One of the common problems faced in collaborative environments and scientific visualization applications, is termed the ‘Display docking’ or the ‘Display Pushing’ problem where the visualization or the presentation generated on one or more computers, has to be distributed to remote sites for viewing and/or image post-processing by a group of collaborators. A typical image source in such a case could be computers ranging from laptops showing presentations, to compute clusters number crunching terabytes of data and rendering high resolution visualizations on tiled displays. This thesis presents a scalable platform independent solution which is capable of transmitting multiple high resolution video streams from such video sources to one or more destinations. The unique capability of this concept is that it is a flexible hardware oriented solution, where no special software/hardware has to be installed on the source or destination machines to enable them to transmit their video. The load of streaming these high-resolution video streams is also off-loaded to dedicated systems, leaving the source machines, generating the graphics, free for pure computation. The multiple streams video streams handled by the system can either be independent of each other or they might be component streams of a video system, such as a tiled display or stereoscopic display. Also, results will be presented from various tests on high speed dedicated long haul networks, and local area gigabit LANs with different Layer 4 protocols.

MS Thesis Committee: Assistant Professor Andrew Johnson, Associate Professor Jason Leigh and Post-Doc Research Associate Luc Renambot.

URL