Importance-based scheduling for predictable real-time systems using MART

The paper proposes a method for improving the predictability of highly dynamic real time systems using Modification and Adjustment of Real time Tasks (MART) (N. Kosugi et al., 1994). We defined the predictability as “each task priority under constant conditions can be predicted”. Highly dynamic systems mean that the number of tasks and task characteristics in them are always changing. So it is extremely difficult to understand or predict the task set´s future states. A new parameter “importance” which indicates the semantic importance is introduced into each task and each task can always be assigned a priority which is possibly suitable for its importance during dynamic changing of the task set. Therefore, each task priority can be managed to a certain extent and as a result, the predictability defined in the paper can be improved. Finally, simulation results are also provided