Research On Optimization Method Of Caching Resource Scheduling Based On 5G Internet Of Vehicles

With the rapid development of mobile communication technology and internet of vehicles(IoV)technology,new applications of IoV,such as intelligent transportation and smart drive,continue to emerge.And the number of vehicular networks mobile device connections and mobile data traffic demand has shown explosive growth.Due to limited backhaul bandwidth and restrictions on spectrum resources,traditional mobile networks are difficult to meet the requirement for high-density access equipment for high-bandwidth and low-latency services.To solve the above problems,5th generation mobile networks(5G)supports mobile edge computing(MEC)to sink functions such as caching,calculation,and processing to the edge of the network.It is greatly enhancing the data distribution and processing capabilities of the IoV.MEC caching can proactive caching popular content to the edge of the network close to vehicle users,thereby supporting contents distributing services with high-bandwidth and low-latency.It effectively relieves the pressure on the backhaul bandwidth of the mobile network,reduces network delay,and provides technical support for future highreliability services such as intelligent driving.The high mobility of vehicles on the IoV makes the network topology present complex temporal and spatial dynamics.Also,it causes the wireless communication status of vehicle users in the IoV to show dynamic changes.Existing mobile edge caching schemes are difficult to directly apply to the IoV scenes with high mobility and limited service time for vehicle users.As a result,the demand for high bandwidth and low latency content distribution services required by many emerging applications on the IoV cannot be met.Therefore,the thesis mainly researches the optimization method of edge caching resource scheduling for 5G IoV,realizes the cross-domain joint optimization of the vehicular networks caching resources and wireless communication resources,and addresses the continuous changes in geographic location and dynamic temporal and spatial changes of content popularity caused by the mobility of vehicle users.And the vehicle user quality of experience(QoE)evaluation of multiple bitrate video is studied.The main results achieved in this thesis are as follows:Firstly,a mobile edge caching scheduling strategy based on MEC is proposed.To solve mobile network content distribution delay and system energy consumption,we first study mobile edge caching’s theory and technology.Then,based on MEC mobile edge caching architecture,we proposed a caching scheduling joint optimization method of user average delay and system energy consumption.The method reveals the influence of caching capacity,backhaul bandwidth,and content popularity distribution on the caching scheduling method and mobile networks performance and realizes the cross-domain joint optimization of caching resources and mobile networks communication resources.Finally,numerical results demonstrate that our proposed algorithm converges to the algorithm with the best latency performance at a very low cost of caching resources.And our proposed algorithm effectively saves network operation and deployment costs,reduces energy consumption,and provides a reference solution for green energy supply.Secondly,a mobility-aware caching resource scheduling optimization method on the internet of vehicles is proposed.To meet the low-latency,high-bandwidth services requirements for IoV applications,such as autonomous driving.And the impact of high mobility of vehicle nodes on caching and network performance.Based on the first part’s conclusions,we propose a MEC caching architecture in the IoV and achieve rapid content distribution.Then,formulate a jo int optimization problem that maximizes the average download percentage of vehicle users and caching efficiency.We take the method of parameter control that the problem is transformed into maximizing the average download percentage under the caching cost’s determination.Further,the joint optimization problem is divided into a single extremum sub-problem and a nonlinear integer programming sub-problem.For the single extremum sub-problem,we use the numerical method to solve it.For nonlinear integer programming,our designs a joint optimization for vehicle networks caching(JOVNC)algorithm to solve it.Finally,numerical results illustrate that the proposed method effectively improves the caching efficiency and the average download percentage of vehicle users.Our proposed algorithm effectively achieves the cross-domain joint optimization of caching resources and communication resources in the IoV.Thirdly,a method for optimizing the caching resources of the IoV based on temporal and spatial dynamic popularity-aware is proposed.The method mainly considers the impact of vehicle users’ mobility on the popularity of content in the IoV,especially the temporal and spatial variation of content popularity caused by vehicle users’ movement across regions.Firstly,we analyze the characteristics of content popularity and introduce reinforcement learning algorithms.We then formulate a dynamic content popularity caching resources optimization problem in the IoV that aim to maximize the average download rate of vehicle users.We modeled the optimization problem as a markov decision process(MDP)problem.And proposed an algorithm based on deep reinforcement learning(DRL)to solve the MDP problem.Further,to meet the more rapid variations of content popularity,we improved the DRL algorithm.Finally,numerical results illustrate that our proposed algorithm significantly increases the average download rate of vehicle users and achieve fast content distribution in the temporal and spatial dynamic environment of content popularity in the IoV.Fourth,a QoE-aware caching resource optimization method for vehicular networks is proposed.Vehicle users have different mobile video QoE perceptions due to differences in vehicular equipment devices and wireless communication conditions.The QoE evaluation method of static mobile users in the mobile network can not work for the dynamic scenarios of the IoV.Firstly,a QoE evaluation model of vehicle users for IoV mobile video is proposed.Secondly,because of the trade-off between video quality and video diversity for multiple bitrate video.We propose a deep reinforcement learning caching resource optimization method to maximize vehicle users’ average QoE.We improved the DRL algorithm and effectively improved vehicle users’ average QoE,and enhanced algorithm converges’ performance.Finally,numerical results indicate that the proposed algorithm obtains a higher average vehicle user’s QoE than the other benchmark algorithms.