云边协同系统中基于博弈论的资源分配与任务卸载方案  

Game-Based Resource Allocation and Task Offloading Scheme in Collaborative Cloud-Edge Computing System

作　　者：吴学文[1] 廖婧贤 Wu Xuewen;Liao Jingxian(School of Computer and Information,Hohai University,Nanjing 211100,China)

机构地区：[1]河海大学计算机与信息学院,江苏南京211100

出　　处：《系统仿真学报》2022年第7期1468-1481,共14页Journal of System Simulation

摘　　要：综合考虑时延、能耗和计算资源成本,构建云边协同系统中的效用最大化问题,并将其分解为计算资源分配、上行功率分配和任务卸载策略三个子问题。提出一种基于博弈论的资源分配和任务卸载方案(game-based resource allocation and task offloading,GRATO)以分别解决上述子问题。利用凸优化条件求得计算资源分配最优解;设计一种低复杂度的上行功率分配方法用于降低无线干扰;针对任务卸载策略优化问题,提出一种基于博弈论的分布式任务卸载算法(gamebased distributed task offloading algorithm,GDTOA)。仿真结果表明,GRATO方案在时延和能耗方面的性能优于其他方案,还可以感知用户的优先级,使紧急用户具有更高的效用和更低的时延。Considering the delay,energy consumption and computing resource cost,the utility maximization problem in collaborative cloud-edge system is constructed,and divided into three subproblems:computing resource allocation,uplink power allocation and task offloading strategy.A game-based resource allocation and task offloading(GRATO)scheme is proposed to solve those subproblems.The optimal solution of computing resource allocation is obtained by using convex optimization conditions;a low complexity uplink power allocation method is designed to reduce wireless interfere;a game-based distributed task offloading algorithm(GDTOA)is proposed to optimize the task offloading strategy.Simulation results show that the performance of GRATO is better than other schemes on delay and energy consumption,and it can sense the priority of users,resulting in higher utility and lower latency for emergency users..

关 键 词：边缘计算 资源分配 计算任务卸载 博弈 效用最大化 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] TP391[自动化与计算机技术—计算机科学与技术]