PPSN XII - LNCS 7491-7492 CD-ROM

Network Topology Planning Using MOEA/D with Objective-Guided Operators

Wei Peng¹ and Qingfu Zhang²

¹School of Computer, National University of Defense Technology, Changsha, Hunan, 410073, China
wpeng@nudt.edu.cn

²School of Computer Science & Electronic Engineering, University of Essex, Colchester, Essex, CO4 3SQ, United Kingdom
qzhang@essex.ac.uk

Abstract. Multiobjective evolutionary algorithms (MOEAs) have attracted growing attention recently. Problem-specific operators have been successfully used in single objective evolutionary algorithms and it is widely believed that the performance of MOEAs can be improved by using problem-specific knowledge. However, not much work have been done along this direction. Taking a network topology planning problem as an example, we study how to incorporate problem-specific knowledge into the multiobjective evolutionary algorithm based on decomposition (MOEA/D). We propose objective-guided operators for the network topology planning problem and use them in MOEA/D. Experiments are conducted on two test networks and the experimental results show that the MOEA/D algorithm using the proposed operators works very well. The idea in this paper can be generalized to other multiobjective optimization problems.

Keywords: Multiobjective Optimization, Evolutionary Algorithm, MOEA/D, Network Topology Planning

LNCS 7492, p. 62 ff.

Full article in PDF | BibTeX