A shared resource-aware real-time task allocation algorithm

In order to improve the efficiency of task allocation under partitioned fixed priority scheduling in multi-core real-time systems, we quantitatively analyzed the task blocking time and schedulability losses resulted from FIFO-spin mechanisms. Subsequently, we presented a new task correlation analysis to evaluate the correlation between tasks on different cores, based on which we proposed a novel shared resource-aware task allocation algorithm. The proposed algorithm includes task-grouping policy and task group-splitting policy. The task-grouping policy chooses the tasks that involve in resource contentions to compose relative sub-task-sets, and co-locates as many such sets as possible, so as to avoid remote blockings; while the task group-splitting policy splits the sub-task-sets that can not be allocated to the same core based on the task correlation analysis, and allocates the tasks split from the original sub-task-sets to the least-loaded core, so as to reduce spin-related schedulability losses. Experimental evaluation results indicate that the proposed algorithm outperforms the similar algorithms in terms of both the task-set acceptance ratio and the spin-related (schedulability) losses.