Research And Implementation Of Service Migration Strategy In Mobile Edge Computing Environment

Mobile Edge Computing(MEC)provides a high-performance,low-latency and high-bandwidth carrier-grade service environment by sinking storage and computing power to mobile edge nodes,accelerating the distribution and download of content,services and applications in the network and enabling users to enjoy a higher quality network experience.However,with the rise of mobile applications such as Telematics and autonomous driving,the limited network coverage of edge nodes can lead to poor network connectivity to the local edge cloud when users move to other edge nodes,which can reduce the quality of service(Qo S)or lead to service disruption in some cases.To ensure service continuity and quality of service,service migration is required by means of service scheduling to migrate services to a new nearby server.In order to achieve fast and efficient service migration,this thesis examines both service migration strategies and service migration methods to address the impact of service quality degradation due to service migration.The details are as follows.(1)To address the problems of when to trigger migration in the service migration process and the increase in service overhead due to blind selection of migration targets,this paper proposes a service migration strategy based on a Markovian decision process,where the service migration problem is modelled as a Markovian decision process and a two-dimensional random walk model is used for the user movement model.In this scheme,the three attributes of user movement direction,service migration cost and service transmission cost are combined to construct a benefit function,and the migration policy is updated based on the long-term benefit value of the Bellman equation to derive an optimal migration policy,which can guide whether,when and where to migrate the local MEC service.Simulation results show that the proposed algorithm reduces the total service cost while ensuring service continuity and reduces the impact of user movement on subsequent service quality compared with traditional service migration strategies.(2)To address the problem of downtime due to temporary service interruptions during service migration,this paper proposes a checkpoint-based pre-dump iterative service migration method to support real-time service migration on edge devices hosting virtualised resources(Docker).The approach exploits the tiered storage nature of Docker containers(containers are stacked from multiple bottom image layers and top container layers)and first implements fast end-to-end image migration using incremental synchronisation to reduce the transfer of redundant layers by migrating only the missing image layer data for the target container against the image layer stack architecture.For container-level migration,most of the memory state information is transferred to the target server through multiple iterations before the container freezes,with each iteration transferring only the dirty memory pages that were modified by the service process during the previous pre-dump.When the container freezes,only the dirty pages and changed execution state information generated during the last iteration are transferred to the target server,so the amount of data transferred is minimal and,thanks to the low amount of data transferred during the service freeze,the service downtime is significantly reduced.(3)Based on the above proposed checkpoint-based pre-dump iterative service migration method,an experimental platform is built to verify the feasibility and performance of the scheme.The experimental results show that the scheme can achieve real-time migration of containers,and under the conditions of setting different network bandwidth and dirty page change rate,the performance of the scheme is good,and the service downtime is significantly reduced compared with the traditional migration scheme.