一种多处理器平台上的传感器事务调度算法

如何有效地调度传感器事务以维护数据的时态一致性是信息物理融合系统研究中的一个重要问题。已有的调度算法基本上都是针对单处理器平台来设计的。提出一种多处理器平台上的传感器事务调度算法,算法通过合理地分配和调整事务实例执行所需处理器资源来保证数据的时态一致性约束,通过预先计算出全局重复调度序列来降低运行开销,给出了算法的可调度性分析。实验结果表明,该算法具有较高的调度成功率,其产生的更新负载也较低。

A sensor transaction scheduling algorithm on multiprocessor platforms

How to schedule sensor transactions to maintain data validity is an important research topic for cyber physical systems. Previous studies are generally restricted to uni-processor platforms. We therefore propose a sensor transaction scheduling algorithm on multiprocessor platforms. The algorithm appropriately allocates and adjusts the processors’ resources for each update instance to satisfy temporal consistency constraints. It calculates the global repeating schedule segment off-line in advance to reduce the runtime overhead. The schedulability analysis of the algorithm is also given. Experimental results show that the proposed algorithm performs well in terms of scheduling success ratio and update workload.