Dynamic Multiplexing and High-Performance Modeling in Distributed Simulation

Abstract

This article reviews existing message traffic reduction schemes and proposes a dynamic multiplexing approach as an efficient message traffic reduction scheme. This approach is based on a dynamic encoding of the joint output of sender components into a single message and is applicable to a model involving distributed components moving and interacting in multidimensional space by combining with predictive interest-based quantization. This article applies this approach to a projectile/missile application with realistic multidimensional dynamics. The authors investigate the dependence of simulation accuracy and required network bandwidth on the number of active components and basic time granule. Analysis and empirical data clearly indicate the advantages of dynamic multiplexing in saving communication and computation resources in a large-scale distributed system.