摘要
为了减少软件容错模型中实时调度算法的抢占次数,提出了一种部分抢占调度算法(PPA),该算法不仅考虑了如何尽可能多地执行主部分,还考虑了如何减少抢占次数,采用了类似非抢占最早时限优先算法(EDFA)来调度主部分.对不同CPU利用率和软件错误概率的任务集合进行模拟实验,结果表明,PPA算法在可以获得与目前所知的同类算法近似调度性能的同时,还可以在一定情况下极大地减少任务调度间的抢占次数,从而减少了系统中因抢占次数过多带来的额外运行时调度开销等负面因素.
A scheduling algorithm called partial-preemptive prediction algorithm(PPA)was proposed in the software fault-tolerant model to reduce the preemption. PPA considers how to reduce the preemption during scheduling as well as how to execute the primaries as many as possible. PPA uses a method similar to non-preemptive earliest deadline first algorithm(EDFA) to schedule the primaries. Simulations on task sets with different CPU utilization and fault possibility show that PPA can obtain the similar scheduling performance as well as the well-known algorithms so far. Moreover, PPA can reduce the preemption dramatically than previous algorithms under certain conditions, thus reduces the negative impact introduced by the preemption such as overhead runtime computation time.
出处
《浙江大学学报(工学版)》
EI
CAS
CSCD
北大核心
2009年第6期1047-1052,共6页
Journal of Zhejiang University:Engineering Science
关键词
硬实时系统
软件容错
调度算法
抢占
最早时限优先算法
hard real-time system
software fault-tolerant
scheduling algorithm
preemptive on
EDFA