Research On Computation Offloading And Resource Allocation Policy Optimization In Multi-user And Multi-server Edge Computing Environment

Mobile edge computing(MEC)is an emerging computing paradigm capable of extending computing,communication,and storage facilities to the edge of the radio access network.Resource-constrained terminal devices can perform computation-intensive and latency-sensitive computing tasks through MEC to meet the computation and latency requirements.Computation offloading is one of the key technologies of MEC,and the existing literature has carried out a series of adequate discussions on the computation offloading of MEC.However,due to the resource constraints of the MEC environment,the heterogeneity of software and hardware,and the dynamic and randomness of terminal devices,policy optimization of computation offloading has always been regarded as a key challenge for MEC.In this thesis,the policy optimization problem of computation offloading and resource allocation in a complex MEC environment is studied deeply to allocate MEC system resources reasonably and improve the overall performance of the MEC system.This thesis studies the policy optimization problem of computation offloading and resource allocation in the edge computing environment.The main works are as follows.(1)Research on offloading policy optimization in the edge computing market with multi-server competition.First of all,aiming at the uneven distribution of computing resources in the MEC system caused by free computing services,we study and construct a kind of edge computing market model based on software defined networking with multi-server competition.Secondly,we formulate the maximizing the profit of MEC servers as the optimization objective and optimize the computation offloading policy by jointly optimizing the choice of MEC servers,the size of offloading data for end-users,and the pricing of computing services for MEC servers.Then,we propose a dynamic computation offloading algorithm based on proximal policy optimization and strategy game to solve this problem.Finally,a series of simulation results prove the performance and advantages of the proposed algorithm.(2)Research on computation offloading and resource allocation policy optimization of UAV-assisted edge computing.Considering the special scenario of MEC,we further extend the computation offloading research to the UAV-assisted MEC scenario.First of all,we construct a UAV-assisted MEC system model,in which the UAV is the relay node for communication and computation.Second,we formulate minimizing the weighted total cost of system latency,UAV energy consumption,and the size of discarded task data as the optimization objective to optimize computation offloading and resource allocation policy.Subsequently,we formulate this joint optimization problem as a Markov decision process and propose a dynamic computation offloading algorithm based on Soft Actor-Critic to solve this problem.Finally,a series of simulation results prove the performance and advantages of the algorithm.