Abstract:

In order to solve the problem that embedded real-time systems' miss rate of real-time task is too high and the problem of poor stability of real-time scheduling while the load of the system is uncertain, a scheduling model based on feedback control was proposed. This model consisted of an improved multi-level queue scheduler and three controllers, those were access controller, execution level controller and Proportional-Integral-Derivative (PID) controller. The deviation of task miss rate was feedbacked to the PID controller and generated a corresponding adjustment, the adjustment acted on the other two controllers to adjust the level of real-time tasks. Tasks adjusted were executed by the scheduler. After structural adjustment and improvement, the scheduling model was applied in embedded Configurable operating system (eCos). The experimental results show that the proposed method can reduce miss rate of tasks, and solve the overload problem of the system.

Key words: embedded real-time system, real-time scheduling algorithm, feedback control, processor utilization, task miss rate

摘要：

针对嵌入式实时系统在系统负载模型不确定的情况下系统的实时任务错过率过高以及调度稳定性差的问题,提出了一个基于反馈控制的调度模型.该模型主要由改进的多级队列调度器和3个控制器(准入控制器、执行等级控制器、比例积分微分(PID)控制器)组成.任务的错过率偏差反馈到PID控制器,产生相应的调整量并作用于其他两个控制器,对实时任务的执行等级进行调整, 经过调整的任务被调度器调度执行.在对模型中每个部件进行了一些结构调整和改进设计之后,将该调度模型应用于嵌入式可配置操作系统(eCos)中.实验结果表明,该模型降低了任务的时限错过率,同时解决了系统频频超载的问题.

关键词: 嵌入式实时系统, 实时调度算法, 反馈控制, 处理器利用率, 任务错过率