A Process Based on the Model-Driven Architecture to Enable the Definition of Platform-Independent Simulation Models - Simulation and Modeling Methodologies, Technologies and Application - page 120

5 Conclusions

A wider adoption of the ABMS is still hindered by the lack of approaches able to
fully support the experts of typical ABMS domains (e.g. financial, economic, social,
logistics, chemical, engineering) in the definition and implementation of agent-based
simulation models. In this context, the paper has proposed a solution, centered on the
joint use of the Model-Driven Architecture and AMF-based Platform-Independent
Metamodel, which aims to overcome the main drawbacks of available ABMS lan-
guages, methodologies and tools. In particular, the proposed process (MDA4ABMS)
allows to (automatically) produce Platform-Specific simulation Models (PSMs) start-
ing from a Platform-Independent simulation Model (PIM) obtained on the basis of a
Computation Independent Model (CIM), thus allowing domain experts to exploit
more high-level design abstractions in the definition of simulation models and to ex-
change/update/refine the so obtained simulation models regardless to the target plat-
form chosen for the simulation and result analysis. Moreover, the semi-automatic
model transformations, enabled by the defined metamodels and related mappings,
ease the exploitation of the proposed modeling notation and process, while the adop-
tion of the standard UML notation and the visual modeling tool provided by AMP
reduce the learning curve of the process.
The MDA4ABMS process has been exemplified with reference to the well-known
Demographic Prisoner's Dilemma which is able to represent several social and eco-
nomic complex scenarios thus demonstrating the efficacy of the process and the re-
lated tools in supporting domain experts from the definition of conceptual simulation
models to their concrete implementation on different target ABMS platforms.
Ongoing research efforts are devoted to: (i) define and extensive experiment a full-
fledged ABMS methodology based on the MDA4ABMS process and able to seam-
lessly guide domain experts from the analysis of a complex system to its agent-based
modeling and simulation; (ii) look for frameworks different from AMF (e.g. HLA)
suitable to define PIM metamodels able to support the modeling of simulation scena-
rios with specific requirements such as distribution and/or human participation.

References

1. Agt, H., Bauhoff, G., Cartsburg, M., Kumpe, D., Kutsche, R., Milanovic, N.: Metamode-
ling Foundation for Software and Data Integration. In: Yang, J., Ginige, A., Mayr, H.C.,
Kutsche, R.-D. (eds.) UNISCON 2009. LNBIP, vol. 20, pp. 328-339. Springer, Heidelberg
(2009)
2. Alonso, F., Frutos, S., Martínez, L., Montes, C.: SONIA: A Methodology for Natural
Agent Development. In: Gleizes, M.-P., Omicini, A., Zambonelli, F. (eds.) ESAW 2004.
LNCS (LNAI), vol. 3451, pp. 245-260. Springer, Heidelberg (2005)
3. The AMP project, http://www.eclipse.org/amp/
4. Atkinson, C., Kühne, T.: Model-driven development: A metamodeling foundation. IEEE
Software 20(5), 36-41 (2003)
5. Bauer, B., Müller, J.P., Odell, J.: Agent UML: A Formalism for Specifying Multiagent
Software Systems. In: Ciancarini, P., Wooldridge, M.J. (eds.) AOSE 2000. LNCS,
vol. 1957, pp. 91-103. Springer, Heidelberg (2001)