实时任务的选择性压缩

Compressing real time tasks selectively

实时任务的动态调度仍然是一个值得研究的问题,特别是在一些新的应用背景下.当一个实时系统已经处于满负荷运行,而又有新任务需要插入或老任务需要加速时,就要有老任务降速(受压)以出让带宽.即使如此,什么时刻能开始这一调整是一个关键问题.关于这一调整时刻的计算,现有研究成果都只是基于一个受压任务给出的.基于EDF(Earliest Deadline First)调度算法,提出了在多个可压缩任务中进行选择的思想,给出了相应的选择性压缩算法,引入了立即点百分比进行性能评价.基于这一思想,快速完成插入或加速的概率明显增加.

Dynamic scheduling of real-time tasks still needs to be studied,especially for some new aplications.For a fully loaded real-time system at run time,in order to insert new tasks or accelerate old ones,it is necessary to decelerate（compress）some task to get freed bandwidth.Even though,it is crucial to get the time the adjustment starts.This adjusting time is evaluated on a single compressed task in present documents.Based on the EDF（Earliest Deadline First） algorithm,this paper gives out an new idea to select a task from a number of compressed ones.A corresponding selective compressing algorithm is provided,and the ratio of immediate adjustment is introduced for performance assessment.This scheme increases the probability for quick insertion or acceleration obviously.