Agent-Based Simulation of Data-Driven Fire Propagation Dynamics

John Michopoulos1, Panagiota Tsompanopoulou2, Elias Houstis3,2, and Anupam Joshi4

1Special Projects Group, Code 6303, U.S. Naval Research Laboratory, U.S.A.
john.michopoulos@nrl.navy.mil

2Dept. of Comp. Eng. and Telecommunications, University of Thessaly, Greece
yota@inf.uth.gr

3Computer Sciences Department, Purdue University, U.S.A.
enh@cs.purdue.edu

4Dept. of Comp. Sci. and Electr. Eng., U. of Maryland Baltimore County, U.S.A.

Abstract. Real world problems such as fire propagation prediction, can often be considered as a compositional combination of multiple, simple but coupled subproblems corresponding to analytical and computational behavior models of the systems involved in multiple domains of action. The existence of various computational resources (legacy codes, middleware, libraries, etc.) that solve and simulate the subproblems successfully, the coupling methodologies, the increasing and distributed computer power (GRID etc.) and the polymorphism and plurality of the available technologies for distributed mobile computing, such as Agent Platforms, motivated the implementation of multidisciplinary problem solving environments (MPSE) to overcome the difficulties of their integration and utilization. In this paper we present the onset of the development of computational infrastructure for the simulation of fire propagation in multiple domains using agent platforms as an informal validation of our data-driven environment for multi-physics applications (DDEMA).

LNCS 3038, pp. 732-739.

Last modified: