Combined Discrete Particle Swarm Optimization and Simulated Annealing for Grid Computing Scheduling Problem

The grid scheduling problem is concerted with some tasks assigning to a grid distributed system that the relative tasks have to exchange information on different grids. In the original particle swarm optimization (PSO) algorithm, particles search solutions in a continuous solution space. Since the solution space of the grid scheduling problem is discrete. This paper presents a discrete particle swarm optimization (PSO) that combines the simulated annealing (SA) method to solve the grid scheduling problems. The proposed discrete PSO uses a population of particles through a discrete space on the basis of information about each particle's local best solution and global best solution of all particles. For generating the next solution of each particle, the SA is adopted into the discrete PSO. The objective is to minimize the maximum cost of the grid, which includes computing cost and communication cost. Simulation results show that the grid scheduling problem can be solved efficiently by the proposed method.