A memetic algorithm for reliability-based dynamic scheduling in heterogeneous computing environments

Abstract

Task scheduling in heterogeneous computing environments is one of the widely studied problems in the literature most of which target on static scheduling. There is a few number of work on dynamic scheduling where they mainly target on a single objective: minimization of the schedule length without considering the effect of resource failures. The aim of this paper is two-fold. First we propose a new unified objective of dynamic scheduling problem which considers both schedule length and reliability cost of resources. Then, we present a new memetic algorithm for dynamic scheduling of independent tasks with batch mode by considering our unified objective. The experimental results reveal that makespan and reliability cost could be minimized together according to the weight of the two objectives, and our algorithm outperforms the related work given in the literature according to the unified objective, for most of the test cases.