Noshape

Abstract

This article presents a bio-inspired paradigm for metaphorically modeling agent organizations as adaptive virtual amoebas for the development of large-scale complex multi-agent systems. The presented model is called Noshape inspired from the amoeba, which is a unicellular micro-organism that does not have a definite shape. This article aims to test the performance of Noshape MAS with applications contain higher numbers of agents up to 8000 agents; this number of agents is very huge compared to the current state of the practice of MAS. The performance evaluation results show that Noshape MAS have better long-term performance in terms of service response time compared to present organizational approaches (i.e., federation). In Noshape MAS, the response times of remote agents' interactions will seem to be as those of local interactions thanks to the transparently provided dynamic adaptation behavior which arises from the dynamic overlapping of agent organizations. Further research is recommended to give the focus to performance, resiliency, security, and agent mobility within Noshape MAS.