生产环境下基于遗传算法的实时任务卸载算法

Real-time task offloading algorithm based on genetic algorithm in production environment

移动边缘计算(MEC)模式的出现使得移动设备(MD)可以将移动应用任务卸载至部署在网络边缘的演进型基站(eNB)上执行,从而有效降低MD的能耗与任务的延时。然而,不合理的任务卸载同样会造成MD的高能耗和任务的高延时。基于此,提出了一种实时任务卸载(RTO)算法,该算法能够在满足任务截止时间的限制条件下,最小化MD能耗。RTO在遗传算法(GA)的基础上,引入了动态电压调节技术(DVFS),并在编码策略中考虑了任务的执行位置和MD的计算频率。通过仿真实验,证明了RTO算法的可行性与有效性。

The emergence of Mobile Edge Computing(MEC)mode enables Mobile Device(MD)to offload mobile application tasks directly to evolved NodeBs(eNBs)deployed at the edge of the network over the wireless network,thus effectively reducing MDs′energy consumption and task latency.However,unreasonable task offloading can also cause high energy consumption of MD and high latency of tasks.Based on this situation,a Real-time Task Offloading(RTO)algorithm was proposed,which was capable to minimize the energy consumption of MD while satisfying the constraints of task deadline.Based on the Genetic Algorithm(GA),the Dynamic Voltage Frequency Scaling(DVFS)was introduced by RTO,and the execution position of the task and the computation frequency of MD were taken into account in the coding strategy.The simulation experimental results verified the feasibility and effectiveness of the RTO algorithm.