An Efficient Dynamic Scheduling of Tasks for Multicore Real-Time Systems

Dynamic scheduling of tasks for multi‐core real‐time systems based on optimum energy and throughput

“…The proposed scheduling scheme also finds minimum power levels and execution times for all jobs that are sufficient enough to allocate the jobs into cores. The normalised energy of the system as calculated using the geometric mean formula is ∼16,660 MJ and hence we may consider that model as per work in [18] will consume energy of 16,660 MJ to allocate the above tasks into appropriate cores. Therefore, we may say that the proposed model will save ∼2% of energy in average cases comparable to the model proposed in [18].…”

“…Hence our proposed model shows that almost 98% of all the new random tasks are accommodated into the system at a fixed time period. In [18], it has been shown that the CPU utilisation (i.e. allocation of tasks to cores) of the scheduling scheme is very high comparable to the Rate Monotonic (RM) scheduling and EDF scheduling and 95% of the random new tasks with different periods and execution times are accommodated in the system at a fixed time period.…”

“…In [18], the authors propose a dynamic scheduling model where new tasks are stored in the global queue and are divided into a number of jobs using the pfair concept. Each job is associated with pseudo‐release time and pseudo‐deadline.…”

Reinforcement learning based multi core scheduling (RLBMCS) for real time systems

“…Task migration is used to migrate the rejected jobs from small cores into big cores through dispatcher. K. Baital et al [14] presented a scheduling algorithm where random tasks generated at different time intervals with different periodicity and execution time can be accommodated into a system, which is already running a set of tasks, meeting the deadline criteria of the tasks. Using the concept of Pfair scheduling, random new tasks have been divided to fit into the idle times of the different cores of the system.…”

“…In this algorithm Recurrent Neural Network has been utilized to find both the number of queues and the optimized quantum of each queue. K. Baital et al [18] presents a scheduling algorithm where random tasks generated at different time intervals with different periodicity and execution time can be accommodated into a system, which is already running a set of tasks, meeting the deadline criteria of the tasks. A hybrid limited-preemption real-time scheduling algorithm is derived in [19] by M. Bertogna et al, that aims to have low runtime overhead while scheduling all systems that can be scheduled by fully preemptive algorithms.…”