Grid-Filtered Region-Based Data Distribution Management in Large-Scale Distributed Simulation Systems
Simulation Symposium, Annual (2005)
San Diego, California, USA
Apr. 4, 2005 to Apr. 6, 2005
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/ANSS.2005.23
Azzedine Boukerche , University of Ottawa
Nathan J. McGraw , University of Ottawa
Caron Dzermajko , University of Ottawa
Kaiyuan Lu , University of Ottawa
In a large-scale distributed simulation, participating federates receive and send messages to share state data and update subscription and publication regions. The focus of Data Distribution Management (DDM), a High Level Architecture (HLA) Run-Time Infrastructure (RTI) service, is limiting and controlling the volume of data, regarding simulated entities, exchanged between participating hosts. One of the key factors in many large-scale distributed simulations is the ability to see or be seen by other participants in the simulation. Several DDM methods have been introduced, but time performance, message volume and resource usage continue to be the factors in the practical application of these methods. In an effort to offer a more efficient and more scalable solution to DDM, we propose an algorithm which we refer to as Grid-Filtered Region-Based DDM, that utilizes a grid overlay on the virtual space, determines the percentage of grid covered by the subscription or publication region and further filters, based on a percentage threshold, using a matching technique, like that of the Region-Based DDM scheme. We present the implementation details of our scheme and report on our set of experiments we have carried out to evaluate its performance.
A. Boukerche, N. J. McGraw, K. Lu and C. Dzermajko, "Grid-Filtered Region-Based Data Distribution Management in Large-Scale Distributed Simulation Systems," Proceedings. 38th Annual Simulation Symposium(ANSS), San Diego, CA, USA, 2005, pp. 259-266.