EXTENDED ABSTRACTDistributed Virtual Environments (DVE) such as military or civil protection distributed simulations and massively multiplayer online games (MMOG), for instance World of Warcraft or Second Life, are currently gaining increasingly attention in the software market.Currently the client server model is still the reference computational model for this kind of applications, but approaches based on the P 2P model [4,2,1,3] have recently been proposed. The definition of a fully distributed architecture for DV E is an actual challenge because of the complexity of these applications which integrate networks, graphics and AI programming. On the other hand, the adoption of a distributed computational model is mandatory to overcome the low scalability of client server architectures.In a DV E, a set of active entities, represented by avatars, interact with each other and with a set of passive objects, like weapons, potions, etc. Each avatar moves continuously within the DV E and in general it interacts with the avatars and the objects located in its surroundings only. This locality property which is modeled through the notion of Area of Interest (AOI) of an avatar, should be properly exploited by a communication support to reduce the amount of messages exchanged through the P 2P overlay. On the other hand, since the view of each avatar is constrained by its AOI, a mechanism to dynamically acquire knowledge of avatars and of passive objects beyond the AOI is required.The management of the state of the passive objects of the DV E is a further critical issue for the development of P 2P supports because it requires to manage the concurrent updates that may occur when sevPermission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. eral avatars concurrently update the same object. The adoption of classical solutions like those based on logical timestamps is not suitable in this case, due to the high number of messages required to implement these mechanisms. Furthermore, since P 2P networks are highly dynamic, a set of mechanisms to guarantee object persistence should be defined as well.We are currently investigating the feasibility of exploiting Voronoi Tessellations [5] to model DV E applications. Given a set of sites S = s 1 ...s n in a 2D space, a Voronoi tessellation is a space partition that assigns to each site s i the region V oro(s i ), that includes the set of points which are closer to s i than to any other site s j ∈ S, i = j, according to a given metric. Standard euclidean metric is exploited in classical Voronoi tessellations. Two sites are Voronoi Neighbours iff the borders of their region overlaps. A Delaunay triangulation is a graph that connects the Voron...