Sharing One Pheromone Matrix:A New Approach to Parallel ACO

GAO Yan-Ming,QIAN Pei-De
DOI: https://doi.org/10.1360/aas-007-0418
2007-01-01
ACTA AUTOMATICA SINICA
Abstract:This paper proposes and implements a new ap- proach to parallel ant colony optimization(ACO)algorithms. The principal idea is to make multiple ant colonies share and utilize only one pheromone matrix.We call our approach shar- ing one pheromone matrix(SHOP).This paper briefly describes how to parallelize ACS and MMAS by SHOP strategy,and tries to hybridize these two in parallel.By tackling symmetric travel- ling salesman problems,this paper compares SHOP-ACO imple- mentation with the relevant sequential ACO algorithms under fair computing environment,as well as with the existing parallel ACO algorithms.The experimental results indicate that SHOP strategy is superior to the sequential ACO algorithms and the existing parallel strategies.
What problem does this paper attempt to address?