一种面向移动云计算的多目标任务卸载算法

Multi-objective task offloading algorithm for mobile cloud computing

计算能力和资源受限的移动设备可将待处理的密集型任务卸载到云端执行,从而增强移动设备的计算能力并减少电池能源消耗(EC)。然而,现有研究在卸载任务时不能较好地均衡移动端的应用完成时间(FT)和EC。提出了基于分解的多目标进化算法(MOEA/D)来同时优化应用FT和EC,并将动态电压频率调整技术引入MOEA/D中,在不增加应用FT的前提下,调节移动设备的CPU时钟频率以进一步降低移动设备的EC。仿真结果表明,与多个算法相比,所提出的算法在多目标性能上更优。

Mobile devices with limited computing power and resources can offload intensive tasks to the cloud for execution,thus improving the computing capacity of mobile devices and reducing battery energy consumption.However,the existing researches cannot properly balance the application finish time and energy consumption of the mobile terminal when offloading tasks.An MOEA/D based algorithm was proposed to optimize the application finish time and energy consumption,and dynamic voltage frequency scaling technology was introduced into the MOEA/D to adjust the CPU clock frequency of mobile devices to further decrease the energy consumption without increasing the application finish time.The simulation results demonstrate that the proposed algorithm outperforms a number of existing algorithm in terms of the multi-objective performance.