| With the development of wireless mobile communication technology and the continuous emergence of burgeoning communication services,a large number of latency-sensitive computing services have been generated,which greatly challenge the weak computing capability of wireless devices.In order to meet this challenge,mobile edge computing(MEC)technology has adopted cloud-like computing distribution architecture,sunk conventional computing resources to servers located at the edge of wireless access networks,and provided low latency computing services for proximal users.However,the offloading process of MEC technology depends on multiple access of wireless access networks.The existing orthogonal multiple access(OMA)transmission architecture incurs tens of milliseconds latency in physical layer transmission,which is far beyond the ultra-low latency which approximates to millisecond expected by MEC technology and has become one of the main constraints to the application of MEC technology.In order to address this problem,non-orthogonal multiple access(NOMA)technology is integrated with MEC technology to formulate the NOMA-driven MEC technology architecture,termed as NOMA-MEC,which can achieve parallel task transmission in the code/power domain.Compared with conventional OMA-driven MEC systems,NOMA-MEC systems greatly reduce the latency of task offloading and improve the spectral efficiency for the task offloading of MEC technology.In order to implement efficient transmission of multi-information flow,existing NOMAMEC systems optimize the power allocation and task splitting by designing the computing resource scheduling scheme,with the aim of bringing offloading performance gains in terms of the latency,the energy efficiency and other performance indicators of NOMA-MEC systems.However,the NOMA-MEC offloading scheme needs to meet the condition that users have to simultaneously complete task offloading to servers.The NOMA-MEC offloading scheme ignores the computing capability difference between the cloud/edge server,resulting in insufficient computing resource utilization of cloud/edge servers and an increase in the task completion latency.When integrating NOMA with OMA,termed as the hybrid NOMA technology,a better latency performance can be achieved by hybrid NOMA systems than OMA systems.This thesis aims at NOMA-MEC systems,and realizes the goal of minimizing the system latency from two perspectives: resource scheduling factor optimization and intelligent edge server selection policy.Aiming at this goal,this thesis has completed the following two aspects of work:(1)In the scenario that a fixed edge server is responsible for assisting task execution,the joint optimization problem of power allocation and task scheduling is formulated to minimize the task completion latency of hybrid NOMA-MEC systems.In order to solve the formulated non-convex optimization problem,an adaptive power-task resource allocation iterative algorithm(APTRAIA)is proposed.Simulation results demonstrate that the proposed algorithm can effectively reduce the task completion latency.(2)Considering the user mobility and environmental uncertainty,an edge server selection problem aiming at minimizing the system latency is formulated.In order to solve this problem,an intelligent edge server selection strategy(IESSS)based on Markov decision process(MDP)is proposed.According to the current environment and state information,the proposed IESSS algorithm can select an edge server from multiple idle edge servers to assist task execution,so as to make full use of system computing resources and further reduce the task completion latency.Simulation results demonstrate that the IESSS algorithm can reduce the task completion latency with lower algorithm overhead compared with conventional algorithms. |
