Distribution and partitioning techniques for NVEs: the case of EVE

Abstract

The majority of the systems and platforms developed for supporting distributed virtual environments are based on the concept of distribution from the early beginning of their development. In this paper we present the migration to a distributed virtual environment from a traditional client-server architecture. In particular, this paper describes the case of EVE, a networked virtual environment originally aimed to support small-scale applications. EVE started as a standard client-multi server architecture, which could support multiple concurrent virtual worlds with a maximum number of seventeen simultaneous participants in each of these worlds. However, the need to support larger-scale applications revealed that the traditional architecture, upon which EVE was based, is insufficient to meet the needs of these applications, which are large both in the sense of virtual space and graphics and in regard to the number of concurrent participants. This paper discusses the migration of EVE to a distributed platform, which is able to support large-scale networked virtual environments. In particular, the paper describes the modifications realized in the architectural model of the initial platform for supporting effectively large-scale applications. The basic entities of the distributed model are presented, their operations, as well as the interconnection among them. In addition, the paper presents an initial approach of the algorithm that is adopted for the efficient partitioning of the virtual world and the assignment of the clients to the entities and resources of the distributed platform. The approach presented is space-object driven, in the sense that both the actual size of the virtual space along with the number of objects with which the user can interact is taken into account during the partitioning