S.F. Midkiff

This effort will investigate the network services needed to effectively support (i) demanding virtual reality applications where the VR environment (CAVE) is separated from computational resources (remote workstations and parallel processors), and (ii) distributed VR environments and telepresence (remote virtual presence) applications. These applications are extremely demanding of network services as they require low overhead (especially for distributed computing), real-time delivery (for interactivity and animations), highly variable bandwidth (due to varying tasks and compression), and efficient multicast support (especially for distributed VR and telepresence). While high data rate networks, such as ATM over OC-3 and OC-12 links can provide the required bandwidth, quality of service (QoS) guarantees for these application requirements are hard to achieve with any efficiency and providing appropriate QoS is an open research issue. Existing network and transport protocols, especially the widely used TCP/IP, are deficient in their support for these requirements. In addition, the switched structure of ATM creates problems for multicast operation that must be resolved.

The proposed infrastructure will provide an excellent "laboratory" for investigating these issues. The infrastructure will provide the essential mix of high data rate networking (ATM), VR environment (CAVE), computing resources (remote workstations and parallel machines), and demanding applications (as described elsewhere in this proposal). With these elements, it is possible to understand application-network interaction through measurement and characterization, investigate protocols to meet application needs, and evaluate new network protocols and services.