Expected schedulability based on stochastic model for tasks in soft real-time system
基于随机模型的软实时系统的任务期望可调度性

By studying the characteristics of the task-schedulability of stochastic tasks in a soft real-time system, we develop the expected schedulability conceptually. The expected schedulability is closely related with the arrival time of a real task; hence the proposed method can be applied to investigate the schedulability of a sub-set of tasks in any given time-interval. The condition for expected schedulability is given. If the duration of a real task satisfies this condition, it is schedulable. Because simulation results show a good agreement with analytical ones, we confirm that the real tasks are always schedulable when the loading factor of the soft real-time system is smaller than 69%(as for some deterministic model). This indicates that the proposed method of expected schedulability based on the stochastic model can provide a large threshold value and a better adaptability for soft real-time systems.