A hybrid real-time scheduling mechanism based on multiprocessor for real-time tasks in weakly hard specification

Abstract

In real-time systems, there are three categories which are based on the “seriousness” of missing a deadline, which are hard, soft, and weakly hard. Real-time scheduling algorithms proposed for use can guarantee a bounded allowance of deadline misses in a predictable way dedicated to weakly hard real-time tasks. A number of studies from previous research on multiprocessors in scheduling algorithms for weakly hard tasks in real-time systems used non-optimal heuristics, wherein these cannot guarantee that an allocation of all tasks can be feasibly scheduled. Moreover, the use of a hierarchical scheduling algorithm under the PFair algorithm may cause high scheduling overhead due to frequent preemptions and migrations. This research is done to address the problem of optimization in partitioned scheduling and task migration in global scheduling, that causes scheduling overheads. Therefore, to achieve these objectives, this study proposes a hybrid scheduling mechanism that uses the partitioning and global approaches, which are R-BOUND-MP-NFRNS and RM-US (m/3m-2) with the multiprocessor response time test. Based on the simulation results, when comparing the hybridized scheduling approach and R-BOUND-MP-NFRNS, it is seen that the deadline satisfaction ratio improves by 2.5%. In case of the proposed approach versus multiprocessor response time, the deadline satisfaction ratio has seen an improvement of 5%. The overhead ratio for the proposed hybrid approach versus R-BOUND-MP-NFRNS has reduced by 5%, and in case of the proposed hybrid approach versus multiprocessor response time, it reduces by 7%. According to the results, it can be seen that the proposed hybrid approach achieved a higher percentage in the ratio of deadline satisfaction, and minimized its overhead percentage when compared to the other approaches.