A hybrid genetic/optimization algorithm for a task allocation problem

Abstract

We consider the problem of designing a distributed computing system for handling a set of repetitive tasks on a periodic basis. Tasks assigned to different processors need communication link capacity, tasks executing on the same processor do not. The aim is to develop a design of minimum total cost that can handle all the tasks. We compare the performances of a genetic algorithm, a commercial 0–1 integer programming software and a hybrid approach from the literature, in solving real instances of the problem. Copyright © 1999 John Wiley & Sons, Ltd.