy gradient algorithm(PG)and deep Q-learning algorithm(DQN).