An Improved MOEA/D Algorithm for Multi-objective Multicast Routing with Network Coding

Impact Factor:3.907

DOI number:10.1016/j.asoc.2017.05.033

Affiliation of Author(s):Southwest Jiaotong Univ, Sch Informat Sci & Technol

Journal:Applied Soft Computing

Key Words:Network coding,Multicast,Multi-objective evolutionary algorithm

Abstract:Network coding enables higher network throughput, more balanced traffic, and securer data transmission. However, complicated mathematical operations incur when packets are combined at intermediate nodes, which, if not operated properly, lead to very high network resource consumption and unacceptable delay. Therefore, it is of vital importance to minimize various network resources and end-to-end delays while exploiting promising benefits of network coding. Multicast has been used in increasingly more applications, such as video conferencing and remote education. In this paper the multicast routing problem with network coding is formulated as a multi-objective optimization problem (MOP), where the total coding cost, the total link cost and the end-to-end delay are minimized simultaneously. We adapt the multi-objective evolutionary algorithm based on decomposition (MOEA/D) for this MOP by hybridizing it with a population-based incremental learning technique which makes use of the global and historical information collected to provide additional guidance to the evolutionary search. Three new schemes are devised to facilitate the performance improvement, including a probability-based initialization scheme, a problem-specific population updating rule, and a hybridized reproduction operator. Experimental results clearly demonstrate that the proposed algorithm outperforms a number of state-of-the-art MOEAs regarding the solution quality and computational time. (C) 2017 Elsevier B.V. All rights reserved.

Co-author:Huanlai Xing*,Zhaoyuan Wang,Tianrui Li,Hui Li,Rong Qu

Document Code:10.1016/j.asoc.2017.05.033

Volume:59

Page Number:88-103

ISSN No.:1568-4946

Translation or Not:no

Date of Publication:2017-10-01

Included Journals:SCI