A Voronoi based framework for the definition of P2P distributed virtual environments

The diffusion of wide area networks has lead to the definition of novel applications such as the Distributed Virtual Environments (DVE), for instance massively multiplayer games, militar or civil distributed simulations. In a DV E a set of active entities (avatars), interact with each other and with a set of passive objects located in their surroundings. While most DV E are currently still managed according to the client server model, the P2P model has recently been investigated, even if the killer application for P2P has been till now the file sharing one. This thesis investigates the feasibility of a P2P architecture for Distributed Virtual Environments. Locality of DV E interactions, modeled through the notion of Area of Interest, AOI of each entity, is properly exploited by the P2P communication support to reduce the amount of messages exchanged through the P2P overlay. Furthermore, a mechanism to dynamically acquire knowledge of the state located beyond the AOI and a strategy to preserve consistency of the replicated state is defined. We propose to model a DV E by a Voronoi Tessellation, where the sites correspond to peers and the Delaunay triangulation links define the topology of the P2P overlay. Different solutions for the definition of the overlay network and a set of protocols for the propagation of the heartbeats and the management of the passive objects have been formally defined, implemented and evaluated with respect to consistency and scalability. Finally, this thesis investigates the definition of a hybrid P2P support for the development of a DV E, where a hierarchy of peers is defined according to their computational/communication power. The resulting overlay is modeled through a Weighted Voronoi Diagram.