A New Hybrid Optimization Technique for Scheduling of Periodic and Non-periodic Tasks

Abstract

This article addresses a renowned issue of allocating periodic tasks to a network of heterogeneous processors in distributed computing systems (DCS) subject to timing constraints, tasks precedence, and arbitrary communication among them, in order to lessen the overall busy time whereas guaranteeing the tasks deadlines. A new hybrid optimization (NHO) technique is introduced, a fusion of k-mean clustering (KMC) and Branch-and-Bound (B&B) method for reducing overall normalized busy time (NSBT) of the system. This technique is stationed on B&B method in which each branch grants scheduling solution. K-mean clustering (KMC) technique has been utilized to reduce the complexity of B&B technique by pruning the branches those do not lead feasible solution. A specialized case of non-periodic tasks allocation issue is also studied in this work. The problem is intractable in nature. Finally, a demonstrative example and comparison with some computational experiences are presented. Experimental results reveal that proposed technique achieves better proficiency than other existing techniques in literature. This model is advisable for arbitrary number of processors and tasks.