Multi-user Service Migration In Mobile Edge Computing

Mobile edge computing is an emerging technology for 5G network that leverages resources deployed at the proximity of users to improve network performance.Utilizing mobile edge computing,network services can be deployed to edge servers to reduce service latency.Due to the mobility of users,the services require to be migrated to maintain ultra-low latency performance.However,the single-user service migration strategies are inapplicable to multi-user scenario.Firstly,making service migration decisions for each user independently will suffer from the interference among the users.Moreover,prediction of future state,which is crucial for service migration decisions,becomes intractable when the number of users is large.Thus,we research the multi-user service migration problem as follows.We first investigate the multi-user service migration in heterogeneous denes cellular networks.Quantified the interference among users with the limited storage capacity in edge servers,we formulate the multi-user service migration as a optimization problem with the aim of minimizing the average latency,where the limited energy in edge servers is taken into account.We then develop an efficient multi-user service migration scheme based on Lyapunov optimization to resolve the original problem without predicting the trajectories of the users.Next,we study the multi-user service migration in terminal-edge cooperative networks.Due to the mobility of terminal devices acted as servers,the dynamic network topology complicates the migration decisions.We formulate the multi-user service migration as a optimization problem with the aim of minimizing the average latency,where the limited storage in edge server is taken into account.A future latency and interference aware multi-user service migration algorithm based on hidden markov model and the interference scale is proposed to solve the problem.A series of simulations are conducted to achieve 9.1 % and 8.0 % performance gain over state-of-the-art solutions in terms of average latency,respectively.By studying the multi-user service migration and applying it to various scenarios(such as Internet of Vehicles),the quality of experience of users can be effectively improved.