Research On Computing Migration And Resource Allocation Technology In Mobile Edge Computing Systems

In recent years,the number of mobile devices and computing-intensive applications have increased explosively,which poses a great challenge to the traditional cloud computing architecture.As one of the core technologies of 5G,mobile edge computing(MEC)can sink cloud computing services to the near end and efficiently absorb the computing tasks offloaded from nearby mobile devices,which has achieved the effect of reducing system delay and extending the battery life of mobile devices.This paper combines drone communication,MEC and wireless charging technology to greatly expand the range of services by utilizing the mobility of UAVs and to enable near-end wireless devices to work for a long time,which improves the overall performance of the network.In this paper,three different system models are proposed: multi-to-multi static network model,multi-to-multi dynamic network model and multi-to-single dynamic network mobile UAVs edge computing model.The belief propagation algorithm and Lyapunov optimization algorithm are applied to optimize the computing offloading strategy and resource allocation in these three models.In this paper,the following research results are obtained for these three models:(1)In the small base station network,this paper proposes an optimization algorithm based on belief propagation for multi-to-multi physical scenarios in static networks.The algorithm transforms the centralized optimization into the distributed user for decision making,and iteratively optimizes the suboptimal task offloading decision by using the characteristics of the belief propagation to minimize the weighted sum of the system energy consumption and the delay.The simulation results show that the performance of the proposed algorithm is very close to the optimal algorithm.(2)In the static UAV edge computing system,this paper proposes an online dynamic offloading decision optimization algorithm based on Lyapunov equation for the optimization goal of maximizing long-term average terminal's throughput.The algorithm decouples the optimization objective equation from time slots while considering the randomness of the task arrival and the channel condition.Then it optimizes the optimal offloading strategy,computing resources and charging according to the current system state in each time slot.The simulation results verify the theoretical derivation and demonstrate the impact of different parameters on the system.(3)In the mobile UAV edge computing system,this paper proposes a dynamic user association and path planning joint optimization algorithm based on Lyapunov equation for the optimization goal of maximizing long-term average UAV throughput.In particular,by time decoupling,we optimize the optimal user association,drone path,and server calculation frequency using successive convex approximation in each time slot.The simulation results show that there is a tradeoff between the optimization goal and the queue backlog length.Finally,the thesis summarizes all the work and the shortcomings,then the follow-up study were discussed.