Big Data

Gridification of Genetic Algorithm with Reduced Communication for the Job Shop Scheduling Problem

Date Added: Sep 2010
Format: PDF

This paper presents a parallel hybrid evolutionary algorithm executed in a grid environment. The algorithm executes local searches using Simulated Annealing within a Genetic Algorithm to solve the Job Shop Scheduling Problem. Experimental results of the algorithm obtained in the "Tarantula MiniGrid" are shown. Tarantula was implemented by linking two clusters from different geographic locations in Mexico (Morelos-Veracruz). The technique used to link the two clusters and conFigureure the Tarantula MiniGrid is described. The effects of latency in communication between the two clusters are discussed. It is shown that the evolutionary algorithm presented is more efficient working in Grid environments because it can carry out major exploration and exploitation of the solution space.