Quality Criteria for Multiagent-based Simulations with Conservative Synchronisation Qualitätskriterien für multiagenten-basierte Simulation mit konservativer Synchronisation

The concept of multiagent-based simulation introduces the agent programming paradigm to simulation. Multiagent systems ease the implementation of software systems to control complex business processes, such as supply chain management (SCM). Problem complexity is decreased by abolishing monolithic programs. Instead, decision-making is delegated to software agents as local entities. This allows coping even with processes that cannot be controlled centrally due to their inherent physical distribution. Simulation allows evaluating logistics strategies regarding their applicability in such processes. However, general agent development frameworks are not designed to consider simulation-specific issues. In particular, they provide no means for synchronisation. This paper identifies time model adequacy, causality, and reproducibility as quality criteria that must be ensured by a simulation middleware implementing synchronisation. Furthermore, a formal definition of these quality criteria for conservative synchronisation is presented.