The Agreement as an Adaptive Architecture for Open Multi-Agent Systems

J. Santiago Pérez-Sotelo, Carlos E. Cuesta, and Sascha Ossowski, Actas de los Talleres de las Jornadas de Ingeniería del Software y Bases de Datos,, 2009, []

Multi-Agent Systems (MAS) are increasingly popular in AI to solve complex problems; they are conceived to be flexible and to be able to adapt to different situations. However, this feature is often compromised because of the need to reach internal coordination. This paper suggests that this issue requires an architectural solution, distilled as a new construct, the agreement, designed to be adaptive. The paper starts discussing the coordination problem and some of their traditional solutions, leading to an organization-based approach. This is followed by the description of a base architecture, including organizations and services, capable of supporting the agreement structure. The concept itself, and some requirements, is discussed and outlined in a set of logical layers, which are also defined as intertwining facets. Afterwards a real-world case study, in the domain of medical emergencies, is presented, describing a situation where the agreement model is applied, and some requirements to shift from directed to emergent evolution are highlighted, indicating some lines of future work.