Research On Resource Allocation In Mobile Edge Computing

We live in a world where end devices are ubiquitous and perform computations on a daily basis,however,these devices are limited by battery capacity and computing resources,which may not be able to satisfy smart applications that require large computing power(e.g.,augmented reality and face recognition)in terms of service reliability.Mobile edge computing(MEC)comes into being.It sinks cloud computing capabilities to the edge of the network,close to end users,and brings ultra-low latency and high bandwidth.The terminal device mainly transfers the locally generated tasks to the edge server for processing through task offloading,and thus utilizes the rich sensing,communication,computing,storage,and intelligent resources in the mobile edge computing system.However,a large number of emerging application tasks need to request edge computing services with the development of the 5G and 6G eras,resulting in a mismatch between surging service requests and the limited resource supply in MEC systems.In this case,reasonable resource allocation in MEC systems is particularly important,which can reduce the processing time of tasks,improve the task completion rate,ensure service quality,and improve the efficiency of service providers.This thesis conducts research on task offloading and resource allocation related optimization technologies in MEC systems,mainly including the following three research contents:1)Aiming at the challenges brought by the limited device energy and computing resources in the Io T system to emerging delay-limited applications,it is proposed to use energy harvesting technology to assist the user side to perform task computing,and introduce MEC systems to perform task offloading to utilize its abundant resources.First,an energy-sustainable mobile edge computing-enabled Io T framework is proposed,which integrates energy harvesting technology into Io T devices,and divides tasks into subtasks for parallel processing on Io T devices and edge servers.Second,under the proposed framework,a parallel task offloading model,an energy queue model on user equipment,etc.are constructed,and packet loss and task processing delay are defined.On this basis,the optimization problem of minimizing the long-term task processing delay and packet loss rate is studied,which takes into account the energy stability of the user side and the constraints of communication and computing resources.In order to solve this problem,an online distributed algorithm based on Lyapunov optimization is designed to perform parallel task offloading and resource allocation strategy.Finally,simulation experiments verify the performance advantages of the proposed strategy under different V values(which can be considered as a factor to weigh the importance of delay and energy),task arrival rate,energy acquisition power,task deadline,distance and other factors.2)Aiming at the communication and/or computing resource bottlenecks in the task offloading process in the mobile edge server system,as well as the consequent problems such as large communication/computing queuing delay and poor user service experience in the task offloading process,it is proposed to consider both the user side and the Task offloading and joint resource allocation strategy for edge server-side communication and computing resources.First,a communication-computation tandem queuing model is constructed based on queuing theory to characterize the dynamic resource relationship between mobile devices and connected edge servers to facilitate joint resource allocation in mobile edge computing systems.Second,the queue model of task offloading is constructed as a discretetime Markov decision process and its components are defined to facilitate system analysis and decision-making.Based on this,an optimization problem of maximizing system throughput is proposed to achieve optimal task offloading and resource allocation given resource constraints and user service quality requirements.To solve the optimization problem,a two-stage user-server matching and joint resource allocation algorithm is designed to realize global resource optimization of mobile edge computing system and improve the efficiency of service providers.Finally,simulation experiments verify the performance advantages of the proposed strategy under different factors such as the number of mobile users,task arrival speed,delay limit,maximum transmission power,maximum computing power,and the number of edge servers.3)Aiming at problems such as poor user service experience and insufficient resource utilization on edge servers caused by uneven spatial and temporal distribution of user service requests and workload differences on edge servers,a multi-hop edge computing task offloading strategy by introducing relay nodes is proposed.First,a novel multi-hop mobile edge computing framework is proposed,in which vehicles are introduced as storecarry-forward relay nodes to facilitate the uploading of user tasks in mobile edge computing systems.Second,under the proposed framework,the task transmission and computing process are constructed into a single-queue multi-server model,and a Markov chain is constructed based on this to characterize the queue dynamics and uncertainty of resource availability on relay vehicles and edge servers sex.Based on the constructed model,an end-to-end service delay model is derived and a resource optimization problem to minimize long-term service delay is studied.To solve the optimization problem,a relay node and edge server selection algorithm is designed to achieve load balancing of edge servers and reduce user service delay.Finally,simulation experiments verify the performance advantages of the proposed strategy under different task arrival speeds and different vehicle available resources.