Robust collaboration and interaction in a multi-agent system basing on hierarchical networks

This paper considers the capability of collaboration and interaction between agents in a new architecture of multi-agent systems (MAS) basing on the thought of network behavior and co-evolution. Actually, current development of MAS nearly synchronously accompanies with the development of communication networks ever increasingly because the approaches of information exchange are mainly depended on the modes of data transmission and the spectacular of advanced communication technologies. The real detailed complexity inside the media transmission platform is masked in the proposed architecture of MAS so that the complex processes of data dissemination can be coped with reasonably and simply by way of decentralized intelligent agents. According to the simulation result and the comparison with other typical models of MAS, the presented architecture of MAS and its running mechanism exhibit better advantages such as flexibility, robustness and expansibility suited to the collaboration and interaction among decentralized agents in MAS. Toward the future research, some related optimal algorithms and running mechanisms concerning the integrated performance and the complexity of the MAS's architecture will be investigated and developed.