A Self-organization Model for MAS Based on Trust

This paper focuses on addressing the challenge of maintaining information coherence and robustness within a multi-agent system (MAS) that aggregates information from distributed sources, some of which may be defective intentionally or unintentionally. We propose a self-organizational approach in this context, emphasizing a systemic perspective that considers structural coupling across two levels: direct information gathering and communication. Specifically, we integrate a trust mechanism with local behavioral rules and selective environmental pressures to facilitate the emergence of two co-evolving organizations: one at the social level and the other at the spatial level. The social organization mirrors the trust relationships developed among the agents, while the spatial organization represents the deployment of agents in the environment to encourage exploration. The local behavioral rules encompass three categories: deployment rules, communication rules, and retro-action rules governing communication and deployment. We conduct simulations to experiment with the combination of these behavioral rules, observing the emergence of organizational structures and roles within the system.