A novel task scheduling algorithm for real time systems
Pankaj Kumar, Kirti Sharma · 2013
Earlier Deadline First and Rate Monotonic have been proved to be optimal scheduling algorithms. EDF works efficiently in underload condition and RM works efficiently in overload condition. Whereas D_R_EDF algorithm combines the advantages of both EDF and RM algorithms. It has been found that EDF, RM and D_R_EDF algorithms fail to work for incoming periodic tasks not satisfying the constraints imposed by them. A new algorithm capable of scheduling any type of incoming periodic task is proposed. The algorithm presented in this work is an improved version of D _R_EDF algorithm as it is capable of migrating the task(s) to the co-processor which are otherwise missed by the EDF, RM and D_R_EDF algorithms. The new algorithm is found to be performing successfully for all the periodic tasks with and without the constraints of the existing algorithms.