| With the rapid development of 5G and the Internet of Things,computing-intensive applications such as augmented/virtual reality,telemedicine and unmanned driving are widely used,and user requirements for transmission rate and service experience are increasing exponentially.The mobile device itself has some non-negligible characteristics,such as battery level,memory and transfer data rate,etc.,which can cause some tasks to be difficult or impossible to complete locally.In cloud computing,users can extend battery life and execute complex applications by offloading applications to the cloud.Since cloud servers are far away from users,offloading tasks to cloud servers will inevitably cause a lot of communication delay.Therefore,neither local nor cloud computing can meet the needs of emerging applications.To address this challenge,researchers propose multi-access edge computing.Users offload their tasks to the edge cloud server of the multi-access edge computing system,which can make up for the shortcomings of their own limited resources,reduce the task execution delay and the energy consumption of terminal devices.In real scenarios,a task can be executed in multiple ways,and resources such as computing and storage resources of a server are often requested by many tasks simultaneously.Therefore,how to choose the offloading strategy of tasks has a key impact on the performance of the entire system.In this thesis,we first study the offloading problem of independent tasks in multi-access edge computing.On this basis,the offloading problem of dependent tasks in multi-access edge computing is studied.The main work of this thesis is as follows:(1)For the multi-access edge computing independent task scenario,this thesis proposes a computing offloading scheme considering both the execution delay of the task and the energy consumption of the local device.In this scenario,tasks can not only be calculated locally,but can also be offloaded to edge cloud servers for execution.We model and study the delay and energy cost of tasks in two execution modes,considering communication,queuing,switching,computing and transmission delays.The optimization objective is to minimize the total cost of the system.The cost of each task consists of the weighted sum of the execution delay of the task and the energy consumption of the local device,and the weights of the delay and energy consumption of each task are different.Since the problem is NP-hard,a computational offloading algorithm based on particle swarm optimization is designed to solve this problem.In the proposed algorithm,a transition probability is designed to select the network selection point and the service placement point,and Dijkstra algorithm is used to calculate the optimal path from the network selection point to the service placement point.The simulation results show that the algorithm can significantly reduce the system cost.(2)For the multi-access edge computing dependent task scenario,reasonable arrangements of task execution sequence and execution location are crucial to improve user experience.Aiming at the multi-access edge computing dependent task scenario,this thesis proposes a problem of offloading multi-dependent tasks by considering both task completion time and execution cost in multi-access edge computing.In problem modeling,multiple interdependent subtasks of the user are considered to be executed on the local device,edge cloud server,and remote cloud server respectively,so as to reduce the total completion time and execution time of multi-access edge computing task offloading.Since the research problem is NP-hard,an improved multi-objective particle swarm optimization algorithm based on queue is designed to solve this problem,and pareto optimal relation is introduced to compare the advantages and disadvantages of different solutions in multi-objective.Extensive simulation experiments show that the proposed algorithm can balance the task completion time and execution cost well,so as to make the user experience better. |
PDF Full Text Request