Modeling agents with the CSP formalism: an approach for optimization of distributed problems

Multi-Agent Systems (MAS) are an efficient approach to deal with distributed complex problems. However, they are slightly suited to optimization ones. In order to tackle this problem we present a model to specify Multi-Agent Systems with the Constraint Satisfaction Problem (CSP) formalism. It allows to take advantages from CSP solving algorithms and to deal with optimization problems. For this purpose we specify agents as Constraint Satisfaction and Optimization Problems, i.e. CSOP and consequently the MAS as a Distributed CSOP (DCSOP). We illustrate the interest of our approach on the antennae parameter setting problem which is an example of distributed constrained problem. We implement this specification to validate the optimization function model that we propose.