Research On Task Offloading And Resource Allocation In Collaborative Edge Computing Network

In recent years,the prevalence of 5G and 6G industries has made the Narrow-Band Internet of Things industry a good time,and the ecology of the Internet of Everything has become a common practice.A large number of smart products and smart applications need to rely on servers with computing power for data processing.However,accessing the cloud for auxiliary computing can solve the problem of insufficient local computing power,it cannot meet people's tolerance for delay and energy consumption.Therefore,edge computing emerged as the times require.Edge computing technology will offload the operation of auxiliary computing to the cloud and sink to the user edge,thereby reducing the delay and energy consumption caused by the long transmission path.However,after the introduction of edge computing,we have to admit that the computing power of the edge is indeed smaller than that of the cloud,and that the offloading computing demand of users in this area served by the edge server is too large.In order to ensure that all users can perform offloading operations,the edge server will not provide all computing power to a single user.This leads to the fact that a single edge server may not be sufficient for computing-intensive users to perform independent offload computing.At this point,the model of collaborative edge services kicked off.Therefore,this thesis studies the computational offloading decision and optimal resource allocation scheme for two user request task models in collaborative edge computing networks.For the case where a single user independently issues a computing offload task request,it is suitable for computing tasks required by users in general scenarios.First,the internal components of the user task should be divided,and the internal components of the task should be constructed in the form of a Task Call Graph with random coupling before and after.Secondly,the edge servers should be divided into request edge servers and collaborative edge servers.At this time,the uninstallation of user components There are three locations:the local side,the requesting edge server side,and the collaborative edge server side.Then,according to the division of task components and the offloading position of components,the execution delay and energy consumption of the task are comprehensively considered,and 9 offloading situations are analyzed and summarized.According to the characteristics of time delay,unified modeling of offloading decision and resource allocation is carried out,an optimization function is established,and the two parts of resource allocation and offloading decision are decoupled through mathematical proof,and the solution of resource allocation is solved by using convex optimization method.A search algorithm based on^*is proposed,which can be automatically adjusted by rewriting its heuristic function part,so that the results can reach the optimal faster due to the adaptive heuristic search framework to solve the offloading decision;Finally,the model is extended from two edge servers to n edge servers.At this time,the offloading positions are changed from 3 to n+1,and the search space is also increased exponentially.The solution can also be completed using the algorithm framework proposed in this thesis.The algorithm is compared with other traditional offloading algorithms,which ensures the feasibility of the model proposed in this thesis and the high quality of the algorithm.For the situation where there is a correlation between the request tasks sent by two users,it is suitable for situations where the execution results of others are required to perform task calculation,such as two-person VR games,and the correlation between getting off the car and the car in smart traffic scenarios.First of all,the components of the respective tasks are divided,and they are also constructed in the form of a task call graph to adapt to random and changeable real-world scenarios.The analysis considers 4 different execution positions between the dependent components of two users,and then combines the execution of the internal components of 9 individual user tasks to model the problem,and analyzes the offloading decision and resource allocation.By adjusting different weights,it can adapt to different user groups,decouple the optimization function,and independently consider resource allocation and unloading decisions.Then the weighted value of delay and energy consumption is minimized and solved by Lagrange Multiplier Method.Then,by improving the adaptive heuristic search framework proposed in this thesis,the offloading decision problem is searched and solved,so as to obtain the optimal offloading decision.Finally,it is compared with other search algorithms to prove the feasibility of the model and the effectiveness of the improved adaptive heuristic search algorithm framework.