A SCALEABLE NETWORK MODEL FOR DISTRIBUTED VIRTUAL WORLDS

Abstract

Distributed Virtual Worlds refer to virtual environments where multiple geographically dispersed users connected by a network can interact with one another. Usually, when the number of concurrent users is very high, the network traffic increases drastically and the overall performance of the network and associated virtual worlds becomes unbearably sluggish. This thesis presents: • A scaleable network model which can support a large number of users and is suitable for group-based Virtual Reality applications. It is a Client-Server design where the clients are partitioned into groups. Groups are distributed over multiple servers, when the client space is very large. The interaction among the members of a group is more tightly coupled compared to inter-group communication. • A set of interface routines in the form of a toolkit which can be used to develop networked VR applications, supporting the above network architecture. • A new dead-reckoning technique which reduces network traffic significantly when a large number of users interact simultaneously over network. The server process run on SUN SPARCstation whereas the client's world runs on IBM or compatible PC. The client process can communicate to server through modem and telephone line. A group treasure hunt game has been developed with this network architecture, where a player can see other players of the same group moving around in a large maze. The player can interact with other members in the group and also with shared objects within or across groups.