A novel data distribution management scheme to support synchronization in large-scale distributed virtual environments

Abstract

In large-scale distributed virtual environments - LDVEs - extensive and/or multidimensional virtual 3D environments can be shared by thousands of participant users in applications ranging from multiplayer games to virtual cities, virtual shopping mall, open space military training etc. In such systems, distribution seems inherent. The world geometric data of an extensive region or terrain may need to be divided in areas, which are distributed across the network. By distributing the data, host memory use and processing power are reduced since less data needs to be received, processed and stored. However, these gains can be easily surpassed by the amount of traffic generated by the synchronization messages transmitted, as well as the processing time due to the control infrastructure and the bulky data transmission of parts of the distributed VE. So, good mechanisms for data distribution management (DDM) are an important requirement for LDVEs. This paper describes the grid filtered region-based (GFRB) DDM scheme, which combines the positive aspects of both region-based and grid-based DDM methods to reduce network traffic and latency in LDVEs using a minimum number of multicast groups. GFRB scheme is unique because it provides a finer grained mechanism to exact match publishers to subscribers with intersecting regions. It does that with a reduced number of messages and multicast groups, by taking into consideration the percentage of overlap of a region on a cell. Experimental results show that GFRB scheme can be a potential solution to data synchronization in large distributed virtual environments.