一种改进的自适应多媒体任务调度算法

An Improved Adaptive Task Scheduling for Multimedia Applications

支持多媒体任务调度以满足其性能需求,是一项重要而富有挑战性的工作,一直备受关注,并出现了一些实时任务模型。它们都需要任务提供最坏执行时间(Worst Case Execution Time,WCET),以方便准入控制机制的实现,但这正是多媒体任务难以提供的。那么在WCET未知的前提下,如何实现多媒体任务的调度,而且必须支持准入控制和动态QoS控制机制,支持尽可能多任务的执行,使CPU资源的利用最大化?本文首先提出了一种改进的基于速率的自适应(Adaptive Rate-Based,ARB)任务模型。然后通过理论分析和实验证明了:在WCET未知的情况下,基于ARB任务模型的多媒体任务调度算法、准入控制和自适应QoS控制机制是可行的、有效的,而且可以支持尽可能多任务的执行,达到了预期的目标。

Much research has been done on guarantee the OoS of multimedia systems, and several real-time task models have been studied. But the Worst Case Execution Time （WCET）, which is hard to be provided for multimedia tasks, is needed in these models to make admission controlling possible. Thus, how to effectively schedule multimedia tasks with admission controlling and self-adaptive QoS controlling and more fully resource utilization if WCET is unknown. First, an improved task model, called Adaptive Rate-Based task model （ARB） with WCET unknown, is presented. Then through theoretical analysis and simulation experiment, it is demonstrated to be feasible and effective to schedule multimedia tasks with admission controlling and self-adaptive QoS controlling based on ARB task model.