Adaptive And Efficient Cooperative Caching Scheme Based On Dynamic Traffic Flow

In recent years,with the development of Internet of vehicles and the popularity of intelligent vehicles,new wireless network applications for Internet of vehicles continue to emerge.Among them,such as high-definition map push,traffic condition analysis,short video entertainment,etc.,not only require a large amount of data,but also require very low request delay,so edge caching technology has been widely concerned.Edge cache technology can effectively relieve the working pressure of core network by deploying cache and providing download service at the edge of network.However,in the scenario of Internet of vehicles(iot),especially the Road section with high traffic density,the data demand is large.Due to the limited storage capacity of road Side units(Rsus),it is difficult to cope with the sudden and large cache service demand.Recent studies have proposed that the use of on-board cache resources to provide services for vehicle networking can effectively relieve the pressure of regional cache service.However,this research method still has some challenges: 1)Due to the limited continuous communication time between vehicles in the Internet of vehicles,different content preferences of vehicle users,overlapping service coverage of adjacent cached vehicles and other factors,the selection of cached vehicles and the deployment of cached content become more complicated; 2)Further,in the three-level cache architecture of cloud,road and vehicle,there is a contradiction between the limited backhaul link bandwidth and a large number of diversified user requests,so how to use the cache resources of cloud and road to cooperate with the cache vehicle services is still to be solved.Therefore,this thesis focuses on the impact of vehicle mobility,content popularity,cached vehicle competition on vehicle cache,as well as the collaborative service between cloud-road and cached vehicle,and studies the vehicle dynamic cache scheme and the cloud-road-vehicle three-level collaborative cache strategy based on coded multicast.The specific research is as follows.The main challenge of using vehicles to provide cache service is how to efficiently deploy cache vehicles according to the temporal and spatial distribution characteristics of traffic flow,so that vehicles can relieve the pressure of regional cache service and avoid the waste of cache resources.Therefore,this thesis analyzes the vehicle-to-vehicle(V2V)communication duration,content popularity differences,and competition with other caching service vehicles that affect the efficiency of Vehicle caching.It establishes that maximizing the utility of Vehicle caching service is the goal,and the vehicle caching content placement and service emission power are the decision variables.At the same time,a vehicle cache service algorithm based on Deep Deterministic Policy Gradient(DDPG)is designed to solve the optimization model considering the constraints of vehicle cache space and transmission power.Simulation shows that,under the decision-making conditions of this algorithm,vehicles can adjust content deployment and service scope adaptively according to regional traffic flow information,and obtain higher cache service utility compared with other algorithms.By efficiently utilizing cache resources,the pressure of regional cache service can be relieved.In view of the limited core network backhaul bandwidth in the three-level cache architecture of cloud-road-vehicle,this thesis adopts coded multicast,where content fragments are deployed in the Rsus in a cooperative cache mode.The cloud server collects content requests and encodes the requested data in combination with the cache deployment of Rsus.Only one multicast is needed,and multiple Rsus simultaneously obtain their required cache contents through decoding.And save bandwidth.However,the introduction of coded multicast allows for further coupling of cache deployments between different Rsus,while introducing additional latency because coded multicast requires collecting cache requests from Rsus.Therefore,this thesis proposes a cooperative cache scheme with variable time scale,establishes an optimization model with the goal of maximizing the joint effectiveness of coding multicast and the quality of user service,takes the RSU cooperative cache content placement and the duration of cloud collection of RSU requests as decision variables,and considers the constraints of RSU cache space and user request delay.A multi-level collaborative cache service algorithm based on this model is designed.Simulation results show that the RSU cooperative cache placement strategy designed by this algorithm can realize multicast utility,and the cloud can send fewer multicast packets compared with other algorithms,which can reduce the backhaul bandwidth of the core network and ensure the quality of service for users.