Metaheuristics for solving the biobjective single-path multicommodity communication flow problem

被引:11
作者
Masri, Hela [1 ]
Krichen, Saoussen [2 ]
Guitouni, Adel [3 ]
机构
[1] Univ Tunis, Tunis Business Sch, Tunis, Tunisia
[2] Univ Tunis, Inst Super Gest Tunis, Tunis, Tunisia
[3] Univ Victoria, Peter B Gustavson Sch Business, Victoria, BC, Canada
关键词
multiple-objective programming; metaheuristics; network flows; nonlinear programming; VARIABLE NEIGHBORHOOD SEARCH; ROUTING-PROBLEMS; ASSIGNMENT PROBLEM; MPLS NETWORKS; ALGORITHM;
D O I
10.1111/itor.12378
中图分类号
C93 [管理学];
学科分类号
12 ; 1201 ; 1202 ; 120202 ;
摘要
Single-path multicommodity flow problem (SMCFP) is a well-known combinatorial optimization problem, in which the flow of each commodity can be transmitted using only one path linking its destination to an appropriate origin within the addressed network. In this paper, we study the SMCFP in a multiobjective context by considering the simultaneous optimization of paths' delay and average reliability. The network is modeled as a finite set of nodes that can communicate using preestablished connections where each connection is characterized by a capacity, a lead time, and a reliability. A node can be an information producer or/and information consumer. The contention problem is solved by assigning a path and a dedicated bandwidth to each flow. The problem is formulated as a biobjective nonlinear optimization problem. This biobjective problem has not been considered in the literature. We design three alternative procedures for approximating the Pareto front. We proposed an MGA based on NSGA-II, a multiobjective variable neighborhood search and a new distance-based hybrid metaheuristic. The hybridization integrates a local search into the framework of genetic algorithm to effectively drive the search toward a better approximating of the Pareto front. The propounded algorithms' efficiencies are experimentally investigated on a test bed of instances applied to a planar and a grid network. A comparative study is conducted based on different multiobjective performance indicators.
引用
收藏
页码:589 / 614
页数:26
相关论文
共 41 条
  • [11] MULTIOBJECTIVE ROUTING IN MULTISERVICE MPLS NETWORKS WITH TRAFFIC SPLITTING - A NETWORK FLOW APPROACH
    Girao-Silva, Rita
    Craveirinha, Jose
    Climaco, Joao
    Captivo, M. Eugenia
    [J]. JOURNAL OF SYSTEMS SCIENCE AND SYSTEMS ENGINEERING, 2015, 24 (04) : 389 - 432
  • [12] Multi-objective genetic local search algorithm using Kohonen's neural map
    Hakimi-Asiabar, Mehrdad
    Ghodsypour, Seyyed Hassan
    Kerachian, Reza
    [J]. COMPUTERS & INDUSTRIAL ENGINEERING, 2009, 56 (04) : 1566 - 1576
  • [13] A DUAL ALGORITHM FOR THE CONSTRAINED SHORTEST-PATH PROBLEM
    HANDLER, GY
    ZANG, I
    [J]. NETWORKS, 1980, 10 (04) : 293 - 310
  • [14] Variable neighborhood search: Principles and applications
    Hansen, P
    Mladenovic, N
    [J]. EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 2001, 130 (03) : 449 - 467
  • [15] A multicommodity network-flow problem with side constraints on paths solved by column generation
    Holmberg, K
    Yuan, D
    [J]. INFORMS JOURNAL ON COMPUTING, 2003, 15 (01) : 42 - 57
  • [16] Variable neighborhood search for location routing
    Jarboui, Bassem
    Derbel, Houda
    Hanafi, Said
    Mladenovic, Nenad
    [J]. COMPUTERS & OPERATIONS RESEARCH, 2013, 40 (01) : 47 - 57
  • [17] Multi-objective vehicle routing problems
    Jozefowiez, Nicolas
    Semet, Frederic
    Talbi, El-Ghazali
    [J]. EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 2008, 189 (02) : 293 - 309
  • [18] Kleinberg J. M., 1996, THESIS
  • [19] Koch R., 2005, Approximation and Online Algorithms. Third International Workshop, WAOA 2005. Revised Papers (Lecture Notes in Computer Science Vol. 3879), P244
  • [20] Lanza-Gutiérrez JM, 2013, LECT NOTES COMPUT SC, V8273, P145, DOI 10.1007/978-3-642-45008-2_12