摘要
为了降低移动设备的能耗,移动应用可分解为多个互联模块,构建为任务交互图TIG结构,并卸载至云端资源上执行。多数已有研究仅在任务调度阶段以动态电压/频率调整手段进行能量优化,忽略了任务-资源映射阶段。基于此,在任务-资源映射阶段和资源频率分配阶段中同步进行能量优化,设计一种基于二次分派问题QAP的调度算法。任务-资源映射阶段中,计算最差全局松弛时间,并将其分配至整个TIG中的任务。资源频率分配阶段中,通过动态电压/频率调整,在不超过局部期限的同时减慢任务执行,实现能效优化。实验结果表明,在该算法下进行的任务卸载,平均可以为移动设备节省约35%的能耗。
To reduce the energy consumption in mobile devices,the mobile application can be divided into several interconnected partitions and construct as task interaction graph(TIG)to offload to cloud resources.Most of the existing studies only use dynamic voltage/frequency adjustment to optimize energy in the task scheduling stage,and ignore the task-resource mapping stage.Based on this problem,we attempt to apply energy efficiency optimization in both the task-resource mapping stage and the resource-frequency assignment,and propose a scheduling algorithm based on quadratic assignment problem(QAP).In the task-resource stage,we calculated the worst global slack time,and distributed it to the whole TIG.In the resource-frequency assignment stage,the algorithm achieved energy efficiency optimization by adjusting the dynamic voltage/frequency to slow down the task execution while not exceeding the local deadline.Experimental results show that the task offloading brought by our algorithm can save 35%energy in mobile device.
作者
孙赫勇
文勃
郑丹
周强
Sun Heyong;Wen Bo;Zheng Dan;Zhou Qiang(Dalian Branch Office of CRRC Qingdao Sifang Locomotive and Rolling Stock Co.Ltd.,Dalian 116000,Liaoning,China;Qingdao Branch Office of CRRC Qingdao Sifang Locomotive and Rolling Stock Co.Ltd.,Qingdao 266000,Shandong,China;Jiaozuo Branch Office of CRRC Qingdao Sifang Locomotive and Rolling Stock Co.Ltd.,Jiaozuo 454150,Henan,China;School of Software,Beijing Jiaotong University,Beijing 100044,China)
出处
《计算机应用与软件》
北大核心
2020年第1期8-14,52,共8页
Computer Applications and Software
基金
国家自然科学基金项目(61673049)
关键词
移动云计算
任务交互图
能效
任务-资源映射
调度
任务卸载
Mobile cloud computing
Task interaction graph
Energy efficiency
Task-resource mapping
Scheduling
Task offloading
