• DocumentCode
    3352178
  • Title

    The Shortest Path Parallel Algorithm on Single Source Weighted Multi-level Graph

  • Author

    Peng, Qiang ; Yu, Yulian ; Wei, Wenhong

  • Author_Institution
    Dept. of Comput. Sci., South China Univ. of Technol., Guangzhou, China
  • Volume
    2
  • fYear
    2009
  • fDate
    28-30 Oct. 2009
  • Firstpage
    308
  • Lastpage
    311
  • Abstract
    Aiming at the single source shortest path problem on the single source weighted multi-level graph, this paper brings forward a new and practical parallel algorithm on 2-D mesh network by constructing a model of vector-matrix multiple and choosing the planar evenly partition method. The parallel algorithm is propitious to handle different sizes of multi-level graph structure by the use of its simple and regular characteristic, and our algorithm only needs less computation resources. The parallel algorithm was implemented on MPI. The theoretical analysis and the experimental results prove that the parallel algorithm has the good performances in terms of speed-up ratio and parallel efficiency, and its performance is a little better than the parallel dijkstra algorithm.
  • Keywords
    matrix algebra; parallel algorithms; 2D mesh network; MPI; multi level graph; parallel dijkstra algorithm; parallel efficiency; shortest path parallel algorithm; single source; speed up ratio; vector matrix multiple; Algorithm design and analysis; Computer science; Concurrent computing; Graph theory; Intelligent transportation systems; Mesh networks; Parallel algorithms; Partitioning algorithms; Performance analysis; Shortest path problem; parallel algorithm; single source shortest path; single source weighted multi-level graph; vector-matrix multiple;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Computer Science and Engineering, 2009. WCSE '09. Second International Workshop on
  • Conference_Location
    Qingdao
  • Print_ISBN
    978-0-7695-3881-5
  • Type

    conf

  • DOI
    10.1109/WCSE.2009.819
  • Filename
    5403295