移动云计算环境下基于任务依赖的计算迁移研究

移动终端硬件的资源受限问题可以通过将本地计算任务迁移至云端来缓解。然而,相比远程云端,某些实时要求较高的复杂应用更适合迁移至微云。这类应用中各任务之间的依赖关系也会对迁移方案产生较大影响。结合任务之间的依赖关系及微云的特点,基于遗传算法思想提出一种计算迁移方法。根据不同微云处理不同类型任务时的能力,将微云进行类型划分。根据移动应用中不同任务之间时序与数据的双重依赖关系,结合能耗和响应时间的考量,设计一个计算迁移算法,以获取具有较优效用值的迁移方案。通过仿真验证了该方法的可行性和有效性。

COMPUTING MIGRATION BASED ON TASK DEPENDENCIES IN MOBILE CLOUD COMPUTING ENVIRONMENT

The resource-constrained problem of mobile terminal hardware can be solved by the migration of some computing tasks from the local to the cloud in the mobile cloud computing environment. However, compared with remote clouds, some complex applications with higher real-time requirements are more suitable for migration to micro-cloud. The dependencies between tasks in such applications have a great impact on the migration scheme. Combining the dependencies between tasks and the characteristics of micro-cloud, a computational migration method was proposed based on the idea of genetic algorithm. According to the ability of different micro-cloud to handle different types of tasks, the micro-cloud were classified. According to the dual dependence of timing and data between different tasks in mobile applications, we designed a computational migration algorithm to obtain a migration scheme with better utility values, combining with energy consumption and response time. The feasibility and effectiveness of this method are verified by simulation.