Environment organization of roles using polymorphism

In the field of multi-agent systems, there has lately been a growing interest on ways in which the environment can better be exploited to coordinate agent behavior and manage complex problems. This paper describes an environment that is able to organize and adapt agent roles as conditions warrant. Roles are adapted using polymorphism as directed by the environment. The design combines strategies from game theory and other biologically inspired models to address fault mitigation in large-scale, real-time, distributed systems. It is implemented on a prototype of the data acquisition system for BTeV, a High Energy Physics experiment consisting of 2500 digital signal processors. Results show environment organization of roles for the lightweight agents embedded within each of the individual processors.