Abstract:

Real-time systems require the task to get the correct results before its deadline in the worst-case. If the deadline is exceeded, the task is considered a wrong behavior. Improvement in task scheduling analysis and the task-sets schedulability is particularly important. Generalized scheduling can integrate the advantages of fixed priority scheduling, prevent tasks from unnecessary preemption, reduce additional memory usage, and improve the schedulability of task-sets. But the method of task schedulability analysis is too rough, which can impact worst-case response time analysis of the task and reduce the task-sets' schedulibility. For these problems, we re-establish a task model based on generalized scheduling by increasing the number of task's running stage and propose a method for improving the schedulability by assigning task's preemption threshold, adjusting preemption thresholdand length of task running stage and optimizing the block tolerance of the task. Finally, experimental results show that the proposed scheduling algorithm can effectively improve the feasibility of the task-sets.

Key words: fixed-priority, preemption threshold, running stage, improved generalized scheduling, feasibility