Optimal pipeline decomposition and adaptive network mapping to support distributed remote visualization

Mengxia Zhu, Qishi Wu, Nageswara S.V. Rao, Sitharama Iyengar

Research output: Contribution to journalArticlepeer-review

9 Scopus citations


This paper discusses algorithmic and implementation aspects of a distributed remote visualization system that optimally decomposes and adaptively maps the visualization pipeline to a wide-area shared or dedicated network. The first node of the system typically generates or stores raw data sets, and a remote client resides on the last node equipped with a display device ranging from a personal desktop to a powerwall. Intermediate nodes include workstations, clusters, or rendering engines, which can be located anywhere on the network. We employ a regression method to estimate the effective bandwidth of a transport path. Based on link measurements, node characteristics, and module properties, we strategically organize visualization pipeline modules into groups and dynamically assign the groups to various network nodes to achieve minimal total delay or maximal frame rate. We propose polynomial-time algorithms using the dynamic programming method to compute optimal solutions for the problems of pipeline decomposition and network mapping under different constraint conditions. The proposed remote visualization system is implemented and deployed at several geographically distributed nodes for experimental testing. The proposed decomposition and mapping scheme is generic and can be applied to other distributed applications whose computing components form a linear arrangement.

Original languageEnglish
Pages (from-to)947-956
Number of pages10
JournalJournal of Parallel and Distributed Computing
Issue number8
StatePublished - Aug 2007


  • Bandwidth measurement
  • Distributed computing
  • Dynamic programming
  • Network mapping
  • Remote visualization
  • Visualization pipeline


Dive into the research topics of 'Optimal pipeline decomposition and adaptive network mapping to support distributed remote visualization'. Together they form a unique fingerprint.

Cite this