一种有限优先级的静态优先级分配算法

An Assignment Algorithm of Static Priority for Limited Priority Levels

静态优先级调度在实时系统中得到了广泛应用.然而,静态优先级调度受到系统支持的优先级个数的限制.当任务的个数大于优先级个数时,需要将多个任务映射到同一个优先级.针对优先级个数有限的情况,给出了在截止期限大于周期时任务可调度的充分必要条件,并提出了基于有限优先级的静态优先级分配算法(AGP).AGP算法对于基本任务集合是最优的静态优先级分配算法.其最优性表现在,所需的优先级个数最小,并且若采用AGP算法不可调度某个任务集,则采用其他静态优先级分配算法也不可调度该任务集.模拟结果表明,AGP算法的可调度性要远远大于常量法. AGP算法对于解决在嵌入式实时系统中任务的优先级分配问题具有重要意义.

Static priority scheduling is widely used in real-time systems. But its schedulability will be reduced if priority levels of the system are insufficient. A task set may require more priority levels than the system can support. In this case, more than one task must be grouped into the same priority. This paper presents necessary and sufficient conditions for analyzing the schedulability of static priority algorithms on resources with limited priority levels. A static priority assignment algorithm (AGP) with limited priority levels is developed. As it turns out, AGP is optimal for the basic task set in the sense that the number of priority levels required by AGP is minimal and no other static priority rule can schedule a basic task set which cannot scheduled by AGP. Simulation results show that the schedulability of AGP is much higher than that of Constant Ratio Grid algorithm. AGP is significant for solving the problem of assigning priorities of tasks in embedded real-time systems.