On the Modeling, Refinement and Integration of Decentralized Agent Coordination

The integration of self-organizing processes in distributed software systems allows to equip applications with adaptive features that rise from the (inter-)actions of agents. In this paper, we address the systematic development of decentralized inter-agent processes in Multi-Agent Systems (MAS) that give rise to self-organizing properties. A

systemic

programming model is utilized that comprises a modeling approach to the conception of feedback loop structures and a reference architecture for the automated enactment of process prescriptions. A key design criterion for this tool set is the ability to supplement decentralized adaptations to already constructed software systems. Here, we present and exemplify the systematic design of decentralized coordination. The refinement of coupled feedback loop structures is presented as a design activity that is part of the systematic integration of coordination processes. This development approach is exemplified in two simulation scenarios, where the epidemic dissemination of information is integrated.