A hybrid immune genetic algorithm for scheduling in computational grid

Abstract

Computational grid (CG) provides a wide distributed platform for high end compute intensive applications. Job scheduling in a computational grid is an activity in which the submitted jobs are assigned on the nodes of the grid with the objective to optimise some characteristic parameter. One of such important characteristic parameters is makespan. Makespan depends on the number of jobs, number of nodes and job execution time. As such, scheduling problem is NP-Hard therefore soft computing techniques are often applied to get an optimal schedule. Genetic algorithm (GA), a search procedure based on the evolutionary computation, is often applied to solve a large class of complex optimisation problems. Immune genetic algorithm (IGA) is a hybrid approach which incorporates the features of immune algorithm in GA. GA takes longer to converge and also has the possibility to get trapped in local optima nearing convergence. IGA is more efficient as it explores the solution space based on the affinity, an entropy based property. This work applies IGA to schedule the submitted jobs on the grid nodes for the optimal makespan. Simulation of the proposed model is done to evaluate its performance and it exhibit effective result.