Issues in large scale collaborative virtual environments.

Collaborative Virtual Environments (CVE) are virtual reality spaces that enable participants to collaborate and share objects, as if physically present in the same place. CVE concepts have been used in many systems in the past few years. Applications of such technology range from military combat simulations to various civilian commercial applications. These collaboration spaces have strict performance requirements. Today, there are many such systems developed specifically for collaboration. At the same time, some relatively new standards that address multiuser virtual environments and shared spaces have become available; however, most of these standards have been developed assuming that a small number of users would be interacting at a given time. The architectures available today provide support for a modest number of users but they fail if too many users are "present" together in a small "space" in the Virtual World. In this work, we first evaluate the currently available standards for the case of a very large number of users. An Adaptive Hybrid Architecture for VEry Large Virtual EnvironmenTs (VELVET) is then introduced. VELVET allows a large number of users to interact in a CVE. It also supports small groups of users, but it is in the large environment case that shows its greatest potential. VELVET introduces a novel adaptive area of interest management, which supports heterogeneity amongst the various participants. That allows users in a supercomputer with a high-speed networking connection to successfully collaborate with others in not-so-powerful systems behind a slow dial-up connection. In order to make a Collaborative Virtual Environment more interesting to users, it is possible to "stitch" together copies of areas which users may have interest in from one Virtual World into another. This procedure augments the physical size of a Virtual World, and creates a potentially larger number of users within the World, first because of the "embedding" and second since the added attractions may work as an incentive for some more users to join the "embedded" World. On the other hand this procedure brings up a series of problems related with consistency, which are also addressed in the thesis. We introduce a methodology which ensures that all copies of a given area of a World are kept consistent among them, as well as with the original world. We also apply this methodology in VELVET, as well as in other Architectures. Additionally, we introduce other approaches to be used when a less strict consistency model is sufficient