摘要:云计算和移动互联网的不断融合，促进了移动云计算的产生与发展.在移动云计算环境下，用户可将工作流的任务迁移到云端执行，这样不但能够提升移动设备的计算能力，而且可以减少电池能源消耗.但是不合理的任务迁移会引起大量的数据传输，这不仅损害工作流的服务质量，而且会增加移动设备的能耗.基于此，提出了基于延时传输机制的多目标工作流调度算法MOWS-DTM.该算法基于遗传算法，结合工作流的调度过程，在编码策略中考虑了工作流任务的调度位置和执行排序.由于在用户不断移动的过程中，移动设备的无线网络信号也在不断变化，当传输一定大小的数据时，网络信号越强则需要的时间越少，从而移动设备的能耗也越少.而且工作流结构中存在许多非关键任务，延长非关键任务的执行时间并不会对工作流的完工时间造成影响.因此，在工作流调度过程中融入了延时传输机制DTM，该机制能够同时有效地优化移动设备的能耗和工作流的完工时间.仿真结果表明，相对于MOHEFT算法和RANDOM算法，MOWS-DTM算法在多目标性能上更优.



Abstract:The integration between cloud computing and mobile Internet promotes the development of mobile cloud computing. The tasks of workflow can be migrated to cloud that can not only improve the computing capacity of mobile device, but also reduce the energy consumption of battery. However, a great amount of data transmission introduced by using unreasonable tasks scheduling strategies can damage the QoS (quality of service) of workflow and increase the energy consumption of mobile device. In this paper, a multi-objective workflow scheduling is proposed based on delay transmission mechanism (MOWS-DTM) to optimize execution time of workflow and energy consumption of mobile device in mobile cloud computing environment. MOWS-DTM, derived from genetic algorithm, is combined with the process of workflow scheduling and takes both task scheduling location and execution sequence into consideration in coding strategy. When mobile user is moving, wireless network signal of mobile device is changing with the pace of different location. The stronger the network signal, the less time it takes to transmit data with fixed size, and the less energy the mobile device will consume. Moreover, there are many non-critical tasks reside in workflow, and increasing their execution time will not affect the makespan of workflow. Therefore, the delay transmission mechanism (DTM), incorporated in the process of workflow scheduling, can optimize the energy consumption of mobile device and the makespan of workflow simultaneously. Simulation results demonstrate significant multi-objective performance improvement of MOWS-DTM over the MOHEFT algorithm and RANDOM algorithm.



PDF全文下载地址:

http://jos.org.cn/jos/article/pdf/5479