Research On Service Migration Strategy Based On Trajectory Prediction In Mobile Edge Computing

With the development of science and technology and the continuous improvement of user needs,applications such as autonomous driving,virtual reality,and industrial Internet of Things have emerged.Such applications require fast response and calculation results within milliseconds,and traditional cloud computing cannot get the calculation results within the specified time frame due to its geographical limitation,resulting in poor user experience due to task timeout.To overcome the shortcomings of traditional cloud computing,Mobile Edge Computing(MEC)has emerged.MEC deploys computing power near the demand side,and users can directly offload tasks to the nearest edge server,thus significantly reducing latency and improving user experience.However,MEC also has its shortcomings.Limited by cost,energy,volume,latency and other factors,the edge server computing resources,service scope is limited.Due to the mobility of users and complex and variable task requirements,services need to be migrated between different edge servers to ensure service continuity,so how to select the service migration target becomes an urgent problem.In the process of service migration,different links have different reliability,cost and latency,so how to choose a better service migration path to reduce the impact of service migration is also a problem that cannot be ignored.Therefore,in this paper,we focus on the above issues and conduct research on service migration target selection and service migration transmission path optimization in edge computing,and the main contents of the research include:1)For the migration target selection problem in service migration,this paper proposes a service migration target selection method based on trajectory prediction.Firstly,use the trajectory prediction model to predict the user’s trajectory and obtain the range of user activities in the future period;secondly,obtain the list of edge servers according to the user’s activity range,construct the communication model,load model and energy consumption model at the same time,and convert them into a joint optimization problem considering the delay,cost and load at the same time;Finally,The joint optimization problem is solved using a particle swarm optimization algorithm to derive a service migration target selection scheme,and then the service migration is performed according to the service migration target selection scheme.The experimental results show that the service migration target selection method based on trajectory prediction performs better than the traditional method in several metrics.2)For the service migration path optimization problem,this paper proposes a service migration path optimization method based on trajectory prediction.Firstly,use the traceless Kalman trajectory prediction model to predict users’ future trajectories and determine the target edge servers for service pre-migration;secondly,use the service migration path optimization algorithm to obtain a set of low-cost and high-efficiency service migration paths for service migration.The experimental results show that the path selection method proposed in this paper has certain advantages over other traditional schemes and can obtain cost-effective service migration paths,reduce the service migration overhead and improve the service migration speed at the same time.