HD-GraphViz: highly distributed graph visualization on tiled displays

  • Authors:

  • Sangwon Chae;Aditi Majumder;M. Gopi

  • Affiliations:

  • University of California, Irvine, CA;University of California, Irvine, CA;University of California, Irvine, CA

  • Venue:
  • Proceedings of the Eighth Indian Conference on Computer Vision, Graphics and Image Processing
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we propose a distributed approach for visualizing graph datasets using distributed force-directed layout algorithm on multiple displays each of which is controlled by a compute device. In this distributed network of compute-display nodes, the display devices form a geometrically contiguous tiled display wall, and the data distribution among the compute-display nodes for algorithmic purposes is related to the data distribution to its corresponding display nodes for visualization purposes. This approach collocates computation and visualization within the same node and minimizes network bandwidth requirement for data exchange between nodes. Leveraging the advantages of the above novel system design, we also present a distributed force directed graph layout algorithm. In this distributed algorithm, we address two issues specific to tiled displays: number of edges crossing the display panel boundaries (across two different compute-display clients) and the location of graph vertices close to panel boundaries which may lead to incorrect visual analytic conclusions. We integrate the cost functions expressing the above two effects along with the cost function of the edge-crossings within each display into the graph layout algorithm. Using this distributed approach, for the first time a very large graph has been laid out in a large real estate provided by the tiled display. The proposed system is evaluated on various parameters including the scalability, the impact on display real-estate utilization, and finally, network bandwidth usage.