Reusability and Composability Analysis for an Agent-Based Hierarchical Modelling and Simulation Framework.
Feng Zhu,Yiping Yao,Jin Li,Wenjie Tang
DOI: https://doi.org/10.1016/j.simpat.2018.10.009
IF: 4.199
2018-01-01
Simulation Modelling Practice and Theory
Abstract:Agent-based modeling and simulation has been proved useful in a variety of different complex adaptive systems comprised of autonomous, interacting components. To resolve the shortcoming of current agent-based modeling frameworks with respect to decomposition and modularity, this paper presents a formal hierarchical modeling and simulation framework with three–level architecture to reduce ambiguity as well as improve clarity in the model definition. The bottom level is Component Model (CM), which implements some domain-specific support functionality, such as curve motion in the physical domain, intelligent decision-making in the cognitive domain, etc. The middle level is Agent Model (AM), which describes an agent which can react to the current situation by executing a sequence of CMs. The top level is System Model (SM), which defines that a CAS model consists of several AMs and also the interactions between these AMs. In the hierarchical architecture, one SM can be built up from lower-level models, which were linked in a loosely coupled fashion via an event-driven interface. We then analyze the reusability and composability of lower-level models of this hierarchical framework in a formalized way. To demonstrate the effectiveness of the proposed solution, we develop a graphical composite modelling tool named GraphSim, and the case study concerning two social dynamics system scenarios is also presented.