Research On Optimization Of Edge Computing Task Offloading For Emergency Scenario

Mobile Edge Computing(MEC)is an emerging technology that uses a small cloud computing platform deployed at the edge of a mobile network to support resource-intensive but latency-sensitive applications.However,the existing MEC technology is not suitable for the explosive increase in the number of Mobile Users(MU)or the sparse distribution of network facilities.The general MEC scheme cannot fully handle the user emergency communication and task offloading requirements in the post-disaster emergency communication network.Unmanned Aerial Vehicle(UAV)can play an important role in wireless systems,because it can be flexibly deployed to quickly help disaster areas improve signal coverage and restore communication quality,so that users in disaster areas can normally offload tasks;mobile Ad＿hoc networks is a wireless distributed network composed of multiple independent mobile communication nodes.It can quickly build a mobile communication network at any time and any place without the support of hardware infrastructure network facilities,so it can be applied The network communication in the post-disaster area is quickly restored,and the normal task offloading of users can be guaranteed within a certain period of time.However,the above two methods of assisting MEC to restore post-disaster emergency communication still face many challenges,such as UAV flight trajectory planning and Ad＿hoc network routing planning.Therefore,it is of great significance to study how to plan UAV flight trajectories or Ad＿hoc routing and other auxiliary MEC fast networking in emergency scenarios to shorten the delay while ensuring the quality of service.This thesis studies the deployment of UAV-MEC system and Ad＿hoc-MEC system respectively under the emergency situation of network communication interruption in the region after the disaster,and studies the task unloading scheduling optimization strategy of the two systems in the edge environment,in order to minimize the total delay of task unloading in the system.In UAV-MEC system,we comprehensively consider the flight trajectory of UAV and the task unloading decision of each MU.In the Ad＿hoc-MEC system,we comprehensively consider the routing assignment of each MU and the routing planning of Ad＿hoc.We have deeply analyzed and explored the optimization problems of the two systems,and the following are the specific contents:In the post-disaster scenario of the UAV-MEC system,this thesis aims to minimize the total delay of user task offloading in the UAV-MEC system as an optimization problem.According to the relevant advantages of UAV and MEC technology,we constructed a UAVMEC system model applied in emergency relief scenarios,and defined the target optimization problem as the system total delay when minimizing the problem.Due to the non-convex nature of the problem,there is a high-dimensional state space and continuous action space,and it is difficult to find an optimal solution.This thesis proposes a Split DQN(SDQN)algorithm based on Reinforcement Learning(RL),and uses this algorithm to solve optimization problems.The results of simulation experiments show that compared with the benchmark algorithm and other traditional RL algorithms,the proposed SDQN algorithm can be closer to the global optimal solution of the optimization problem,significantly improve the system decision-making,and thus reduce the total system delay.In the post-disaster scene of the Ad＿hoc-MEC system,the same optimization problem is aimed at minimizing the total delay of user task offloading in the system.According to the outstanding advantages of MEC and Ad＿hoc network,we reasonably proposed and constructed the Ad＿hoc-MEC system model applied in emergency scenarios.Constrained by discrete variables and user communication scope,our goal is to minimize system latency by jointly optimizing user service order,Ad-hoc routing information.Considering that the optimization problem is an NP-hard problem,traditional optimization algorithms are difficult to solve.Based on this,this thesis also uses the SDQN algorithm.Finally,the results of simulation experiments prove that our proposed SDQN algorithm can solve the optimization problem of minimizing the total delay of the system better than other RL algorithms in the Ad＿hoc-MEC system,and effectively meet the timeliness of task offloading in emergency scenarios.