A self-organizing environment for distributed multiagent-based simulation.

摘要

In this work, we propose a design for environments in large-scale, distributed Multi-Agent Based Simulation Systems (MABS). This design applies to domains dealing with agents situated in decentralized, distributed, non deterministic and adaptive environments. We define a novel model of interaction between the environment and context-aware agents, and present a hierarchical structure that allows the environment to self-organize at run-time to accommodate uneven spatial dispersion of agents. We propose a distributed environment architecture to address the problem of scalability, and discuss how self-organization helps balance the utilization of available computing resources. Finally, we present a distributed visualization system that allows multiple users to simultaneously view and interact with the simulation. Our results are discussed in the context of DIVAs, a large scale MABS framework developed at the University of Texas at Dallas.