An insect inspired approach for optimization of tasks scheduling in computational grids

Abstract

The article suggests a novel optimization algorithm named Lepidoptera butterfly approach (LBA) that is inspired from the behavior of insects, American butterflies and their counterparts. This algorithm keenly observes the behavior of the Lepidoptera insects and tries to find an optimal solution through a larger solution space. The proposed algorithm LBA mimics the behavioral aspects of these insects. The insects (butterflies) migrate more often from one land to another in search of food particles and reproduction of offsprings. If they find the food particles and climate of the new land suitable, these insects often reproduce their offsprings in this new land. Hence, the suggested approach classifies the network of grids into two subnetworks (or two different lands) and thereby, generates two sub-populations. The algorithm then considers each individual subnetwork and their subpopulations. In our case, these are the jobs that contribute to each subnetwork and the offsprings that are reproduced are called as tasks. Our algorithm finds the best tasks and best jobs in every subnetwork and finally combines them and try to allocate the tasks/jobs to resources considering the constraints like cost and make-span time. However, this scheduling of tasks is considered as an NP-Complete problem. The algorithm is tested using 30 runs for simulation under these two constraints. This article makes a comparison with different existing optimization techniques like GA, TLBO, etc. The results signifies that our proposed approach of LBA performs better as compared to others. This inspires the authors to study the performance behavior of this approach in optimizing the scheduling problem in a computational grid environment under the constraints like time and cost.