A novel hybrid approach of mean field annealing and genetic algorithm for load balancing problem

Abstract

In this paper, we introduce a new solution for the load balancing problem, which is an important issue in parallel processing. Our novel load balancing technique called MGA is a hybrid algorithm of Mean Field Annealing (MFA) and Simulated annealing-like Genetic Algorithm (SGA). SGA uses the Metropolis criteria for state transition as in simulated annealing to keep the convergence property in MFA. The MGA combines the benefit of both the rapid convergence property of MFA and various and effective genetic operations of SGA. We compare the proposed MGA with MFA and GA. Our experimental results show that our new technique improves performance in terms of communication cost, load imbalance and maximum execution time. © Springer-Verlag Berlin Heidelberg 2004.