Fuzzy-based gang scheduling approach for multiprocessor systems

Different types of task scheduling problems have been formulated due to their importance in operating systems design. However, the problem has been shown to be computationally difficult and it is not possible to obtain optimal solutions. Therefore, more attention has been paid to formulate approximate and heuristic algorithms during recent years. This paper presents an attempt to adopt a modified scheduling algorithm suitable for multiprocessor systems by applying soft computing concepts. It suggests a modified approach for multiprocessor scheduling by combining gang scheduling and fuzzy logic decision maker. The fuzzy decision maker uses incomplete information about current states of the jobs and multiprocessor system to compute a new priority for each runnable task. The proposed scheduling approach is tested on several case studies, and computational results are compared to those obtained with other approaches. The given results proof that the proposed approach achieves the best performance including the computational efficiency, the average waiting time, and the average turnaround time.