A Region Growing Algorithm for Solving CVE Partitioning Problem

Abstract

Collaborative Virtual Environment (CVE) systems offer multi-user support, in which many users share a virtual world and work collaboratively, and in which multiple-server architectures are widely used and multiple servers deployed, with each one handling a part of the virtual world. One of the key issues in the design of scalable CVE systems based on multiple-server architecture is the partitioning problem. In this paper, a new quality function of partitioning is used to evaluate the partitioning method and a region growing (RG) algorithm is proposed for solving the partitioning problem in CVE systems. The RG algorithm regards the partitioning problem as a dynamic assignment of cells to the servers. Some cells are chosen as growing points, and their neighbor cells are merged into the same region. Users in the same region are connected with the same server until the server is overloaded. Through the experiment, we discover that our method can provide better partitioning solutions with shorter execution time than the methods proposed in the previous literature of large-scale CVE systems. Therefore, the proposed implementation of RG algorithm improves scalability and efficiency of CVE systems.