针对分布式控制系统中的周期性实时任务,基于版本复制技术设计一种新的容错调度算法.该算法将主动副版本技术、被动副版本技术和副版本重叠技术融合起来,充分利用它们的优点.算法通过延迟副版本的开始执行时间减少主/副版本执行时间的重叠,从而减少副版本的冗余度并提高处理器的利用率.给出了任务集容错可调度性的判别定理和副版本冗余时间的计算方法.在采用以最小化"最坏情况响应时间"的最佳适应方法和首次适应任务分配方法的基础上,给出启发式任务分配算法.最后对算法进行了仿真实验,实验结果表明了算法的有效性和可行性.
For periodic real-time tasks in distributed control system,a novel fault-tolerant scheduling algorithm is presented.The algorithm fuses active backup-copy technique and passive backup-copy technique and overlapping backup-copy technique in order to take use of their advantages.By differing the time when the backup copy begin running,it reduce overlapping time of primary and backup copies of a task in order to decrease redundancy of backup copy and improve the utilization of processor.The theory for judging schedulability of task set and the method for computing redundant time of backup copy are given.Based on minimizing "worst case response time" and first-fit method,the heuristic task assigning algorithm is given.Finally,simulation experiments are carried out and results prove the feasibility and effectiveness of the algorithm.