Title: "Geoprocessing Optimization in Grid Systems"

When: December 17, 2004 10:00am-12:00 noon

Where: 1st Floor Conference Room, IS Building

Who: Shuo Liu

Committee:
Dr. Hassan Karimi (chair), DIST
Dr. Michael Lewis, DIST
Dr. Vladimir Zadorozhny, DIST
Dr. Ralph Z. Roskies, Pittsburgh Supercomputing Center
Dr. M. Talat Odman, School of Civil and Environmental Engineering, Georgia Institute of Technology


Abstract: Geoprocessing is typically data-intensive and/or compute-intensive which demands high computing power and data storage from computing environments. As the amount of geospatial data increases, such requirements become crucial and may even go beyond the computing capacity a single site can provide currently. As an emerging new computing technology, grid computing offers a novel way to integrate distributed computing resources and support collaboration across networks, which makes it a perfect candidate as the backbone-technology for geoprocessing. Although there have been a few research efforts on applying grid computing to the geospatial domain, none offers a general optimization mechanism for geoprocessing.

In this proposal, a novel optimization technique for geoprocessing in grid systems, GOG (Geoprocessing Optimization in Grids), which improves the system response time, is proposed. GOG tackles the optimization problem from three perspectives: resource selection, parallelism utilization, and dynamic adaptation. Three modules, one for each perspective, will be developed accordingly and will be validated through a proof-of-concept.