Task Offloading And Resource Allocation In Edge Computing System Interconnected By Metro Optical Network

With the emergence of 5G applications and the rapid development of the mobile Internet,network traffic has exploded,and the requirements for latency have become increasingly stringent.The metro networks close to end users are rapidly upgrading and undertake the allocation of different network and IT resources.Meanwhile,the Multi-access Edge Computing(MEC)has been proposed,which can provide cloud computing capabilities at the edge of the network,so that user equipment can access computing resources with low latency,and improve backbone network congestion.Therefore,the MEC system interconnected by the metro optical network is an integration of innovative network architecture and technology,which can provide users with real-time computing capabilities at the network edge.To reduce latency,improve system resource utilization,and achieve optimization of QoS and system performance,it is necessary to design a reasonable and efficient task scheduling and resource allocation solution.This paper takes the novel 5G application architecture of the MEC system interconnected by metro optical network as the background,aiming at the phenomenon that the computing capability of the nodes does not match its workload,and researches the joint task offloading and resource allocation strategy for overloaded MEC server to the light-loaded MEC server.The main work includes:Firstly,the system architecture and resource management characteristics are studied.Secondly,the server overload problem in the MEC system and the solution of task peer offloading are introduced,and the mathematical modeling is established.Then for small-scale networks,with the goal of minimizing the average latency,the MINLP model of joint offloading decision and resource allocation was formulated.Finally,a latency-aware joint task offloading and resource allocation heuristic based on the Genetic Algorithm framework was developed.The algorithm is divided into two parts:task peer offloading decision and resource allocation.The goal is to achieve joint optimization of latency performance and system resource utilization.The simulation results show that the MINLP model can achieve the best latency performance and QoS of the system within the constraints,while the heuristic algorithm can achieve near-optimal latency and blocking performance,as well as effectively optimize the resource utilization.