| With the development of information technology,5G and Internet of Things technology are making progress and many new application services are emerging one after another,such as Automatic driving,Real-time games,High-definition video.These new application services have a large number of computation-intensive or delay-sensitive tasks.However,the processing of tasks and user service experience will be affected due to the limited of computation capacities of mobile devices.Mobile edge computing(MEC)is an important technology to solve this problem.It pushes more computing resources to the edge of the network and uses edge servers to handle computing tasks.By this way,mobile devices can effectively reduce the serving delay of tasks and improve user service quality.In the dynamic network scenario,due to the high mobility of users and the dynamic characteristic of services,how to ensure the success of task offloading and the continuity of task processing is a problem worthy of discussion.Therefore,reliable and efficient task offloading is still a challenge in dynamic scenarios.In view of the above problems,the research contents of this dissertation are as follows:1.Research on dynamic task offloading and scheduling based on mobility prediction: In the MEC network scenario,for users with high-speed mobility in the dynamic network scenario,in order to reduce the total serving delay of task processing in the system and improve the success rate of task offloading,this dissertation proposes a dynamic task offloading and scheduling scheme based on mobility prediction.We first propose a prediction model to predict the residence time of high-speed users in the base station to assist the subsequent optimization.This prediction model is based on long short term memory(LSTM)network.The offloading and scheduling problem of time-dependent application tasks is modeled as a combinatorial optimization problem,which is solved by genetic algorithm.It is verified by simulation that the scheme proposed in this dissertation is superior to other reference schemes in the performances of system serving delay and can effectively reduce task offloading failure rate.2.Research on dynamic task offloading and scheduling based on two-layer genetic algorithm: In the MEC network scenario,considering the task dynamics and processing continuity of intelligent mobile devices,this dissertation proposes a dynamic task offloading and scheduling scheme based on two-layer genetic algorithm.In order to reduce the network cost of the system,we use the combination of rolling window rescheduling strategy and two-layer genetic algorithm to optimize the offloading and scheduling of dynamic arrived tasks.Firstly,the rolling horizon rescheduling strategy is used to decompose the complex long-term dynamic problem into a series of static periodic optimization problems.To effectively solve the optimal offloading and scheduling problem of each rolling horizon,the static optimization problem is further decomposed into two subproblems: task offloading decision optimization and task scheduling order optimization,which can be solved by twolayer genetic algorithm.Simulation results show that the proposed scheme can effectively reduce the network overhead of the system and is better than other comparison schemes in performance improvement.3.Design and verification of MEC service dynamic offloading system: This dissertation establishes a MEC service dynamic offloading system based on the GUI interface of MATLAB,which is divided into the MEC offloading part of dynamic users and the MEC offloading part of dynamic arrived services.The system verifies the functions of the above two schemes and outputs the interface results.The verification results show that the system designed in this dissertation has clear functions and simple operation,which can assist in task offloading scheduling in dynamic scenarios,help users better understand the schemes and realize system optimization and rational allocation of resources. |
PDF Full Text Request