一种动态传感器事务调度算法

A Dynamic Sensor Transaction Scheduling Algorithm

实时数据库系统必须合理调度执行传感器事务以维护实时数据对象的有效性.现有算法不能很好地解决最坏执行时间远大于平均执行时间时的事务调度.提出一种动态传感器调度算法DS-FP-DA.算法预先确定各事务在DS-FP调度下的预留时间.在系统运行过程中,算法通过接纳控制机制来选择合适的实例执行.实验表明,算法能有效降低数据的加权平均无效时间.

In real-time database systems, sensor transactions should be effectively scheduled to maintain the temporal validity of real-time data objects. The algorithms proposed so far are not suitable for the cases in which the worst case execution time is much larger than the average execution time. A dynamic scheduling algorithm called DS-FP-DA is proposed to solve the problem. The CPU time preserved for each instance in the DS-FP schedule is computed at first, and then an admission control scheme is used at running time to choose the incoming instances for execution. Experiments show that DS-FP-DA can significantly reduce the weighted average invalid time of real-time data objects.