基于任务间依赖关系的小小区协作卸载策略  

Collaborative Computation Offloading Exploring Task Dependencies in Small Cell Networks

作　　者：康曼聪 李曦[1] 纪红[1] 张鹤立 KANG Man-cong;LI Xi;JI Hong;ZHANG He-li(School of Information and Communication Engineering,Beijing University of Posts and Telecommunications,Beijing 100876,Chin)

机构地区：[1]北京邮电大学信息与通信工程学院,北京100876

出　　处：《北京邮电大学学报》2021年第1期72-78,共7页Journal of Beijing University of Posts and Telecommunications

基　　金：国家自然科学基金项目(61771070)。

摘　　要：在密集部署的小小区网络中,考虑到小小区基站(SBS)的计算资源有限,提出了基于任务间串并依赖关系的协作卸载策略,以降低计算卸载任务的整体完成时延.首先,考虑将可以同时执行的并行任务卸载至不同的SBS,利用计算资源的分布式特点来降低整体时延,同时最大化单个SBS上的串行任务数量,以减小所需SBS的数目;然后,根据网络的负载均衡情况对2种场景进行讨论,联合考虑任务间的依赖关系、不同SBS的可用计算资源量和SBS与用户间的信道质量,分别引入最长路径理论和图着色算法以确定最佳任务卸载方案.仿真结果表明,与已有策略相比,所提策略可降低计算卸载任务的整体完成时延.In dense small cell networks,a task-dependency-based collaborative offloading scheme is proposed to cope with the limited computational resources in small cell base station (SBS),which can further reduce the total execution delay. Firstly,the parallel tasks are offloaded to different SBSs to reduce computing delay,while sequential tasks are offloaded to the same SBSs to reduce the number of required SBS. Then,based on whether different SBSs have the same load pressure or not,the optimal offloading strategies are proposed in two different scenarios by introducing the longest path theory and the graph coloring theory,respectively. The strategies jointly considerate the user energy constraint,dependencies among tasks,the amount of available computational resources and channel conditions of different SBSs.Simulation shows that the proposed strategies can reduce total execution delay compared with existing strategies in both scenarios.

关 键 词：小小区网络 移动边缘计算 协作卸载 时延优化 

分 类 号：TN929.5[电子电信—通信与信息系统]