Research On Computation Offloading Strategy For 5G Low Latency And High Reliability In-vehicle Services

With the development of Internet of Vehicles(Io V),vehicles have higher requirements for driving safety,efficiency and entertainment.In the upcoming 5G network,the Internet of vehicle is a typical application of Ultra Reliable and Low Latency Communication scenario(URLLC).5G Io V will support high-level autonomous driving,ultra high definition video streaming,online games and other high-quality applications.These applications require network support,and need low latency and high reliability to ensure user experience quality.Cloud computing,which could provide powerful computing,storage and massive network resources to users,is considered as a widely used computing model in recent years.However,cloud computing based Io V faces many problems.For example,cloud data centre is deployed far away from users,which leads high transmission latency on the core network,and if cloud data centres are serving a lot of users' requests,it has many shortcomings such as it needs much bandwidth and it is hard to improve robustness.Therefore,it is difficult for cloud computing to effectively support low latency and high reliability applications in5 G Io V.In order to solve the above problems of cloud computing,this paper researches on Mobile/Multi-access Edge Computing(MEC)and focus on the excellent characteristics and supporting application services of MEC.The edge network architecture and computation offloading under the Io V are analyzed emphatically.Besides,the low latency computation offloading modeling and the reliability of computation offloading are studied systematically,which help us propose a low latency and high reliability computation offloading strategy.The proposed strategy improved reliability of task processing under the latency requirement of real-time in-vehicle application.The specific contents of this paper are as follows:In the Io V scenario,the latency of data transmission is too high for the cloud computing architecture.To solve the problem,this paper researched a Io V architecture(ME-VANET)combined with mobile/multi-access edge computing.By directly offloading the real-time service of the terminal vehicle to the edge MEC device for processing,the high transmission latency of data on the core network is reduced.ME-VANET uses the combination of Software Defined Network(SDN)and MEC to provide flexible network control and centralized resource management for the Io V.In the ME-VANET architecture,the execution latency of the computation offloading of in-vehicle service is theoretically modeled,and the partial offloading and multi-device joint distributed offloading technology are used to jointly process the offloaded-task using multiple MEC devices,further reducing execution latency.When dealing with random task offloading,the proposed computation offloading strategy improves the latency performance by 72.4% and 55.8%,respectively compared with the traditional terminal device processing and single MEC device processing,which can more efficiently support low-latency in-vehicle services.For the high demand for high reliability of lo V,this paper studied reliability in low-latency computation offloading.In the computation offloading of real-time vehicle services,a lowlatency and high reliability computation offloading strategy is proposed which not only embraced the low latency requirement of services but also analyzed reliability.In complex Io V system,computing nodes and links may occur faults,resulting in the failure of the task.This paper simulated possible task executions on edge nodes before computation offloading,and used retransmission and reallocation fault-tolerant mechanism to ensure the successful execution of task.By analyzing the probability of each execution cases and whether the task's latency constraint is satisfied,this paper comprehensively evaluates the reliability of a computation offloading strategy,and proposed a theoretical model of high reliable computation offloading under the latency constraint of task,so that the task is still reliable when some faults occur.In addition,an improved particle swarm optimization algorithm(SAPSORO)combining simulated annealing and constrained particle swarm optimization is used to solve the optimization problem in the high reliability offloading model.The simulation results show that the proposed offloading strategy based on SAPSO-RO algorithm can provide low latency and high reliability for latency sensitive in-vehicle services in ME-VANET architecture,which can meet the requirements of low latency and maximize the probability of successful execution of the services.