The Edge Server Based Collaborative Caching Strategy Study For Vehicular Networks

With the development of 5G and next generation Internet technology,the main point of the Internet of vehicles(IOV)has gradually changed from vehicular ad-hoc network(VANET)to cooperative vehicle Infrastructure system(CVIS).By improving the carrying capacity and service response capacity of IOV,we can meet the needs of large-scale data computing and high reliability,low delay and differentiated business requirements in CVIS.This is the core issue facing the development of IOV.Multi-access edge computing(MEC)technology is the key to solve the above problems.It sinks the cache resources,cloud computing capacity and the IOV local applications to the edge of IOV,enabling vehicle users to perform task calculation and service response near the end.However,there are still challenges in making efficient use of the MEC cache resources in IOV to improve the efficiency of CVIS.Firstly,there is a lack of collaboration among MECs in IOV,and the cache resources of adjacent nodes are not fully utilized.Secondly,the cache strategy is not fully combined with the sense ability of CVIS for further optimization,resulting in the mismatch between the resource allocation of cache space and the IOV business requirements.In view of above problems,based on the layered network architecture of IOV,this paper focuses on optimizing the scheduling and allocation of the MEC cache resources,aiming at improving the carrying capacity of IOV,reducing the response delay of the IOV services and differentially processing the IOV services.The main work contents and innovations are as follows:(1)A MEC-based collaborative caching system for IOV is proposed.The system is deployed in the MEC server of IOV,which can make full use of the sense ability of CVIS.The obtained multi-dimensional sense information is used to optimize the cache decision.According to the collaborative service response process,the system can use the resources in the MEC collaboration domain to respond to the vehicle content requests.This approach deepens the cache collaboration between MEC servers,enables more vehicle requests to respond at the edge,and then reduces the response delay of the IOV service.To a certain extent,it solves the problem of insufficient carrying capacity and service response capacity of IOV.(2)A MEC-based collaborative cache update strategy for the IOV service awareness is proposed.The strategy comprehensively considers the priority weight of the IOV service,the benefit of collaborative cache placement and the cost of cache replacement.The deep Q network algorithm and greedy algorithm are used to solve the problem.This approach solves the problem that the resource allocation of cache space does not match the IOV business requirements.In addition,by innovating the reward and exploration strategies in the reinforcement learning algorithm framework,we can further improve the accuracy and long-term performance of the cache update decision.(3)The traffic and communication simulation scene of IOV is established based on the actual deployment of IOV in Shougang Park,which hosts the Beijing 2022 Olympic Winter Games.Based on the simulation scenario,the system and algorithm mentioned in the scheme are implemented and tested.Firstly,the simulation results not only verify the functional integrity of the cache system,but also verify the effectiveness of the connection between the cache system and the IOV system.Secondly,the performance evaluation results show that the proposed algorithm improves the long-term cache hit rate and reduces the service response delay to a certain extent.Finally,the proposed algorithm can improve the cache hit rate of content with high service priority,so as to verify the functional effectiveness of differentiated cache according to the IOV business requirements.